article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the self - gravitating gas in thermal equilibrium has been thoroughly studied since many years@xcite . as a consequence of the long range attractive newton force
, the selfgravitating gas admits a consistent thermodynamic limit @xmath4 with @xmath7 fixed . in this limit , extensive thermodynamic quantities like energy , free energy , entropy are proportional to @xmath3@xcite .
in ref.@xcite we investigated how a cosmological constant affects the properties of the non - relativistic self - gravitating gas in thermal equilibrium by mean field methods .
the mean field approximation becomes exact in the limit when the number of particles becomes infinity . in the present paper
we study the stability properties of the self - gravitating gas in the presence of a cosmological constant by mean field and monte - carlo methods .
the use of monte - carlo simulations is particularly useful since they are like real experiments and allow to unambiguously determine the stability or instability of the system .
we compute both by monte carlo and mean field methods several physical quantities : the equation of state , the particle density and the average particle distance for different values of the dark energy .
an excellent agreement between both approaches is found except very close to the collapse transition [ see figs .
the difference between both approaches turns to be , as expected , of the order @xmath8 where the number of particles @xmath3 we choose in the monte carlo simulations was @xmath9 .
the slightly larger split between monte carlo and mean field near the collapse is due to the fact that there the corrections to the mean field become singular @xcite .
we find that the onset of instability in the canonical ensemble coincides with the point where the isothermal compressibility diverges . at this point
we find that the dimensionless parameter @xmath10 is maximal .
( @xmath11 was introduced in ref .
@xcite for @xmath12 ) .
we find that the domain of stability of the gas increases for increasing cosmological constant .
the dark energy has an anti - gravity effect that disfavours the collapse pushing apart the particles . in absence of cosmological constant
the particle density @xmath13 is a * decreasing * function of the distance in the self - gravitating gas ( see for example @xcite ) . in the presence of a positive @xmath0
we find that @xmath13 * decreases * with @xmath14 when the self - gravity dominates over the dark energy .
this happens for @xmath15 . for @xmath16
the cosmological constant dominates over the self - gravity and the the particle density turns to * increase * with the distance .
this is a consequence of the repulsive character of the dark energy .
the monte - carlo study of the self - gravitating gas shows the validity of the dilute thermodynamic limit introduced in refs.@xcite .
namely , for @xmath4 with @xmath7 and @xmath6 fixed , a consistent thermodynamic limit is reached where extensive thermodynamic quantities like energy , free energy , entropy are shown to be proportional to @xmath3 .
furthermore , the monte - carlo simulations allow us to study the condensed phase which turns to be an extremely dense body .
the phase transition to collapse is found to be of zeroth order .
the outline of the paper is as follows .
section ii presents non - relativistic selfgravitating particles in the presence of the cosmological constant , their statistical mechanical treatment and the mean field approach . in section iii
we analyze the stability of the selfgravitating gas for zero and nonzero cosmological constant @xmath17 while in section iv we give the physical picture of the gas and its particle density as a function of @xmath17 .
section v contains the results of our monte - carlo simulations on the phase diagram , the collapse transition , the particle density and the properties of the collapsed phase .
we recall some results about the self - gravitating gas in the presence of the cosmological constant @xcite . in the newtonian limit the einstein equations of general relativity become@xcite @xmath18 where @xmath2 stands for the gravitational potential , @xmath19 for the density of massive particles and @xmath17 for the cosmological constant .
eq.([poil ] ) determines the weak and static gravitational field produced by non relativistic matter in the presence of the cosmological constant . for zero cosmological constant
we recover the usual poisson equation , as it must be .
the hamiltonian for such set of self - gravitating particles in the presence of the cosmological constant can be written in the center of mass frame as@xcite @xmath20 where @xmath21 stands for the mass of the particle at the point @xmath22 with momentum @xmath23 and @xmath24 since we choose center of mass coordinates .
the cosmological constant contribution to the potential energy grows negative for increasing values of the particle distances @xmath25 to the center of mass .
therefore , the gravitational effect of the cosmological constant on particles amount to push them outwards .
it can be then said that the cosmological constant has an * anti - gravitational * effect .
we present here the statistical mechanics of the self - gravitating gas in the presence of the cosmological constant in the canonical ensemble . for simplicity we shall consider @xmath3 particles with identical mass @xmath27 .
the hamiltonian is given by eq.([hamil ] ) and therefore the classical partition function of the gas is then , @xmath28 it is convenient to introduce the dimensionless coordinates @xmath29 .
the momenta integrals are computed straightforwardly .
hence , the partition function factorizes as the partition function of a perfect gas times a coordinate integral .
@xmath30 where @xmath31 the dimensionless parameters which characterize the strength of self - gravity and dark energy are respectively , @xmath32 @xmath33 can be interpreted as the ratio of the self - gravitating energy per particle to the kinetic energy per particle .
we define the ratio of dark energy and self - gravity @xmath34 it is the ratio between the dark energy contained inside the volume @xmath2 and the mass of ordinary matter . in the large @xmath3 limit
the @xmath35-uple integral in eq.([part ] ) can be approximated by a functional integral over the density . in the saddle point approximation
we find that the density @xmath36 obeys to the differential equation @xcite @xmath37 the density has to be normalized as @xmath38 in hydrostatic equilibrium we obtain the same equation @xcite .
therefore , hydrostatics and mean field are equivalent in the @xmath39 limit . in the limiting case @xmath40 eq.([poisson ] ) becomes the well known lane - emden equation in the absence of cosmological constant ( see refs .
@xcite ) . for spherically symmetric configurations
the mean field equation ( [ poisson ] ) becomes an ordinary non - linear differential equation .
@xmath41 the various thermodynamic quantities are expressed in terms of their solutions .
we work in a unit sphere , therefore the radial variable runs in the interval @xmath42 , @xmath43 .
the density of particles @xmath44 has to be normalized according to eq.([normalisation ] ) . integrating eq.([poissonrad ] ) from @xmath45 to @xmath46 yields , @xmath47 setting , @xmath48 @xmath49 and @xmath50 , the saddle - point equation ( [ poissonrad ] ) simplifies as , @xmath51 in order to have a regular solution at origin we have to impose @xmath52 we find from eqs.([emx])-([transf ] ) for fixed values of @xmath53 and @xmath54 , @xmath55 eqs.([up0 ] ) and ( [ emx2 ] ) provide the boundary conditions for the nonlinear ordinary differential equation ( [ reducedeq ] ) . in particular , they impose the dependence of @xmath56 on @xmath57 and @xmath58 .
we find from eq.([reducedeq ] ) for small @xmath59 the expansion of @xmath60 in powers of @xmath59 with the result , @xmath61 where we imposed eq.([up0 ] ) .
the value of @xmath62 follows by imposing eq.([emx2 ] ) .
using eqs.([x ] ) , ( [ phi ] ) and ( [ transf ] ) we can express the density of particles in terms of the solution of eq.([reducedeq ] ) @xmath63\ ; .\ ] ] we choose to express the physical quantities in terms of @xmath53 [ eq.([etaxi ] ) ] and the ratio @xmath64 of dark energy to self - gravity [ eq.([x ] ) ] .
the local pressure obeys the ideal gas equation of state in local form @xmath65 using eqs.([dens ] ) the pressure at the boundary is given by @xmath66\ ; \equiv \frac{n \ ; t}{v } \ ; f(x,\eta).\ ] ] the contrast @xmath67 between the pressure at the center and at the boundary can be written as , @xmath68 we plot in fig .
[ ps ] the density at the boundary versus @xmath53 for different values of @xmath64 .
the self - gravitating system can be in one of two phases : gaseous or highly condensed .
mean field theory describes the gas phase and gives exactly the physical quantities in the thermodynamic limit .
the equilibrium density @xmath70 solution of the mean field equation ( [ poisson ] ) minimizes the helmholtz free energy in the * canonical * ensemble .
this density is the most probable distribution which becomes absolutely certain in the infinite @xmath3 limit .
all thermodynamics quantities follows from this density @xmath70 .
this continuous density @xmath70 solution of the mean field equation ( [ poisson ] ) fails to describe the condensed phase .
the condensed phase have been found by monte - carlo simulations for @xmath69 @xcite . in the microcanonical ensemble
the center collapses in a very dense core surrounded by a halo of particles . in the canonical ensemble all the particles collapse in a very dense body @xcite .
the stability of the gaseous phase in the mean field approach can be analyzed looking for the extrema of the dimensionless parameter@xcite @xmath71 when the temperature @xmath72 , the pressure @xmath73 and the number of particles @xmath3 are kept fixed in the canonical ensemble .
the gas phase becomes instable ( for @xmath69 ) when @xmath74 is maximal@xcite .
using eqs.([etaxi ] ) , ( [ f ] ) and ( [ zeta ] ) we find that @xmath75 where @xmath76 is the external density in absence of dark energy .
a better physical insight is obtained by looking to the the isothermal compressibility @xmath77 using eqs .
( [ etaxi ] ) and ( [ f ] ) we find that the dimensionless isothermal compressibility @xmath78 takes the form @xmath79 we see from eqs.([zetaet ] ) and ( [ kappat ] ) that @xmath80 hence , @xmath81 * diverges * at the extrema of @xmath11 .
the isothermal compressibility @xmath81 is positive from @xmath82 till @xmath83 . at this point @xmath81 as well as the specific heat at constant
pressure diverge and change their signs@xcite .
moreover , at this point the speed of sound at the center of the sphere becomes imaginary@xcite .
therefore , small density fluctuations will grow exponentially in time instead of exhibiting oscillatory propagation .
such a behaviour leads to the jeans instability and collapse@xcite .
monte - carlo simulations confirmed the presence of this instability at @xmath83 in the canonical ensemble for @xmath69 @xcite .
we now compute the dimensionless isothermal compressibility for the self - gravitating gas with @xmath85 .
introducing the dimensionless parameter @xmath86 the dimensionless isothermal compressibility @xmath87 takes the form : @xmath88 the jeans instability happens when the isothermal compressibility diverges .
this happens for @xmath89 verifying : @xmath90 thus , for a fixed @xmath91 , @xmath92 is the value of @xmath53 which maximizes the quantity @xmath93 as for the self - gravitating gas with @xmath94 is maximal when the specific heat at constant pressure diverges and changes its sign .
we plot @xmath92 versus @xmath64 in fig .
[ etao ] using eq.([alpha ] ) .
monte carlo simulations confirm that the onset of the jeans collapse happens at @xmath95 . we can write @xmath33 [ eq.([etaxi ] ) ] as @xmath96 where @xmath97 is the total mass of the gas .
therefore , the collapse condition @xmath98 can be written as @xmath99 the gas collapses when its mass takes a larger value than the critical one @xmath100 .
we can consider @xmath101 as the generalization of the jeans mass in the presence of the cosmological constant .
notice that the presence of @xmath0 * increases * the jeans mass with respect the @xmath12 case [ see fig .
[ etao ] ] .
the effects of self - gravitation and dark energy go in opposite directions .
self - gravitating forces are attractive , while dark energy produces repulsion .
if the ratio of dark energy and self - gravity is @xmath102 [ defined by eq.([x ] ) ] , self - gravitation dominates over dark energy . if @xmath103 the effect of self - gravitation is exactly compensated by the dark energy . if @xmath16 dark energy dominates over self - gravitation .
we illustrate these three cases plotting @xmath104 versus @xmath53 for @xmath105 in fig .
[ ps ] * * first case : @xmath102 : self - gravity dominates * + we illustrate this case choosing @xmath106 .
we see in fig .
[ ps ] that @xmath104 versus @xmath53 exhibits a form analogous to @xmath107 versus @xmath53 in absence of dark energy .
@xmath104 has two riemann sheets as a function of @xmath53 . in the first sheet @xmath104 monotonically decreases with @xmath53 for fixed @xmath108 . *
* second case : @xmath103 : exact compensation * + the effect of self - gravitation is compensated by the effect of dark energy .
we see in fig .
[ ps ] that @xmath109 .
an exactly homogeneous sphere @xmath110 is a solution ( see below ) . in such special case
the self - gravitating gas behaves as a perfect gas ( with @xmath111 everywhere ) . * * third case : @xmath16 : dark energy dominates * + we illustrate this case choosing @xmath112 .
we see in fig .
[ ps ] that @xmath113 is an increasing function of @xmath53 for fixed @xmath114 .
the jeans instability happens in the three cases . the value of @xmath115 where the gas collapses increases with @xmath64 ( fig .
[ etao ] ) .
the gas collapses for @xmath116 at a higher density @xmath117 for a given temperature or for a lower temperature for a given density @xmath117 than for @xmath40 .
that is , the domain of stability of the gas increases for increasing @xmath64 .
the dark energy has an anti - gravity effect that disfavours the collapse pushing the particles toward the boundary of the sphere .
notice that the point of jeans instability is in the first riemann sheet of the @xmath53-plane for @xmath118 .
we study the density @xmath44 as a function of the radial coordinate @xmath119 .
the density @xmath44 of the self - gravitating gas in absence of dark energy ( @xmath40 ) is always a * decreasing * function of @xmath119 .
this follows from the attractive character of gravitation ( see for example ref .
@xcite ) .
the @xmath119-dependence of the density @xmath44 of a self - gravitating gas in the presence of dark energy is more involved because the dark energy opposes to the attraction by gravity .
the functions @xmath44 and @xmath60 have the same qualitative dependence in @xmath119 since they are related by exponentiation [ see eq .
( [ dens ] ) ] .
the behaviour at the center of the sphere ( @xmath45 ) is governed by the sign of @xmath62 since @xmath120 [ eq . ( [ up0 ] ) ] and
we find from eq.([dl ] ) that @xmath121 .
hence , @xmath122 = -\mbox{sign } [ u_0 ] \ ; .\ ] ] therefore , @xmath44 decreases ( increases ) at @xmath45 if @xmath123 ( @xmath124 ) .
the behaviour at the boundary of the sphere ( @xmath125 ) is governed by the sign of @xmath126 since according to eq.([emx2 ] ) @xmath127 = -\mbox{sign } [ \eta^{r}-\xi^{r } ] = \mbox{sign}(x-1 ) \ ; .\ ] ] therefore , @xmath44 decreases ( increases ) at @xmath128 if @xmath102 ( @xmath16 ) .
the particle density exhibits for the stable solutions ( namely , for @xmath129 ) one of the following behaviours illustrated in fig .
[ rho ] * * decreasing * : @xmath102 and @xmath123 .
the density @xmath44 decreases from the center of the sphere till the boundary .
we plot in fig .
[ rho ] the density @xmath44 versus @xmath119 for @xmath106 and @xmath130 .
the self - gravitation dominates over dark energy . * * increasing * : @xmath16 and @xmath124 .
the density @xmath44 increases from the center of the sphere till the boundary .
we plot in fig .
[ rho ] the density @xmath44 versus @xmath119 for @xmath112 and @xmath130 .
the dark energy dominates over self - gravitation . * * homogeneous * : @xmath103 and @xmath131 .
the density is homogeneous : @xmath110 .
the effect of self - gravity of particles is exactly compensated by the effect of dark energy .
we plot in fig .
[ rho ] the density @xmath44 versus @xmath119 for @xmath103 and @xmath130 .
eq.([poisson ] ) for @xmath132 is similar to the equation that describes in two space dimensions ( multi)-vortices in the ginsburg - landau or higgs model in the limit between superconductivity of type i and ii @xcite .
however , for the vortex case one has @xmath133 since like charges repel each other in electrodynamics while masses attract each other in gravity .
this opposite nature of the forces is responsible of the fact that eq.([poisson ] ) for @xmath132 only has trivial stable solutions while a host of non - trivial solutions appear in the vortex case@xcite .
the standard metropolis algorithm @xcite was first applied to self - gravitating gas in absence of dark energy in ref.@xcite .
we apply this standard metropolis to the self - gravitating gas in presence of dark energy .
we perform it in a volume @xmath2 in the canonical ensemble at temperature @xmath72 .
we compute in this way the external pressure , the energy , the average density , the average particle distance and the average squared particle distance as function of @xmath53 for a given value of the ratio of dark energy and self - gravity @xmath134 .
we implement the metropolis algorithm in the following way .
we start from a random distribution of @xmath3 particles in the chosen volume .
we update such configuration choosing a particle at random and changing at random its position .
we then compare the energies of the former and the new configurations .
we use the standard metropolis test to choose between the new and the former configurations .
the energy of the configurations are calculated performing the exact sums as in eq.([upun ] ) .
we use as statistical weight for the metropolis algorithm in the canonical ensemble , @xmath135 [ see eqs.([part ] ) and ( [ upun ] ) ] .
the number of particles go up to @xmath136 .
we introduce a short distance cutoff in the newtons potential [ see eq.([upun ] ) ] @xmath137 with @xmath138 where @xmath139 is the short distance cut - off .
the presence of the short distance cut - off prevents the collapse ( here unphysical ) of the self - gravitating gas ( for more details see @xcite ) .
we apply the metropolis algorithm to the self - gravitating gas in presence of dark energy in a sphere and in a cube . for @xmath40 ( absence of dark energy )
the monte carlo results in a sphere and in a cube with identical volume give practically the same results @xcite .
the geometry does not influence the physics of the system
. however , for @xmath140 ( presence of dark energy ) the monte carlo results in a sphere and in a cube turn to be different .
this effect can be traced to the repulsive character of the dark energy that pushes the particles towards the boundary . as a result ,
the physical quantities becomes sensible to the geometry .
we thus compared the mean field results with spherical symmetry from the previous sections and ref.@xcite with monte carlo calculations performed on a sphere .
we recall that our self - gravitating system exhibits three different behaviours according the value of @xmath64 , the relative rate of dark energy and self - gravity . in all three cases two different phases
show up : for @xmath141 we have a non perfect gas and for @xmath142 it is a very condensed system with negative pressure .
the transition between the two phases is very sharp . * for @xmath102
the self - gravity dominates over dark energy .
we plot the external density @xmath143 versus @xmath53 for @xmath106 in fig .
[ fmc0p3 ] . in the gas phase @xmath144 monotonically decreases with @xmath53 from @xmath82 till the collapse point @xmath145 .
+ the average distance between particles @xmath146 and the average squared distance between particles @xmath147 monotonically decrease with @xmath53 ( fig .
[ rmc ] ) .
when the gas collapses at @xmath148 , @xmath146 and @xmath147 exhibit a sharp drop . * for @xmath103 the effect of dark energy
exactly compensates the effect of self - gravity .
we have @xmath149 in the gas phase .
the system behaves like a perfect gas from @xmath82 till the collapse point @xmath145 . * for @xmath16
the dark energy dominates over self - gravity .
we plot in fig .
[ fmc1p5 ] the external density @xmath150 versus @xmath53 for @xmath112 . in the gas phase @xmath144 monotonically increases with @xmath53 from @xmath82 till the collapse point @xmath145 .
we find that the monte carlo calculations in the sphere accurately reproduce the mean field results for @xmath144 in the gas phase with spherical symmetry .
we then consider a cube and a sphere of identical volume .
the monte carlo calculations turn to give different results in these two cases : a ) the external pressure is larger in the cube than in the sphere .
b ) the gas phase is more stable in the cube .
c ) the collapse value @xmath151 in the cube is larger than its value for the same @xmath64 in the sphere .
the phase transition happens in the monte carlo simulations at @xmath145 .
it must be compared with the point of jeans instability @xmath89 according to mean field [ see iii.c ] where the isothermal compressibility diverges and changes its sign .
we see in that @xmath151 is * very close * of @xmath152 .
they are probably the same point .
for @xmath106 we have @xmath153 and @xmath154 .
for @xmath112 we have @xmath155 and @xmath156 .
we conclude that the collapse of the system observed in the monte carlo simulations is due to the jeans instability when the isothermal compressibility diverges and is well described by the mean field approach . for @xmath157
we see that the gaseous phase is metastable . before collapsing the system
stay for a long monte carlo time in the metastable gaseous phase in the monte carlo computations .
the monte carlo time for collapse increases with increasing value of @xmath64 .
this is due to the repulsive effect of dark energy .
the gas phase is less unstable in the presence of dark energy .
we illustrate the two behaviours of the self - gravitating gas in the presence of the cosmological constant plotting the average density as a function of two cartesian coordinates , the third coordinate being fixed . for simplicity
we depict the densities in the cube . for @xmath16
we see that the density is larger on the boundary that at the center of the cube ( fig .
[ densinc ] ) .
the dark energy dominates over the self - gravity and pushes the particles toward the boundary of the cube . for @xmath102
we see that the density is larger at the center than on the boundary of the cube ( fig .
[ densdec ] ) .
the self - gravity dominates over dark energy and attracts the particles to the center of the cube .
we illustrate now the two phases ( gaseous and condensed ) plotting the average particle distribution in a cube . fig .
[ profgas ] and fig .
[ profcoll ] depict the average particle distribution from monte carlo calculations with @xmath158 particles for @xmath106 at both sides of the collapse point .
[ profgas ] corresponds to the gaseous phase and fig .
[ profcoll ] to the collapsed phase .
we find that the monte carlo simulations ( describing thermal equilibrium ) are much more efficient than the @xmath3-body simulations integrating newton s equations of motion .
[ indeed , the integration of newton s equations provides much more information than thermal equilibrium investigations ] . actually , a few hundreds of particles are enough to get quite accurate results in the monte carlo simulations ( except near the collapse points )
. moreover , the monte carlo results turns to be in excellent agreement with the mean field calculations up to corrections of the order @xmath8 . in the condensed phase all the particles collapse in a very dense body ( see fig .
[ profcoll ] ) .
the self - gravity contribution of the potential energy dominates overwhelmingly the dark energy contribution in the condensed phase . using eqs .
( [ hamil ] ) , ( [ etaxi ] ) and ( [ x ] ) and the virial theorem @xcite we obtain that the external pressure is expressed as @xmath159 the monte carlo results indicate that @xmath160 .
thus in this condensed phase the external pressure can be approximated by @xmath161 where @xmath162 .
since @xmath163 has a jump at the transition , the gibbs free energy is discontinuous and we have a phase transition of the * zeroth * order as in the absence of cosmological constant @xcite .
the behaviour of the self - gravitating gas is significantly influenced by the cosmological constant ( or not ) depending on the value of the ratio @xmath64 defined by eq.([x ] ) , @xmath164 where @xmath165 is the average mass density in the volume @xmath2 and where the subscript @xmath166 indicates mass and cosmological constant inside the volume @xmath2 .
@xmath64 takes small values for astrophysical objects except for clusters of galaxies where @xmath167 and especially for the universe as a whole where @xmath168 .
the smallness of @xmath64 stems from the tiny value of the present cosmological constant @xmath169g/@xmath170@xcite .
the particle mass density is much larger than @xmath17 in all situations except for the universe as a whole .
however , the non - relativistic and equilibrium treatment does not apply for the universe as a whole .
certainly , the galaxy distribution can be considered as a non - relativistic self - gravitating gas in the presence of the cosmological constant .
but the gas of galaxies is certainly not yet at thermal equilibrium today .
perhaps , thermal equilibrium may be reached in some billions of years . incidentally , at thermal equilibrium the particle density is regular at the center of the distribution in the whole range of @xmath171 .
namely , there is no cusp at the center in a situation of thermal equilibrium . | the self - gravitating gas in the presence of a positive cosmological constant @xmath0 is studied in thermal equilibrium by monte carlo simulations and by the mean field approach .
we find excellent agreement between both approaches already for @xmath1 particles on a volume @xmath2 [ the mean field is exact in the infinite @xmath3 limit ] .
the domain of stability of the gas is found to increase when the cosmological constant increases .
the particle density is shown to be an increasing ( decreasing ) function of the distance when the dark energy dominates over self - gravity ( and vice - versa ) .
we confirm the validity of the thermodynamic limit : @xmath4 with @xmath5 and @xmath6 fixed . in such dilute limit extensive thermodynamic quantities like energy , free energy , entropy turn to be proportional to @xmath3 .
we find that the gas is stable till the isothermal compressibility diverges . beyond this point
the gas becomes a extremely dense object whose properties are studied by monte carlo . |
core - collapse supernovae ( sne ) are triggered by massive stars at the end of their lives .
thanks to the high luminosity , sne can be detected in distant galaxies . by measuring the cosmic occurrence rate of core - collapse sne , one can study the overall formation rate of the massive star population through cosmic time .
with upcoming or planned large - scale optical and near - infrared ( nir ) surveys , sne could be detected to @xmath7 or higher .
it is thus important to derive a realistic estimate for the detectability in order to make an efficient survey strategy for such observational programmes .
@xcite study the detectability of core - collapse sne with the _ large synoptic survey telescope _
( lsst , @xcite ) .
they show that the wide survey with lsst ( with the 5 @xmath8 @xmath9-band limiting magnitude of 25 mag ) can detect type ii sne , core - collapse sne with hydrogen features , up to @xmath10 .
magnification by gravitational lensing will boost the number of detectable sne @xcite .
type ii sne at higher redshifts can be detected by an extremely deep nir survey @xcite with future satellites such as _ james webb space telescope _ ( jwst ) .
@xcite show that up to several thousands type ii sne can be detected at @xmath11 if the survey depth reaches the flux limit of 3 njy ( 30 mag in ab magnitude ) , which can be achievable with long ( @xmath12 seconds ) integration with jwst .
it may be more promising to aim at detecting sne that are bright in the rest frame optical to ultra - violet ( uv ) wavelengths .
type iin sne , a type of sne with narrow hydrogen emission lines , are ideal targets to detect at high - redshifts @xcite .
type iin sne are thought to be powered by the interaction of the sn ejecta with the circumstellar medium @xcite , and are bright in the rest frame optical to uv .
@xcite reported the discovery of type iin sne at @xmath13 and 2.0 in the archival data of the legacy survey with canada - france - hawaii telescope .
there is another type of possibly luminous sne worth mentioning here .
pair - instability sne ( pisne ) are thought to be triggered by very massive stars with mass in the range of 140 - 260 @xmath14 at zero metallicity @xcite , and are also expected to be very luminous @xcite .
pisne might also occur even in the local universe , as suggested by the recent discovery of sn 2007bi @xcite .
however , pisne are not very luminous at rest uv wavelengths @xcite , and thus they can be detected only at @xmath15 even with lsst @xcite .
it is possible , in principle , to detect pisne at @xmath16 with an extremely deep nir survey with @xmath1728.7 mag ( @xcite , see also recent papers by @xcite ) , but the occurence rate of pisne ( or the formation rate of extremely massive stars ) at such high redshifts is highly uncertain .
interestingly , recent theoretical studies and detailed simulations of the first generation of stars suggest that the characteristic mass of the first stars is not as large as the progenitor of a pisn @xcite .
then observing core - collapse sne may be a more promising route to study the massive star population in the early universe .
recently , superluminous sne were discovered such as sn 2006gy @xcite and sn 2005ap @xcite , whereas other superluminous sne include sne 2003ma @xcite , 2006 oz @xcite , 2008es @xcite , 2008fz @xcite , 2008ma @xcite , 2010gx ( ptf10cwr , * ? ? ?
* ; * ? ? ?
* ) , scp 06f6 @xcite .
ptf09atu , ptf09cnd , ptf09cwl , @xcite , ps1 - 10ky , ps1 - 10awh @xcite , and css100217 @xcite .
see also @xcite for other earlier candidates .
these sne have an absolute magnitude of about @xmath18 mag in optical .
some of them are classified as type iin ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , or type iil , which is a subclass of type ii sne with the linear light curve decline @xcite , and others are hydrogen poor ( type ibc , e.g. , * ? ? ?
* ; * ? ? ?
the nature of these superluminous sne is poorly known .
some of them are powered by the circumstellar interaction , or by the shock breakout from the dense circumstellar medium @xcite , as suggested by the presence of narrow emission lines in superluminous type iin sne ( e.g. , * ? ? ?
it is also argued that superluminous sne could be powered by a large amount of @xmath19ni which is synthesized as a result of energetic core - collapse sne @xcite .
other scenarios include the interaction between shells ejected by the pulsational pair - instability @xcite , and sne powered by a magnetar @xcite .
although the origin of the high luminosity is not yet fully understood , the progenitors of the superluminous sne are thought to be very massive .
the long time - scale of the light curve is suggestive of a long diffusion times of optical photons and hence of a large mass of ejecta ( for a given kinetic energy ) .
at least one progenitor of type iin sn is directly identified to be as luminous as luminous blue variables with the zero - age main sequence mass of @xmath20 @xcite .
superluminous sne may provide an important information on the star formation in the high - redshift universe .
because superluminous sne are likely to be triggered by the very massive end of the stellar population , the relative occurence rate of such sne with respect to the overall star formation rate can be a sensitive probe of the slope of the stellar initial mass function ( imf ) .
theoretical studies suggest that the shape of the imf may be different in the early universe , or could even be of top - heavy one ( e.g. , * ? ? ?
* ) . detecting superluminous sne at high redshifts will give a direct evidence for such non - standard imf .
in this paper , we study the detectability of superluminous sne at high redshifts with the upcoming wide - field surveys .
@xcite noted that superluminous sne are bright enough that they can be detectable even at @xmath0 with 8m - class telescopes , such as subaru and lsst .
we derive a realistic estimate for the detectability for a few particular set of surveys . for the optical survey
, we focus on the proposed survey with the subaru / hyper suprime - cam ( hsc , @xcite ) .
the method of mock observations is described in section [ sec : setup ] .
we show that with a realistic survey strategy we can detect superluminous sne to @xmath21 ( section [ sec : opt ] ) .
more emphasis is made on the detectability by upcoming nir survey utilizing future satellites , such as euclid , the wide - field infrared survey telescope ( wfirst ) , and wide - field imaging surveyor for high - redshift ( wish ) ( section [ sec : nir ] ) .
we study how the expected number of detection at @xmath6 is affected by the slope of the imf at the massive end .
details of the sample selection are described in section [ sec : selection ] .
finally , we give our concluding remarks in section [ sec : conclusions ] . | observations of high - redshift supernovae ( sne ) open a novel opportunity to study the massive star population in the early universe .
we study the detectability of superluminous sne with upcoming optical and near - infrared ( nir ) surveys .
our calculations are based on the cosmic star formation history , the sn occurence rate , the characteristic colour and the light curve of the sne that are all calibrated by available observations .
we show that 15 - 150 sne up to @xmath0 will be discovered by the proposed subaru / hyper suprime - cam deep survey : 30 deg@xmath1 survey with 24.5 ab mag depth in @xmath2-band for 3 months . with its ultra - deep layer ( 3.5 deg@xmath1 with 25.6 ab mag depth in @xmath2-band for 4 months )
, the highest redshift can be extended to @xmath3 .
we further explore the detectability by upcoming nir survey utilizing future satellites such as euclid , wfirst , and wish .
the wide - field nir surveys are very efficient to detect high - redshift sne . with a hypothetical deep nir survey for 100 deg@xmath1 with 26 ab mag depth at 1 - 4 @xmath4 m ,
at least @xmath5 50 sne will be discovered at @xmath6 in half a year .
the number of the detected sne can place a strong constraint on the stellar initial mass function or its slope especially at the high - mass end .
superluminous sne at high redshifts can be distinguished from other types of sne by the long time - scale of their light curves in the observer s frame , the optical colours redder than other core - collapse sne and the nir colours redder than any other types of sne .
[ firstpage ] stars : luminosity function , mass function
supernovae : general early universe |
the science of networks is a modern discipline spanning the natural , social and computer sciences , as well as engineering @xcite .
networks , or graphs , consist of _ vertices _ and _ edges_. an edge typically connects a pair of vertices .
networks occur in an huge variety of contexts .
facebook , for instance , is a large social network , where more than one billion people are connected via virtual acquaintanceships .
another famous example is the internet , the physical network of computers , routers and modems which are linked via cables or wireless signals ( fig .
[ figgennet ] ) .
many other examples come from biology , physics , economics , engineering , computer science , ecology , marketing , social and political sciences , etc .. most networks of interest display _ community structure _ , i. e. , their vertices are organised into groups , called _ communities _ , _ clusters _ or _ modules_. in fig .
[ figsfi ] we show a collaboration network of scientists working at the santa fe institute ( sfi ) in santa fe , new mexico .
vertices are scientists , edges join coauthors .
edges are concentrated within groups of vertices representing scientists working on the same research topic , where collaborations are more natural .
likewise , communities could represent proteins with similar function in protein - protein interaction networks , groups of friends in social networks , websites on similar topics on the web graph , and so on .
identifying communities may offer insight on how the network is organised .
it allows us to focus on regions having some degree of autonomy within the graph .
it helps to classify the vertices , based on their role with respect to the communities they belong to .
for instance we can distinguish vertices totally embedded within their clusters from vertices at the boundary of the clusters , which may act as brokers between the modules and , in that case , could play a major role both in holding the modules together and in the dynamics of spreading processes across the network .
community detection in networks , also called _ graph _ or _ network clustering _ , is an ill - defined problem though .
there is no universal definition of the objects that one should be looking for .
consequently , there are no clear - cut guidelines on how to assess the performance of different algorithms and how to compare them with each other . on the one hand , such ambiguity leaves a lot of freedom to propose diverse approaches to the problem , which often depend on the specific research question and ( or ) the particular system at study . on the other hand , it has introduced a lot of noise into the field , slowing down progress . in particular , it has favoured the diffusion of questionable concepts and convictions , on which a large number of methods are based .
this work presents a critical analysis of the problem of community detection , intended to practitioners but accessible to readers with basic notions of network science .
it is not meant to be an exhaustive survey .
the focus is on the general aspects of the problem , especially in the light of recent findings .
also , we discuss some popular classes of algorithms and give advice on their usage . more info on network clustering can be found in several review articles @xcite .
the contents are organised in three main sections .
section [ sec - comm ] deals with the concept of community , describing its evolution from the classic subgraph - based notions to the modern statistical interpretation .
next we discuss the critical issue of validation ( section [ sec - valid ] ) , emphasising the role of artificial benchmarks , the importance of the choice of partition similarity scores , the conditions under which clusters are detectable , the usefulness of metadata and the structural peculiarities of communities in real networks .
section [ sec - meth ] hosts a critical discussion of some popular clustering approaches .
it also tackles important general methodological aspects , such as the determination of the number of clusters , which is a necessary input for several techniques , the possibility to generate robust solutions by combining multiple partitions , the main approaches to discover dynamic communities , as well as the assessment of the significance of clusterings . in section
[ sec - soft ] we indicate where to find useful software .
the concluding remarks of section [ sec - ol ] close the work .
we start with a subgraph @xmath0 of a graph @xmath1 .
the number of vertices and edges are @xmath2 , @xmath3 for @xmath1 and @xmath4 , @xmath5 for @xmath0 , respectively .
the adjacency matrix of @xmath6 is @xmath7 , its element @xmath8 equals @xmath9 if vertices @xmath10 and @xmath11 are neighbours , otherwise it equals @xmath12 .
we assume that the subgraph is connected because communities usually are .
other types of group structures do not require connectedness ( section [ sec - mv ] ) .
the subgraph is schematically illustrated in fig .
[ figstruct ] .
its vertices are enclosed by the dashed contour .
the magenta dots are the external vertices connected to the subgraph , while the black ones are the remaining vertices of the network .
the blue lines indicate the edges connecting the subgraph to the rest of the network . the _ internal _ and _ external _ degree @xmath13 and @xmath14 of a vertex @xmath10 of the network with respect to subgraph @xmath0 are the number of edges connecting @xmath10 to vertices of @xmath0 and to the rest of the graph , respectively .
both definitions can be expressed in compact form via the adjacency matrix @xmath7 : @xmath15 and @xmath16 , where the sums run over all vertices @xmath11 inside and outside @xmath17 , respectively . naturally , the degree @xmath18 of @xmath10 is the sum of @xmath13 and @xmath14 : @xmath19 .
if @xmath20 and @xmath21 @xmath10 has neighbours only within @xmath0 and is an _ internal vertex _ of @xmath0 ( dark green dots in the figure ) . if @xmath22 and @xmath21 @xmath10 has neighbours outside @xmath0 and is a _ boundary vertex _ of @xmath0 ( bright green dots in the figure ) . if @xmath23 , instead , the vertex is disjoint from @xmath0 .
the _ embeddedness _
@xmath24 is the ratio between the internal degree and the degree of vertex @xmath10 : @xmath25 .
the larger @xmath26 , the stronger the relationship between the vertex and its community .
the _ mixing parameter _
@xmath27 is the ratio between the external degree and the degree of vertex @xmath10 : @xmath28 . by definition ,
@xmath29 .
now we present a number of variables related to the subgraph as a whole .
we distinguish them in three classes .
the first class comprises measures based on internal connectedness , i. e. , on how cohesive the subgraph is .
the main variables are : * _ internal degree _ @xmath30 .
the sum of the internal degrees of the vertices of @xmath0 .
it equals twice the number @xmath5 of internal edges , as each edge contributes two units of degree . in matrix form , @xmath31
* _ average internal degree _ @xmath32 .
average degree of vertices of @xmath0 , considering only internal edges : @xmath33 . *
_ internal edge density _ @xmath34 .
the ratio between the number of internal edges of @xmath0 and the number of all possible internal edges : @xmath35 we remark that @xmath36 is the maximum number of internal edges that a simple graph with @xmath4 vertices may have .
the second class includes measures based on external connectedness , i. e. , on how embedded the subgraph is in the network or , equivalently , how separated the subgraph is from it .
the main variables are : * _ external degree _ , or _ cut _ , @xmath37 .
the sum of the external degrees of the vertices of @xmath0 .
it gives the number of external edges of the subgraph ( blue lines in fig .
[ figstruct ] ) . in matrix form , @xmath38 . * _ average external degree _ , or _ expansion _ , @xmath39 .
average degree of vertices of @xmath0 , considering only external edges : @xmath40 . *
_ external edge density _ , or _ cut ratio _ , @xmath41 .
the ratio between the number of external edges of @xmath0 and the number of all possible external edges : @xmath42 finally , we have hybrid measures , combining internal and external connectedness .
notable examples are : * _ total degree _ , or _
volume _ , @xmath43 .
the sum of the degrees of the vertices of @xmath0 . naturally , @xmath44 . in matrix form , @xmath45 . * _ average degree _ @xmath46 .
average degree of vertices of @xmath0 : @xmath47 . *
_ conductance _ @xmath48 .
the ratio between the external degree and the total degree of @xmath0 : @xmath49 all definitions we have given hold for the case of undirected and unweighted networks .
the extension to weighted graphs is straightforward , as it suffices to replace the number of edges " with the sum of the weights carried by every edge .
for instance , the internal degree @xmath50 of a vertex @xmath51 becomes the _ internal strength _
@xmath52 , which is the sum of the weights of the edges joining @xmath51 with the vertices of subgraph @xmath17 . for the internal and external edge densities of eqs .
( [ eq1 ] ) and ( [ eq2 ] ) one would have to replace the numerators with their weighted counterparts and multiply the denominators by the average edge weight @xmath53 , where @xmath54 is the element of the _ weight matrix _
, indicating the weight of the edge joining vertices @xmath10 and @xmath11 ( @xmath55 if @xmath10 and @xmath11 are disconnected ) and @xmath3 the total number of graph edges . in tables
[ tab : metricsv ] and [ tab : metrics ] we list all variables we have presented along with their extensions to the case of weighted networks . in directed networks
one would have to distinguish between incoming and outgoing edges .
extensions of the metrics are fairly simple to implement , though their usefulness is unclear .
figure [ figcom ] shows how scholars usually envision community structure .
the network has three clusters and in each cluster the density of edges is comparatively higher than the density of edges between the clusters .
this can be summarised by saying that communities are dense subgraphs which are well separated from each other .
this view has been challenged , recently @xcite , as we shall see in section [ ncp ] .
communities may overlap as well , sharing some of the vertices .
for instance , in social networks individuals can belong to different circles at the same time , like family , friends , work colleagues .
figure [ figoverl ] shows an example of a network with overlapping communities .
communities are typically supposed to be overlapping at their boundaries , as in the figure .
recent results reveal a different picture , though @xcite ( section [ ncp ] ) . a subdivision of a network into overlapping communities
is called _ cover _ and one speaks of _ soft clustering _ , as opposed to _ hard clustering _ , which deals with divisions into non - overlapping groups , called
_ partitions_. the generic term _ clustering _ can be used to indicate both types of subdivisions .
covers can be _ crisp _ , when shared vertices belong to their communities with equal strength , or _ fuzzy _
, when the strength of their membership can be different in different clusters .
the oldest definitions of community - like objects were proposed by social network analysts and focused on the internal cohesion among vertices of a subgraph @xcite .
the most popular concept is that of _ clique _ @xcite .
a clique is a _ complete graph _ , that is , a subgraph such that each of its vertices is connected to all the others .
it is also a maximal subgraph , meaning that it is not included in a larger complete subgraph . in modern network science
it is common to call clique any complete graph , not necessarily maximal .
triangles are the simplest cliques .
finding cliques is an * np*-complete problem @xcite ; a popular technique is the bron
kerbosch method @xcite .
the notion of cliques , albeit useful , can not be considered a good candidate for a community definition .
while a clique has the largest possible internal edge density , as all internal edges are present , communities are not complete graphs , in general .
moreover , all vertices have identical role in a clique , while in real network communities some vertices are more important than others , due to their heterogeneous linking patterns .
therefore , in social network analysis the notion has been relaxed , generating the related concepts of _ n - cliques _
@xcite , _ n - clans _ and _ n - clubs _ @xcite .
other definitions are based on the idea that a vertex must be adjacent to some minimum number of other vertices in the subgraph .
@xmath56-plex _ is a maximal subgraph in which each vertex is adjacent to all other vertices of the subgraph except at most @xmath56 of them @xcite .
details on the above definitions can be found in specialised books @xcite . for a proper community definition
, one should take into account both the internal cohesion of the candidate subgraph and its separation from the rest of the network .
a simple idea that has received a great popularity is that a community is a subgraph such that the number of internal edges is larger than the number of external edges " .
this idea has inspired the following definitions .
an _ @xmath57-set
_ @xcite , or _ strong community _
@xcite , is a subgraph such that the internal degree of each vertex is greater than its external degree .
a relaxed condition is that the internal degree of the subgraph exceeds its external degree [ _ weak community _
@xcite ] to be a weak community it is not necessary that the number of internal edges @xmath5 exceeds that of external edges @xmath37 . since the internal degree @xmath58 ( section [ sec - var ] ) the actual condition is @xmath59 . ] .
a strong community is also a weak community , while the converse is not generally true .
a drawback of these definitions is that one separates the subgraph at study from the rest of the network , which is taken as a single object .
but the latter can be in turn divided into communities .
if a subgraph @xmath0 is a proper community , it makes sense that each of its vertices is more strongly attached to the vertices of @xmath0 than to the vertices of any other subgraph .
this concept , proposed by hu et al .
@xcite , is more in line ( though not entirely ) with the modern idea of community that we discuss in the following section .
it has generated two alternative definitions of strong and weak community .
a subgraph @xmath0 is a strong community if the internal degree of any vertex within @xmath0 exceeds the internal degree of the vertex within any other subgraph , i. e. , the number of edges joining the vertex to those of the subgraph ; likewise , a community is weak if its internal degree exceeds the ( total ) internal degree of its vertices within every other community . a strong ( weak ) community la radicchi et al .
is a strong ( weak ) community also in the sense of hu
et al .. the opposite is not true , in general ( fig .
[ figsw ] ) .
in particular , a subgraph can be a strong community in the sense of hu et al . even though all of its vertices have internal degree smaller than their respective external degree .
the above definitions of communities use _ extensive _ variables : their value tends to be the larger , the bigger the community ( e. g. , the internal and external degrees ) .
but there are also variables discounting community size .
an example is the internal cluster density @xmath60 of eq .
( [ eq1 ] ) .
one could assume that a subgraph @xmath0 with @xmath56 vertices is a cluster if @xmath60 is larger than a threshold @xmath61 .
setting the size of the subgraph is necessary because otherwise any clique would be among the best possible communities , including trivial two - cliques ( simple edges ) or triangles . as we have seen in the previous section , traditional definitions of community
rely on counting edges ( internal , external ) , in various ways .
but what one should be really focusing on is the _ probability _ that vertices share edges with a subgraph .
the existence of communities implies that vertices interact more strongly with the other members of their community than they do with vertices of the other communities .
consequently , there is a preferential linking pattern between vertices of the same group .
this is the reason why edge densities end up being higher within communities than between them .
we can formulate that by saying that vertices of the same community have a higher probability to form edges with their partners than with the other vertices .
let us suppose that we estimated the edge probabilities between all pairs of vertices , somehow .
we can define the groups by means of those probabilities .
it is a scenario similar to the classic one we have seen in section [ sec - defs ] , where we add and compare probabilities , instead of edges .
natural definitions of strong and weak community are : * a _ strong community _ is a subgraph each of whose vertices has a higher probability to be linked to every vertex of the subgraph than to any other vertex of the graph . * a _ weak community _ is a subgraph such that the average edge probability of each vertex with the other members of the group exceeds the average edge probability of the vertex with the vertices of any other group . ] .
the difference between the two definitions is that , in the concept of strong community , the inequality between edge probabilities holds at the level of every pair of vertices , while in the concept of weak community the inequality holds only for averages over groups .
therefore , a strong community is also a weak community , but the opposite is not true , in general .
now we can see why the former definitions of strong and weak community @xcite are not satisfactory .
suppose to have a network with two subgraphs @xmath7 and @xmath62 of very different sizes , say with @xmath63 and @xmath64 vertices ( fig .
[ figexsw ] ) . and @xmath62 and edge probabilities @xmath65 between vertices of the subgraphs and @xmath66 between vertices of @xmath7 and @xmath62 .
the red circle is a representative vertex of subgraph @xmath7 , the smaller blue circle represents the rest of the vertices of @xmath7 .
the subgraphs are both strong and weak communities in the probabilistic sense , but they may be neither strong nor weak communities according to the classic definitions by radicchi et al .
@xcite and hu et al .
@xcite , if @xmath62 is sufficiently larger than @xmath7 . ]
the network is generated by a model where the edge probability is @xmath65 between vertices of the same group and @xmath67 for vertices of different groups .
the two subgraphs are communities both in the strong and in the weak sense , according to the probability - based definitions above .
the expected internal degree of a vertex of @xmath7 is @xmath68 : since there are @xmath63 possible internal neighbours , but for simplicity one allows for the formation of self - edges , from a vertex to itself .
results obtained with and without self - edges are basically undistinguishable , when community sizes are much larger than one .
we shall stick to this setup throughout the paper . ] .
likewise , the expected external degree of a vertex of @xmath7 is @xmath69 .
the expected internal and external degrees of @xmath7 are @xmath70 and @xmath71 .
for any two values of @xmath65 and @xmath67 one can always choose @xmath72 sufficiently larger than @xmath63 that @xmath73 , which also implies that @xmath74 .
in this setting the subgraphs are neither strong nor weak communities , according to the definitions proposed by radicchi et al . and
hu et al .. how can we compute the edge probabilities between vertices ? this is still an ill - defined problem , unless one has a model stating how edges are formed .
one can make many hypotheses on the process of edge formation .
+ for instance , if we take social networks , we can assume that the probability that two individuals know each other is a decreasing function of their geographical distance , on average @xcite .
each set of assumptions defines a model . for our purposes
, eligible models should take into account the possible presence of groups of vertices , that behave similarly .
the most famous model of networks with group structure is the _
stochastic block model _ ( sbm )
suppose we have a network with @xmath2 vertices , divided in @xmath75 groups .
the group of vertex @xmath10 is indicated with the integer label @xmath76 .
the idea of the model is very simple : the probability @xmath77 that vertices @xmath10 and @xmath11 are connected depends exclusively on their group memberships : @xmath78 .
therefore , it is identical for any @xmath10 and @xmath11 in the same groups .
the probabilities @xmath79 form a @xmath80 symmetric matrix , called the _
stochastic block matrix_. the diagonal elements @xmath81 ( @xmath82 ) of the stochastic block matrix are the probabilities that vertices of block @xmath56 are neighbours , whereas the off - diagonal elements give the edge probabilities between different blocks between blocks @xmath83 and @xmath84 is fixed ( @xmath85 ) , instead of the edge probabilities . if @xmath86 the two models are fully equivalent if the edge probabilities @xmath87 are chosen such that the expected number of edges running between @xmath83 and @xmath84 coincides with @xmath88 . ] . for @xmath89 , with @xmath90 , we recover community structure , as the probabilities that vertices of the same group are connected exceed the probabilities that vertices of different groups are joined ( fig . [
fig : stylized - hotdog ] ) .
it is also called _ assortative structure _ , as it privileges bonds between vertices of the same group .
the model is very versatile , though , and can generate various types of group structure . for @xmath91 , with @xmath90 , we have _ disassortative structure _ , as edges are more likely between the blocks than inside them ( fig .
[ fig : stylized - bipartite ] ) . in the special case in which @xmath92 we recover _ multipartite structure _ , as there are edges only between the blocks . if @xmath93 , @xmath94 , we have _ core - periphery structure _ : the vertices of the first block ( core ) are relatively well - connected amongst themselves as well as to a peripheral set of vertices that interact very little amongst themselves ( fig .
[ fig : stylized - coreper ] ) .
if all probabilities are equal , @xmath95 , @xmath96 , we recover the classic random graph la erds and rnyi @xcite ( fig .
[ fig : stylized - random ] ) . here
any two vertices have identical probability of being connected , hence there is no group structure .
this has become a fundamental axiom in community detection , and has inspired some popular techniques like , e. g. , modularity optimisation @xcite ( section [ sec - modopt ] ) .
random graphs of this type are also useful in the validation of clustering algorithms ( section [ art - bench ] ) .
alternative community definitions are based on the interplay between network topology and dynamics .
diffusion is the most used dynamics .
random walks are the simplest diffusion processes .
a simple random walk is a path such that the vertex reached at step @xmath97 is a random neighbour of the vertex reached at step @xmath98 .
a random walker would be spending a long time within communities , due to the supposedly low number of routes taking out of them @xcite .
the evolution of random walks does not depend solely on the number or density of edges , in general , but also on the structure and distribution of paths formed by consecutive edges , as paths are the routes that walkers can follow .
this means that random walk dynamics relies on _ higher - order structures _ than simple edges , in general .
such relationship is even more pronounced when one considers markov dynamics of second order or higher , in which the probability of reaching a vertex at step @xmath99 of the walk does not depend only on where the walker sits at step @xmath97 , but also on where it was at step @xmath98 and possibly earlier @xcite .
indeed , one could formulate the network clustering problem by focusing on higher order structures , like _ motifs _
( e. g. , triangles ) @xcite .
the advantage is that one can preserve more complex features of the network and its communities , which typically get lost when one uses network models solely based on edge probabilities , like sbms .
the drawback is that calculations become more involved and lengthy .
is a definition of community really necessary ? actually not , most techniques to detect communities in networks do not require a precise definition of community .
the problem can be attacked from many angles .
for instance , one can remove the edges separating the clusters from each other , that can be identified via some particular feature @xcite . but
defining clusters beforehand is a useful starting point , that allows one to check the reliability of the final results .
in this section we will discuss the crucial issue of validation of clustering algorithms .
validation usually means checking how precisely algorithms can recover the communities in benchmark networks , whose community structure is known .
benchmarks can be computer - generated , according to some model , or actual networks , whose group structure is supposed to be known via non - topological features ( metadata ) . the lack of a universal definition of communities makes the search for benchmarks rather arbitrary , in principle .
nevertheless , the best known artificial benchmarks are based on the modern definition of clusters presented in section [ sec - mv ] .
we shall present some popular artificial benchmarks and show that partition similarity measures have to be handled with care .
we will see under which conditions communities are detectable by methods , and expose the interplay between topological information and metadata .
we will conclude by presenting some recent results on signatures of community structure extracted from real networks .
the principle underneath stochastic block models ( section [ sec - mv ] ) has inspired many popular benchmark graphs with group structure .
community structure is recovered in the case in which the probability for two vertices to be joined is larger for vertices of the same group than for vertices of different groups ( fig .
[ fig : stylized - hotdog ] ) . for simplicity ,
let us suppose that there are only two values of the edge probability , @xmath65 and @xmath100 , for edges within and between communities , respectively .
furthermore , we assume that all communities have identical size @xmath101 , so @xmath102 , where @xmath75 is the number of communities . in this version ,
the model coincides with the _ planted l - partition model _ , introduced in the context of graph partitioning ) ] . ]
the expected internal and external degrees of a vertex are @xmath103 and @xmath104 , respectively , yielding an expected ( total ) vertex degree @xmath105 .
girvan and newman @xcite set @xmath106 , @xmath107 ( for a total number of vertices @xmath108 ) and fixed the average total degree @xmath109 to @xmath110 .
this implies that @xmath111 and @xmath65 and @xmath112 are not independent parameters .
the benchmark by girvan and newman is still the most popular in the literature ( fig .
[ figgn ] ) .
performance plots of clustering algorithms typically have , on the horizontal axis , the expected external degree @xmath113
. for low values of @xmath113 communities are well separated is higher , to keep the total degree constant . ] and most algorithms do a good job at detecting them . by increasing @xmath113 , the performance declines . still , one expects to do better than by assigning memberships to the vertices at random , as long as @xmath114 , which means for @xmath115 . in section [ detectab ]
we will see that the actual threshold is lower , due to random fluctuations .
the benchmark by girvan and newman , however , is not a good proxy of real networks with community structure .
for one thing , all vertices have equal degree , whereas the degree distribution of real networks is usually highly heterogeneous @xcite . in addition , most clustering techniques find skewed distributions of community sizes @xcite .
for this reason , lancichinetti , fortunato and radicchi proposed the _ lfr benchmark _ , having power - law distributions of degree and community size @xcite ( fig .
[ figlfr ] ) .
the mixing parameters @xmath116 of the vertices ( section [ sec - var ] ) are set equal to a constant @xmath117 , which estimates the quality of the partition is actually only the average of the mixing parameter over all vertices .
in fact , since degrees are integer , it is impossible to tune them such to have exactly the same value of @xmath117 for each vertex , and keep the constraint on the degree distribution at the same time . ] .
lfr benchmark networks are built by joining stubs at random , once one has established which stubs are internal and which ones are external with respect to the community of the vertex attached to the stubs . in this respect
, it is basically a configuration model @xcite with built - in communities .
clearly , when @xmath117 is low , clusters are better separated from each other , and easier to detect .
when @xmath117 grows , performance starts to decline .
but for which range of @xmath117 can we expect a performance better than random guessing ?
let us suppose that the group structure is detectable for @xmath118 $ ] .
the upper limit @xmath119 should be such that the network is random for @xmath120 . the network is random when stubs are combined at random , without distinguishing between internal and external stubs , which yields the standard configuration model .
there the expected number of edges between two vertices with degrees @xmath18 and @xmath121 is @xmath122 , @xmath3 being the total number of network edges .
let us focus on a generic vertex @xmath10 , belonging to community @xmath17 .
we denote with @xmath123 and @xmath124 the sum of the degrees of the vertices inside and outside @xmath17 , respectively .
clearly @xmath125 .
in a random graph built with the configuration model , vertex @xmath10 would have an expected internal degree . ]
@xmath126 and an expected external degree @xmath127 . since , by construction , @xmath128 and @xmath129 , the community @xmath17 is not real when @xmath130 and @xmath131 , which implies @xmath132 .
we see that @xmath123 depends on the community @xmath17 : the larger the community , the lower the threshold is
. therefore , not all clusters are detectable at the same time , in general . for this to happen
, @xmath117 must be lower than the minimum of @xmath133 over all communities : @xmath134 . if communities are all much smaller than the network as a whole , @xmath135 and @xmath119 could get very close to the upper limit @xmath9 of the variable @xmath117 .
however , it is possible that the actual threshold is lower than @xmath119 , due to the perturbations of the group structure induced by random fluctuations ( section [ detectab ] ) .
anyway , in most cases the threshold is going to be quite a bit higher than @xmath136 , the value which is mistakenly considered as the threshold by some scholars .
the lfr benchmark turns out to be a special version of the recently introduced degree - corrected stochastic block model @xcite , with the degree and the block size distributed according to truncated power laws , which is local , a global parameter @xmath137 is used , estimating how strong the community structure is . ] .
the lfr benchmark has been extended to directed and weighted networks with overlapping communities @xcite .
the extensions to directed and weighted graphs are rather straightforward .
overlaps are obtained by assigning each vertex to a number of clusters and distributing its internal edges equally among them .
recently , another benchmark with overlapping communities has been introduced by ball , karrer and newman @xcite .
it consists of two clusters @xmath7 and @xmath62 , with overlap @xmath17 .
vertices in the non - overlapping subsets @xmath138 and @xmath139 set edges only between each other , while vertices in @xmath17 are connected to vertices of both @xmath7 and @xmath62 .
the expected degree of all vertices is set equal to @xmath140 .
the authors considered various settings , by tuning @xmath109 , the size of the overlap and the sizes of @xmath7 and @xmath62 , which may be uneven .
however , the fact that all vertices have equal degree ( on average ) makes the model less realistic and flexible than the lfr benchmark . following the increasing availability of evolving time - stamped network data sets ,
the analysis and modelling of temporal networks have received a lot of attention lately @xcite . in particular ,
scholars have started to deal with the problem of detecting evolving communities ( section [ sec - dynclus ] ) .
a benchmark designed to model dynamic communities was proposed by granell et al .
it is based on the planted @xmath141-partition model , just like the benchmark of girvan and newman , where @xmath65 and @xmath67 are the edge probabilities within communities and between communities , respectively .
communities may grow and shrink ( fig .
[ figdb]a ) , they may merge with each other or split into smaller clusters ( fig .
[ figdb]b ) , or do all of the above ( fig .
[ figdb]c ) .
the dynamics unfold such that at each time the subgraphs are proper communities in the probabilistic sense discussed in section [ sec - mv ] . in the merge - split dynamics ,
clusters actually merge before the inter - community edge probability @xmath112 reaches the value @xmath65 of the intra - community edge probability , due to random fluctuations ( section [ detectab ] ) . .
reprinted figure with permission from @xcite .
2015 , by the american physical society . ] in section [ sec - mv ] we have shown why random graphs can not have a meaningful group structure .
that means that they can be employed as _ null benchmarks _ , to test whether algorithms are capable to recognise the absence of groups .
many methods find non - trivial communities in such random networks , so they fail the test .
we strongly encourage doing this type of exam on new algorithms @xcite .
the accuracy of clustering techniques depends on their ability to detect the clusters of networks , whose community structure is known .
that means that the partition detected by the method(s ) has to match closely the planted partition of the network .
how can the similarity of partitions be computed ?
this is an important problem , with no unique solution . in this section
we discuss some issues about partition similarity measures .
more information can be found in @xcite , @xcite and @xcite .
let us consider two partitions @xmath142 and @xmath143 of a network @xmath6 , with @xmath144 and @xmath145 clusters , respectively .
let @xmath2 be the total number of vertices , @xmath146 and @xmath147 the number of vertices in clusters @xmath148 and @xmath149 and @xmath150 the number of vertices shared by clusters @xmath148 and @xmath149 : @xmath151 .
the @xmath152 matrix @xmath153 whose entries are the overlaps @xmath150 is called _ confusion matrix _ , _ association matrix _ or _
contingency table_. most similarity measures can be divided in three categories : measures based on _ pair counting _ ,
_ cluster matching _ and _ information theory_. pair counting means computing the number of pairs of vertices which are classified in the same ( different ) clusters in the two partitions .
let @xmath154 indicate the number of pairs of vertices which are in the same community in both partitions , @xmath155 ( @xmath156 ) the number of pairs of elements which are in the same community in @xmath157 ( @xmath158 ) and in different communities in @xmath158 ( @xmath157 ) and @xmath159 the number of pairs of vertices that are in different communities in both partitions .
several measures can be defined by combining the above numbers in various ways .
a famous example is the _ rand index _
@xcite @xmath160 which is the ratio of the number of vertex pairs correctly classified in both partitions ( i. e. either in the same or in different clusters ) , by the total number of pairs .
another notable option is the _ jaccard index _
@xcite , @xmath161 which is the ratio of the number of vertex pairs classified in the same cluster in both partitions , by the number of vertex pairs classified in the same cluster in at least one partition .
the jaccard index varies over a broader range than the rand index , due to the dominance of @xmath159 in @xmath162 , which typically confines the rand index to a small interval slightly below @xmath9 .
both measures lie between @xmath12 and @xmath9 .
if we denote with @xmath163 and @xmath164 the sets of vertex pairs with are members of the same community in partitions @xmath157 and @xmath158 , respectively , the jaccard index is just the ratio between the intersection and the union of @xmath163 and @xmath164 .
such concept can be used as well to determine the similarity between two clusters @xmath7 and @xmath62 @xmath165 the score @xmath166 is also called jaccard index and is the most general definition of the score , for any two sets @xmath7 and @xmath62 @xcite . measuring
the similarity between communities is very important to determine , given different partitions , which cluster of a partition corresponds to which cluster(s ) of the other(s ) .
for instance , the cluster @xmath149 of @xmath158 corresponding to cluster @xmath148 of @xmath158 is the one maximising the similarity between @xmath148 and @xmath149 , e. g. , @xmath167 .
this strategy is also used to track down the evolution of communities in temporal networks @xcite .
the rand and the jaccard indices , as defined in eqs .
( [ eqt04 ] ) and ( [ eqt20 ] ) , have the disturbing feature that they do not take values in the entire range @xmath168 $ ] .
for this reason , adjusted versions of both indices exist , in that a baseline is introduced , yielding the expected values of the score for all pairs of partitions @xmath169 and @xmath170 obtained by randomly assigning vertices to clusters such that @xmath169 and @xmath170 have the same number of clusters and the same size for all clusters of @xmath157 and @xmath158 , respectively @xcite .
the baseline is subtracted from the unadjusted version , and the result is divided by the range of this difference , yielding @xmath9 for identical partitions and @xmath12 as expected value for independent partitions .
but there are problems with these definitions as well . the null model used to compute the baseline relies on the assumption that the communities of the independent partitions have the same number of vertices as in the partitions whose similarity is to be compared .
but such assumption usually does not hold , in practical instances , as algorithms sometimes need the number of communities as input , but they never impose any constraint on the cluster sizes . adjusted indices have also the disadvantage of nonlocality @xcite : the similarity between partitions differing only in one region of the network depends on how the remainder of the network is subdivided
. moreover , the adjusted scores can take negative values , when the unadjusted similarity lies below the baseline .
a better option is to use _ standardised indices _
@xcite : for a given score @xmath171 the value of the null model term @xmath116 is computed along with its standard deviation @xmath172 over many different randomisations of the partitions @xmath157 and @xmath158 . by computing the @xmath173-score @xmath174
we can see how non - random the measured similarity score is , and assess its significance .
it can be shown that the @xmath173-scores for the jaccard , rand and adjusted rand indices coincide @xcite , so the measures are statistically equivalent .
since the actual values @xmath171 of these indices differ for the same pair of partitions , in general , we conclude that the magnitudes of the scores may give a wrong perception about the effective similarity . cluster matching aims at establishing a correspondence between pairs of clusters of different partitions based on the size of their overlap .
a popular measure is the _ fraction of correctly detected vertices _ , introduced by girvan and newman @xcite .
a vertex is correctly classified if it is in the same cluster as at least half of the other vertices in its cluster in the planted partition .
if the detected partition has clusters given by the merger of two or more groups of the planted partition , all vertices of those clusters are considered incorrectly classified . the number of correctly classified vertices is then divided by the number @xmath2 of vertices of the graph , yielding a number between @xmath12 and @xmath9 . the recipe to label vertices as correctly or incorrectly classified is somewhat arbitrary .
the fraction of correctly detected vertices is similar to @xmath175 where @xmath176 is the index of the best match @xmath177 of cluster @xmath178 @xcite .
a common problem of this type of measures is that partitions whose clusters have the same overlap would have the same similarity , regardless of what happens to the parts of the communities which are unmatched .
the situation is illustrated schematically in fig .
[ figsimclus ] . partitions @xmath179 and @xmath180
are obtained from @xmath181 by reassigning the same fraction of their elements to the other clusters .
their overlaps with @xmath181 are identical and so are the corresponding similarity scores . however , in partition @xmath180 the unmatched parts of the clusters are more scrambled than in @xmath179 , which should be reflected in a lower similarity score .
similarity can be also estimated by computing , given a partition , the additional amount of information that one needs to have to infer the other partition .
if partitions are similar , little information is needed to go from one to the other .
such extra information can be used as a measure of dissimilarity . to evaluate the shannon information content @xcite of a partition , we start from the community assignments @xmath182 and @xmath183 , where @xmath184 and @xmath185 indicate the cluster labels of vertex @xmath10 in partition @xmath157 and @xmath158 , respectively .
the labels @xmath186 and @xmath187 are the values of two random variables @xmath188 and @xmath189 , with joint distribution @xmath190 , so that @xmath191 and @xmath192 . the _ mutual information _
@xmath193 of two random variables is @xmath194 , where @xmath195 is the shannon entropy of @xmath188 and @xmath196 is the conditional entropy of @xmath188 given @xmath189 .
the mutual information is not ideal as a similarity measure : for a given partition @xmath157 , all partitions derived from @xmath157 by splitting ( some of ) its clusters would all have the same mutual information with @xmath157 , even though they could be very different from each other . in this case
the mutual information equals the entropy @xmath197 , because the conditional entropy is zero .
it is then necessary to introduce an explicit dependence on the other partition , that persists even in those special cases .
this has been achieved by introducing the _ normalized mutual information _ ( nmi ) , obtained by dividing the mutual information by the arithmetic average and @xmath198 @xcite .
] of the entropies of @xmath157 and @xmath158 @xcite @xmath199 the nmi equals @xmath9 if and only if the partitions are identical , whereas it has an expected value of @xmath12 if they are independent . since the first thorough comparative analysis of clustering algorithms @xcite ,
the nmi has been regularly used to compute the similarity of partitions in the literature .
however , the measure is sensitive to the number of clusters @xmath200 of the detected partition , and may attain larger values the larger @xmath200 , even though more refined partitions are not necessarily closer to the planted one .
this may give wrong perceptions about the relative performance of algorithms @xcite . a more promising measure , proposed by meil @xcite is the _ variation of information _
( vi ) @xmath201 the vi defines a metric in the space of partitions as it has the properties of distance ( non - negativity , symmetry and triangle inequality ) .
it is a local measure : the vi of partitions differing only in a small portion of a graph depends on the differences of the clusters in that region , and not on how the rest of the graph is subdivided .
the maximum value of the vi is @xmath202 , which implies that the scores of an algorithm on graphs of different sizes can not be compared with each other , in principle .
one could divide @xmath203 by @xmath202 @xcite , to force the score to be in the range @xmath168 $ ] , but the actual span of values of the measure depends on the number of clusters of the partitions .
in fact , if the maximum number of communities is @xmath204 , with @xmath205 , @xmath206 . consequently , in those cases where it is reasonable to set an upper bound on the number of clusters of the partitions , the similarities between planted and detected partitions on different graphs become comparable , and it is possible to assess both the performance of an algorithm and to compare algorithms across different benchmark graphs .
we stress , however , that the measure may not be suitable when the partitions to be compared are very dissimilar from each other @xcite and that it shows unintuitive behaviour in particular instances @xcite .
so far we discussed of comparing partitions .
what about covers ?
extensions of the measures we have presented to the case of overlapping communities are usually not straightforward .
the _ omega index _
@xcite is an extension of the adjusted rand index @xcite .
let @xmath157 and @xmath158 be covers of the same graph to be compared .
we denote with @xmath207 the number of pairs of vertices occurring together in exactly @xmath11 communities in both covers .
it is a natural generalisation of the variables @xmath159 and @xmath154 we have seen above , where @xmath11 can also be larger than @xmath9 since a pair of vertices can now belong simultaneously to multiple communities .
the variable @xmath208 is the fraction of pairs of vertices belonging to the same number of communities in both covers ( including the case @xmath209 , which refers to the pairs not being in any community together ) .
the omega index is defined as @xmath210 where @xmath211 is the expected value of @xmath212 according to the null model discussed earlier , in which vertex labels are randomly reshuffled such to generate covers with the same number and size of the communities .
the nmi has also been extended to covers by lancichinetti , fortunato and kertsz @xcite .
the definition is non - trivial : the community assignments of a cover are expressed by a vectorial random variable , as each vertex may belong to multiple clusters at the same time .
the measure overestimates the similarity of two covers , in special situations , where intuition suggests much lower values .
the problem can be solved by using an alternative normalisation , as shown in @xcite .
unfortunately neither the definition by lancichinetti , fortunato and kertsz nor the one by mcdaid , greene and hurley are proper extensions of the nmi , as they do not coincide with the classic definition of eq .
( [ eqt08 ] ) when partitions in non - overlapping clusters are compared . however , the differences are typically small , and one can rely on them in practice .
esquivel and rosvall have proposed an actual extension @xcite . following the comparative analysis performed in @xcite
, the nmi by lancichinetti , fortunato and kertsz has been regularly used in the literature , also in the case of regular partitions , without overlapping communities . if covers are fuzzy ( section [ sec - defs ] ) , the similarity measures above can not be used , as they do not take into account the degree of membership of vertices in the communities they belong to .
a suitable option is the _ fuzzy rand index _
@xcite , which is an extension of the adjusted rand index . both the fuzzy rand index and the omega index coincide with the adjusted rand index when communities do not overlap . for temporal networks
, a nave approach would be comparing partitions ( or covers ) corresponding to configurations of the system in the same time window , and to see how this score varies across different time windows .
however , this does not tell if the clusters are evolving in the same way , as there would be no connection between clusterings at different times .
a sensible approach is comparing sequences of clusterings , by building a confusion matrix that takes into account multiple snapshots .
this strategy allows one to define dynamic versions of various indices , like the nmi and the vi @xcite . in conclusion ,
while there is no clear - cut criterion to establish which similarity measure is best , we recommend to use measures based on information theory .
in particular , the vi seems to have more potential than others , for the reasons we explained , modulo the caveats in refs .
there are currently no extensions of the vi to handle the comparison of covers , but it would not be difficult to engineer one , e. g. , by following a similar procedure as in @xcite , though this might cost the sacrifice of some of its nice features
. one should keep in mind that the choice of one similarity index or another is a sensitive one , and warped conclusions may be drawn when different measures are adopted . in fig .
[ figcompsim ] we show the accuracy of two algorithms on the lfr benchmark ( section [ art - bench ] ) : _ ganxis _ , a method based on label propagation @xcite and _ linkcommunities _ , a method based on grouping edges instead of vertices @xcite ( section [ sec - lc ] ) .
the accuracy is estimated with the nmi by lancichinetti , fortunato and kertsz @xcite ( left diagram ) and with the omega index [ eq . ( [ eq_omega1 ] ) ] ( right diagram ) .
from the left plot one would think that ganxis clearly outperforms linkcommunities , whereas from the right plot ganxis still prevails for @xmath117 until about @xmath213 ( though the curves are closer to each other than in the nmi plot ) and linkcommunities is better for larger values of @xmath117 .
vertices , average degree @xmath214 , maximum degree @xmath215 , exponents @xmath216 and @xmath9 for the degree and community size distributions and range @xmath217 $ ] for the community size.,title="fig : " ] vertices , average degree @xmath214 , maximum degree @xmath215 , exponents @xmath216 and @xmath9 for the degree and community size distributions and range @xmath217 $ ] for the community size.,title="fig : " ] in validation procedures one assumes that , if the network has clusters , there must be a way to identify them . therefore ,
if we do not manage , we have to blame the specific clustering method(s ) adopted . but are we certain that clusters are always detectable ?
most networks of interest are _ sparse _ , i. e. ,
their average degree is much smaller than the number of vertices .
this means that the number of edges of the graph is much smaller than the number of possible edges @xmath218 .
a more precise way to formulate this is by saying that a graph is sparse when , in the limit of infinite size , the average degree of the graph remains finite .
a number of analytical calculations can be carried out by using network sparsity .
many algorithms for community detection only work on sparse graphs . on the other hand
, sparsity can also give troubles .
due to the very low density of edges , small amounts of noise could perturb considerably the structure of the system .
for instance , random fluctuations in sparse graphs could trick algorithms into finding groups that do not really exist ( section [ sec - sign ] ) .
likewise , they could make actual groups undetectable .
let us consider the simplest version of the assortative stochastic block model , which matches the planted partition model ( section [ art - bench ] ) .
there are @xmath75 communities of the same size @xmath219 , and only two values for the edge probability : @xmath65 for pairs of vertices in the same group and @xmath112 for pairs of vertices in different groups .
since the graphs are sparse , @xmath65 and @xmath112 vanish in the limit of infinite graph size .
so we shall use the expected internal and external degrees @xmath220 and @xmath221 , which stay constant in that limit . by construction ,
the groups are communities so long as @xmath222 or , equivalently , for @xmath223 .
but that does not mean that they are always detectable . in principle
, dealing with the issue of detectability involves examining all conceivable clustering techniques , which is clearly impossible .
fortunately , it is not necessary , because we know what model has generated the communities of the graphs we are considering .
the most effective technique to infer the groups is then fitting the stochastic block model on the data ( _ a posteriori block modelling _ ) .
this can be done via the maximum likelihood method @xcite . in recent work
@xcite , decelle et al .
have shown that , in the limit of infinite graph size , the partition obtained this way is correlated with the planted partition whenever @xmath224 which implies @xmath225 so , given a value of @xmath113 , when @xmath226 is in the range @xmath227 $ ] the probability @xmath228 of classifying a vertex correctly is not larger than the probability @xmath229 of assigning the vertex to a randomly chosen group , although the groups are communities , according to the model .
we stress that this result only holds when the graphs are sparse : if @xmath65 and @xmath112 remain non - zero in the large - n limit ( dense graph ) , the classic detectability threshold @xmath114 is correct . a fortiori , no clustering technique can detect the clusters better than random assignment when the inference of the model parameters fails to do so .
if communities are searched via the spectral optimisation of newman - girvan s modularity @xcite , one obtains the same threshold of eq .
( [ eqdetect1 ] ) @xcite , provided the network is not too sparse . for the benchmark of girvan and newman ( section [ art - bench ] )
@xcite it has long been unclear where the actual detectability limit sits .
girvan - newman benchmark graphs are not infinite , their size being set to @xmath230 , so there is no proper detectability transition , but rather a smooth crossover from a regime in which clusters are frequently detectable to a regime where they are frequently undetectable . for this reason there
can not be a sharp threshold separating the two regimes .
still it is useful to have an idea of where the pattern changes . in the following
we shall still use the term threshold to refer to the crossover point . in the beginning
, scholars thought that clusters are detectable as long as they satisfy the definition of strong community by radicchi et al .
@xcite ( section [ sec - defs ] ) , i. e. , as long as the expected internal degree exceeds the expected external degree , yielding a threshold @xmath231 @xcite .
since the expected total degree of a vertex is set to @xmath110 , communities are detectable as long as @xmath232 .
it soon became obvious that the actual threshold should be the one of the modern " definition of community we have presented in section [ sec - mv ] , according to which the condition coincide .
] is @xmath114 , that is @xmath233 .
however , numerical calculations reveal that algorithms tend to fail long before that limit . from eq .
( [ eqdetect1 ] ) we see that for the case of four infinite clusters and total expected degree @xmath234 , the theoretical detectability limit is @xmath235 . in fig .
[ figdet ] we see the performance on the benchmark of three well - known algorithms : louvain @xcite , a greedy optimisation technique of newman girvan modularity @xcite ( section [ sec - modopt ] ) ; infomap , which is based on random walk dynamics @xcite ( section [ sec - dynmet ] ) ; oslom , that searches for clusters via a local optimisation of a significance score @xcite .
the accuracy is estimated via the fraction of correctly detected vertices ( section [ sim - me ] ) .
the three thresholds @xmath236 , @xmath237 and @xmath238 are represented by vertical lines .
the performance of all methods becomes comparable with random assignment well before @xmath237 .
the theoretical limit @xmath238 appears to be compatible with the performance curves .
indicates the threshold corresponding to the concept of strong community la radicchi et al . , the dashed line at @xmath239 the threshold according to the probability - based definition of strong community we have given in section [ sec - mv ] .
the baseline of random assignment is @xmath240 ( horizontal dashed line ) .
all algorithms do not do better than random assignment already before @xmath239 .
the theoretical detectability limit is at @xmath241 , in the limit of groups of infinite size . ]
graph sparsity is a necessary condition for clusters to become undetectable , but it is not sufficient .
the symmetry of the model we have considered plays a major role too .
clusters have equal size and vertices have equal degree .
this helps to confuse " algorithms .
if communities have unequal sizes and the degree of vertices are correlated with the size of their communities , so that vertices have larger degree , the bigger their clusters , community detection becomes easier , as the degrees can be used as proxy for group membership . in this case , the non - trivial detectability limit disappears when there are four clusters or fewer , while it persists up to a given extent of group size inequality when there are more than four clusters @xcite .
other types of block structure , like core - periphery , do not suffer from detectability issues @xcite .
lfr benchmark graphs are more complex models than the one studied in @xcite and it is not clear whether there is a non - trivial detectability limit , though it is unlikely , due to the big heterogeneity in the distribution of vertex degree and community size .
another standard way to test clustering techniques is using real networks with known community structure .
knowledge about the memberships of the vertices typically comes from _ metadata _ , i. e. , non - structural information .
if vertices are annotated communities are assumed to be groups of vertices with identical tags .
examples are user groups in social networks like livejournal and product categories for co - purchasing networks of products of online retailers such as amazon . in fig .
[ zach ] we show the most popular of such benchmark graphs , zachary karate club network @xcite .
it consists of @xmath242 vertices , the members of a karate club in the united states , who were observed over a period of three years .
edges connect individuals interacting outside the activities of the club .
eventually a conflict between the club president ( vertex @xmath242 ) and the instructor ( vertex @xmath9 ) led to the fission of the club in two separate groups , whose members supported the instructor and the president , respectively ( indicated by the colours ) .
indeed , the groups make sense topologically : vertices @xmath9 and @xmath242 are hubs , and most members are directly connected to either of them .
most algorithms of community detection have been tested on this network , as well as others , e. g. , the american college football network @xcite or lusseau s network of bottlenose dolphins @xcite .
the idea is that the method doing the best job at recovering groups with identical annotations would also be the most reliable in applications . such idea , however , is based on a questionable principle , i. e. , that the groups corresponding to the metadata are also communities in the topological sense we have discussed in section [ sec - comm ] .
communities exist because their vertices are supposed to be similar to each other , in some way .
the similarity among the vertices is then revealed topologically through the higher edge probability among pairs of members of the same group than between pairs of members of different groups , whose similarity is lower .
hence , when one is provided with annotations or other sources of information that allows to classify vertices based on their similarity , one expects that such similarity - based classes are also the best communities that structure - based algorithms may detect .
indeed , for some small networks like zachary s karate club this seems to be the case .
but for quite some time scholars could not test this hypothesis , due to the limited number of suitable data sets . over the past few years
this has finally become possible , due to the availability of several large network data sets with annotated vertices @xcite .
it turns out that the alignment between the communities found by standard clustering algorithms and the annotated groups is not good , in general . in fig .
[ tmhric ] we show the similarity between the topological partitions found by different methods and the annotated partitions , for several social , information and technological networks @xcite .
the heights of the vertical bars are the values of the normalised mutual information ( nmi ) @xcite .
groups of contiguous bars represent the scores for a given data set . to the left of the vertical dashed line
we see the results for classic benchmarks , like lfr graphs ( section [ art - bench ] ) @xcite , zachary karate club , etc . , and the scores are generally good . but for the large data sets to the right of the line the scores are rather low , signalling a significant mismatch between topological and annotated communities . for amazon
co - purchasing network @xcite , in which vertices are products and edges are set between products often purchased by the same customer(s ) , the similarity is quite a bit higher than for the other networks .
this is because the classification of amazon products is hierarchical ( e. g. , ` books / fiction / fantasy ` ) , so there are different levels of annotated communities , and the reported scores refer to the one which is most similar to the structural ones detected by the algorithms , while the other levels would give lower similarity scores .
low similarity at the partition level does not rule out that some communities of the structural partition significantly overlap with their annotated counterparts , but precision and recall scores show that this is not the case .
results depend more on the network than on the specific method adopted , none of which appears to be particularly good on any ( large ) data set .
so the hypothesis that structural and annotated clusters are aligned is not warranted , in general .
there can be multiple reasons for that .
the attributes could be too general or too specific , yielding communities which are too large or too small to be interesting .
moreover , while the best partition of the network delivered by an algorithm can be poorly correlated with the metadata , there may be alternative topological divisions that also belong to a set of valid solutions , according to the algorithm , but happen to be better correlated with the annotations @xcite .
the fact that structural and annotated communities may not be strongly correlated has important consequences .
scholars have been regularly testing their algorithms on small annotated graphs , like zachary s karate club , by tuning parameters such to obtain the best possible performance on them .
this is not justified , in general , as it makes sense only when there is a strong correspondence , which is a priori unknown .
also , forcing an alignment with annotations on one data set does not guarantee that there is going to be a good alignment with the annotations of a different network .
besides , one of the reasons why people use clustering algorithms is to provide an improved classification of the vertices , by using structure .
if one obtained the same thing , why bother ?
the right thing to do is using structure _ along with _ the annotations , instead of insisting on matching them . this way the information coming from structure and metadata can be combined and we can obtain more accurate partitions , if there is any correspondence between them .
recent approaches explicitly assume that the metadata ( or a portion thereof ) are either exactly or approximately correlated with the best topological partition @xcite .
a better approach is not assuming a priori that the metadata correlate with the structural communities .
the goal is quantifying the relationship between metadata and community and use it to improve the results .
if there is no correlation , the metadata would be ignored , leaving us with the partition derived from structure alone .
methods along these lines have been developed , using stochastic block models .
newman and clauset @xcite have proposed a model in which vertices are initially assigned to clusters based on metadata , and then edges are placed between vertices according to the degree - corrected stochastic block model @xcite .
hric et al . have designed a similar model @xcite , in which the interplay between structure and metadata is represented by a multilayer network ( fig .
[ dameta ] ) .
the generative model is an extension of the hierarchical stochastic block model ( sbm ) @xcite with degree - correction for the case with edge layers @xcite . at ( 0,0 ) ; at ( 0,0 ) ; here the metadata is not supposed to correspond simply to a partition of the vertices .
the majority of data sets contain rich metadata , with vertices being annotated multiple times , and often few vertices possess the exact same annotations and can be thus associated to the same group .
in addition , while the number of communities is required as input by the method of newman and clauset , here it is inferred from the data .
finally , it is also possible to assess the metadata in its power to predict the network structure , not only their correlation with latent partitions . this way it is possible to predict _ missing vertices _ of the network , i. e. , to infer the connections of a vertex from its annotations only .
we stress that neither method requires that all vertices are annotated .
applications of the method by hric et al .
@xcite reveal that in many data sets there are statistically significant correlations between the annotations and the network structure , while in some cases the metadata seems to be largely uncorrelated with structural clusters .
we conclude that network metadata should not be used indiscriminately as ground truth for community detection methods .
even when the metadata is strongly predictive of the network structure , the agreement between the annotations and the network division tends to be complex , and very different from the one - to - one mapping that is more commonly assumed .
moreover , data sets usually contain considerable noise in their annotations , and some metadata tags are essentially random , with no relationship to structure .
artificial benchmark graphs are certainly very useful to assess the performance of clustering algorithms .
however , one could always question whether the model of community structure they propose is reliable .
how can we assess this ? in order to characterise real " communities we have to find them first .
but that can only be done via some algorithm , and different algorithms search for different types of objects , in general .
still , one may hope that general properties of communities can be consistently uncovered across different methods and data sets , while other features are more closely tied to the specific method(s ) used to detect the clusters and ( or ) the specific data set at study ( or classes thereof ) .
a seemingly robust feature of communities in real networks is the heterogeneity of their size distribution .
most clustering techniques find skewed distributions of cluster sizes in many networks .
so , there appears to be no characteristic size for a community : small communities usually coexist with large ones .
this feature is rather independent of the type of network ( fig .
[ figcomsize ] ) .
it may signal a hierarchy among communities , with small clusters included in large ones .
methods unable to distinguish between hierarchical levels might find blended " partitions , consisting of communities of different levels and hence of very different sizes .
the lfr benchmark was the first graph model to take explicitly into account the heterogeneity of community sizes ( section [ art - bench ] ) .
another interesting question is how the quality of communities depends on their size .
leskovec et al .
@xcite carried out a systematic analysis of clusters in many large networks , including traditional and on - line social networks , technological , information networks and web graphs . instead of considering partitions , they focused on individual communities , which are derived by optimising conductance ( section [ sec - var ] ) around seed vertices .
we remind that the conductance of a cluster is the ratio between the number of external edges and the total degree of the cluster .
minimising conductance effectively combines the two main community demands , i. e. , good separation from the rest of the graph ( low numerator ) and large number of internal edges ( high denominator ) .
the measure is also relatively insensitive to the size of the clusters , as both the numerator and the denominator are typically proportional to the number of vertices of the community
. therefore one could use it to compare the quality of clusters of different sizes . for any given size @xmath56 leskovec et al . identified the subgraph with @xmath56 vertices with the lowest conductance .
this way , for each network one can draw the _ network community profile _ ( ncp ) , showing the minimum conductance as a function of community size .
the ncps of all networks studied by leskovec et al .
have a characteristic shape : they go downwards till @xmath243 vertices , and then they rise monotonically for larger subgraphs [ fig .
[ figlesk ] ( left ) ] .
alternative shapes have been recently found for other data sets @xcite . for networks characterised by ncps like the one in fig .
[ figlesk ] ( left ) the most pronounced communities are fairly small in size .
such small clusters are weakly connected to the rest of the network , often by a single edge ( in this case they are called _ whiskers _ ) , and form the _ periphery _ of the graph .
large clusters have comparatively lower quality and melt into a big _ core_. large communities can often be split in parts with lower conductance , so they can be considered conglomerates of smaller communities . a schematic picture of the resulting network structure is shown in fig .
[ figlesk ] ( right ) .
the shape of the ncp is fairly independent of the specific technique adopted to identify the subgraphs with minimum conductance .
the different shapes of the ncps encountered in data suggest that core - periphery is not the only model of group structure of real networks @xcite .
the ncp is a signature that can be used to select generative mechanisms of community structure . indeed
, many standard models typically yield ncp sloping steadily downwards , at odds with the ones encountered in many social and information networks .
stochastic block models ( section [ sec - mv ] ) are sufficiently versatile that they can reproduce the ncp shape of fig .
[ figlesk ] ( left ) , by suitably tuning the parameters . in the standard lfr benchmark ( section [ art - bench ] ) the mixing parameters are tightly concentrated about a value @xmath117 by construction ,
hence all clusters have approximately conductance @xmath117 , yielding a roughly flat ncp , it can be easily shown that the conductance of the subgraph is exactly equal to @xmath117 . ]
however , the model can be easily modified by making @xmath117 community - dependent and a large variety of ncps are attainable , including the one of fig .
[ figlesk ] ( left ) .
the main problem of working with ncps is that they are based on extreme statistics , as one systematically reports the minimum conductance for a given cluster size . how representative is this extremal subgraph of the population of subgraphs with the same size
? there may be just a few clusters of a given size with low conductance .
it may happen that many subgraphs have conductance near the minimum corresponding to their size(s ) , which would then be representative .
alternatively most subgraphs might have much larger conductance than the minimum but low enough that they can be still considered communities . in this case
one should conclude that communities of that size are not of very high quality , on average .
the above scenarios might lead to different conclusions about the actual community structure of the system .
in general , even if one could produce a version of the ncp where the trend refers to representative samples of communities of equal size ( whatever that means ) , the actual values of the conductance are as important as the shape of the curve .
if conductance is sufficiently low for all cluster sizes , it means that there are good communities of any size .
the fact that small clusters could be of higher quality does not undermine the role of large clusters .
the observation that large clusters consist of smaller clusters of higher quality may just be evidence of hierarchical structure in the network , which is a trademark of many complex systems @xcite . in that case
high levels of the hierarchy are not less important than low ones , a priori .
in fact , the actual relative importance of communities should not only come from the sheer value of specific metrics , like conductance , but also from their statistical significance ( section [ sec - sign ] ) . that notwithstanding , we strongly encourage analyses like the one by leskovec et al . , as they provide a statistical characterisation of community structure , in a way that is only weakly algorithm - dependent .
one has to define operationally what a cluster is , but in a simple intuitive way that allows us to draw conclusions about the structure of the graph . in principle
one could do the same by analysing the clusters delivered by any algorithm , but there would be two important drawbacks .
first , the clusters may not be easy to interpret , as most clustering algorithms usually do not require a clear - cut definition of community .
second , one would have to handle a partition of the network in communities , instead of probing locally the group structure of the network .
therefore , for a given vertex one would have only one cluster ( or a handful , if communities overlap ) , while a local exploration allows to analyse a whole population of candidate subgraphs , which gives more information .
the local subgraphs recovered this way do not need to be strongly matching the clusters delivered by any algorithm , but they provide useful signatures that allow to restrict the set of possible model explanations for the network s group structure .
such investigation can be replicated on any model graph to check whether the results match ( e. g. , whether the ncps coincide ) .
[ fig : noover ] + + another approach to infer properties of clusters of real networks is using annotations . while we have shown
that annotated clusters do not necessarily coincide with structural ones ( section [ topmet ] ) , general features can be still derived , provided they are consistently found across different data sets and annotations .
a recent analysis by yang and leskovec has questioned the common picture of networks with overlapping communities @xcite .
scholars usually assume that clusters overlap at their boundaries , hence edge density should be larger in the non - overlapping parts ( fig .
[ figoverl ] ) . instead , by analysing the overlaps of annotated clusters in large social and information networks , yang and leskovec found that the probability that two vertices are connected is larger in the overlaps , and grows with the number of communities sharing that pair of vertices .
in addition , _ connector vertices _ , i. e. , vertices with the largest number of neighbours within a community , are more likely to be found in the overlaps .
these findings suggest that the overlaps may play an important role in the community structure of networks . in fig .
[ fig : agm ] we compare the conventional view with the one resulting from the analysis . the _ community - affiliation graph model _ ( agm ) @xcite and the _ cluster affiliation model for big networks _ ( bigclam ) @xcite are clustering techniques based on generative models of networks featuring communities with dense overlaps .
the models are based on the principle that vertices are more likely to be neighbours the more the communities sharing them , in line with the empirical finding of @xcite .
actual overlapping communities exist in many contexts .
however , it is unclear whether soft clustering is statistically founded . a recent analysis aiming at identifying suitable stochastic block models to describe real network data indicate that in many cases hard partitions ought to be preferred , as they give simpler descriptions of the group structure of the data than soft partitions @xcite .
this could be due to the fact that the underlying models are based on placing edges independently of each other , neglecting higher order structures between vertices , like motifs . by adopting approaches that take into account higher - order structures things may change and community overlaps might become a statistically robust feature .
the pervasive overlaps found by yang and leskovec in annotated data can be found if higher order effects are considered @xcite , without ad hoc hypotheses .
there are many algorithms to detect communities in graphs
. they can be grouped in categories , based on different criteria , like the actual operational method @xcite , or the underlying concept of community @xcite . in most applications , however , just a few popular algorithms are employed . in this section
we present a critical analysis of these methods .
we show the advantages of knowing the number of clusters before - hand and how it is possible to derive robust solutions from partitions delivered by stochastic clustering techniques .
we discuss approaches to the problem of detecting communities in evolving networks and how to assess the significance of the detected clustering .
we conclude by suggesting the methods that currently appear to be most promising . in general , the only preliminary information available to any algorithm
is the structure of the network , i. e. , which pairs of vertices are connected to each other and which are not ( possibly including weights ) .
any insight about community structure is supposed to be given as output of the procedure .
naturally , it would be valuable to have some information on the unknown division of the network beforehand , as one could reduce considerably the huge space of possible solutions , and increase the chance of successfully identifying the communities . among all the possible pre - detection inputs ,
the number @xmath75 of clusters plays a prominent role .
many popular classes of algorithms require the specification of @xmath75 before they run , like methods imported from data clustering or parametric statistical inference approaches ( section [ sec - inference ] ) .
other methods are capable to infer @xmath75 as they can choose among partitions into different numbers of communities .
but even such methods could benefit from a preliminary knowledge of @xmath75 @xcite . in fig .
[ fignumclus ] we report standard accuracy plots of two algorithms on the planted partition model ( section [ art - bench ] ) with two clusters of equal size .
the algorithms are modularity optimisation via simulated annealing @xcite and the absolute potts model ( apm ) @xcite ( section [ sec - dynmet ] ) .
there are two performance curves for each method : one comes from the standard application of the method , without constraints ; the other is obtained by forcing the method to explore only the subset of partitions with the correct number of clusters @xmath93 .
vertices , which are grouped in two equal - sized communities .
the accuracy is measured via the fraction of correctly detected vertices ( section [ sim - me ] ) .
the horizontal line indicates the accuracy of random guessing , the dashed vertical line the theoretical detectability limit ( section [ detectab ] ) . for each algorithm
we show two curves , referring to the results of the method in the absence of any information on the number of clusters , and when such information is fed into the model as initial input . in both cases , knowing the number of clusters beforehand leads to a much better performance . ]
we see that the accuracy improves considerably when @xmath75 is known .
this is particularly striking in the case of modularity optimisation , which is known to have a limited resolution , preventing the method from identifying the correct scale of the communities , even when the latter are very pronounced @xcite ( section [ sec - modopt ] ) .
knowing @xmath75 and constraining the optimisation of the measure to partitions with fixed @xmath75 , the problem can be alleviated @xcite .
but how do we know how many clusters there are ? here
we briefly discuss some heuristic techniques , for statistically principled methods we defer the reader to section [ sec - inference ] .
it has been recently shown that in the planted partition model @xmath75 can be correctly inferred all the way up to the detectability limit from the spectra of two matrices : the _ non - backtracking matrix _ @xmath244 @xcite and the _ flow matrix _ @xmath245 @xcite .
they are @xmath246 matrices , where @xmath3 is the number of edges of the graph .
each edge is considered in both directions , yielding @xmath247 directed edges and indicated with the notation @xmath248 , meaning that the edge goes from vertex @xmath10 to vertex @xmath11 .
their elements read @xmath249 and @xmath250 in eq .
( [ eqfm ] ) @xmath18 is the degree of vertex @xmath10 .
so the elements of @xmath245 are basically the elements of @xmath244 , normalised by vertex degrees .
this is done to account for the heterogeneous degree distributions observed in most real networks .
both matrices have non - zero elements only for each pair of edges forming a directed path from the first vertex of one edge to the second of the other edge .
to do that , edges have to be incident at one vertex . as a matter of fact ,
the non - backtracking matrix @xmath244 is just the adjacency matrix of the ( directed ) edges of the graph .
the name of the matrix @xmath251 is due to a connection with the properties of non - backtracking walks .
non - backtracking walk _
@xcite is a path across the edges of a graph that is allowed to return to a vertex visited previously only after at least two other vertices have been visited ; immediate returns like @xmath252 are forbidden .
the elements of the @xmath56-th power of @xmath251 yield the number of non - backtracking walks of length @xmath56 from a ( directed ) edge of the graph to another and the trace of the power matrix the number of closed non - backtracking walks of length @xmath56 starting from any given ( directed ) edge .
a remarkable property of both matrices is that on networks with homogeneous groups ( i. e. , of similar size and internal edge density ) most eigenvalues , which are generally complex , are enclosed by a circle centred at the origin , and that the number of eigenvalues lying outside of the circle is a good proxy of the number of communities of the network @xcite .
for @xmath244 the circle s radius is given by the square root @xmath253 of the leading eigenvalue @xmath137 , which may diverge for networks with heterogeneous degree distributions ( e. g. , power laws ) ; for @xmath245 it equals @xmath254 , which is never greater than @xmath9 .
unfortunately , computing the eigenvalues of the non - backtracking or the flow matrix is lengthy .
both are @xmath246 matrices .
the adjacency matrix @xmath255 has @xmath256 elements , so @xmath244 and @xmath245 are larger by a factor of @xmath257 , where @xmath109 is the average degree of the network
. an approximate but reliable computation of the spectra requires a time which scales superlinearly ( approximately quadratic ) with the network size @xmath2 .
so the problem is intractable for graphs with number of edges of the order of millions or higher .
also , if communities have diverse sizes and edge densities , as it happens in most networks encountered in applications , the bulk of eigenvalues may not have a circular shape , and it may become problematic to identify eigenvalues falling outside of the bulk .
besides , non - backtracking walks must contain cycles , hence trees dangling off the graph do not affect the spectrum of @xmath251 , which remains unchanged if all dangling trees are removed .
this is a disturbing feature , as tree - like regions of the graph may play a role in the network s community structure , and most methods would find different partitions if trees are kept or removed .
the spectrum of the flow matrix , instead , changes when dangling trees are kept or removed @xcite . in the limiting case in which the network itself is a tree , all eigenvalues of @xmath251 and @xmath258 are zero and even if there were a community structure one gets no relevant information .
the number of clusters can also be deduced by studying how the eigenvectors of graph matrices rotate when the adjacency matrix of the graph is subjected to random perturbations @xcite .
on stochastic block models this approach infers the correct value of @xmath75 up to a threshold preceding the detectability limit .
the method is also computationally expensive . in general ,
if one can identify a set ( range ) of promising @xmath75-values , from preliminary information or via calculations like the ones described above or in section [ sec - inference ] , it is better to run constrained versions of clustering methods , searching for solutions only among partitions with those numbers of communities , than letting the methods discover @xmath75 by themselves , which may lead to solutions of lower quality .
many clustering techniques are stochastic in character and do not deliver a unique answer .
a common scenario is when the desired solution corresponds to extrema of a cost function , that can only be found via approximation techniques , with results depending on random seeds and on the choice of initial conditions .
techniques not based on optimisation sometimes have the same feature , when tie - break rules are adopted in order to choose among multiple equivalent options encountered along the calculation .
what to do with all these partitions ? sometimes there are objective criteria to sort out a specific partition and discard all others .
for instance , in algorithms based on optimisation , one could pick the solution yielding the largest ( smallest ) value of the function to optimise . for other techniques
there is no clear - cut criterion .
a promising approach is combining the information of the different outputs into a new partition .
consensus clustering @xcite is based on this idea .
the goal is searching for a _ consensus _
_ partition _ , that is better fitting than the input partitions .
consensus clustering is a difficult combinatorial optimisation problem .
an alternative greedy strategy @xcite relies on the _ consensus matrix _ , which is a matrix based on the co - occurrence of vertices in communities of the input partitions ( fig .
[ figcons ] ) .
the consensus matrix is used as an input for the graph clustering technique adopted , leading to a new set of partitions , which produce a new consensus matrix , etc . , until a unique partition is finally reached , which is not changed by further iterations . . the combination of ( i ) , ( ii ) , ( iii ) and ( iv ) yields the weighted consensus matrix illustrated on the right .
the thickness of each edge is proportional to its weight . in the consensus matrix
the community structure of the original network is more visible : the two communities have become cliques , with `` heavy '' edges , whereas inter - community edges are rather weak .
this improvement is obtained despite the presence of two inaccurate partitions in three clusters ( iii and iv ) . reprinted figure with permission from @xcite .
2012 , by the nature publishing group . ]
the steps of the procedure are listed below .
the starting point is a network @xmath6 with @xmath2 vertices and a clustering algorithm @xmath259 . 1 .
apply @xmath259 on @xmath6 @xmath260 times , yielding @xmath260 partitions .
compute the consensus matrix @xmath261 : @xmath262 is the number of partitions in which vertices @xmath10 and @xmath11 of @xmath6 are assigned to the same community , divided by @xmath260 .
all entries of @xmath261 below a chosen threshold @xmath263 are set to zero and use the full matrix all along @xcite . ] .
4 . apply @xmath259 on @xmath261 @xmath260 times , yielding @xmath260 partitions .
if the partitions are all equal , stop . otherwise go back to 2 .
since the consensus matrix is in general weighted , the algorithm @xmath259 must be able to handle weighted networks , even if the graph at study is binary .
fortunately many popular algorithms have natural extensions to the weighted case .
the integration of consensus clustering with popular existing techniques leads to more accurate partitions than those delivered by the methods alone on lfr benchmark graphs @xcite .
interestingly , this holds even for methods whose direct application gives poor results on the same graphs , like modularity optimisation ( section [ sec - modopt ] ) .
the variability of the partitions , rather than being a problem , becomes a factor of performance enhancement .
the outcome of the procedure depends on the choice of the threshold parameter @xmath263 and the number of input partitions @xmath260 , which can be selected by testing the performance on benchmark networks @xcite .
consensus clustering is also a promising technique to detect communities in evolving networks ( section [ sec - dynclus ] ) .
spectral graph clustering is an approach to detect clusters using spectral properties of the graph @xcite .
the eigenvalue spectrum of several graph matrices ( e. g. , the adjacency matrix , the laplacian , etc . ) typically consists of a dense bulk of closely spaced eigenvalues , plus some outlying eigenvalues separated from the bulk by a significant gap .
the eigenvectors corresponding to these outliers contain information about the large - scale structure of the network , like community structure .
spectral clustering consists in generating a projection of the graph vertices in a metric space , by using the entries of those eigenvectors as coordinates .
the @xmath10-th entries of the eigenvectors are the coordinates of vertex @xmath10 in a @xmath56-dimensional euclidean space , where @xmath56 is the number of eigenvectors used .
the resulting points can be grouped in clusters by using standard partitional clustering techniques like @xmath56-means clustering @xcite .
spectral clustering is not always reliable , however .
when the network is very sparse ( section [ sec - mv ] ) the separation between the eigenvalues of the community - related eigenvectors and the bulk is not sharp .
eigenvectors corresponding to eigenvalues outside of the bulk may be correlated to high - degree vertices ( hubs ) , instead of group structure .
likewise , community - related eigenvectors can be associated to eigenvalues ending up inside the bulk . in these situations , selecting eigenvectors based on whether their associated eigenvalues are inside or outside the bulk yields a heterogeneous set , containing information both on communities and on other features ( e. g. , hubs ) .
using those eigenvectors for the spectral clustering procedure renders community detection more difficult , sometimes impossible .
unfortunately , many of the networks encountered in practical studies are very sparse and can lead to this type of problems . indeed on sparse networks constructed with the planted partition model spectral methods relying on standard matrices [ adjacency matrix , laplacian , modularity matrix @xcite , etc . ]
fail before the theoretical detectability limit ( section [ detectab ] ) @xcite . the non - backtracking matrix @xmath251 of eq . ( [ eqnbtm ] ) was introduced to address this problem @xcite . on the planted partition model
the associated eigenvalues of the community - related eigenvectors of @xmath251 are separated from the bulk until the theoretical detectability limit , so spectral methods using the top eigenvectors of @xmath251 are capable to find communities as long as they are detectable , modulo the caveats we expressed in section [ sec - tools ] .
soft clustering , where communities may overlap , is an even harder problem than hard clustering , where there is no community overlap .
the possibility of having multiple memberships for the vertices introduces additional degrees of freedom in the problem , causing a huge expansion of the space of possible solutions .
it has been pointed out that overlapping communities , especially in social networks , reflect different types of associations between people @xcite .
two actors could be co - workers , friends , relatives , sport mates , etc .. actor @xmath7 could be a work colleague of @xmath62 and a friend of @xmath17 , so she would sit in the overlap between the community of colleagues of @xmath62 and the community of friends of @xmath17 .
for this reason , it has been suggested that an effective way to recover overlapping clusters is to group edges , rather than vertices . in the example above , the edges connecting @xmath7 with @xmath62 and @xmath7 with @xmath17 would be placed in different groups , and since they both have @xmath7 as endpoint , the latter turns out to be an overlapping vertex .
moreover , edge clustering is claimed to have the additional advantage of reconciling soft clustering with hierarchical community structure @xcite .
if there is hierarchy , communities are nested within each other as many times as there are hierarchical levels .
hierarchical structure is often represented via _
dendrograms _ , with the network being divided in clusters , which are in turn divided in clusters , and so on until one ends up with singleton clusters , consisting of one vertex each .
but this can be done only if communities do not share vertices .
overlapping vertices should be assigned to multiple clusters of lower hierarchical levels , yielding multiple copies of them in the dendrogram .
instead , one could build a dendrogram displaying edge communities , where each edge is assigned to a single cluster , but clusters can still overlap because edges in different clusters may share one endpoint @xcite .
some remarks are in order .
first , there may still be overlapping communities even if there were a single type of association between the vertices . for instance , if we keep only the friendship relationships within a given population of actors , there are many social circles and there could be active actors with multiple ties within two circles , or more .
second , in the traditional picture of networks with community structure ( fig .
[ figcom ] ) , the edges connecting two different groups may be assigned to one of the communities they join or they could be put together in a separate group . either way , they would signal an overlap between the communities , which is artificial .
this happens even in the extreme case of a single edge connecting vertices @xmath7 and @xmath62 of two groups , as that edge will have to be assigned to a group , which inevitably forces @xmath7 and @xmath62 into a common cluster .
third , if we rely on the picture emerging from the analysis by yang and leskovec @xcite ( section [ ncp ] ) overlaps between clusters could be much denser than we expect , hence not only vertices but also edges may be shared among different groups , and edge dendrograms would have the same problem as classic vertex dendrograms .
fourth , the computational complexity of the calculation can rise substantially , as in networks of interest there are typically many more edges than vertices . finally , there is nothing revealing that there is a conceptual or algorithmic advantage in grouping edges versus vertices , other than works showing that a specific edge clustering technique outperforms some vertex clustering techniques on a specific set of networks
. to shed some light on the situation , we performed the following test .
we took some network data sets with annotated vertices , giving an indication about what the communities of those networks could be that metadata are not necessarily correlated with topological clusters .
we used data sets for which there is some correlation . ] . for each network
@xmath6 we derived the corresponding _ line graph _ @xmath264 , which is the graph whose vertices are the edges of @xmath6 , while edges are set between pairs of vertices of @xmath264 whose corresponding edges in @xmath6 are adjacent at one of their endpoints .
vertex communities of @xmath264 are then edge communities of the original network @xmath6 .
the question is whether by working on @xmath264 the detection improves or not .
we searched for overlapping communities with oslom @xcite .
we applied oslom on the original graphs and on their line graphs .
the covers found on the line graphs were turned into covers of the vertices of @xmath6 , by replacing each vertex of @xmath264 with the pair of vertices of the corresponding edge of @xmath6 .
the results can be seen in fig .
[ figlg ] , showing how similar the covers found on the original networks and on the line graphs are with respect to the covers of annotated vertices .
neither approach is very accurate , as expected ( see section [ topmet ] and fig .
[ tmhric ] ) , but vertex communities show a greater association to the annotated clusters than edge communities , except in a few instances where the similarity is very low .
analyses carried out on lfr benchmark graphs ( not shown ) lead to the same conclusion .
we stress that traditional line graphs have the problem that edges adjacent to a hub vertex in the original graph turn into vertices who are all connected to each other , forming giant cliques , which might dominate the structure of the line graph , misleading clustering techniques .
the procedure can be refined by introducing weights for the edges of the line graphs , that can be computed in various ways , e. g. , based on the similarity of the neighbourhoods of adjacent edges in the original network @xcite .
still we believe that our tests provide some evidence that edge clustering is no better than vertex clustering , in general .
the superiority of algorithms based on either approach should be assessed a posteriori , case by case , and the answer may depend on the specific data sets under investigation .
statistical inference provides a powerful set of tools to tackle the problem of community detection .
the standard approach is to fit a generative network model on the data @xcite .
the stochastic block model ( sbm ) is by far the most used generative model of graphs with communities ( see section [ sec - mv ] and references therein ) .
we have seen that it can describe other types of group structure , like disassortative and core - periphery structure ( fig .
[ figsbm ] ) .
the unnormalised maximum log - likelihood that a given partition @xmath265 in @xmath75 groups of the network @xmath6 is reproduced by the standard sbm reads @xcite @xmath266 where @xmath88 is the number of edges running from group @xmath83 to group @xmath84 , @xmath267 ( @xmath268 ) the number of vertices in @xmath83 ( @xmath84 ) and the sum runs over all pairs of groups ( including when @xmath269 ) .
this version of the model , however , does not account for the degree heterogeneity of most real networks , so it does a poor job at describing the group structure of many of them .
therefore , karrer and newman proposed the _ degree - corrected stochastic block model _ ( dcsbm ) @xcite , in which the degrees of the vertices are kept constant , on average , via the introduction of additional suitable parameters .
the unnormalised maximum log - likelihood for the dcsbm is @xmath270 where @xmath271 ( @xmath272 ) is the sum of the degrees of the vertices in @xmath83 ( @xmath84 ) .
the most important drawback of this type of approach is the need to specify the number @xmath75 of groups beforehand , which is usually unknown for real networks .
this is because a straight maximisation of the likelihoods of eqs .
( [ eqllsbm ] ) and ( [ eqlldc ] ) over the whole set of possible solutions yields the trivial partition in which each vertex is a cluster ( _ overfitting _ ) . in section [ sec - tools ]
we have seen ways to extract @xmath75 from spectral properties of the graph .
but it would be better to have statistically principled methods , to be consistent with the approach used to perform the inference .
a possibility is _ model selection _ , for instance by choosing the model that best compresses the data @xcite .
the extent of the compression can be estimated via the total amount of information necessary to describe the data , which includes not only the fitted model , but also the information necessary to describe the model itself , which is a growing function of the number of blocks @xmath75 @xcite .
this quantity , that we indicate with @xmath273 , is called the _
description length_. minimising the description length naturally avoids overfitting . partitions with large @xmath75 are associated to heavy " models in terms of their information content , and do not represent the best compression . on the other hand , partitions with low @xmath75 have high information content , even if the model itself is not loaded with parameters .
hence the minimum description length corresponds to a non - trivial number of groups and it makes sense to minimise @xmath273 to infer the block structure of the graph .
it turns out that this approach has a limited resolution on the standard sbm : the maximum number of blocks that can be resolved scales as @xmath274 for a fixed average degree @xmath109 , where @xmath2 is the number of vertices of the network .
this means that the minimum size of detectable blocks scales as @xmath274 , just as it happens for modularity maximisation ( section [ sec - modopt ] ) . a more refined method of model selection , consisting in a nested hierarchy of stochastic block models , where an upper level of the hierarchy serves as prior information to a lower level , brings the resolution limit down to @xmath202 , enabling the detection of much smaller blocks @xcite .
other techniques to extract the number of groups have been proposed @xcite .
optimisation techniques have received the greatest attention in the literature .
the goal is finding an extremum , usually the maximum , of a function indicating the quality of a clustering , over the space of all possible clusterings .
quality functions can express the goodness of a partition or of single clusters .
the most popular quality function is the _ modularity _ by newman and girvan @xcite .
it estimates the quality of a partition of the network in communities .
the general expression of modularity is @xmath275 where @xmath3 is the number of edges of the network , the sum runs over all pairs of vertices @xmath10 and @xmath11 , @xmath8 is the element of the adjacency matrix , @xmath276 is the _
null model term _ and in the kronecker delta at the end @xmath277 and @xmath278 indicate the communities of @xmath10 and @xmath11 .
the term @xmath276 indicates the average adjacency matrix of an ensemble of networks , derived by randomising the original graph , such to preserve some of its features .
therefore , modularity measures how different the original graph is from such randomisations .
the concept was inspired by the idea that by randomising the network structure communities are destroyed , so the comparison between the actual structure and its randomisation reveals how non - random the group structure is .
a standard choice is @xmath279 , @xmath18 and @xmath121 being the degrees of @xmath10 and @xmath11 , and corresponds to the expected number of edges joining vertices @xmath10 and @xmath11 if the edges of the network were rewired such to preserve the degree of all vertices , on average .
this yields the classic form of modularity @xmath280 other choices of the null model term allow us to incorporate specific features of network structure , like bipartiteness @xcite , correlations @xcite , signed edges @xcite , space embeddedness @xcite , etc .. the extension of eq .
( [ eq : mod ] ) and of its variants to the case of weighted networks is straightforward @xcite . for simplicity
we focus on unweighted graphs here , but the issues we discuss are general . because of the delta ,
the only contributions to the sum come from vertex pairs belonging to the same cluster , so we can group these contributions together and rewrite the sum over the vertex pairs as a sum over the clusters @xmath281 .
\label{eq : mod1}\ ] ] here @xmath282 the total number of edges joining vertices of community @xmath17 and @xmath283 the sum of the degrees of the vertices of @xmath17 ( section [ sec - var ] ) . the first term of each summand in eq .
( [ eq : mod1 ] ) is the fraction of edges of the graph falling within community @xmath17 , whereas the second term is the expected fraction of edges that would fall inside @xmath17 if the graph were taken from the ensemble of random graphs preserving the degree of each vertex of the original network , on average .
the difference in the summand would then indicate how non - random " subgraph @xmath17 is .
the larger the difference the more confident we can be that the placement of edges within @xmath17 is not random ( fig .
[ mod - ill ] ) .
large values of @xmath284 are then supposed to indicate partitions with high quality .
modularity maximisation is np - hard @xcite .
therefore one can realistically hope to find only decent approximations of the modularity maximum and a wide variety of approaches has been proposed .
due to its simplicity , the prestige of its inventors and early results on the benchmark of girvan and newman ( section [ art - bench ] ) and on small real benchmark networks , like zachary karate club network ( fig .
[ zach ] ) , modularity has become the best known and most studied object in network clustering .
in fact , soon after its introduction , it seemed to represent the essence of the problem , and the key to its solution .
however , it became quickly clear that the measure is not as good as it looks .
for one thing , there are high - modularity partitions even in random graphs without groups @xcite .
this seems counterintuitive , given that modularity has been designed to capture the difference between random and non - random structure .
modularity is a sort of distance between the actual network and _ an average _ over random networks , ignoring altogether the distribution of the relevant community variables , like the fractions of edges within the clusters , over all realisations generated by the configuration model .
if the distribution is not strongly peaked , the values of the community variables measured on the original graph may be found in a large number of randomised networks , even though the averages look far away from them . in other words
, we should pay more attention to the _ significance _ of the maximum modularity value @xmath285 , than to the value itself .
how can we estimate the significance of @xmath285 ?
a natural way is maximising @xmath284 over all partitions of every randomised graph .
one then computes the average @xmath286 and the standard deviation @xmath287 of the resulting values .
the statistical significance of @xmath285 is indicated by the distance of @xmath285 from the null model average @xmath286 in units of the standard deviation @xmath287 , i. e. , by the @xmath173-score @xmath288 if @xmath289 , @xmath285 indicates strong community structure .
this approach has problems , though .
the main drawback is that the distribution of @xmath290 over the ensemble of null model random graphs , though peaked , is not gaussian .
therefore , one can not attribute to the values of the @xmath173-score the significance corresponding to a gaussian distribution , and one ought to compute the statistical significance for the correct distribution . also , the @xmath173-score depends on the network size , so the same values may indicate different levels of significance for networks differing considerably in size .
next , it is not true that the modularity maximum always corresponds to the most pronounced community structure of a network . in fig .
[ reslim ] we show the well - known example of the ring of cliques @xcite .
the network consists of @xmath110 cliques with four vertices each .
every clique has two neighbouring cliques , connected to it via a single edge .
intuition suggests that the graph has a natural community structure , with @xmath110 communities , each corresponding to one clique .
indeed , the @xmath284-value of this partition is @xmath291 , pretty close to @xmath9 , which is the upper bound of modularity .
however , there are partitions with larger values , like the partition in @xmath292 clusters indicated by the dashed contours , whose modularity is @xmath293 .
this is due to the fact that @xmath284 has a preferential scale for the communities , deriving from the underlying null model and revealed by its explicit dependence on the number of edges @xmath3 of the network [ eq .
( [ eq : mod1 ] ) ] . according to the configuration model ,
the expected number @xmath294 of edges running between two subgraphs @xmath7 and @xmath62 with total degree @xmath295 and @xmath296 , respectively , is approximately @xmath297 .
consequently , if @xmath295 and @xmath296 are of the order of @xmath298 or smaller , @xmath294 could become smaller than @xmath9 .
this means that in many randomisations of the original graph @xmath6 , subgraphs @xmath7 and @xmath62 are disconnected and even a single edge joining them in @xmath6 signals a non - random association . in these cases , modularity is larger when @xmath7 and @xmath62 are put together than when they are treated as distinct communities , as in the example of fig .
[ reslim ] .
the modularity scale depends only on the number of edges @xmath3 , and it may have nothing to do with the size of the actual communities of the network .
the resolution limit questions the usefulness of modularity in practical applications @xcite .
many attempts have been made to mitigate the consequences of this disturbing feature .
one approach consists in introducing a resolution parameter @xmath299 into modularity s formula @xcite . by tuning @xmath299
it is possible to arbitrarily vary the resolution scale of the method , going from very large to very small communities
. we shall discuss such multi - resolution approaches in section [ sec - dynmet ] .
here we emphasize that multi - resolution versions of modularity do not provide a reliable solution to the problem .
this is because modularity maximisation has an additional bias : large subgraphs are usually split in smaller pieces @xcite .
this problem has the same source as the resolution limit , namely the choice of the null model . since modularity has a preferential scale for the communities , when a subgraph is too large it is convenient to break it down , to increase the modularity of the partition .
so , when there is no characteristic scale for the communities , like when there is a broad cluster size distribution , large communities are likely to be broken , and small communities are likely to be merged .
since multi - resolution versions of modularity can only shift the resolution scale of the measure back and forth , they are unable to correct both effects at the same time @xcite . in addition
, tuning the resolution parameter in the search for good partitions is usually computationally very demanding , as in many cases the optimisation procedure has to be repeated over and over for all @xmath299-values one desires to investigate .
we stress that the resolution limit is a feature of modularity itself , not of the specific way adopted to maximise it .
therefore , there is no magic heuristic that can circumvent this issue .
the louvain method @xcite has been held as one such magic heuristic .
the method performs a greedy optimisation of @xmath284 in a hierarchical manner , by assigning each vertex to the community of their neighbours yielding the largest @xmath284 , and creating a smaller weighted super - network whose vertices are the clusters found previously .
partitions found on this super - network hence consist of clusters including the ones found earlier , and represent a higher hierarchical level of clustering .
the procedure is repeated until one reaches the level with largest modularity . in the comparative analysis of clustering algorithms performed by lancichinetti and fortunato on the lfr benchmark @xcite ,
the louvain algorithm was the second best - performing method , after infomap @xcite .
this has given the impression that the peculiar strategy of the method solves the resolution problems above , which is not true .
the reason why the performance is so good is that lancichinetti and fortunato adopted the lowest partition of the hierarchy , the one with the smallest clusters @xcite . by using the partition with highest modularity performance
degrades considerably ( fig .
[ louvainl ] ) , as expected . as suggested by the developers of the algorithm themselves , using the lowest level helps avoiding unnatural community mergers ; as an example , they showed that the natural partition of the ring of cliques ( fig .
[ reslim ] ) can be recovered this way . and @xmath300 vertices ) and of the range of community sizes ( label @xmath301 indicates that communities have between @xmath302 and @xmath215 vertices , label @xmath62 that they have between @xmath303 and @xmath304 vertices ) .
accuracy is calculated via the normalised mutual information ( nmi ) , in the version by lancichinetti , fortunato and kertsz @xcite .
results are heavily depending on the hierarchical level one chooses at the end of the procedure .
when one picks the top level ( diamonds ) , which is the one with largest modularity , the accuracy is poor , as expected , especially when communities are smaller .
when one goes for the bottom level ( squares ) , which has lower modularity and smaller clusters than the top level partition , there is a far better agreement with the planted partition and the performance gets closer to that of infomap ( circles ) .
the squares follow the performance curves used in the comparative analysis by lancichinetti and fortunato @xcite .
courtesy from andrea lancichinetti . ] however , the bottom level has lower modularity than the top level , so we face a sort of contradiction , in that users are encouraged to use suboptimal partitions , even though one assumes that the best clustering corresponds to the highest value of the quality function , which is what the method is supposed to find .
there is no guarantee that the bottom level yields the most meaningful solution .
on the other hand , users have the option of choosing among a few partitions and a slightly higher chance to find what they search for
. moreover , the modularity landscape is characterised by a larger than exponential of graph vertices . ]
number of distinct partitions , whose modularity values are very close to the global maximum @xcite . this explains why many heuristic methods of modularity maximisation are able to come very close to the global maximum of @xmath284 , but it also implies that the global maximum is basically impossible to find .
in addition , high-@xmath284 partitions are not necessarily similar to each other , despite the proximity of their modularity scores .
the optimal structural partition , which may not correspond to the modularity maximum due to problems exposed above , may however have a large @xmath284-value .
therefore the optimal partition is basically indistinguishable from a huge number of high - modularity partitions , which are in general structurally dissimilar from it .
the large structural diversity of high - modularity partitions implies that one can not rely on any of them , at least in principle .
reliable solutions could be singled out when the domain user imposes some constraints on the clustering of the system , or when she expects it to have specific features . in the absence of additional information or expectations ,
consensus clustering could be used to derive more robust partitions .
indeed , it has been shown that the consensus of many high - modularity partitions , combined with a hierarchical approach , could help to solve resolution problems and to avoid to find communities in random graphs without groups @xcite . as of today
, modularity optimisation is still the most used clustering technique in applications .
this may appear odd , given the serious issues of the method and the fact that nowadays more powerful techniques are available , like a posteriori stochastic block modelling ( section [ sec - wmt ] ) .
indeed newman has proven that optimising modularity is equivalent to maximising the likelihood that the planted partition model reproduces the network @xcite .
but the planted partition model is a very specific case of the general stochastic block model , in that the intra - group edge probabilities are all equal to the same value @xmath65 and the inter - group edge probabilities are all equal to the same value @xmath112 .
there is no reason to limit the inference to this specific case , when one could use the full model .
optimising partition quality functions may lead to resolution problems , just like it happens for modularity , i. e. , the partition found when the detection is performed only on @xmath301 @xcite .
this is possible when the coefficients of the summand corresponding to each community does not depend on global properties of the graph .
even those functions have their own preferential community scale , though . ] .
instead , one could try to optimise _ cluster quality functions_. one starts with some function @xmath305 expressing how community - like " a subgraph is and with a seed vertex @xmath84 .
the goal is to build a cluster @xmath306 including @xmath84 such that @xmath307 is maximum .
this is usually done by exploring the neighbours of the temporary subgraph @xmath306 , starting from the neighbours of @xmath84 when @xmath306 includes only @xmath84 .
the neighbouring vertex whose inclusion yields the largest increase of @xmath75 is added to the subgraph .
when a new vertex is included , the structure of the subgraph is altered and the other vertices can be examined again , as it might be advantageous to knock some of them out .
the process stops when the quality @xmath307 can not be increased anymore via the inclusion or the exclusion of vertices .
the optimisation of cluster quality functions offers a number of advantages over the optimisation of partition quality functions .
first , it is consistent with the idea that communities are local structures , which are sensitive to what happens in their neighbourhood , but are fairly unaffected by the rest of the network : the structure of a social circle in europe is hardly influenced by the dynamics of social circles in australia , though they are parts of the same global social network of humans .
consequently , if a network undergoes structural changes in one region , community structure is altered and is to be recovered only in that region , while the clustering of the rest of the network remains the same . by optimising partition quality functions , instead
, any little change may have an effect on every community of the graph .
second , since cluster quality functions do not embody any global scale , severe resolution problems are usually avoided .
moreover , one can investigate only parts of the network , which is particularly valuable when the graph is large and a global analysis would be out of reach , computationally .
the local exploration of the graph allows to reach vertices already assigned to clusters , so overlaps can be naturally detected . in the last years several algorithms based on the optimisation of cluster quality functions
have been designed @xcite .
communities can also be identified by running dynamical processes on the network , like diffusion @xcite , spin dynamics @xcite , synchronisation @xcite , etc .. in this section we focus on diffusion and spin dynamics , that inform most approaches .
random walk dynamics is by far the most exploited in community detection . if communities have high internal edge density and are well - separated from each other , random walkers would be trapped in each cluster for quite some time , before finding a way out and migrating to another cluster .
we briefly discuss two broad classes of algorithms : methods based on vertex similarity and methods based on the map equation .
the first class of techniques consists in using random walk dynamics to estimate the similarity between pairs of vertices .
for instance , in the popular method _ walktrap _ the similarity between vertices @xmath10 and @xmath11 is given by the probability that a random walker moves from @xmath10 to @xmath11 in a fixed number of steps @xmath97 @xcite .
the parameter @xmath97 has to be large enough , to allow for the exploration of a significant portion of the graph , but not too big , as otherwise one would approach the stationary limit in which transition probabilities trivially depend on the degrees of the vertices .
if there is a pronounced community structure , pairs of vertices in the same cluster are much more easily reachable by a random walk than pairs of vertices in different clusters , so the vertex similarity is expected to be considerably higher within groups than between groups of the graph , whose element @xmath308 equals the probability that a random walker , sitting at @xmath11 , moves to @xmath10 ; raising the elements of the resulting matrix to a power , such that the larger values are enhanced with respect to the smaller ones , many of which are set to zero to lighten the calculations , while the remaining ones are normalised by dividing them by the sum of elements of their column , yielding a new transfer matrix .
the process eventually reaches a stationary state , corresponding to the matrix of a disconnected graph , whose connected components are the sought clusters . ] . in that case , clusters can be readily identified via standard hierarchical or partitional clustering techniques @xcite .
this class of methods have a high computational complexity , higher than quadratic in the number @xmath2 of vertices ( on sparse graphs ) , so they can not be used on large networks . besides , they are often parameter - dependent .
the map equation stems from a seminal paper by rosvall and bergstrom @xcite , who asked what is the most parsimonious way to describe an infinitely long random walk taking place on the graph .
the information content of any description is given by the total number of bits required to indicate the various stages of the process .
the simplest description is obtained by listing sequentially all vertices reached by the random walker , each vertex being described by a unique codeword . however ,
if the network has a community structure , there may be a more compact description , which follows the principle of geographic maps , where there are multiple cities and streets with the same name across regions .
vertex codewords could be recycled among different communities , which play the role of regions / states , and vertices with identical name are distinguished by specifying the community they belong to .
if clusters are well separated from each other , transitions between clusters are infrequent , so it is advantageous to use the map , with the communities as regions , because in the description of the random walk the codewords of the clusters will not be repeated many times , while there is a considerable saving in the description due to the limited length of the codewords used to denote the vertices ( fig .
[ figinfomap ] ) .
the map equation yields the description length of an infinite random walk consists of two terms , expressing the shannon entropy of the walk within and between clusters . the best partition is the one yielding the minimum description length .
this method , called infomap , can be applied to weighted networks , both undirected and directed . in the latter case ,
random walk dynamics is modified by introducing a teleportation probability , as in the pagerank process @xcite , to ensure that a non - trivial stationary state is reached .
it has been successively extended to the detection of hierarchical community structure @xcite and of overlapping clusters @xcite . in classic random
walks the probability of reaching a vertex only depends on where the walker stands , not on where it is coming from .
the map equation has also been extended to random walks whose transition probabilities depend on earlier steps too ( higher - order markov dynamics ) @xcite , retaining memory of the ( recent ) past .
applications show that in this way one can recover overlapping communities more easily than by using standard first - order random walk dynamics , especially pervasive overlaps , which are usually out of reach for most clustering algorithms ( section [ ncp ] ) .
infomap and its variants usually return different partitions than structure - based methods ( e. g. , modularity optimisation ) .
this is because they are based on flows running across the system , as opposed to structural variables like number of edges , vertex degrees , etc .. the difference is particularly striking on directed graphs @xcite , where edge directions heavily constrain the possible flows .
structural features obviously play a major role on the dynamics of processes running on graphs , but dynamics can not be generally reduced to an interplay of structural elements , at least not simple ones like , e. g. , vertex degrees . sometimes structural and dynamic approaches are equivalent , though . for instance , newman - girvan s modularity is a special case of a general quality function , called _ stability _ , expressing the persistence of a random walk within communities @xcite .
the methods we have discussed so far are global , in that they aim at finding the whole community structure of the system .
however , random walks along with other dynamical processes can be used as well to explore the network locally , starting from seed vertices @xcite .
good communities correspond to bottlenecks of the dynamics and depend on the choice of the seed vertices , the time scale of the dynamics , etc .. such local perspective enables to identify community overlaps in a natural way , due to the possibility of reaching vertices multiple times , even if they are already classified .
spin dynamics @xcite are also regularly used in network clustering .
the first step is to define a spin model on the network , consisting of a set of spin variables @xmath309 , assigned to the vertices and a hamiltonian @xmath310 , expressing the energy of the spin configuration @xmath311 . for community detection ,
spins are usually integers : @xmath312 .
contributions to the energy are usually given by spin - spin interactions .
the coupling of a spin - spin interaction can be _
ferromagnetic _ ( negative ) or
_ antiferromagnetic _ ( positive ) , if the energy is lower when the spins are equal or not , respectively .
the goal is to find those spin configurations that minimise the hamiltonian @xmath310 .
if couplings are all ferromagnetic , the minimum energy would be trivially obtained for the configurations where all vertices have identical spin values .
instead , one would like to have identical spins for vertices of the same cluster , and different spins for vertices in different clusters , to identify the community structure .
therefore , hamiltonians feature both ferromagnetic and antiferromagnetic interactions [ _ spin glass dynamics _ @xcite ] .
a popular model consists in rewarding edges between vertices in the same cluster , as well as non - edges between vertices in different clusters , and penalising edges between vertices of different clusters , along with non - edges between vertices in the same cluster . this way , if the edge density within communities is appreciably larger than the edge density between communities , as it often happens , having equal spin values for vertices in the same cluster would considerably lower the energy of the configuration . on the other hand , to bring the energy further down the spins of vertices in different clusters should be different , as many such vertices would be disjoint from each other , and such non - edges would increase the energy of the system if the corresponding spin variables were equal .
a general expression for the hamiltonian along these lines is @xcite @xmath313\delta(s_i , s_j)\ , , \label{eqrbgen}\ ] ] where @xmath8 is the element of the adjacency matrix , @xmath314 are arbitrary coefficients , and the kronecker delta selects only the pairs of vertices with the same spin value . a popular model is obtained by setting @xmath315 and @xmath316 , where @xmath299 is a tunable parameter and @xmath276 a null model term , expressing the expected number of edges running between vertices @xmath10 and @xmath11 under a suitable randomisation of the graph structure .
the resulting hamiltonian is @xcite @xmath317 if @xmath318 and @xmath279 , @xmath18 ( @xmath121 ) being the degree of @xmath10 ( @xmath11 ) and @xmath3 the total number of graph edges , the hamiltonian of eq .
( [ eqrb ] ) coincides with the modularity by newman and girvan [ eq . ( [ eq : mod ] ) ] , up to an irrelevant multiplicative constant .
consequently , modularity can be interpreted as the hamiltonian of a spin glass as well . by setting @xmath319 and @xmath320
we obtain the _ absolute potts model _ ( apm ) @xcite , whose hamiltonian reads @xmath321\delta(s_i , s_j)\ , .
\label{eqapm}\ ] ] here , there is no null model term .
the models of eqs .
( [ eqrb ] ) and ( [ eqapm ] ) can be trivially extended to weighted graphs @xcite .
they allow to explore the network at different resolutions , by suitably tuning the parameter @xmath299 .
however , there usually is no information about the community sizes , so it is not possible to decide _ a priori _ the proper value(s ) of @xmath299 for a specific graph .
a common heuristic is to estimate the _ stability _ of partitions as a function of @xmath299 .
it is plausible that partitions recovered for a given @xmath299-value will not be disrupted if @xmath299 is varied a little .
so , the whole range of @xmath299 can be split into intervals , each interval corresponding to the most frequent partition detected in it .
good candidates for the unknown community structure of the system could be the partitions found in the widest intervals of @xmath299 , as they are likely to be more stable ( or robust ) than the other partitions .
however , the results of the algorithm do not usually have a linear relationship with @xmath299 , hence the width of the intervals is not necessarily correlated with stability , as intervals of the same width but centred at different values of @xmath299 may have rather different importance . a good operational definition of stability is based on the stochastic character of optimisation methods , which typically deliver different results for the same system and set of parameters , by changing initial conditions and/or random seeds . if a partition is robust in a given range of @xmath299-values , most partitions delivered by the algorithm will be very similar . on the other hand ,
if one explores a @xmath299-region in between two strong partitions , the algorithm will deliver the one or the other partition and the individual replicas will be , on average , not so similar to each other .
so , by calculating the similarity @xmath322 of partitions found by the method at a given resolution parameter @xmath299 ( for different choices of initial conditions and random seeds ) , stable communities are revealed by peaks of @xmath322 @xcite . since clustering in large graphs can be very noisy , peaks may not be well resolved .
noise can be reduced by working with consensus partitions of the individual partitions returned by the method for a given @xmath299 ( section [ sec - consensus ] ) .
these manipulations are computationally costly , though .
besides , multi - resolution techniques may miss relevant cluster sizes , as it happens for multi - resolution modularity @xcite ( section [ sec - modopt ] ) .
due to the increasing availability of time - stamped network data , there is currently a lot of activity on the development of methods to analyse temporal networks @xcite . in particular , the problem of detecting dynamic communities has received a lot of attention @xcite .
clustering algorithms used for static graphs can be ( and often are ) used for dynamic networks as well .
what needs to be established is how to handle the evolution .
typically one can describe it as a succession of _ snapshots _ @xmath323 , where each snapshot @xmath324 corresponds to the configuration of the graph in a given time window .
there are two possible strategies .
the simplest approach is to detect the community structure for each individual snapshot , which is a static graph @xcite .
next , pairs of communities of consecutive windows are associated .
a standard procedure is finding the cluster @xmath325 in window @xmath97 that is most similar to cluster @xmath326 in window @xmath99 , for instance by using jaccard similarity score [ eq . ( [ eqt20 ] ) ] @xcite .
this way every community has an image in each phase of the network evolution and one can track its dynamics .
various scenarios are possible .
communities may disappear at some point and new communities may appear , following the exclusion or the introduction of vertices and edges , respectively .
furthermore , a cluster may fragment into smaller ones or merge with others .
however , since snapshots are handled separately , this strategy often produces significant variations between partitions close in time , especially when the data sets are noisy , as it usually happens in applications
. it would be preferable to have a unified framework , in which communities are inferred both from the current structure of the graph and from the knowledge of the community structure at previous times .
an interesting implementation of this strategy is _ evolutionary clustering
_ @xcite .
the goal of the approach is finding a partition that is both faithful to the system configuration at snapshot @xmath97 and close to the partition derived for the previous snapshot @xmath98 .
a cost function is introduced , whose optimisation yields a tradeoff between such two constraints .
there is ample flexibility on how this can be done , in practice .
many known clustering techniques normally used for static graphs can be reformulated within this evolutionary framework . some interesting algorithms based on evolutionary clustering
have been proposed @xcite .
mucha et al .
have also presented a method that couples the system s configurations of different snapshots , within a modularity - based framework @xcite . in the resulting quality function ( _ multislice modularity _ ) , all configurations are simultaneously taken into account and the coupling between them is expressed by a tunable parameter .
the approach can handle general multilayer networks @xcite , where layers are either networks whose vertices are connected by a specific edge type ( e. g. , friendship , work relationships , etc .
, in social networks ) , or networks whose vertices have connections ( interactions , dependencies ) with the vertices of other networks / layers . on the other hand , since the approach is based on modularity optimisation , it has the drawbacks exposed in section [ sec - modopt ] .
consensus clustering ( section [ sec - consensus ] ) is a natural approach to find stable dynamic clusterings by combining multiple snapshots .
let us suppose we have a time range going from @xmath327 to @xmath328 , that we want to divide into @xmath329 windows of size @xmath330 . for the sake of stability , one should consider sliding windows , i. e. , overlapping time intervals .
this way consecutive partitions will be based on system configurations sharing a lot of vertices and edges , and change is ( typically ) smooth . in order to have exactly @xmath329 frames ,
each of them has to be shifted by an interval @xmath331 with respect to the previous one .
so we obtain the windows @xmath332 $ ] , @xmath333 $ ] , @xmath334 $ ] , ... , @xmath335 $ ] .
the community structure of each snapshot can be found via any reliable static clustering technique .
next , the consensus partition from the clusterings of @xmath83 consecutive snapshots , with @xmath83 suitably chosen , is derived @xcite .
again , one could consider sliding windows : for instance , the first window could consist of the first @xmath83 snapshots , the second one by those from @xmath216 to @xmath336 , and so on until the interval spanned by the last @xmath83 snapshots . in fig .
[ dyncons ] we show an application of this procedure on the citation network of papers published in journals of the american physical society ( aps ) .
years , except at the right end of each diagram : since there is no data after @xmath337 , the last windows must have @xmath337 as upper limit , so their size shrinks ( @xmath338 , @xmath339 , @xmath340 , @xmath341 ) .
each vertical bar represents a consensus partition combining pairs of consecutive snapshots .
a color uniquely identifies a community , the width of the links between clusters is proportional to the number of papers they have in common .
the rapid growth of the field _
complex networks _ is clearly visible , as well as its later split into a number of smaller subtopics , like _ community structure _ , _ epidemic spreading _ , _ robustness _ , etc .. reprinted figure with permission from @xcite .
2012 , by the nature publishing group . ]
an alternative way to uncover the evolution of communities by accounting for the correlation between configurations of neighbouring time intervals is to use probabilistic models @xcite .
if the system is large and its structure is updated in a stream fashion , instead of working on snapshots one could detect the clustering _ online _
, every time the configuration of the system varies due to new information , like the addition of a new vertex or edge @xcite .
an advantage of this approach is that change is due to the effect that the small variation in the network structure has on the system , and it can be tracked by simply adjusting the partition of the previous configuration , which can be usually done rather quickly .
let us suppose that we have identified the communities , somehow .
are we done ?
unfortunately , things are not that simple . in fig .
[ figrand0 ] ( top left ) we show the adjacency matrix of the random graph la erds - rnyi illustrated in fig . [
fig : stylized - random ] .
the graph has @xmath342 vertices , so the matrix is @xmath343 .
black dots indicate the existence of an edge between the corresponding vertices , while missing edges are represented in white . by construction
, there is no group structure .
however , we can rearrange the elements of the matrix , by reordering the vertex labels . in fig .
[ figrand0 ] ( top right ) we see that , by doing that , one can generate a group structure , of the assortative type , with two blocks of equal size .
if we increase the number of blocks to three fig .
[ figrand0 ] ( bottom left ) and ten fig .
[ figrand0 ] ( bottom right ) we can make the matrix look more and more modular .
this is why many clustering techniques detect communities in random networks as well , though they should not .
where do the groups come from ?
since they can not be real by construction , they must be generated by random fluctuations in the network construction process .
random fluctuations are particularly relevant on sparse graphs ( section [ detectab ] ) .
the lesson we learn from this example is that it is not sufficient to identify groups in the network , but one should also ask _ how significant _ , or non - random , they are .
unfortunately , most clustering algorithms are not able to assess the significance of their results .
if the groups are compatible with random fluctuations , they are not proper groups and should be disregarded . the lower the chance that they are generated by randomness , the more confident we can be that the blocks reflect some actual group structure .
naturally , this can be done only if one has a reliable _ null model _ , describing how the structure of the network at study can be randomised and allowing us to estimate how likely it is that the candidate group structure is generated this way .
the configuration model @xcite is a popular null model in the literature .
it generates all possible configurations preserving the number of vertices and edges of the network at study , and the degrees of its vertices .
one may compute some variables of the original network , and estimate the probability that the model reproduces them , or _
p - value _ , i. e. , the fraction of model configurations yielding values of the variables compatible with those measured on the original graph .
if the p - value is sufficiently low ( @xmath344 is a standard threshold ) , one concludes that the property at study can not be generated by randomness only . for community structure , one can compute various properties of the clusters , e. g. , their internal density , and compare them with the model values .
some clustering algorithms , like oslom @xcite are based on this principle . along the same lines
, @xmath173-scores can be used as well [ see the example of eq .
( [ eq : zscore ] ) ] .
degree - corrected stochastic block models @xcite ( section [ sec - inference ] ) also include the configuration model , which corresponds to the case without group structure ) . ] . in this case
significance can be estimated by doing model selection between the versions with and without blocks ( section [ sec - inference ] ) .
a concept very related to significance is that of _
robustness_. if clusters are significant it means that they are resilient if the network structure is perturbed , to some extent .
one way to quantitatively assess this is introducing into the system a controlled amount of noise , and checking how much noise it takes to disrupt the group structure
. the greater the required perturbation , the more robust the communities . for instance , a perturbation could be rewiring a fraction of randomly chosen edges @xcite . after the network is perturbed , the community structure is derived and compared to the one of the original network . the trend of the partition similarity shows how the group structure responds to perturbations .
a similar approach consists in sampling network configurations from a population which the original network is supposed to belong to ( _ bootstrapping _ ) , and comparing the clusterings found in those configurations , to check how frequently subsets of vertices are clustered together in different samples , which is an index of the robustness ( significance ) of their clusters @xcite . at the end of the day , what most people want to know is : which method shall i use on my data ?
since the clustering problem is ill - defined , there is no clear - cut answer to it .
popular techniques are based on similar ideas of communities , like the ones we reviewed in sections [ sec - defs ] and [ sec - mv ] .
what makes the difference is the way clusters are sought .
the specific procedure affects the reliability of the results ( e. g. , because of resolution problems ) and the time complexity of the calculation , determining the scope of the method and constraining its applicability .
most methods propose a universal recipe , that is supposed to hold on every data set . in so doing ,
one neglects the peculiarities of the network at study , which is valuable information that could orient the method towards more reliable solutions . but algorithms are usually not so flexible to account for specific network features .
for instance , in some cases , there is no straightforward extension capable to handle high - level features like edge direction or overlapping communities .
validation of algorithms , like the comparative analysis of @xcite , have allowed to identify a set of methods that perform well on artificial benchmarks .
there are two important issues , though .
first , we do not know how well real networks are described by currently used benchmark models .
therefore , there is no guarantee that methods performing well on benchmarks also give reliable results on real data sets .
structural analyses like the ones discussed in section [ ncp ] might allow to identify more promising benchmark models .
second , if we rely so much on current benchmarks , which are versions of the stochastic block model ( sbm ) , we already know what the best method is : a posteriori block modelling , i. e. , fitting a sbm on the data .
indeed , there are several advantages to this approach .
it is more general , it does not only discover communities but several types of group structures , like disassortative groups ( fig .
[ fig : stylized - bipartite ] ) and core - periphery structure ( fig . [ fig : stylized - coreper ] ) .
it can also capture the existence of hierarchies among the clusters .
moreover , it yields much richer results than standard clustering algorithms , as it delivers the entire set of parameters of the most likely sbm , with which one can construct the whole network , instead of just grouping vertices .
sbms are very versatile as well .
they can be extended to a variety of contexts , e. g. , directed networks @xcite , networks with weighted edges @xcite , with overlapping communities @xcite , with multiple layers @xcite , with annotations @xcite . besides
, the procedure can be applied to any network model with group structure , not necessarily sbms .
the choice between alternative models can be done via model selection .
a posteriori block modelling is not among the fastest techniques available .
networks with millions of vertices and edges could be investigated this way , but very large networks remain out of reach . fortunately , many networks of interest can be attacked .
the biggest problem of this class of methods , i. e. , the determination of the number of clusters , seems to be solvable ( section [ sec - inference ] ) .
we recommend to exploit the power of this approach in applications .
algorithms based on the optimisation of cluster quality functions should be considered as well ( section [ sec - modopt ] ) , because they may avoid resolution problems and explore the network locally , which is often the only option when the system is too large to be studied as a whole .
algorithms based on the optimisation of partition quality functions , like modularity maximisation , are plagued by the problems we discussed in section [ sec - modopt ] .
nevertheless , if one knows , or discovers , the correct number of clusters @xmath75 , and the optimisation is constrained on the subset of partitions having @xmath75 clusters , such algorithms become competitive @xcite .
we also encourage to use approaches based on dynamics ( section [ sec - dynmet ] ) . in principle , the resulting clustering depends on the specific dynamics adopted . in practice
, there often is a substantial overlap between the clusters found with different dynamics .
an important question is whether dynamics may uncover groups that are not recoverable from network structure alone .
differences in the clusterings found via dynamical versus structural approaches could be due to the fact that dynamical processes are sensitive to more complex structural elements than edges ( e. g. , paths , motifs ) @xcite ( section [ sec - mv ] ) .
however , even if that were true , dynamical approaches could be more natural ways to handle such higher - order structures , and to make sense of the resulting community structure . in general
, however , the final word on the reliability of a clustering algorithm is to be given by the user , and any output is to be taken with care .
intuition and domain knowledge are indispensable elements to support or disregard solutions .
in this section we provide a number of links where one can find the code of clustering algorithms and related techniques and models .
* _ artificial benchmarks_. code to generate lfr benchmark graphs ( section [ art - bench ] ) can be found here .
the code for the dynamic benchmark by granell et al .
@xcite is available at . *
_ partition similarity measures_. many partition similarity measures have their own function in r , python and matlab and are easy to find . the variant of the nmi for covers proposed by lancichinetti et al .
@xcite can be found at , the one by esquivel and rosvall @xcite at . *
_ consensus clustering_. the technique proposed by lancichinetti and fortunato @xcite to derive consensus partitions from multiple outputs of stochastic clustering algorithms can be downloaded from . *
_ spectral methods_. the spectral clustering method by krzakala et al .
@xcite , based on the non - backtracking matrix ( sections [ sec - tools ] and [ sec - spectral ] ) , can be downloaded here : . *
_ edge clustering_. the code for the edge clustering technique by ahn et al .
@xcite can be found here : .
the link to the stochastic block model based on edge clustering by ball et al .
@xcite is provided below . * _ methods based on statistical inference_. the code to perform the inference of the degree - corrected stochastic block model we have pointed to techniques to infer the number of clusters beforehand .
] by karrer and newman is available at .
the weighted stochastic block model by aicher et al .
@xcite can be found at .
the code for the overlapping stochastic block model based on edge clustering by ball et al .
@xcite is at .
the model combining structure and metadata by newman and clauset @xcite is coded at .
the program to infer the bipartite stochastic block model by larremore et al .
@xcite can be found at .
+ the algorithms for the inference of community structure developed by tiago peixoto are implemented within the python module graph - tool and can be found at .
they allow us to perform model selection of various kinds of stochastic block models : degree - corrected @xcite , with overlapping groups @xcite , and for networks with layers , with valued edges and evolving in time @xcite .
the hierarchical block model of @xcite , that searches for clusters at high resolution , is also available .
all such variants can be combined at ease by selecting a suitable set of options .
+ the algorithms for the inference of overlapping communities via the community - affiliation graph model ( agm ) @xcite and the cluster affiliation model for big networks ( bigclam ) @xcite ( section [ ncp ] ) can be found in the package . * _ methods based on optimisation_. there is a lot of free software for modularity optimisation . in the igraph library
( ) there are several functions , both in the @xmath345 and in the python package : cluster_fast_greedy ( r ) and community_fastgreedy ( python ) , implementing the fast greedy optimisation by clauset et al .
@xcite ; cluster_leading_eigen ( r ) and community_leading_eigenvector ( python ) for the optimisation based on the leading eigenvector of the modularity matrix @xcite ; cluster_louvain ( r ) and community_multilevel ( python ) are the implementations of the louvain method @xcite ; cluster_optimal ( r ) and community_optimal_modularity turn the task into an integer programming problem @xcite ; cluster_spinglass ( r ) and community_spinglass ( python ) optimise the multi - resolution modularity proposed by reichardt and bornholdt @xcite
. + some methods based on the optimisation of cluster quality functions are also available .
the code for the optimisation of the local modularity by clauset @xcite can be found at .
the code for oslom is downloadable from the dedicated website . *
_ methods based on dynamics_. infomap @xcite is currently a very popular algorithm and its code can be found in various places .
it has a dedicated website , where several extensions can be downloaded , including hierarchical community structure @xcite , overlapping clusters @xcite and memory @xcite .
infomap has also its own functions on igraph , both in the r and in the python package ( cluster_infomap and community_infomap , respectively ) .
_ walktrap _
@xcite , another popular method based on random walk dynamics , is available on igraph , via the functions cluster_walktrap ( r ) and community_walktrap ( python ) .
the local community detection algorithms proposed in @xcite can be downloaded from . *
_ dynamic clustering_. the code to optimise the multislice modularity by mucha et al .
@xcite is available at .
detection of dynamic communities can be performed as well with consensus clustering ( section [ sec - consensus ] ) and via stochastic block models ( section [ sec - inference ] ) .
links to the related code have been provided above .
as long as there will be networks , there will be people looking for communities in them .
so it is of uttermost importance to have a set of reliable concepts and principles guiding scholars towards promising solutions for network clustering .
we have presented established views of the main aspects of the problem , and exposed the strengths as well as the limits of popular notions and approaches .
what s next ?
we believe that there will be a trend towards the development of domain - dependent algorithms , exploiting as much as possible information and peculiarities of network data sets .
generalist methods could still be used to get first indications about community structure and orient the investigation in promising directions .
some existing approaches are sufficiently flexible to accommodate various features of networks and community structure ( section [ sec - wmt ] ) . at the same time , we believe that it is necessary to find accurate models of networks with community structure , both for the purpose of designing realistic benchmark graphs for validation , and for a more precise inference of the groups and of their features .
investigations of real networks at the level of subgraphs , along the lines of those discussed in section [ ncp ] , are instrumental to the definition of such models .
while benchmark graphs can be improved , there is one test that one can rely on to assess the performance of clustering algorithms : applying methods on random graphs without group structure .
we know that many popular techniques find groups in such graphs as well , failing the test . on a related note , it is critical to determine how non - random the clusters detected on real networks are , i. e. , to estimate their significance ( section [ sec - sign ] ) .
we stress that this exposition is by no means complete .
the emphasis is on the fundamental aspects of network clustering and on main stream approaches .
we discussed works and listed references which are of more immediate relevance to the topics discussed .
a number of topics have not been dealt with .
still we hope that this work will help practitioners to design more and more reliable methods and domain users to extract useful information from their data .
we thank alex arenas , florian kimm , tiago peixoto , mason porter and martin rosvall for a careful reading of the manuscript and many valuable comments .
we gratefully acknowledge multiplex , grant no .
317532 of the european commission . | community detection in networks is one of the most popular topics of modern network science .
communities , or clusters , are usually groups of vertices having higher probability of being connected to each other than to members of other groups , though other patterns are possible . identifying communities is an ill - defined problem .
there are no universal protocols on the fundamental ingredients , like the definition of community itself , nor on other crucial issues , like the validation of algorithms and the comparison of their performances .
this has generated a number of confusions and misconceptions , which undermine the progress in the field .
we offer a guided tour through the main aspects of the problem .
we also point out strengths and weaknesses of popular methods , and give directions to their use . |
classical statistical theory grants validity of statistical tests and confidence intervals assuming a wall of separation between the selection of a model and the analysis of the data being modeled . in practice , this separation rarely exists , and more often a model is `` found '' by a data - driven selection process . as a consequence inferential guarantees derived from classical theory are invalidated . among model selection methods that are problematic for classical inference , _ variable selection _ stands out because it is regularly taught , commonly practiced and highly researched as a technology .
even though statisticians may have a general awareness that the data - driven selection of variables ( predictors , covariates ) must somehow affect subsequent classical inference from @xmath0- and @xmath1-based tests and confidence intervals , the practice is so pervasive that it appears in classical undergraduate textbooks on statistics such as @xcite .
the reason for the invalidation of classical inference guarantees is that a data - driven variable selection process produces a model that is itself stochastic , and this stochastic aspect is not accounted for by classical theory .
models become stochastic when the stochastic component of the data is involved in the selection process .
( in regression with fixed predictors the stochastic component is the response . )
models are stochastic in a well - defined way when they are the result of formal variable selection procedures such as stepwise or stagewise forward selection or backward elimination or all - subset searches driven by complexity penalties ( such as @xmath2 , aic , bic , risk - inflation , lasso@xmath3 ) or prediction criteria such as cross - validation , or more recent proposals such as lars and the dantzig selector ; for an overview see , for example , @xcite .
models are also stochastic but in an ill - defined way when they are informally selected through visual inspection of residual plots or normal quantile plots or other regression diagnostics .
finally , models become stochastic in an opaque way when their selection is affected by human intervention based on post - hoc considerations such as `` in retrospect only one of these two variables should be in the model '' or `` it turns out the predictive benefit of this variable is too weak to warrant the cost of collecting it . '' in practice , all three modes of variable selection may be exercised in the same data analysis : multiple runs of one or more formal search algorithms may be performed and compared , the parameters of the algorithms may be subjected to experimentation and the results may be critiqued with graphical diagnostics ; a round of fine - tuning based on substantive deliberations may finalize the analysis .
posed so starkly , the problems with statistical inference after variable selection may well seem insurmountable . at a minimum
, one would expect technical solutions to be possible only when a formal selection algorithm is ( 1 ) well - specified ( 1a ) in advance and ( 1b ) covering all eventualities , ( 2 ) strictly adhered to in the course of data analysis and ( 3 ) not `` improved '' on by informal and post - hoc elements .
it may , however , be unrealistic to expect this level of rigor in most data analysis contexts , with the exception of well - conducted clinical trials .
the real challenge is therefore to devise statistical inference that is valid following _ any _ type of variable selection , be it formal , informal , post hoc or a combination thereof .
meeting this challenge with a relatively simple proposal is the goal of this article .
this proposal for valid _
_ po__st-__s__election _ _ i__nference , or `` _ _ posi _ _ '' for short , consists of a large - scale family - wise error guarantee that can be shown to account for all types of variable selection , including those of the informal and post - hoc varieties . on the other hand , the proposal is no more conservative than necessary to account for selection , and in particular it can be shown to be less conservative than scheff s simultaneous inference .
the framework for our proposal is in outline as follows
details to be elaborated in subsequent sections : we consider linear regression with predictor variables whose values are considered fixed , and with a response variable that has normal and homoscedastic errors .
the framework does not require that any of the eligible linear models is correct , not even the full model , as long as a valid error estimate is available .
we assume that the selected model is the result of some procedure that makes use of the response , but the procedure does not need to be fully specified .
a crucial aspect of the framework concerns the use and interpretation of the selected model : we assume that , after variable selection is completed , the selected predictor variables and only they will be relevant ; all others will be eliminated from further consideration .
this assumption , seemingly innocuous and natural , has critical consequences : it implies that statistical inference will be sought for the coefficients of the selected predictors only and in the context of the selected model only .
thus the appropriate targets of inference are the best linear coefficients within the selected model , where each coefficient is adjusted for the presence of all other included predictors but not those that were eliminated .
therefore the coefficient of an included predictor generally requires inference that is specific to the model in which it appears . summarizing in a motto , a difference in adjustment implies a difference in parameters and hence in inference .
the goal of the present proposal is therefore simultaneous inference for all coefficients within all submodels .
such inference can be shown to be valid following any variable selection procedure , be it formal , informal , post hoc , fully or only partly specified .
problems associated with post - selection inference were recognized long ago , for example , by @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite .
more recently specific problems have been the subject of incisive analyses and criticisms by the `` vienna school '' of ptscher , leeb and schneider ; see , for example , leeb and ptscher ( @xcite ) , @xcite , @xcite , @xcite , ptscher and schneider ( @xcite ) , as well as @xcite and @xcite .
important progress was made by @xcite and @xcite .
this article proceeds as follows : in section [ secreinterpretation ] we first develop the `` submodel view '' of the targets of inference after model selection and contrast it with the `` full model view '' ( section [ secsubmodel ] ) ; we then introduce assumptions with a view toward valid inference in `` wrong models '' ( section [ secassumptions ] ) .
section [ secestimation ] is about estimation and its targets from the submodel point of view .
section [ secvalidcis ] develops the methodology for posi confidence intervals ( cis ) and tests . after some structural results for the posi problem in section [ secstruct ] , we show in section [ secasymp ] that with increasing number of predictors @xmath4 the width of posi cis can range between the asymptotic rates @xmath5 and @xmath6 .
we give examples for both rates and , inspired by problems in sphere packing and covering , we give upper bounds for the limiting constant in the @xmath6 case .
we conclude with a discussion in section [ secconc ] .
some proofs are deferred to the , and some elaborations to the online appendix in the supplementary material [ @xcite ] .
computations will be described in a separate article .
simulation - based methods yield satisfactory accuracy specific to a design matrix up to @xmath7 , while nonasymptotic universal upper bounds can be computed for larger @xmath4 .
it is a natural intuition that model selection distorts inference by distorting sampling distributions of parameter estimates : estimates in selected models should tend to generate more type i errors than conventional theory allows because the typical selection procedure favors models with strong , hence highly significant predictors .
this intuition correctly points to a multiplicity problem that grows more severe as the number of predictors subject to selection increases .
this is the problem we address in this article .
model selection poses additional problems that are less obvious but no less fundamental : there exists an ambiguity as to the role and meaning of the parameters in submodels . in one view , the relevant parameters are always those of the full model , hence the selection of a submodel is interpreted as estimating the deselected parameters to be zero and estimating the selected parameters under a zero constraint on the deselected parameters . in another view , the submodel has its own parameters , and the deselected parameters are not zero but nonexistent .
these distinctions are not academic as they imply fundamentally different ideas regarding the targets of inference , the measurement of statistical performance , and the problem of post - selection inference .
the two views derive from different purposes of equations : * underlying the full model view of parameters is the use of a full equation to describe a `` data generating '' mechanism for the response ; the equation hence has a causal interpretation . *
underlying the submodel view of parameters is the use of any equation to merely describe association between predictor and response variables ; no data generating or causal claims are implied . in this article
we address the latter use of equations .
issues relating to the former use are discussed in the online appendix of the supplementary material [ @xcite , section b.1 ] . in what follows we elaborate three points that set the submodel interpretation of coefficients apart from the full model interpretation , with important consequences for the rest of this article : the full model has no special status other than being the repository of available predictors .
the coefficients of excluded predictors are not zero ; they are not defined and therefore do not exist .
the meaning of a predictor s coefficient depends on which other predictors are included in the selected model .
\(1 ) the full model available to the statistician often can not be argued to have special status because of inability to identify and measure all relevant predictors .
additionally , even when a large and potentially complete suite of predictors can be measured , there is generally a question of predictor redundancy that may make it desirable to omit some of the measurable predictors from the final model .
it is a common experience in the social sciences that models proposed on theoretical grounds are found on empirical grounds to have their predictors entangled by collinearities that permit little meaningful statistical inference .
this situation is not limited to the social sciences : in gene expression studies it may well occur that numerous sites have a tendency to be expressed concurrently , hence as predictors in disease studies they will be strongly confounded .
the emphasis on full models may be particularly strong in econometrics where there is a `` notion that a longer regression has a causal interpretation , while a shorter regression does not '' [ @xcite , page 59 ] . even in causal models ,
however , there is a possibility that included adjuster variables will `` adjust away '' some of the causal variables of interest . generally , in any creative observational study involving novel predictors
, it will be difficult a priori to exclude collinearities that might force a rethinking of the predictors . in conclusion , whenever predictor redundancy is a potential issue , it can not a priori be claimed that the full model provides the parameters of primary interest .
\(2 ) in the submodel interpretation of parameters , claiming that the coefficients of deselected predictors are zero does not properly describe the role of predictors .
deselected predictors have no role in the submodel equation ; they become no different than predictors that had never been considered .
the selected submodel becomes the vehicle of substantive research irrespective of what the full model was .
as such the submodel stands on its own .
this view is especially appropriate if the statistician s task is to determine which predictors are to be measured in the future .
\(3 ) the submodel interpretation of parameters is deeply seated in how we teach regression .
we explain that the meaning of a regression coefficient depends on which of the other predictors are included in the model : `` the slope is the average difference in the response for a unit difference in the predictor , _ at fixed levels of all other predictors in the model_. '' this `` ceteris paribus '' clause is essential to the meaning of a slope . that there is a difference in meaning when there is a difference in covariates is most drastically evident when there is a case of simpson s paradox . for example , if purchase likelihood of a high - tech gadget is predicted from _ age _ , it might be found against expectations that younger people have lower purchase likelihood , whereas a regression on _ age _ and _ income _ might show that at fixed levels of income younger people have indeed higher purchase likelihood .
this case of simpson s paradox would be enabled by the expected positive collinearity between _ age _ and _ income_. thus the marginal slope on _ age _ is distinct from the _ income_-adjusted slope on _ age _ as the two slopes answer different questions , apart from having opposite signs . in summary ,
_ different models result in different parameters with different meanings_. must we use the full model with both predictors ? not if _ income _
data is difficult to obtain or if it provides little improvement in @xmath8 beyond _ age_. the model based on _ age _ alone can not be said to be a priori `` wrong . '' if , for example , the predictor and response variables have jointly multivariate normal distributions , then every linear submodel is `` correct . ''
these considerations drive home , once again , that sometimes no model has special status . in summary ,
a range of applications call for a framework in which the full model is not the sole provider of parameters , where rather each submodel defines its own .
the consequences of this view will be developed in section [ secestimation ] .
we state assumptions for estimation and for the construction of valid tests and cis when fitting arbitrary linear equations .
the main goal is to prepare the ground for valid statistical inference after model selection_not _ assuming that selected models are correct .
we consider a quantitative response vector @xmath9 , assumed random , and a full predictor matrix @xmath10 , @xmath11 , assumed fixed .
we allow @xmath12 to be of nonfull rank , and @xmath13 and @xmath4 to be arbitrary .
in particular , we allow @xmath14 . throughout the article we let @xmath15 due to frequent reference we call @xmath16 ( @xmath17 ) `` _ _ the classical case__. '' it is common practice to assume the full model @xmath18 to be correct . in the present framework , however , first - order correctness , @xmath19 = { { \mathbf{x}}}{{\bolds{\beta}}}$ ] , will not be assumed . by implication
, first - order correctness of any submodel will not be assumed either .
effectively , @xmath20 \in{\mathbb{r}}^n\ ] ] is allowed to be unconstrained and , in particular , need not reside in the column space of @xmath12 .
that is , the model given by @xmath12 is allowed to be `` first - order wrong , '' and hence we are , in a well - defined sense , serious about g.e.p .
box s famous quote .
what he calls `` wrong models , '' we prefer to call `` approximations '' : all predictor matrices @xmath12 provide approximations to @xmath21 , some better than others , but the degree of approximation plays no role in the clarification of statistical inference .
the main reason for elaborating this point is as follows : after model selection , the case for `` correct models '' is clearly questionable , even for `` consistent model selection procedures '' [ @xcite , page 101 ] ; but if correctness of submodels is not assumed , it is only natural to abandon this assumption for the full model also , in line with the idea that the full model has no special status . as we proceed with estimation and inference guarantees in the absence of first - order correctness we will rely on assumptions as follows : * for estimation ( section [ secestimation ] ) , we will only need the existence of @xmath22 $ ] . * for testing and ci guarantees ( section [ secvalidcis ] )
, we will make conventional second - order and distributional assumptions , @xmath23 the assumptions ( [ eqassumption ] ) of homoscedasticity and normality are as questionable as first - order correctness , and we will report elsewhere on approaches that avoid them .
for now we follow the vast model selection literature that relies on the technical advantages of assuming homoscedastic and normal errors . accepting the assumption ( [ eqassumption ] )
, we address the issue of estimating the error variance @xmath24 , because the valid tests and cis we construct require a valid estimate @xmath25 of @xmath24 that is independent of ls estimates . in the classical case ,
the most common way to assert such an estimate is to assume that the full model is first - order correct , @xmath26 in addition to ( [ eqassumption ] ) , in which case the mean squared residual ( msr ) @xmath27 of the full model will do .
however , other possibilities for producing a valid estimate @xmath25 exist , and they may allow relaxing the assumption of first - order correctness : * exact replications of the response obtained under identical conditions might be available in sufficient numbers .
an estimate @xmath25 can be obtained as the msr of the one - way anova of the groups of replicates . * in general , a larger linear model than the full model might be considered as correct ; hence @xmath25 could be the msr from this larger model .
* a different possibility is to use another dataset , similar to the one currently being analyzed , to produce an independent estimate @xmath25 by whatever valid estimation method .
* a special case of the preceding is a random split - sample approach whereby one part of the data is reserved for producing @xmath25 and the other part for estimating coefficients , selecting models and carrying out post - model selection inference . *
a different type of estimate , @xmath25 , may be based on considerations borrowed from nonparametric function estimation [ @xcite ] .
the purpose of pointing out these possibilities is to separate , at least in principle , the issue of first - order model incorrectness from the issue of error estimation under assumption ( [ eqassumption ] ) .
this separation puts the case @xmath28 within our framework as the valid and independent estimation of @xmath24 is a problem faced by all `` @xmath28 '' approaches .
following section [ secsubmodel ] , the value and meaning of a regression coefficient depends on what the other predictors in the model are .
an exception occurs , of course , when the predictors are perfectly orthogonal , as in some designed experiments or in function fitting with orthogonal basis functions . in this case
a coefficient has the same value and meaning across all submodels .
this article is hence a story of ( partial ) collinearity
. we will give meaning to ls estimators and their targets in the absence of any assumptions other than the existence of @xmath22 $ ] , which in turn is permitted to be entirely unconstrained in @xmath29 .
besides resolving the issue of estimation in `` first - order wrong models , '' the major purpose here is to elaborate the idea that the slope of a predictor generates different parameters in different submodels . as each predictor appears in @xmath30 submodels ,
the @xmath4 regression coefficients of the full model generally proliferate into a plethora of as many as @xmath31 distinct regression coefficients according to the submodels they appear in .
to describe the situation we start with notation .
to denote a submodel we use the ( nonempty ) index set @xmath32 of the predictors @xmath33 in the submodel ; the size of the submodel is @xmath34 and that of the full model is @xmath35 .
let @xmath36 denote the @xmath37 submatrix of @xmath12 with columns indexed by @xmath38 .
we will only allow submodels @xmath38 for which @xmath39 is of full rank , @xmath40 we let @xmath41 be the unique least squares estimate in @xmath38 , @xmath42 now that @xmath41 is an estimate , what is it estimating ? following section [ secsubmodel ] , we will not interpret @xmath41 as estimates of the full model coefficients and , more generally , of any model other than @xmath38 .
thus it is natural to ask that @xmath41 define its own target through the requirement of unbiasedness , @xmath43 = \bigl({{\mathbf{x}}}_{\mathrm{m}}^t { { \mathbf{x}}}_{\mathrm{m}}\bigr)^{-1 } { { \mathbf{x}}}_{\mathrm{m}}^t { { \mathbf{e}}}[{{\mathbf{y } } } ] = { \mathop{\arg\min}}_{{{\bolds{\beta } } } ' \in{\mathbb{r}}^m } \bigl\| { { \bolds{\mu}}}- { { \mathbf{x}}}_{\mathrm{m}}{{\bolds{\beta } } } ' \bigr\| ^2.\ ] ] this definition requires no other assumption than the existence of @xmath22 $ ] .
in particular there is no need to assume first - order correctness of @xmath38 or @xmath44 . nor does it matter to what degree @xmath38 provides a good approximation to @xmath21 in terms of approximation error @xmath45 . in the classical case @xmath46
, we can define the target of the full - model estimate @xmath47 as a special case of ( [ eqtarget ] ) with @xmath48 , @xmath49 = \bigl({{\mathbf{x}}}^t { { \mathbf{x}}}\bigr)^{-1 } { { \mathbf{x}}}^t { { \mathbf{e}}}[{{\mathbf{y}}}].\ ] ] in the general ( including the nonclassical ) case , let @xmath50 be any ( possibly nonunique ) minimizer of @xmath51 ; the link between @xmath50 and @xmath52 is as follows : @xmath53 thus the target @xmath52 is an estimable linear function of @xmath50 , without first - order correctness assumptions
. equation ( [ eqcontrast ] ) follows from @xmath54 .
_ notation _ : to distinguish the regression coefficients of the predictor @xmath55 relative to the submodel it appears in , we write @xmath56 $ ] for the components of @xmath57 $ ] with @xmath58 .
an important convention is that indexes are always elements of the full model , @xmath59 , for what we call `` full model indexing . ''
@xmath60the regression coefficient @xmath61 is conventionally interpreted as the `` average difference in the response for a unit difference in @xmath62 , ceteris paribus in the model @xmath38 . ''
this interpretation no longer holds when the assumption of first - order correctness is given up . instead
, the phrase `` average difference in the response '' should be replaced with the unwieldy phrase `` average difference in the response approximated in the submodel @xmath38 . ''
the reason is that the target of the fit @xmath63 in the submodel @xmath38 is @xmath64 , hence in @xmath38 we estimate unbiasedly not the true @xmath21 but its ls approximation @xmath65 .
a second interpretation of regression coefficients is in terms of adjusted predictors : for @xmath58 define the @xmath38-adjusted predictor @xmath66 as the residual vector of the regression of @xmath55 on all other predictors in @xmath38 .
multiple regression coefficients , both estimates @xmath67 and parameters @xmath61 , can be expressed as simple regression coefficients with regard to the @xmath38-adjusted predictors , @xmath68 the left - hand formula lends itself to an interpretation of @xmath69 in terms of the well - known leverage plot which shows @xmath70 plotted against @xmath66 and the line with slope @xmath67 .
this plot is valid without first - order correctness assumption .
a third interpretation can be derived from the second : to unclutter notation let @xmath71 be any adjusted predictor @xmath72 , so that @xmath73 and @xmath74 are the corresponding @xmath67 and @xmath61 .
introduce ( 1 ) case - wise slopes through the origin , both as estimates @xmath75 and as parameters @xmath76 , and ( 2 ) case - wise weights @xmath77 .
equations ( [ eqadjusted ] ) are then equivalent to the following : @xmath78 hence regression coefficients are weighted averages of case - wise slopes , and this interpretation holds without first - order assumptions .
we consider inference for @xmath41 and its target @xmath52 . following section [ secassumptions ]
we require a normal homoscedastic model for @xmath70 , but we leave its mean @xmath22 $ ] entirely unspecified : @xmath79 .
we then have equivalently @xmath80 again following section [ secassumptions ] we assume the availability of a valid estimate @xmath25 of @xmath24 that is independent of all estimates @xmath67 , and we further assume @xmath81 for @xmath82 degrees of freedom . if the full model is assumed to be correct , @xmath83 and @xmath84 , then @xmath85 . in the limit @xmath86
we obtain @xmath87 , the case of known @xmath88 , which will be used starting with section [ secasymp ] .
let @xmath89 denote a @xmath1-ratio for @xmath61 that uses @xmath90 irrespective of @xmath38 , @xmath91 where @xmath92 refers to the diagonal element corresponding to @xmath55 .
the quantity @xmath93 has a central @xmath1-distribution with @xmath82 degrees of freedom .
essential is that the standard error estimate in the denominator of ( [ eqt ] ) does _ not _ involve the msr @xmath94 from the submodel @xmath38 , for two reasons : * we do not assume that the submodel @xmath38 is first - order correct ; hence @xmath95 would , in general , have a distribution that is a multiple of a noncentral @xmath96 distribution with unknown noncentrality parameter . *
more disconcertingly , @xmath95 would be the result of selection , @xmath97 ; see section [ secmodel - selection ] .
not much of real use is known about its distribution ; see , for example , @xcite and @xcite .
these problems are avoided by using one valid estimate @xmath25 that is independent of all submodels . with this choice of @xmath90 , confidence intervals for @xmath98
take the form @xmath99_{jj}^{{1/2 } } { { \hat{\sigma}}}\bigr ] \nonumber\\[-8pt]\\[-8pt ] & = & \bigl [ { { \hat{\beta}}}_{j \cdot{\mathrm{m } } } \pm k { { \hat{\sigma}}}/\|{{\mathbf{x}}}_{j \cdot{\mathrm{m}}}\| \bigr ] . \nonumber\end{aligned}\ ] ] if @xmath100 is the @xmath101 quantile of a @xmath1-distribution with @xmath82 degrees of freedom , then the interval is marginally valid with a @xmath102 coverage guarantee @xmath103 \stackrel{(\ge ) } { = } 1-\alpha.\ ] ] this holds if the submodel @xmath38 is _ not _ the result of variable selection . in practice
, the model @xmath38 tends to be the result of some form of model selection that makes use of the stochastic component of the data , which is the response vector @xmath70 ( @xmath12 being fixed , section [ secassumptions ] ) .
this model should therefore be expressed as @xmath104 . in general
we allow a variable selection _ procedure _ to be any ( measurable ) map @xmath105 where @xmath106 is the set of all full - rank submodels @xmath107 thus the procedure @xmath108 is a discrete map that divides @xmath29 into as many as @xmath109 different regions with shared outcome of model selection .
data dependence of the selected model @xmath108 has strong consequences : * most fundamentally , the selected model @xmath104 is now random .
whether the model has been selected by an algorithm or by human choice , if the response @xmath70 has been involved in the selection , the resulting model is a random object because it could have been different for a different realization of the random vector @xmath70 . *
associated with the random model @xmath110 is the parameter vector of coefficients @xmath111 , which is now randomly chosen also : * * it has a random dimension @xmath112 : @xmath113 . * * for any fixed @xmath114 , it may or may not be the case that @xmath115 . *
* conditional on @xmath115 , the parameter @xmath116 changes randomly as the adjuster covariates in @xmath110 vary randomly .
thus the set of parameters for which inference is sought is random also . with randomness of the selected model and its parameters in mind , what is a desirable form of post - selection coverage guarantee for confidence intervals ?
a natural requirement would be a @xmath102 confidence guarantee for the coefficients of the predictors that are selected into the model , @xmath117 \ge1-\alpha.\ ] ] several points should be noted : * the guarantee is family - wise for all selected predictors @xmath118 , though the sense of `` family - wise '' is unusual because @xmath119 is random . * the guarantee has nothing to say about predictors @xmath120 that have been deselected , regardless of the substantive interest they might have .
predictors of overarching interest should be protected from variable selection , and for these one can use a modification of the posi approach which we call `` posi1 ; '' see section [ secposi1 ] .
* because predictor selection is random , @xmath104 , two realized samples @xmath121 from @xmath70 may result in different sets of selected predictors , @xmath122 .
it would be a fundamental misunderstanding to wonder whether the guarantee holds for both realizations .
instead , the guarantee ( [ eqposiguarantee ] ) is about the _ procedure _
@xmath123 for the long run of independent realizations of @xmath70 ( by the lln ) , and not for any particular realizations @xmath124 .
a standard formulation used to navigate these complexities after a realization @xmath125 of @xmath70 has been analyzed is the following : `` for @xmath118 we have @xmath102 _ confidence _ that the interval @xmath126 contains @xmath127 . ''
* marginal guarantees for individual predictors require some care because @xmath128 does not exist for @xmath120 .
this makes @xmath129 an incoherent statement that does not define an event .
guarantees are possible if the condition @xmath118 is added with a conjunction or is being conditioned on : the marginal and conditional probabilities @xmath130 \quad\mbox{and}\quad { { \mathbf{p}}}\bigl[\beta_{j \cdot{{\hat{\mathrm{m } } } } } \in{\operatorname{ci}}_{j \cdot{{\hat{\mathrm{m}}}}}(k_{j \cdot})| j \in{{\hat{\mathrm{m}}}}\bigr],\ ] ] respectively , are both well - defined and can be the subject of coverage guarantees ; see the online appendix of the supplementary material [ @xcite , section b.4 ] .
finally , we note that the smallest constant @xmath131 that satisfies the guarantee ( [ eqposiguarantee ] ) is specific to the procedure @xmath108 .
thus different variable selection procedures would require different constants .
finding procedure - specific constants is a challenge that will be intentionally bypassed by the present proposals .
the `` posi '' procedure proposed here produces a constant @xmath131 that provides universally valid post - selection inference _ for all model selection procedures _
@xmath108 , @xmath132 \ge1-\alpha\qquad\forall{{\hat{\mathrm{m}}}}.\ ] ] universal validity irrespective of the model selection procedure @xmath108 is a strong property that raises questions of whether the approach is too conservative .
there are , however , some arguments in its favor : \(1 ) universal validity may be desirable or even essential for applications in which the model selection procedure is not specified in advance or for which the analysis involves some ad hoc elements that can not be accurately pre - specified .
even so , we should think of the actually chosen model as part of a `` procedure '' @xmath133 , and though the ad hoc steps are not specified for @xmath70 other than the observed one , this is not a problem because our protection is irrespective of what a specification might have been .
this view also allows data analysts to change their minds , to improvise and informally decide in favor of a model other than that produced by a formal selection procedure , or to experiment with multiple selection procedures .
\(2 ) there exists a model selection procedure that requires the full strength of universally valid posi , and this procedure may not be entirely unrealistic as an approximation to some types of data analytic activities : `` significance hunting , '' that is , selecting that model which contains the statistically most significant coefficient ; see section [ secspar ] .
\(3 ) there is a general question about the wisdom of proposing ever tighter confidence and retention intervals for practical use when in fact these intervals are valid only under tightly controlled conditions
. it might be realistic to suppose that much applied work involves more data peeking than is reported in published articles .
with inference that is universally valid after any model selection procedure , we have a way to establish which rejections are safe , irrespective of unreported data peeking as part of selecting a model .
\(4 ) related to the previous point is the fact that today there is a realization that a considerable fraction of published empirical work is unreproducible or reports exaggerated effects ; well known in this regard is @xcite .
a factor contributing to this problem might well be liberal handling of variable selection and absent accounting for it in subsequent inference .
the concerns over posi s conservative nature can be alleviated somewhat by introducing a degree of flexibility to the posi problem with regard to the universe of models being searched .
such flexibility is additionally called for from a practical point of view because it is not true that all submodels in @xmath106 ( [ eqmall ] ) are always being searched .
rather , the search is often limited in a way that can be specified a priori , without involvement of @xmath70 . for example
, a predictor of interest may be forced into the submodels of interest , or there may be a restriction on the size of the submodels . indeed , if @xmath4 is large , a restriction to a manageable set of submodels is a computational necessity . in much of what follows we can allow the universe @xmath134 of allowable submodels to be an ( almost ) arbitrary but pre - specified nonempty subset of @xmath106 ; w.l.o.g .
we can assume @xmath135 . because we allow only nonsingular submodels [ see ( [ eqmall ] ) ] we have @xmath136 @xmath137 , where as always @xmath138 .
selection procedures are now maps @xmath139 the following are examples of model universes with practical relevance ; see also @xcite , section 1.1 , example 1 .
submodels that contain the first @xmath140 predictors ( @xmath141 ) : @xmath142 .
classical : @xmath143 .
example : forcing an intercept into all models .
submodels of size @xmath144 or less ( `` sparsity option '' ) : @xmath145 .
classical : @xmath146
. submodels with fewer than @xmath144 predictors dropped from the full model : @xmath147 .
classical : @xmath148 .
nested models : @xmath149 .
example : selecting the degree up to @xmath151 in a polynomial regression . models dictated by an anova hierarchy of main effects and interactions in a factorial design .
this list is just an indication of possibilities . in general , the smaller the set @xmath152 is , the less conservative the posi approach is , and the more computationally manageable the problem becomes . with sufficiently strong restrictions , in particular using the sparsity option ( 2 ) and assuming the availability of an independent valid estimate @xmath90 , it is possible to apply posi in certain nonclassical @xmath153 situations . further reduction of the posi problem is possible by pre - screening adjusted predictors _ without the response _
@xmath70 . in a fixed - design regression , any variable selection procedure that does _
not _ involve @xmath70 does _ not _ invalidate statistical inference .
for example , one may decide not to seek inference for predictors in submodels that impart a `` variance inflation factor '' ( @xmath154 ) above a user - chosen threshold : @xmath155 if @xmath55 is centered , hence does not make use of @xmath70 , and elimination according to @xmath156 does not invalidate inference .
we show that universally valid post - selection inference ( [ eqbciuniv ] ) follows from simultaneous inference in the form of family - wise error control for all parameters in all submodels .
the argument depends on the following lemma that may fall into the category of the `` trivial but not immediately obvious . ''
[ lemtrivial ] for any model selection procedure @xmath157 , the following inequality holds for all @xmath9 : @xmath158 this is a special case of the triviality @xmath159 , where @xmath160 . the right - hand max-@xmath161 bound of the lemma is sharp in the sense that there exists a variable selection procedure @xmath108 that attains the bound ; see section [ secspar ] .
next we introduce the @xmath162 quantile of the right - hand max-@xmath161 statistic of the lemma : let @xmath131 be the minimal value that satisfies latexmath:[\[\label{eqkdef } { { \mathbf{p}}}\bigl[\max_{{\mathrm{m}}\in{{\mathcal{m } } } } \max_{j \in{\mathrm{m } } } this value will be called `` the posi constant . ''
it does not depend on any model selection procedures , but it does depend on the design matrix @xmath12 , the universe @xmath134 of models subject to selection , the desired coverage @xmath162 , and the degrees of freedom @xmath82 in @xmath90 , hence @xmath164 .
[ thmsim ] for all model selection procedures @xmath165 we have @xmath166 \ge1-\alpha,\ ] ] where @xmath164 is the posi constant .
this follows immediately from lemma [ lemtrivial ] .
although mathematically trivial we give the above the status of a theorem as it is the central statement of the reduction of universal post - selection inference to simultaneous inference .
the following is just a repackaging of theorem [ thmsim ] : [ corcov ] `` simultaneous post - selection confidence guarantees '' hold for any model selection procedure @xmath167 , @xmath168 \ge1-\alpha,\ ] ] where @xmath164 is the posi constant .
simultaneous inference provides strong family - wise error control , which in turn translates to strong error control for tests following model selection .
[ corstrong ] `` strong post - selection error control '' holds for any model selection procedure @xmath169 , @xmath170 \le\alpha,\ ] ] where @xmath164 is the posi constant and @xmath171 is the @xmath1-statistic for the null hypothesis @xmath172 .
the proof is standard ; see the online appendix of the supplementary material [ @xcite , section b.3 ] .
the corollary states that , with probability @xmath162 , in a selected model _ all _ posi - significant rejections have detected true alternatives .
several portions of the following treatment are devoted to a better understanding of the structure and value of the posi constant @xmath173 . except for very special choices it does not seem possible to provide closed form expressions for its value .
however , the structural geometry and other properties to be described later do enable a reasonably efficient computational algorithm .
r - code for computing the posi constant for small to moderate values of @xmath4 is available on the authors web pages .
this code is accompanied by a manuscript that will be published elsewhere describing the computational algorithm and generalizations . for the basic
setting involving @xmath106 the algorithm will conveniently provide values of @xmath174 for matrices @xmath12 of rank @xmath175 , or slightly larger depending on available computing speed and memory .
it can also be adapted to compute @xmath131 for some other families contained within @xmath106 , such as some discussed in section [ secrestrictions ] .
realizing the idea that the ls estimators in different submodels are generally unbiased estimates of different parameters , we generated a simultaneous inference problem involving up to @xmath176 linear contrasts @xmath61 .
in view of the enormous number of linear combinations for which simultaneous inference is sought , one should wonder whether the problem is not best solved by scheff s method [ @xcite ] which provides simultaneous inference for _ all _ linear combinations . to accommodate rank - deficient @xmath12 , we cast scheff s result in terms of @xmath1-statistics for arbitrary nonzero @xmath177 : @xmath178 the @xmath1-statistics in ( [ eqt ] ) are obtained for @xmath179 .
scheff s guarantee is @xmath180 = 1-\alpha,\ ] ] where the scheff constant is @xmath181 it provides an upper bound for _ all _ posi constants : [ propscheffe ] @xmath182 .
thus for @xmath118 a parameter estimate @xmath183 whose @xmath1-ratio exceeds @xmath184 in magnitude is universally safe from having the rejection of `` @xmath185 '' invalidated by variable selection .
the universality of the scheff constant is a tip - off that it may be too loose for some predictor matrices @xmath12 , and obtaining the sharper constant @xmath186 may be worthwhile .
an indication is given by the following comparison as @xmath187 : * for the scheff constant it holds @xmath188 . * for orthogonal designs it holds @xmath189 .
( for orthogonal designs see section [ secorthopt ] . )
thus the posi constant @xmath190 is much smaller than @xmath184 .
the large gap between the two suggests that the scheff constant may be too conservative at least in some cases
. we will study certain nonorthogonal designs for which the posi constant is @xmath191 in section [ secexch ] . on the other hand
, the posi constant can approach the order @xmath192 of the scheff constant @xmath184 as well , and we will study an example in section [ secrootp ] .
even though in this article we will give asymptotic results for @xmath193 and @xmath194 only , we mention another kind of asymptotics whereby @xmath82 is held constant while @xmath193 : in this case @xmath184 is in the order of the product of @xmath195 and the @xmath102 quantile of the inverse - root - chi - square distribution with @xmath82 degrees of freedom . in a similar way ,
the constant @xmath190 for orthogonal designs is in the order of the product of @xmath196 and the @xmath102 quantile of the inverse - chi - square distribution with @xmath82 degrees of freedom .
there exists a model selection procedure that requires the full protection of the simultaneous inference procedure ( [ eqkdef ] ) .
it is the `` significance hunting '' procedure that selects the model containing the most significant `` effect '' : @xmath197 we name this procedure `` spar '' for `` _ _ single predictor adjusted regression__. '' it achieves equality with the `` significant triviality bound '' in lemma [ lemtrivial ] and is therefore the worst case procedure for the posi problem . in the submodel @xmath198
, the less significant predictors matter only in so far as they boost the significance of the winning predictor by adjusting it accordingly .
this procedure ignores the quality of the fit to @xmath70 provided by the model .
while our present purpose is to point out the existence of a selection procedure that requires full posi protection , spar could be of practical interest when the analysis is centered on strength of `` effects , '' not quality of model fit .
sometimes a regression analysis is centered on a predictor of interest , @xmath55 , and on inference for its coefficient @xmath199 .
the other predictors in @xmath38 act as controls , so their purpose is to adjust the primary predictor for confounding effects and possibly to boost the primary predictor s own `` effect . ''
this situation is characterized by two features : * variable selection is limited to models that contain the primary predictor .
we therefore define for any model universe @xmath134 a sub - universe @xmath200 of models that contain the primary predictor @xmath55 , @xmath201 so that for @xmath202 we have @xmath58 iff @xmath203 .
* inference is sought for the primary predictor @xmath55 only , hence the relevant test statistic is now @xmath204 and no longer @xmath205 .
the former statistic is coherent because it is assumed that @xmath206 .
we call this the `` posi1 '' situation in contrast to the unconstrained posi situation . similar to posi , posi1 starts with a `` significant triviality bound '' : [ lemtrivial1 ] for a fixed predictor @xmath55 and model selection procedure @xmath207 , it holds that @xmath208 for a `` proof , '' the only thing to note is @xmath209 by the assumption @xmath210 . we next define the `` posi1 '' constant for the predictor @xmath55 as the @xmath102 quantile of the max-@xmath161 statistic on the right - hand side of the lemma : let @xmath211 be the minimal value that satisfies @xmath212 \ge1 - \alpha.\ ] ] importantly , this constant is dominated by the general posi constant , @xmath213 for the obvious reason that the present max-@xmath161 is smaller than the general posi max-@xmath161 due to @xmath214 and the restriction of inference to @xmath55 .
the constant @xmath215 provides the following `` posi1 '' guarantee shown as the analog of theorem [ thmsim ] and corollary [ corcov ] folded into one : [ thmsim1r ] let @xmath216 be a selection procedure that always includes the predictor @xmath55 in the model
. then we have @xmath217 \ge1 - \alpha,\ ] ] and accordingly we have the following post - selection confidence guarantee : @xmath218 \ge1-\alpha.\ ] ] inequality ( [ eqsimjr ] ) is immediate from lemma [ lemtrivial1 ] .
the `` triviality bound '' of the lemma is attained by the following variable selection procedure which we name `` spar1 '' : @xmath219 it is a potentially realistic description of some data analyses when a predictor of interest is determined a priori , and the goal is to optimize _ this _ predictor s `` effect . ''
this procedure requires the full protection of the posi1 constant @xmath215 .
in addition to its methodological interest , the posi1 situation addressed by theorem [ thmsim1r ] is of theoretical interest : even though the posi1 constant @xmath215 is dominated by the unrestricted posi constant @xmath131 , we will construct in section [ secrootp ] an example of predictor matrices for which the posi1 constant increases at the scheff rate and is asymptotically more than 63% of the scheff constant @xmath184 .
it follows that near - scheff protection can be needed even for spar1 variable selection .
we can reduce the dimensionality of the posi problem from @xmath220 to @xmath221 , where @xmath222 , by introducing scheff s canonical coordinates .
this reduction is important both geometrically and computationally because the posi coverage problem really takes place in the column space of @xmath12 .
let @xmath223 be any orthonormal basis of the column space of @xmath12 .
note that @xmath224 is the orthogonal projection of @xmath70 onto the column space of @xmath12 even if @xmath12 is not of full rank .
we call @xmath225 and @xmath226 canonical coordinates of @xmath12 and @xmath227
. we extend the notation @xmath39 for extraction of subsets of columns to canonical coordinates @xmath228 .
accordingly slopes obtained from canonical coordinates will be denoted by @xmath229 to distinguish them from the slopes obtained from the original data @xmath230 , if only to state in the following proposition that they are identical .
[ propcancoord ] properties of canonical coordinates : @xmath231 .
@xmath232 and @xmath233 .
@xmath234 for all submodels @xmath235 .
@xmath236 , where @xmath237 .
@xmath238 , where @xmath239 and @xmath240 is the residual vector of the regression of @xmath241 onto the other columns of @xmath228 .
@xmath242 . in the classical case
@xmath243 , @xmath244 can be chosen to be an upper triangular or a symmetric matrix .
the proofs of ( 1)(6 ) are elementary . as for ( 7 ) , an upper triangular @xmath244 can be obtained from a qr - decomposition based on a gram schmidt procedure , @xmath245 , @xmath246 . a symmetric @xmath244
is obtained from a singular value decomposition , @xmath247 , @xmath248 , @xmath249 .
canonical coordinates allow us to analyze the posi coverage problem in @xmath250 . in what follows
we will freely assume that all objects are rendered in canonical coordinates and write @xmath12 and @xmath70 for @xmath244 and @xmath251 , implying that the predictor matrix is of size @xmath252 and the response is of size @xmath253 .
we simplify the posi coverage problem ( [ eqkdef ] ) as follows : due to pivotality of @xmath1-statistics , the problem is invariant under translation of @xmath21 and rescaling of @xmath88 ; see equation ( [ eqt ] ) . hence it suffices to solve coverage problems for @xmath254 and @xmath255 .
in canonical coordinates this implies @xmath256={{\mathbf{0}}}_d$ ] , hence @xmath257 .
for this reason we use the more familiar notation @xmath258 instead of @xmath251 .
the random vector @xmath259 has a @xmath260-dimensional @xmath1-distribution with @xmath82 degrees of freedom , and any linear combination @xmath261 with a unit vector @xmath262 has a one - dimensional @xmath1-distribution .
letting @xmath72 be the adjusted predictors in canonical coordinates , estimates ( [ eqadjusted ] ) and their @xmath1-statistics ( [ eqt ] ) simplify to @xmath263 which are linear functions of @xmath258 and @xmath259 , respectively , with `` posi coefficient vectors '' @xmath264 and @xmath265 that equal @xmath66 up to scale @xmath266 as we now operate in canonical coordinates , we have @xmath267 and @xmath268 , the unit sphere in @xmath250 . to complete the structural description of the posi problem
we let @xmath269 if @xmath270 , we omit the second argument and write @xmath271 . [ propmaxlin ] the posi problem ( [ eqkdef ] ) is equivalent to a @xmath260-dimensional coverage problem for linear functions of the multivariate @xmath1-vector @xmath259 , @xmath272 = { { \mathbf{p}}}\bigl[\max _ { { { \bar{\mathbf{l}}}}\in{\mathcal{l}}({{\mathbf{x}}},{{\mathcal{m } } } ) } \bigl|{{\bar{\mathbf{l}}}}^t{{\mathbf{z}}}/{{\hat{\sigma}}}\bigr| \leq k \bigr ] \stackrel { ( \ge ) } { = } 1 - \alpha.\ ] ] the set @xmath273 of unit vectors @xmath274 has interesting geometric structure which is the subject of this and the next subsection .
the following proposition ( proof in appendix [ appproporth ] ) elaborates the fact that @xmath275 is essentially the predictor vector @xmath55 orthogonalized with regard to the other predictors in the model @xmath38 .
vectors will always be assumed in canonical coordinates and hence @xmath260-dimensional .
[ proporth ] orthogonalities in @xmath273 : the following statements hold assuming that the models referred to are in @xmath134 ( hence are of full rank ) .
adjustment properties : @xmath276 the following vectors form an orthonormal `` gram schmidt '' series : @xmath277 other series are obtained using @xmath278 in place of @xmath279 .
vectors @xmath274 and @xmath280 are orthogonal if @xmath281 , @xmath58 and .
classical case @xmath243 and @xmath270 : each vector @xmath274 is orthogonal to @xmath282 vectors @xmath280 ( not all of which may be distinct ) . the cardinality of orthogonalities in the classical case and @xmath270 is as follows : if the predictor vectors @xmath55 have no orthogonal pairs among them , then @xmath283 .
if there exist orthogonal pairs , then @xmath284 is less .
for example , if there exists exactly one orthogonal pair , then @xmath285 .
when @xmath12 is a fully orthogonal design , then @xmath286 .
coverage problems can be framed geometrically in terms of probability coverage of polytopes in @xmath250 . for the posi problem the polytope with half - width @xmath131
is defined by @xmath287 henceforth called the `` posi polytope . '' the posi coverage problem ( [ eqtcoverage ] ) is equivalent to calibrating @xmath131 such that @xmath288 = 1-\alpha.\ ] ] the simplest case of a posi polytope , for @xmath289 , is illustrated in figure 1 in the online appendix of the supplementary material [ @xcite , section b.7 ] .
more general polytopes are obtained for arbitrary sets @xmath290 of unit vectors , that is , subsets @xmath291 of the unit sphere in @xmath250 .
for the special case @xmath292 the `` polytope '' is the `` scheff ball '' with coverage @xmath293 as @xmath194 : @xmath294 = f_{{{\mathrm{f}}}_{d , r } } \bigl(k^2/d\bigr).\ ] ] many properties of the polytopes @xmath295 are not specific to posi because they hold for polytopes ( [ eqposipolytope ] ) generated by simultaneous inference problems for linear functions with arbitrary sets @xmath290 of unit vectors .
these polytopes : form scale families of geometrically similar bodies : @xmath296 ; are point symmetric about the origin : @xmath297 ; contain the scheff ball : @xmath298 ; are intersections of `` slabs '' of width @xmath299 : @xmath300 have @xmath301 faces ( assuming @xmath302 ) , and each face is tangent to the scheff ball @xmath303 with tangency points @xmath304 @xmath305 .
specific to posi are the orthogonalities described in proposition [ proporth ] . in orthogonal designs , adjustment has no effect : @xmath306 for all @xmath206 , hence @xmath307 and @xmath308 .
the polytope @xmath295 is therefore a hypercube .
this observation implies an optimality property of orthogonal designs if the submodel universes @xmath134 are sufficiently rich to force @xmath273 to contain an orthonormal basis of @xmath250 : the polytope generated by an orthonormal basis is a hypercube ; hence the polytope @xmath309 is contained in this hypercube ; thus @xmath309 has maximal extent if and only if it is equal to this hypercube , which is the case if and only if @xmath273 is this orthonormal basis and nothing more ; that is , @xmath12 is an orthogonal design .
a simple sufficient condition for @xmath134 to grant the existence of an orthonormal basis in @xmath273 is the existence of a maximal nested sequence of submodels such as @xmath310 , @xmath311 in @xmath134 .
it follows according to item ( 2 ) in proposition [ proporth ] that there exists an orthonormal gram
schmidt basis in @xmath273 .
we summarize : [ proporthopt ] among predictor matrices with @xmath312 and model universes @xmath134 that contain at least one maximal nested sequence of submodels , orthogonal designs with @xmath313 columns yield : * the maximal coverage probability @xmath314 $ ] for fixed @xmath131 and * the minimal posi constant @xmath131 satisfying @xmath315 = 1-\alpha$ ] for fixed @xmath316 , @xmath317 .
the proposition holds not only for multivariate @xmath1-vectors and their gaussian limits but for arbitrary spherically symmetric distributions .
optimality of orthogonal designs translates to optimal asymptotic behavior of their constant @xmath318 for large @xmath260 : [ proporthasy ] consider the gaussian limit @xmath194 . for @xmath12 and @xmath134 as in proposition [ proporthopt ] , the asymptotic lower bound for the constant @xmath131 as @xmath319
is attained for orthogonal designs for which the asymptotic rate is @xmath320 by proposition [ proporthopt ] the posi problem is bounded below by orthogonal designs , and by proposition [ propscheffe ] it is loosely bounded above by the scheff ball ( both for all @xmath316 , @xmath260 , and @xmath82 ) .
the question of how close to the scheff bound posi problems can get for @xmath194 will occupy us in section [ secrootp ] . unlike the infimum problem , the supremum problem does not appear to have a unique optimizing design @xmath12 uniformly in @xmath316 , @xmath260 and @xmath82 . in the classical case @xmath243 and @xmath270 there
exists a duality for posi vectors @xmath271 which we will use in section [ secexch ] below but which is also of independent interest .
some preliminaries : letting @xmath321 be the full model , we observe that the ( unnormalized ) posi vectors @xmath322 form the rows of the matrix @xmath323 ; see ( [ eqadjusted ] ) and ( [ eqcontrast ] ) . in a change of perspective
, we interpret the transpose matrix @xmath324 as a predictor matrix , to be called the `` dual design '' of @xmath12 .
it is also of size @xmath325 in canonical coordinates , and its columns are the posi vectors @xmath326 .
it turns out that @xmath327 and @xmath12 pose identical posi problems if @xmath270 : [ thmdual ] @xmath328 .
recall that @xmath271 and @xmath329 contain the normalized versions of the respective adjusted predictor vectors .
the theorem follows from the following lemma which establishes the identities of vectors between @xmath329 and @xmath271 .
we extend obvious notation from @xmath12 to @xmath327 as follows : @xmath330 submodels for @xmath327 will be denoted @xmath331 , but they , too , will be given as subsets of @xmath332 which , however , refer to columns of @xmath327 .
finally , the normalized version of @xmath333 will be written as @xmath334 .
[ lemdual ] for two submodels @xmath38 and @xmath331 that satisfy @xmath335 and @xmath336 , we have @xmath337 the proof is in appendix [ applemdual ] .
the assertion about norms is really only needed to exclude collapse of @xmath338 to zero .
a special case arises when the predictor matrix ( in canonical coordinates ) is chosen to be symmetric according to proposition [ propcancoord](7 ) : if @xmath339 , then @xmath340 , and hence : [ symdualk ] if @xmath12 is symmetric in canonical coordinates , then @xmath341
we consider examples in the classical case @xmath243 and @xmath270 .
also , we work with the gaussian limit @xmath187 , that is , @xmath24 known , and w.l.o.g . @xmath342 . in exchangeable designs
all pairs of predictor vectors enclose the same angle . in
canonical coordinates a convenient parametrization of a family of symmetric exchangeable designs is @xmath343 where @xmath344 , and @xmath345 is a matrix with all entries equal to @xmath346 .
the range restriction on @xmath347 assures that @xmath348 is positive definite .
we will write @xmath349 depending on which parameter matters in a given context .
we will make use of the fact that @xmath350 is also an exchangeable design .
the function @xmath351 maps the interval @xmath352 onto itself , and it holds @xmath353 , @xmath354 as @xmath355 , and vice versa .
exchangeable designs include orthogonal designs for @xmath356 , and they extend to two types of strict collinearities : for @xmath357 the predictor vectors collapse to a single dimension @xmath358 , and for @xmath359 they collapse to a subspace @xmath360 of dimension @xmath361 , where @xmath362 .
as collinearity drives the fracturing of the regression coefficients into model - dependent quantities @xmath61 , it is of interest to analyze @xmath363 as @xmath364 moves from orthogonality at @xmath356 toward either of the two types of collinearity . here is what we find : unguided intuition might suggest that the collapse to rank 1 calls for larger @xmath186 than the collapse to rank @xmath151 .
this turns out to be entirely wrong : collapse to rank 1 or rank @xmath151 has identical effects on @xmath186 .
the reason is duality ( section [ secdual ] ) : for exchangeable designs , @xmath364 collapses to rank 1 if and only if @xmath365 collapses to rank @xmath151 , and vice versa , while @xmath366 according to corollary [ symdualk ] .
we next address the asymptotic behavior of @xmath367 for increasing @xmath4 . as noted in section [ secscheffe ] ,
there is a wide gap between orthogonal designs with @xmath368 and the full scheff protection with @xmath369 .
the following theorem shows how exchangeable designs fall into this gap : [ thmex ] posi constants of exchangeable design matrices @xmath370 [ defined in ( [ eqex ] ) above ] have the following limiting behavior : @xmath371 the proof can be found in appendix [ appthmex ] .
the theorem shows that for exchangeable designs the posi constant remains much closer to the orthogonal case than the scheff case .
thus , for this family of designs it is possible to improve on the scheff constant by a considerable margin .
the following detail of geometry for exchangeable designs has a bearing on their posi constants : the angle between pairs of predictor vectors as a function of @xmath347 is @xmath372 .
as the vectors fall into the rank-@xmath373 collinearity at @xmath359 , the cosine becomes @xmath374 , which converges to zero as @xmath375 .
thus , as @xmath376 , exchangeable designs approach orthogonal designs even at their most collinear extreme . for further illustrative materials related to exchangeable designs ,
see figures 2 and 3 in the online appendix of the supplementary material [ @xcite , section b.7 ] .
the following is a situation in which the asymptotic upper bound for @xmath377 is @xmath6 , hence equal to the rate of the scheff constant @xmath378 .
perhaps surprisingly , it is sufficient to consider posi1 ( section [ secposi1 ] ) whose constant is dominated by that of full posi .
let the posi1 predictor of interest be @xmath379 , so the search is over all models @xmath380 , but inference is sought only for @xmath381 . consider the following upper triangular @xmath325 design matrix in canonical coordinates : @xmath382 where @xmath383 is the primary predictor , and the canonical basis vectors @xmath384 are the controls .
the vector @xmath385 has unit length , and hence the parameter @xmath386 is the correlation between the primary predictor and the controls .
it is constrained to @xmath387 , so @xmath388 has full rank .
for @xmath389 the primary predictor @xmath385 becomes fully collinear with the controls , and it is on the approach to this boundary where the rate of the following theorem is attained : [ thmrootp ] for @xmath24 known , the designs ( [ eqppcdesign ] ) have posi__1 _ _ constants @xmath390 with the following asymptotic rate : @xmath391 the proof is in appendix [ appthmrootp ] . as @xmath392 the theorem provides a lower bound on the rate of the full posi constant .
the value 0.6363 is not maximal , and we have indications that the supremum over all designs may exceed 0.78 . together with the upper bound of corollary [ corbound ] , this would provide a narrow asymptotic range for worst - case posi .
most importantly , the example shows that for some designs posi constants can be much larger than the @xmath393 @xmath161-quantiles used in common practice .
the following is a rough asymptotic upper bound on all posi constants @xmath394 .
it has the scheff rate but with a multiplier that is strictly less than scheff s .
the bound is loose because it ignores the rich structure of the sets @xmath273 ( section [ secorthgeo ] ) and only uses their cardinality @xmath395 ( @xmath31 in the classical case @xmath243 and ) .
[ thmbound ] denote by @xmath396 arbitrary finite sets of @xmath260-dimensional unit vectors , @xmath397 , such that @xmath398 where @xmath399 .
denote by @xmath400 the @xmath401-quantile of @xmath402 .
then the following describes an asymptotic worst - case bound for @xmath400 and its attainment : @xmath403 the proof of theorem [ thmbound ] ( see appendix [ appthmbound ] ) is an adaptation of wyner s ( @xcite ) techniques for sphere packing and sphere covering . the worst - case bound ( @xmath404 )
is based on a surprisingly crude bonferroni - style inequality for caps on spheres .
attainment of the bound ( @xmath405 ) makes use of the artifice of picking the vectors @xmath406 randomly and independently .
applying the theorem to posi sets @xmath407 in the classical case @xmath243 , we have @xmath408 , hence @xmath409 , so the theorem applies with @xmath410 : [ corbound ] in the classical case @xmath16 a universal asymptotic upper bound for the posi constant @xmath411 is @xmath412 the corollary shows that the asymptotic rate of the posi constant , if it reaches the scheff rate , will always have a multiplier that is strictly below that of the scheff constant .
we do not know whether there exist designs for which the bound of the corollary is attained , but the theorem says the bound is sharp for unstructured sets @xmath290 .
we investigated the post - selection inference or `` posi '' problem for linear models whereby valid statistical tests and confidence intervals are sought after variable selection , that is , after selecting a subset of the predictors in a data - driven way .
we adopted a framework that does _ not _ assume any of the linear models under consideration to be correct .
we allowed the response vector to be centered at an arbitrary mean vector but with homoscedastic and gaussian errors .
we further allowed the full predictor matrix @xmath413 to be rank - deficient , @xmath414 , and we also allowed the set @xmath134 of models @xmath38 under consideration to be largely arbitrary . in this framework
we showed that valid post - selection inference is possible via simultaneous inference .
an important enabling principle is that submodels have their own regression coefficients ; put differently , @xmath61 and @xmath415 are generally different parameters if @xmath416 .
we showed that simultaneity protection for all parameters @xmath61 provides valid post - selection inference . in practice
this means enlarging the constant @xmath417 used in conventional inference to a constant @xmath418 that provides simultaneity protection for up to @xmath31 parameters @xmath61 .
we showed that the constant depends strongly on the predictor matrix @xmath12 as the asymptotic bound for @xmath173 with @xmath419 ranges between the minimum of @xmath420 achieved for orthogonal designs on the one hand , and a large fraction of the scheff bound @xmath195 on the other hand .
this wide asymptotic range suggests that computation is critical for problems with large numbers of predictors . in the classical case
@xmath243 our current computational methods are feasible up to about @xmath7 .
we carried out post - selection inference in a limited framework .
several problems remain open , and many natural extensions are desirable : * among open problems is the quest for the largest fraction of the asymptotic scheff rate @xmath195 attained by posi constants .
so far we know this fraction to be at least 0.6363 , but no more than 0.8660 in the classical case @xmath243 .
when the size of models @xmath421 is limited as a function of @xmath4 ( `` sparse models '' ) , better rates can be achieved , and we will report these results elsewhere . *
computations for @xmath422 are a challenge .
straight enumeration of the set of up to @xmath31 linear combinations should be replaced with heuristic shortcuts that yield practically useful upper bounds on @xmath423 , @xmath134 , @xmath424 that are specific to @xmath12 and the set of submodels @xmath134 , unlike the 0.8660 fraction of the scheff bound which is universal .
* situations to which the posi framework should be extended include generalized linear models , mixed effects models , models with random predictors , as well as prediction problems .
results for the last two situations will be reported elsewhere .
* it would be desirable to devise post - selection inference for specific selection procedures for cases in which a strict model selection protocol is being adhered to .
@xmath425 code for computing the posi constant for up to @xmath426 can be obtained from the authors web pages ( a manuscript describing the computations is available from the authors ) .
\(1 ) the matrix @xmath427 has the vectors @xmath264 as its columns .
thus @xmath428 .
orthogonality @xmath429 for @xmath430 follows from @xmath431 .
the same properties hold for the normalized vectors @xmath274 .
\(2 ) the vectors @xmath432 form a gram
schmidt series with normalization , hence they are an o.n .
basis of @xmath433 .
\(3 ) for @xmath281 , @xmath58 , @xmath434 , we have @xmath435 because they can be embedded in an o.n .
basis by first enumerating @xmath38 and subsequently @xmath436 , with @xmath114 being last in the enumeration of @xmath38 and @xmath437 last in the enumeration of @xmath438 .
\(4 ) for any @xmath439 , @xmath440 , there are @xmath441 ways to choose a partner @xmath442 such that either @xmath443 or @xmath444 , both of which result in @xmath445 by the previous part .
the proof relies on a careful analysis of orthogonalities as described in proposition [ proporth ] , part ( 3 ) . in what follows we write
@xmath446 $ ] for the column space of a matrix @xmath447 , and @xmath446^\perp$ ] for its orthogonal complement .
we show first that , for @xmath448 , @xmath336 , the vectors @xmath449 and @xmath274 are in the same one - dimensional subspace , hence are a multiple of each other . to this end
we observe : @xmath450,\qquad { { \bar{\mathbf{l}}}}_{j \cdot{\mathrm{m } } } \in [ { { \mathbf{x}}}_{{\mathrm{m}}\setminus j}]^\perp , & \\ & \displaystyle { { \bar{\mathbf{l}}}}^*_{j \cdot{\mathrm{m}}^ * } \in\bigl[{{\mathbf{x}}}^*_{{\mathrm{m}}^*}\bigr],\qquad { { \bar{\mathbf{l}}}}^*_{j \cdot{\mathrm{m}}^ * } \in\bigl[{{\mathbf{x}}}^*_{{\mathrm{m}}^ * \setminus j}\bigr]^\perp , & \\ \label{dualproof3 } & \displaystyle \bigl[{{\mathbf{x}}}^*_{{\mathrm{m}}^*}\bigr ] = [ { { \mathbf{x}}}_{{\mathrm{m}}\setminus j}]^\perp,\qquad \bigl[{{\mathbf{x}}}^*_{{\mathrm{m}}^ * \setminus j}\bigr]^\perp= [ { { \mathbf{x}}}_{{\mathrm{m}}}].&\end{aligned}\ ] ] the first two lines state that @xmath274 and @xmath449 are in the respective column spaces of their models , but orthogonalized with regard to all other predictors in these models . the last line , which can also be obtained from the orthogonalities implied by @xmath451 ,
establishes that the two vectors fall in the same one - dimensional subspace , @xmath452 \cap[{{\mathbf{x}}}_{{\mathrm{m}}\setminus j}]^\perp = \bigl[{{\mathbf{x}}}^*_{{\mathrm{m}}^*}\bigr ] \cap\bigl[{{\mathbf{x}}}^*_{{\mathrm{m}}^ * \setminus j } \bigr]^\perp\ni{{\bar{\mathbf{l}}}}^*_{j \cdot{\mathrm{m}}^*}.\ ] ] since they are normalized , it follows @xmath453 .
this result is sufficient to imply all of theorem [ thmdual ] .
the lemma , however , makes a slightly stronger statement involving lengths which we now prove . in order to express @xmath264 and @xmath333
according to ( [ eqposivecs ] ) , we use @xmath454 as before and we write @xmath455 for the analogous projection onto the space spanned by the columns @xmath456 of @xmath327 .
the method of proof is to evaluate @xmath457 .
the main argument is based on @xmath458 which follows from these facts : @xmath459 which in turn are consequences of ( [ dualproof3 ] ) and @xmath460 .
we also know from ( [ eqposivecs ] ) that @xmath461 putting together ( [ dualproof5 ] ) , ( [ dualproof6 ] ) and ( [ eqposivecs ] ) , we obtain @xmath462 because the two vectors are scalar multiples of each other , we also know that @xmath463 putting together ( [ dualproof7 ] ) and ( [ dualproof8 ] ) we conclude @xmath464 this proves the lemma and the theorem .
the parameter @xmath347 in equation ( [ eqex ] ) can range from @xmath465 to @xmath466 , but because of duality there is no loss of generality in considering only the case in which @xmath467 , and we do so in the following .
let @xmath468 and @xmath58 .
consider first the case @xmath469 , hence @xmath470 : we have @xmath471 , the @xmath114th column of @xmath12 , and @xmath472 . for any @xmath473 it follows=1 @xmath474=0 consider next the case @xmath475 , and for notational convenience let @xmath476 and @xmath477 where @xmath478 .
the following results can then be applied to arbitrary @xmath38 and @xmath58 by permuting coordinates . the projection of @xmath479 on the space spanned by @xmath480 must be of the form @xmath481 where the constant @xmath386 satisfies @xmath482 .
this follows from symmetry , and no calculation of projection matrices is needed to verify this .
let @xmath483 .
then @xmath484 some algebra starting from @xmath485 yields @xmath486 the term @xmath487 is nonnegative , maximal wrt @xmath488 for @xmath489 , and thereafter maximal wrt @xmath347 for @xmath490 , whence @xmath491 and finally @xmath492 this fact will make the term @xmath487 in ( [ eqbdexl ] ) asymptotically irrelevant . using and @xmath493 as well as ( [ eqbdexl ] ) and ( [ eqbdexda ] ) we obtain @xmath494\\[-8pt ] & \le & \|{{\mathbf{z}}}\|_\infty+ \|{{\mathbf{z}}}\|_\infty+ \biggl{\vert}\frac{1}{2 \sqrt{p } } \sum_{j=1}^p z_j \biggr{\vert}.\nonumber\end{aligned}\ ] ] combining ( [ eqbdexs1 ] ) and ( [ eqbdexs2 ] ) we obtain for @xmath495 the following :
@xmath496 this verifies that @xmath497 it remains to prove that equality holds in ( [ eqbdexineq ] ) . to this end
let @xmath498 denote the order statistics of @xmath499 , @xmath500 .
fix @xmath488 .
we have in probability @xmath501 note that @xmath502 for a given @xmath258 we choose @xmath503 such that @xmath504 is the index of @xmath505 and @xmath506 includes @xmath507 as well as the set of indices of @xmath508 for @xmath509 . from ( [ eqbdexl ] ) we then obtain in probability @xmath510 choosing @xmath488 arbitrarily large and combining this with ( [ eqbdexineq ] ) yields the desired conclusion .
recall from ( [ eqppcdesign ] ) the designs @xmath511 where @xmath512 is the primary predictor .
the matrix @xmath12 will be treated according to posi1 ( section [ secposi1 ] ) , and hence we will examine the distribution of @xmath513 ( assuming @xmath342 known ) .
we determine @xmath514 for a fixed model @xmath38 ( @xmath4 ) with @xmath515 , @xmath516 therefore , @xmath517 for fixed @xmath488 we can explicitly maximize the sum on the right - hand side , @xmath518 where @xmath519 is the @xmath114th order statistic of @xmath499 , @xmath520 , omitting @xmath521 .
we can also explicitly maximize the factor @xmath522 in ( [ eqrootp1 ] ) , @xmath523 and equality is attained as @xmath524 .
therefore , for fixed @xmath488 , we can continue from ( [ eqrootp1 ] ) as follows : @xmath525 the reason for writing the two sums in this manner is that we will interpret them as approximations to riemann sums . to this end
we borrow from @xcite the following approximations for @xmath526 : @xmath527 reparametrizing @xmath528 , the anticipated riemann approximation is @xmath529 therefore , @xmath530 and similarly @xmath531 summarizing , @xmath532 the function @xmath533 is maximized at @xmath534 with@xmath535 .
therefore , @xmath536 the bound is sharp because it is attained by the models that include the first or last @xmath537 order statistics of @xmath258 when @xmath538 and @xmath539 . from ( [ eqk1 ] )
we conclude that @xmath540 .
we show that if @xmath541 , then : * we have a uniform asymptotic worst - case bound , @xmath542 * which is attained when @xmath543 and @xmath544 are i.i.d .
@xmath545 independent of @xmath258 , @xmath546 these facts imply the assertions about @xmath401-quantiles @xmath547 of@xmath548 in theorem [ thmbound ] .
we decompose @xmath549 where @xmath550 and @xmath551 are independent . due to
@xmath552 it is sufficient to show the following : * uniform asymptotic worst - case bound , @xmath553 * attainment of the bound when @xmath554 and @xmath544 are i.i.d .
@xmath545 independent of @xmath555 , @xmath556 to show ( [ equbound ] ) , we upper - bound the noncoverage probability and show that it converges to zero for @xmath557 . to this end
we start with a bonferroni - style bound , as in @xcite , @xmath558 & = & { { \mathbf{p}}}\bigcup_{{{\bar{\mathbf{l}}}}\in{\mathcal{l } } } \bigl [ \bigl|{{\bar{\mathbf{l}}}}^t { { \mathbf{u}}}\bigr| > k ' \bigr ] \nonumber \\ & \le & \sum_{{{\bar{\mathbf{l}}}}\in{\mathcal{l } } } { { \mathbf{p}}}\bigl [ \bigl|{{\bar{\mathbf{l}}}}^t { { \mathbf{u}}}\bigr| > k'\bigr ] \\ & = & |{\mathcal{l}}_p| { { \mathbf{p}}}\bigl [ |u| > k'\bigr],\nonumber\end{aligned}\ ] ] where @xmath559 is any coordinate of @xmath555 or projection of @xmath555 onto a unit vector .
we will show that bound ( [ eqbonf ] ) converges to zero .
we use the fact that @xmath560 , hence @xmath561 = \frac{1}{{{\mathrm{b}}}(1/2,(p-1)/2 ) } \int _ { k'^2}^1 x^{-1/2 } ( 1-x)^{(p-3)/2 } \,dx.\ ] ] we bound the beta function and the integral separately , @xmath562 where we used @xmath563 ( a good approximation , really ) and @xmath564 .
@xmath565 where we used @xmath566 on the integration interval . continuing with the chain of bounds from ( [ eqbonf ] ) we have @xmath567 \le\frac{1}{k ' }
\biggl(\frac{2}{(p-1)\pi } \biggr)^{1/2 } \bigl ( |{\mathcal{l}}_p|^{1/(p-1 ) } \sqrt{1-k'^2 } \bigr)^{p-1}.\ ] ] since @xmath568 , the right - hand side converges to zero at geometric speed if @xmath569 , that is , if @xmath557 .
this proves ( [ equbound ] ) . to show ( [ eqlbound ] ) , we upper - bound the coverage probability and show that it converges to zero for @xmath570 .
we make use of independence of @xmath544 , as in @xcite , @xmath571 & = & \prod_{{{\bar{\mathbf{l}}}}\in{\mathcal{l}}_p } { { \mathbf{p}}}\bigl [ \bigl| { { \bar{\mathbf{l}}}}^t { { \mathbf{u}}}\bigr| \le k ' \bigr ] = { { \mathbf{p}}}\bigl [ | u | \le k ' \bigr]^{|{\mathcal{l}}_p| } \nonumber \\ & = & \bigl(1 - { { \mathbf{p}}}\bigl [ | u | > k ' \bigr ] \bigr)^{|{\mathcal{l}}_p| } \\ & \le & \exp\bigl ( - |{\mathcal{l}}_p| { { \mathbf{p}}}\bigl [ | u | > k ' \bigr ] \bigr).\nonumber\end{aligned}\ ] ] we will lower - bound the probability @xmath572 $ ] recalling ( [ eqbeta ] ) and again deal with the beta function and the integral separately , @xmath573 where we used @xmath574 ( again , a good approximation ) .
@xmath575 where we used @xmath576 . putting it all together we bound the exponent in ( [ eqexp ] ) , @xmath567 \ge\frac{\sqrt{p/2 - 3/4}}{\sqrt{\pi } ( p-1)/2 } \bigl ( |{\mathcal{l}}_p|^{1/(p-1 ) } \sqrt{1-k'^2 } \bigr)^{p-1}.\ ] ] since @xmath568 , the right - hand side converges to @xmath577 at nearly geometric speed if @xmath578 , that is , @xmath570 .
this proves ( [ eqlbound ] ) .
@xmath579we thank e. candes , l. dicker , m. freiman , e. george , a. krieger , m. low , z. ma , e. pitkin , l. shepp , n. sloane , p. shaman and m. traskin for very helpful discussions .
the acronym `` spar '' is due to m. freiman .
we are indebted to an anonymous reviewer for extensive and constructive criticism that influenced the positioning of this article . | it is common practice in statistical data analysis to perform data - driven variable selection and derive statistical inference from the resulting model .
such inference enjoys none of the guarantees that classical statistical theory provides for tests and confidence intervals when the model has been chosen a priori .
we propose to produce valid `` post - selection inference '' by reducing the problem to one of simultaneous inference and hence suitably widening conventional confidence and retention intervals .
simultaneity is required for all linear functions that arise as coefficient estimates in all submodels . by purchasing `` simultaneity insurance '' for all possible submodels ,
the resulting post - selection inference is rendered universally valid under all possible model selection procedures .
this inference is therefore generally conservative for particular selection procedures , but it is always less conservative than full scheff protection .
importantly it does _ not _ depend on the truth of the selected submodel , and hence it produces valid inference even in wrong models .
we describe the structure of the simultaneous inference problem and give some asymptotic results . , , , + |
the study of non - zero degree maps between closed , oriented manifolds has become very active over the last few decades @xcite .
the existence of a non - zero degree map , @xmath1 , defines a transitive relation on the set of homotopy types of closed , oriented manifolds .
whenever such a map exists we say that @xmath2 dominates @xmath3 and write @xmath4 . in this case
@xmath2 is at least as complicated as @xmath3 .
for example , the induced maps in rational homology are surjective , thus , in particular , the betti numbers of @xmath3 are bounded from above by those of @xmath2 .
also , @xmath4 implies that the fundamental group of @xmath2 surjects onto a finite index subgroup of the fundamental group of @xmath3 . in dimension two ,
the domination relation coincides with the ordering given by the genus , but in higher dimensions it fails to be an ordering .
we illustrate this by the following two examples in dimension three .
the examples have obvious generalizations to higher dimensions .
the two three - manifolds @xmath5 and @xmath6 satisfy @xmath7 and @xmath8 , but fail to be homotopy equivalent .
let @xmath2 be a hyperbolic homology three - sphere , and @xmath9 .
then @xmath3 has larger first betti number than @xmath2 , and so @xmath10
. we also have @xmath11 since the fundamental group of @xmath3 is infinite cyclic , and so can not surject onto the fundamental group of a closed negatively curved manifold , for example by preissmann s theorem .
thus @xmath2 and @xmath3 are not comparable under the domination relation . in this paper ,
we study domination by products for three - manifolds .
this is motivated by the work of lh and the first author in @xcite , where strong restrictions were found for certain manifolds with large universal coverings to be dominated by products .
in fact , the results of those papers show that three - manifolds dominated by products can not have hyperbolic or @xmath12-geometry , and must often be prime .
however , in this paper we will not use those earlier results , but follow a more direct approach .
this is possible since in dimension three the only product manifolds are those with a circle factor , and this gives much stronger constraints than the consideration of arbitrary products .
the main result we prove here is the following : [ t : topargument ] a closed , oriented , connected three - manifold @xmath3 is dominated by a product @xmath13 if and only if 1 .
either @xmath3 is finitely covered by a product @xmath14 , for some aspherical surface @xmath15 , or 2 .
@xmath3 is finitely covered by a connected sum @xmath16 . as usual ,
the empty connected sum corresponding to @xmath17 is @xmath18 .
the proof of theorem [ t : topargument ] falls naturally into two parts . on the one hand
, we have to prove that all three - manifolds not listed in the statement of the theorem can not be dominated by products . on the other hand
, we have to prove that the manifolds listed in the theorem are indeed dominated by products .
this is obvious for manifolds finitely covered by products , but it is not obvious for the connected sums occurring in the second statement
. here the proof proceeds by constructing certain maps of non - zero degrees as branched coverings .
this construction , which also has a high - dimensional generalization , is of independent interest .
previously , many non - trivial results have been proved about the domination relation in dimension three using a variety of tools different from the ones we use here , such as thurston s geometries , gromov s simplicial volume , and the seifert volume .
a survey of the state of the art at the beginning of the last decade is given in @xcite .
for more recent results , especially related to the issue of finiteness of sets of mapping degrees between three - manifolds , see for example @xcite and the papers quoted there .
our proofs here are independent of this earlier work , and in fact clarify certain claims made there , cf . subsection [ ss : w ] below .
it is not immediately obvious to what extent theorem [ t : topargument ] really depends on the assumption that the domains of our dominant maps are products , and one could try to replace these products by fibered three - manifolds . for this purpose surface bundles over the circle are not interesting , since every three - manifold is dominated by such a bundle by a result of sakuma @xcite .
however , considering non - trivial circle bundles over surfaces we obtain a result parallel to theorem [ t : topargument ] : [ t : topbundle ] a closed , oriented , connected three - manifold @xmath3 is dominated by a non - trivial circle bundle over a surface if and only if 1 .
either @xmath3 is finitely covered by a non - trivial circle bundle over some aspherical surface , or 2 .
@xmath3 is finitely covered by a connected sum @xmath16 . in section [ s : ess ]
we discuss the notion of rational essentialness in the case of three - manifolds . while this is not logically necessary for the proofs of our main results , we find it convenient , following @xcite , to use this concept as an organizing principle . in section [ s : proof1 ] , respectively section [ s : proof2 ] , we then prove theorems [ t : topargument ] and [ t : topbundle ] for rationally essential , respectively inessential , three - manifolds . in section [ s : algargum ] we reformulate these theorems in terms of thurston geometries and in purely algebraic terms .
finally , in section [ s : groups ] we determine the three - manifold groups presentable by products , and in section [ s : dis ] we make some further remarks .
these last two sections contain two new characterizations of ( aspherical ) seifert manifolds .
the obstructions for domination by products found in @xcite are applicable to rationally essential manifolds in the sense of the following definition going back to gromov @xcite : a closed , oriented , connected @xmath19-manifold @xmath3 is called rationally essential if @xmath20)\neq 0\in h_n(b\pi_1(n);{\mathbb{q } } ) \ , \ ] ] where @xmath21 classifies the universal covering of @xmath3 . for three - manifolds
, this definition can be interpreted in terms of the kneser - milnor prime decomposition @xcite .
recall that this says that a closed oriented connected three - manifold @xmath3 has an essentially unique prime decomposition @xmath22 under the connected sum operation .
each prime summand @xmath23 is either aspherical , is @xmath24 , or has finite fundamental group .
we now have the following : [ t : ess ] for a closed oriented connected three - manifold @xmath3 the following conditions are equivalent : 1 .
@xmath3 is rationally essential , 2 .
@xmath3 has an aspherical summand @xmath23 in its prime decomposition , 3 .
@xmath3 is not finitely covered by a connected sum @xmath16 , 4 .
@xmath3 is compactly enlargeable , 5 .
@xmath3 does not admit a metric of positive scalar curvature .
the last two items are not relevant to the main results of this paper , so we will only discuss them briefly . a connected sum is rationally essential if and only if at least one of the summands is . since @xmath24 and manifolds with finite fundamental group are not rationally essential , this proves the equivalence of ( @xmath25ess ) and ( asph ) .
it is obvious that ( asph ) implies ( nfree ) . for the converse
assume that @xmath3 contains no aspherical summands in its prime decomposition , i.e. that @xmath3 has the form @xmath26 where the empty connected sum ( @xmath27 ) denotes the 3-sphere @xmath18 .
the summands @xmath28 have infinite cyclic fundamental groups and the summands @xmath29 have finite fundamental groups @xmath30 , @xmath31 .
thus , the fundamental group of @xmath3 is the free product @xmath32 where @xmath33 is a free group on @xmath34 generators .
we project this free product to the direct product of the @xmath35 to obtain the following exact sequence : @xmath36 by the kurosh subgroup theorem , @xmath37 is a free group @xmath38 .
since it has finite index in @xmath39 , we see that @xmath3 has a finite covering whose fundamental group is free . by kneser s
prime decomposition theorem and grushko s theorem , we deduce that this covering is a connected sum of @xmath19 copies of @xmath28 .
this means that @xmath3 is finitely covered by a connected sum @xmath16 , where @xmath19 is the number of generators of the free group @xmath40 in the exact sequence ( [ eq.1 ] ) . to see that ( asph ) implies ( enl ) it is enough to show that any aspherical three - manifold @xmath3 is compactly enlargeable
this was proved by gromov and lawson ( * ? ? ?
* theorem 6.1 ) under the assumptions that @xmath39 is residually finite and contains an infinite surface group .
it is now known that all three - manifold groups are residually finite @xcite .
( this reference treats only manifolds satisfying thurston s geometrisation conjecture , which has now been verified by perelman @xcite . ) furthermore , if @xmath39 contains no infinite surface group , then @xmath3 is atoroidal , and so is hyperbolic by perelman s work @xcite .
since hyperbolic manifolds are compactly enlargeable by ( * ? ? ?
3.3 ) , we conclude that ( asph ) implies ( enl ) .
gromov and lawson ( * ? ? ?
* theorem 3.7 ) proved that ( enl ) implies ( npsc ) .
( recall that all oriented three - manifolds are spin . ) finally , ( npsc ) implies ( asph ) because @xmath24 has positive scalar curvature , and so do all three - manifolds with finite fundamental group by perelman s proof of the poincar conjecture @xcite
. a connected sum of manifolds with positive scalar curvature also has positive scalar curvature by the construction of gromov and lawson ; cf .
* theorem 5.4 ) .
it was proved by hanke and schick @xcite that ( enl ) implies ( @xmath25ess ) in all dimensions .
the converse is not true in dimensions @xmath41 by a recent result of brunnbauer and hanke @xcite .
in view of theorem [ t : ess ] , the proofs of theorems [ t : topargument ] and [ t : topbundle ] split into two cases , depending on whether @xmath3 contains an aspherical summand @xmath23 in its prime decomposition , or not . in this section
we deal with the case where an aspherical summand does appear .
the first part of theorem [ t : topargument ] corresponds to the following statement : [ p : ess ] a rationally essential closed oriented three - manifold @xmath3 is dominated by a product if and only if it is finitely covered by a product @xmath14 , with @xmath15 an aspherical surface .
a manifold finitely covered by a product is of course dominated by that product . for the converse assume that a product @xmath13 of a closed , oriented , connected surface @xmath42 with the circle dominates a closed , oriented , connected rationally essential three - manifold @xmath3 and let @xmath43 be a map of non - zero degree
. then @xmath13 must be rationally essential , and so @xmath42 is of positive genus . by replacing @xmath3 by a finite covering if necessary
, we may assume that @xmath44 is @xmath45-surjective .
let @xmath22 be the kneser - milnor prime decomposition of @xmath3 .
each prime summand @xmath23 is either aspherical , is @xmath24 , or has finite fundamental group ; cf . @xcite . by theorem [ t : ess ]
the rational essentialness of @xmath3 is equivalent to the existence of an aspherical summand @xmath23 .
composing @xmath44 with the degree one map @xmath46 collapsing the connected summands other than @xmath23 , we obtain a dominant map @xmath47 between aspherical three - manifolds
. this can not factor through @xmath42 , implying that @xmath48 must be non - trivial on the central @xmath49-subgroup generated by the @xmath50 factor .
but then @xmath51 is a non - trivial central subgroup in @xmath39 , and so this group is freely indecomposable .
thus we may assume that @xmath3 itself is prime and aspherical , for we can either appeal to perelman s proof of the poincar conjecture @xcite to conclude @xmath52 , or we can argue that the assumption @xmath53 depends only on the homotopy type of @xmath3 , which does not change if we replace a manifold by its connected sum with a homotopy sphere .
we have shown that @xmath3 is aspherical , and that its fundamental group has infinite center .
if @xmath3 is haken , then it follows from a result of waldhausen @xcite that @xmath3 is seifert fibered .
in fact , even without the haken condition , @xmath3 must be seifert fibered , by the proof of the seifert fiber space conjecture ( stated in @xcite and proved by casson
jungreis @xcite and gabai @xcite ) .
therefore , after lifting @xmath44 to a suitable covering space , we may assume that @xmath3 is a circle bundle over an aspherical surface .
it remains to show that the euler number of this circle bundle is zero .
we will prove this in the following lemma , thereby completing the proof of proposition [ p : ess ] .
[ lem ] let @xmath54 be an oriented circle bundle with non - zero euler number over a closed aspherical surface
. then every continuous map @xmath55 has degree zero .
since @xmath3 is aspherical , we may assume that @xmath42 has positive genus .
from the discussion above we may assume that @xmath56 is an element of infinite order in the center of @xmath39 . since on a seifert manifold elements of the center of the fundamental group are , up to taking multiples , fibers of seifert fibrations , cf .
@xcite and @xcite , we may assume that @xmath56 is a multiple of the homotopy class of the fiber in @xmath3 .
( the fibration of @xmath3 is unique , cf .
thus the composition @xmath57 kills the homotopy class of the @xmath50-factor in @xmath13 .
since @xmath15 is aspherical , this implies that @xmath57 is homotopic to a map that factors through the projection @xmath58 . by the homotopy lifting property of @xmath54
, the homotopy of @xmath57 can be lifted to a homotopy of @xmath44 , so we may assume that @xmath59 for some continuous map @xmath60 .
since @xmath54 has non - zero euler number , @xmath61 induces the zero map on @xmath62 , and the fundamental class of @xmath15 is not in the image .
as @xmath45 is surjective on @xmath63 , the equation @xmath59 shows that @xmath64 .
now consider the pullback of @xmath3 under @xmath65 : @xmath66 the map @xmath55 factors through @xmath67 as follows : @xmath68 for any pullback of an oriented bundle , the degree of the map between total spaces is the same as the degree of the map of base spaces under which the bundle is pulled back . in our situation
this says that the degree of @xmath69 equals the degree of @xmath65 , which vanishes .
thus @xmath44 factors through a degree zero map , and we finally have @xmath70 .
the next proposition covers the first part of theorem [ t : topbundle ] .
[ p : essbd ] a rationally essential closed oriented three - manifold @xmath3 is dominated by a non - trivial circle bundle over a surface if and only if it is finitely covered by a non - trivial circle bundle over some aspherical surface .
let @xmath71 be a map of non - zero degree , with @xmath2 a non - trivial circle bundle over a surface @xmath72 of genus @xmath73 .
after replacing @xmath3 by a suitable covering , we may assume that @xmath44 is @xmath45-surjective .
since @xmath3 is assumed to be rationally essential , @xmath74 must be infinite , and so @xmath75 .
this means that @xmath2 is aspherical and we have a non - trivial central extension @xmath76 the prime decomposition of @xmath3 contains an aspherical summand @xmath23 by theorem [ t : ess ] .
composing @xmath44 with the degree one map @xmath46 collapsing the connected summands other than @xmath23 , we obtain a dominant map @xmath77 between aspherical three - manifolds .
this can not factor through @xmath72 , implying that @xmath48 must be non - trivial on the central @xmath49-subgroup generated by the circle fibers in @xmath2 .
but then @xmath51 is a non - trivial central subgroup in @xmath39 , and so @xmath3 is prime and therefore irreducible and aspherical itself .
as in the proof of proposition [ p : ess ] we conclude that @xmath3 is seifert fibered . after replacing @xmath2 and @xmath3 by suitable coverings , we may assume that @xmath3 is also a circle bundle .
it remains to prove that it has non - trivial euler class .
now @xmath48 sends the element of @xmath74 represented by the circle fibers in @xmath2 to a non - trivial element of the center of @xmath39 .
this group is torsion - free , so this non - trivial element has infinite order .
some multiple of it is the fiber of a seifert fibration of @xmath3 , cf . @xcite . as mentioned before
, we may assume that this seifert fibration is a circle bundle .
since the fiber in @xmath2 has finite order in homology because the euler class of @xmath2 was non - zero , it follows that the circle fiber in @xmath3 , being , up to taking multiples , the image under @xmath78 of the circle fiber in @xmath2 , also has finite order in homology , and so the euler class of @xmath3 must be non - zero .
in this section we prove theorems [ t : topargument ] and [ t : topbundle ] in the case of rationally inessential manifolds , i. e. those with no aspherical summand in their prime decomposition .
the proof is constructive , exhibiting certain dominant maps as branched coverings .
the second part of theorem [ t : topargument ] corresponds to the following statement : [ p : iness ] every rationally inessential three - manifold is dominated by a product .
since we have shown in the proof of theorem [ t : ess ] that rationally inessential three - manifolds are finitely covered by connected sums of copies of @xmath24 , it suffices to prove the following : [ p : branchcov ] let @xmath79 be a closed , oriented surface of genus @xmath19 . for every @xmath19 the manifold @xmath80 is a @xmath45-surjective branched double covering of @xmath16 .
@xmath81 at 300 87 @xmath82 at 590 87 .,title="fig:",width=453 ] the @xmath83-torus @xmath84 is a branched double covering of @xmath85 with four branch points .
we denote this branched covering , which is the quotient map for an involution on @xmath84 , by @xmath86 ; see figure [ f : branchcov1 ] .
( the letter @xmath87 stands either for `` pillowcase '' , or for the weierstrass @xmath88-function . ) we multiply @xmath87 by the identity map on @xmath50 to obtain a branched double covering @xmath89 this is the case @xmath90 in the claim . at 251 143 , with two branched points , is an annulus in @xmath84.,title="fig:",width=302 ] now let @xmath91 be a @xmath83-ball in @xmath85 that contains exactly two branch points of @xmath87 in its interior , as shown in figure [ f : branchcov2 ] , and let @xmath92 be an interval in @xmath50 .
the product @xmath93 is a @xmath94-ball @xmath95 in @xmath28 .
the preimage of this ball under @xmath96 is @xmath97 where @xmath98 is an annulus in @xmath84 ; see figure [ f : branchcov2 ] .
we remove this @xmath95 from @xmath28 and its preimage from @xmath99 to obtain a branched double covering @xmath100 where @xmath101 . taking the double of we obtain
a branched double covering @xmath102 which is @xmath45-surjective by construction .
this gives the case @xmath103 in the claim .
finally note that , for arbitrary @xmath19 , the connected sum @xmath16 is an @xmath104-sheeted unramified covering of @xmath105 .
taking the fiber product with , we obtain the desired @xmath45-surjective branched double covering of @xmath16 by @xmath80 .
this completes the proof .
proposition [ p : branchcov ] together with proposition [ p : ess ] completes the proof of theorem [ t : topargument ] . for theorem [ t : topbundle ]
we need the following : [ p : inessbundles ] every rationally inessential three - manifold is dominated by a non - trivial circle bundle over a surface .
this , together with proposition [ p : essbd ] , completes the proof of theorem [ t : topbundle ] . since every rationally inessential three - manifold
is finitely covered by some @xmath16 by the proof of theorem [ t : ess ] , proposition [ p : inessbundles ] is a consequence of the following statement .
[ p : branchcovbundles ] for every @xmath19 the connected sum @xmath16 admits a @xmath45-surjective branched double covering by a non - trivial circle bundle over a surface . for @xmath17 ,
the empty connected sum is , by convention , the three - sphere @xmath18 , which , via the hopf fibration , is a non - trivial circle bundle over @xmath85 . pulling back the hopf fibration under a branched double cover @xmath106 , we obtain the desired double branched cover of @xmath18 . for @xmath90
we prove that the total space @xmath2 of the circle bundle with euler number @xmath107 over @xmath84 is a @xmath45-surjective branched double covering of @xmath28 .
start by considering @xmath2 as the mapping torus of the linear torus diffeomorphism given by the matrix @xmath108 and recall that the double branched cover @xmath109 in figure [ f : branchcov1 ] is the quotient map for the involution @xmath110 since @xmath111 commutes with @xmath112 , it induces a fiber - preserving involution , also denoted @xmath111 , of the mapping torus @xmath113 .
the quotient @xmath114 is the mapping torus of the diffeomorphism of @xmath115 induced by @xmath112 .
this diffeomorphism is orientation - preserving , and so @xmath116 .
the projection @xmath117 given by the quotient map for @xmath111 is the desired @xmath45-surjective double branched cover . on every fiber it coincides with @xmath87 . to deal with the case @xmath118
, we revert to thinking of @xmath2 as a circle bundle over @xmath84 , and we fiber sum @xmath19 copies of this circle bundle to obtain a circle bundle with euler number @xmath119 over @xmath79 .
we can perform this fiber sum in such a way that the branched double covering maps @xmath120 on the different summands fit together to give the desired @xmath45-surjective branched double covering of @xmath16 . recall that the circles of the circle fibration of @xmath2 over @xmath84 are contained in the fibers of the mapping torus projection @xmath121 .
pick one such fiber , and thicken it to an annulus @xmath98 contained in a fiber of @xmath61 whose image under @xmath87 is a disk in @xmath85 containing precisely two branch points of @xmath87 , as shown in figure [ f : branchcov2 ] .
a fibered neighbourhood of our circle fiber in @xmath2 is the product of @xmath98 with an interval in @xmath50 , and the image under @xmath120 of this fibered neighbourhood in @xmath28 is a three - ball @xmath95 .
now we can perform the connected sum of two copies of @xmath28 along this @xmath95 , and simultaneously fiber sum two copies of @xmath2 by removing the fibered neighbourhood and gluing the boundary tori in a fiber - preserving way that matches up the branch loci .
this completes the proof for @xmath103 , and the general case follows by iterating the construction .
we now reformulate theorems [ t : topargument ] and [ t : topbundle ] and their proofs to obtain equivalent formulations in terms of thurston geometries and in terms of purely algebraic properties of fundamental groups .
the following is the geometric reformulation of theorem [ t : topargument ] .
[ t : geoargument ] a closed , oriented , connected three - manifold @xmath3 is dominated by a product @xmath13 if and only if 1 .
either @xmath3 possesses one of the geometries @xmath122 or @xmath123 , or 2 .
@xmath3 is a connected sum of manifolds possessing the geometries @xmath124 or @xmath18 .
let @xmath3 be a closed oriented three - manifold dominated by a product @xmath13 .
if the prime decomposition of @xmath3 contains an aspherical summand , then we have seen in the proof of theorem [ t : topargument ] that @xmath3 itself is aspherical , and is finitely covered by a product @xmath14 , with @xmath15 of positive genus . in addition , @xmath3 is seifert fibered since its finite covering @xmath14 is , cf .
moreover , @xmath3 carries the same thurston geometry as this covering , namely the @xmath122 geometry if @xmath15 has genus one , or the @xmath123 geometry if the genus of @xmath15 is at least @xmath83 .
conversely , every manifold with one of these geometries is indeed finitely covered , and , therefore , dominated by a product @xmath14 , cf .
@xcite .
if the prime decomposition of @xmath3 does not contain an aspherical summand , then each prime summand is either @xmath24 , with geometry @xmath125 , or has finite fundamental group , and thus carries the @xmath18 geometry by the work of perelman @xcite .
for all connected sums with only these summands we have proved in the proof of theorem [ t : topargument ] that they are dominated by products . finally , we give an algebraic formulation , in terms of properties of the fundamental group of the target .
[ t : algargument ] a closed , oriented , connected three - manifold @xmath3 is dominated by a product @xmath13 if and only if 1 .
either @xmath39 is virtually @xmath126 , for some aspherical surface @xmath15 , or 2 .
@xmath39 is virtually free .
if @xmath3 is a closed oriented three - manifold dominated by a product @xmath13 , and the prime decomposition of @xmath3 contains an aspherical summand , then we have seen in the proof of theorem [ t : topargument ] that @xmath39 is virtually @xmath126 .
conversely , if @xmath3 has a finite covering @xmath127 with fundamental group @xmath126 , then this covering is prime as its fundamental group is freely indecomposable . since @xmath15 is not @xmath85 , it follows that @xmath127 is irreducible and aspherical @xcite .
thus @xmath127 is homotopy equivalent to @xmath14 , proving that @xmath3 is dominated by a product .
if the prime decomposition of @xmath3 does not contain an aspherical summand , then we have seen that @xmath39 is virtually free .
conversely , if @xmath3 has virtually free fundamental group , then it is finitely covered by a three - manifold with free fundamental group . kneser s prime decomposition theorem and grushko s theorem
imply that this covering must be a connected sum of copies of @xmath28 , where the number of summands is the number of generators of its fundamental group .
the analogous reformulations can also be carried out for theorem [ t : topbundle ] .
the geometric formulation is : [ t : geobundle ] a closed , oriented , connected three - manifold @xmath3 is dominated by a non - trivial circle bundle over a surface if and only if 1 .
either @xmath3 possesses one of the geometries @xmath128 or @xmath129 , or 2 .
@xmath3 is a connected sum of manifolds possessing the geometries @xmath124 or @xmath18 .
we only have to prove the equivalence between the first cases of this theorem and of theorem [ t : topbundle ] . in one direction , if @xmath3 has one of the geometries @xmath128 or @xmath129 , then it is finitely covered by a non - trivial circle bundle over an aspherical surface @xcite .
conversely , if @xmath3 is finitely covered by a non - trivial circle bundle over an aspherical surface , then it is a seifert manifold carrying the same thurston geometry as this finite covering @xcite .
the algebraic version of theorem [ t : topbundle ] reads as follows .
[ t : algbundle ] a closed , oriented , connected three - manifold @xmath3 is dominated by a non - trivial circle bundle over a surface if and only if 1 .
either @xmath39 has a finite index subgroup @xmath130 which fits into a central extension @xmath131 with non - zero euler class for some aspherical surface @xmath15 , or 2 .
@xmath39 is virtually free .
again we only have to prove the equivalence between the first cases of this theorem and of theorem [ t : topbundle ] . in one direction , if @xmath3 is finitely covered by a non - trivial circle bundle over an aspherical surface , then its fundamental group has a finite index subgroup admitting the required central extension .
conversely , if @xmath39 has a finite index subgroup @xmath130 fitting into such a central extension , then the corresponding finite covering has to be prime , irreducible and aspherical , and is therefore homotopy equivalent to the total space of the corresponding circle bundle over @xmath15 .
as an algebraic counterpart of our topological results about domination by products for three - manifolds we now want to determine which fundamental groups of three - manifolds are presentable by products .
first we recall the definition : [ repgroupsdef](@xcite ) an infinite group @xmath130 is _ presentable by a product _ if there is a homomorphism @xmath132 onto a subgroup of finite index , such that both factors @xmath133 have infinite image @xmath134 . without loss of generality
one can replace each @xmath133 by its image in @xmath130 under the restriction of @xmath112 , so that one can assume the factors @xmath133 to be subgroups of @xmath130 and @xmath112 to be multiplication in @xmath130 .
it is obvious that a group with infinite center @xmath135 is presentable by a product just take @xmath136 and @xmath137 .
the property of ( not ) being presentable by a product is preserved under passage to finite index subgroups .
this property was introduced in @xcite and further studied in @xcite because , according to @xcite , it is a property that the fundamental groups of rationally essential manifolds dominated by products must have . for a closed three - manifold @xmath2 with infinite fundamental group the following three properties are equivalent : 1 .
@xmath74 is presentable by a product , 2 .
@xmath74 has a finite index subgroup with infinite center , 3 .
@xmath2 is a seifert manifold .
it is clear that ( 3 ) implies ( 2 ) .
the converse is the celebrated seifert fiber space conjecture , the final cases of which were resolved by casson - jungreis @xcite and by gabai @xcite .
as noted above , it is also clear that ( 2 ) implies ( 1 ) for any group .
we now prove the converse for three - manifold groups . by (
9.2 ) the only non - trivial free product that is presentable by a product is @xmath138 , which is virtually @xmath49 and so satisfies ( 2 ) .
thus , we may assume that @xmath74 is freely indecomposable , and @xmath2 is prime . if @xmath74 is not virtually @xmath49 , then @xmath2 is irreducible and aspherical by the sphere theorem , cf .
@xcite . in particular
, @xmath74 is torsion - free . by (
3.2 ) , a torsion - free group @xmath130 which is presentable by a product has one of the following properties : * either @xmath130 has a finite index subgroup with infinite center , or * some finite index subgroup splits as a direct product of infinite groups . applying this to our @xmath74
, we have to see that the second alternative in fact implies the first .
it is a theorem of epstein @xcite that if the fundamental group of a closed three - manifold splits as a direct product of infinite groups , then one of the factors has to be infinite cyclic .
but then this factor is central in the whole group .
[ [ section ] ] the main result of @xcite was that for rationally essential manifolds , in any dimension , domination by a product implies that the fundamental group is presentable by a product .
theorem [ t : topargument ] shows that the converse is not true already in dimension three : seifert manifolds carrying one of the geometries @xmath128 or @xmath129 are aspherical and have fundamental groups presentable by products , but are not dominated by products .
( these are the only counterexamples to the converse in dimension three . )
the geometry @xmath129 has another interesting feature relevant to our discussion : @xmath129 is quasi - isometric to @xmath123 , cf .
* iv.48 ) .
compact manifolds with the latter geometry are finitely covered by products , whereas those with the former geometry are not even dominated by products , although the fundamental groups are presentable by products in both cases . it was noted in ( * ? ? ?
10.2 ) that presentability by products is not a quasi - isometry invariant property of finitely generated groups .
this , together with the contrast between manifolds with the geometries @xmath129 and @xmath123 , shows that domination by products can not be detected by coarse methods , neither at the level of groups nor at the level of universal coverings of aspherical manifolds .
one of the standard characterizations of closed seifert manifolds is through the property of being finitely covered by circle bundles . in the rationally essential case
this can be weakened by replacing finite coverings by arbitrary dominant maps : our discussion in section [ s : proof1 ] shows that there are no maps of non - zero degree between trivial and nontrivial circle bundles over aspherical surfaces .
this statement already appeared in the work of wang twenty years ago , see ( * ? ? ? * theorem 2 ) .
however , the proof given there is hard to follow . in particular
, there is no argument there for the case covered by our lemma [ lem ] . at the corresponding place in the proof ,
compare @xcite , in particular equation ( iii ) , wang seems to argue that a group that is presentable by a product must itself be a product , which is of course false .
the fundamental groups of seifert manifolds with non - zero euler number are presentable by products , but are not virtually products .
j. hempel , _ residual finiteness for @xmath94-manifolds _ , in _ combinatorial groups theory and topology _ , ed .
s. m. gersten and j. r. stallings , annals of math .
studies vol . 111 , princeton univ . press 1987 . | we determine which three - manifolds are dominated by products .
the result is that a closed , oriented , connected three - manifold is dominated by a product if and only if it is finitely covered either by a product or by a connected sum of copies of @xmath0 .
this characterization can also be formulated in terms of thurston geometries , or in terms of purely algebraic properties of the fundamental group .
we also determine which three - manifolds are dominated by non - trivial circle bundles , and which three - manifold groups are presentable by products . |
fourier - transform infrared ( ftir ) spectroscopy is a widely utilized method to investigate the optical response of gasses , liquids and solids @xcite . in general ,
steady - state properties are measured , however , numerous approaches have been developed over the years to explore time - dependent phenomena by fourier - transform interferometry @xcite , many of them optimized for a certain time regime .
standard rapid - scan techniques are limited by the mirror velocity to a fraction of a second ( typically 10 ms ) , depending on the spectral resolution @xmath0 required : @xmath1 . since this is often not sufficient , step - scanning
is nowadays implemented in several high - end commercial fourier - transform instruments because there is no inherent time limit .
it covers the largest dynamical range with a time resolution of the order of nanoseconds determined by the current detector and electronics technology @xcite and still achieving a high spectral resolution @xmath2 .
the only restriction is the repeatability of the process : depending on the resolution , spectral range and desired signal - to - noise ratio , it has to be executed hundreds of times .
in contrast to the continuously moving interferometer mirror in the rapid - scan configuration , the advantage of the step - scan method is the continuous recording of the signal at a fixed mirror position .
this is repeated after the mirror has moved to the next position , eventually composing the complete interferogram .
the step - scan technique is mainly applied in biophysics and polymer chemistry where , for instance , the photolysis processes of chemical reactions @xcite , bacteria systems @xcite and the time - dependent reorientation of liquid crystals under the influence of a short electric field @xcite are studied . but
also temperature- and light - induced phase transitions can be investigated @xcite .
furthermore , it is used to examine the characteristics of lasers and their mode spectra @xcite and for photo - reflection measurements of semiconducting materials as well as quantum wells @xcite . here
we want to present and discuss our experimental setups employed to investigate the dynamics of phase transitions . tracing the vibrational spectra of molecules
, we probe the configurational changes after the transition has been triggered either by a short laser pulse ( sec . [ sec : opticalswitching ] ) or by an electrical pulse ( sec .
[ sec : electricalswitching ] ) .
in the left panel of fig . [ fig:1-stepscan ] the data acquisition of a step - scan measurement is schematically depicted . at each mirror position @xmath3 along the traveling distance of the mirror
the temporal varying reflection signal @xmath4 is recorded .
the complete interferogram is sampled for various times and retardations by successive stepwise moving of the mirror . a subsequent fourier transformation for each measured time point @xmath5 in the intensity `` matrix '' @xmath6
derives the time - resolved spectrum @xmath7 .
the process can be repeated several times to improve the signal - to - noise ratio whereas at each @xmath8 position the time - dependent signal is averaged ( typically 10 to 50 times ) .
additionally , the stability of the mirror influences the signal - to - noise ratio significantly @xcite , therefore , one must take care that the spectrometer is located in a vibration - free and silent environment .
for this reason we place our ftir - spectrometer on a heavy optical bench mounted on air attenuators to decouple the system from the environment .
furthermore , the vacuum pumps were placed in a separated room to reduce vibrations and acoustic noise . this way we reached a mirror stability better than 3 nm .
the detector signal @xmath9 is recorded as a function of time .
after all mirror positions were passed through , the time - dependent signal is recieved by the execution of a fourier transformation for each measured time point @xmath5 .
( right panel ) illustration of the signal sequence within a step - scan experiment .
( a ) external trigger signal t1 given by the external perturbation source , for example a pulse generator or laser .
( b ) the second signal t2 is generated by the spectrometer and sent to the interferometer motor to move the mirror to the next position .
( c ) after a certain predefined stabilization time the third signal t3 waits for the next external trigger signal t1 and rises afterwards immediately .
it stays high as long as all data points are captured .
( d ) signal t4 corresponds to each recorded time point within the signal t3.,title="fig:",scaledwidth=47.0% ] the detector signal @xmath9 is recorded as a function of time .
after all mirror positions were passed through , the time - dependent signal is recieved by the execution of a fourier transformation for each measured time point @xmath5 .
( right panel ) illustration of the signal sequence within a step - scan experiment .
( a ) external trigger signal t1 given by the external perturbation source , for example a pulse generator or laser .
( b ) the second signal t2 is generated by the spectrometer and sent to the interferometer motor to move the mirror to the next position .
( c ) after a certain predefined stabilization time the third signal t3 waits for the next external trigger signal t1 and rises afterwards immediately .
it stays high as long as all data points are captured .
( d ) signal t4 corresponds to each recorded time point within the signal t3.,title="fig:",scaledwidth=47.0% ] [ fig:2-trigger ] since the data acquisition at each mirror position @xmath3 has to start always at the same time , the data recording must be synchronized with the external stimulation source ( laser or pulse generator ) .
thus , the trigger signal sequence is very important and crucial in a step - scan measurement .
the temporal sequence of the ttl control signals is visualized in the right panel of fig .
[ fig:2-trigger ] .
an external or an internal trigger signal t1 , which is correlated with the beginning of the reaction , controls the data acquisition . yet , before the recording starts , the spectrometer sends a signal t2 to the interferometer , so that the mirror is moved to the next position .
there , it is stabilized for a few milliseconds . after the stabilization procedure ( @xmath10 ms ) the next arising trigger signal t1 is used as a starting signal for the record window t3 .
the signal t3 stays high for the total recording time @xmath11 ( @xmath12 total number of time slices ) which is defined at the beginning of each measurement .
as soon as the signal t3 is on , at each ttl - signal t4 the detector signal is captured . depending on the number of averaging spectra this procedure
is repeated several times starting again with the t3 signal .
subsequently , the mirror moves to the next position .
the time resolution @xmath13 depends mainly on the response time of the detector .
standard pc - mct - detectors , operating in the photo - current mode , have a minimum response time of @xmath14s .
their disadvantage is that the measured current becomes nonlinear above a certain incident threshold intensity .
a photovoltaic ( pv ) detector has a two order of magnitudes shorter response time due to the small detector area .
furthermore , the measured signal is always proportional to the incident light intensity . for the measurement a pc - mct of the model d316 ( bruker optics , ettlingen ) with a time resolution of about 1 @xmath15s and a pv - mct kmpv11 - 11 - 1-j1 from kolmar technologies with a theoretical rise time of 25 ns
are utilized .
the time resolution also depends on the amplifier and the a / d - converter .
there are two amplifiers , the build - in amplifier of the spectrometer and the ka100-a1 from kolmar technologies with a bandwidth of 250 mhz . as an a / d - converter the internal converter of the spectrometer with its time resolution of 6 @xmath15s , with a dynamical range of 24 bit and a maximal input voltage of @xmath16 v , can be used , or a transient recorder m3i.4142 from spectrum systementwicklung microelectronic gmbh , grosshansdorf with a bandwidth of 400 mhz at 16 bit and @xmath17 v. the standard interferogram of a fourier - transform spectrometer consists of an ac- and dc - component : while only the ac - signal contains the important spectral information the dc - signal is usually removed by an electronic high - pass filter .
the phase correction can only be performed directly for a rapid - scan measurement . in a step - scan experiment
the spectrum can include positive as well as negative features .
thus , a phase correction with the raw time - resolved ac - signal does not work .
two options exist to resolve this complication : first , the simultaneously recording of the dc - signal and the ac - signal ; the dc - component provides the right phase correction for the ac - component .
the second possibility is to use the phase of a previous rapid - scan measurement @xcite .
electrically induced phenomena occur mainly in semiconductor leading to non - linear conductivity or negative differential resistance @xcite .
high electric fields can also trigger phase transitions , for instance in charge - density wave systems @xcite or correlated insulators @xcite . in organic conductors
electrical switching is subject of research for quite some time @xcite . as an example of electric switching
, we here have investigated the polarization - dependent vibrational modes of liquid crystals , which allow us to trace the orientation of the molecule upon applying a voltage by time - resolved infrared measurements . in these molecules
several vibrational modes exist which can be assigned to different parts of the molecules . by static and time - resolved polarization - dependent measurements we gain on the one hand information about the orientation of the individual molecular building blocks in static positions and on the other hand about the temporal evolution of the electric switching process .
hence , we learn about the rotation of the different molecular constituents with the electric field .
liquid crystals are widely used for electronic displays due to the possibility to manipulate the orientation of the complex organic molecules by applying an electric field and this way to control the transmitted light .
we have chosen the felix 017/100 mixture from the clariant as a ferroelectric liquid crystal that constitutes a chiral smectic c phase ( smc@xmath18 ) at ambient conditions creating a helical structure with @xmath19 , as depicted in fig .
[ fig:3-lc](a ) . the spontaneous electric polarization @xmath20 is perpendicular to the director vector @xmath21 and the layer normal vector @xmath22 .
while in general the different orientations average and the total polarization is zero , a uniaxial rubbed surface aligns the molecules parallel to each other , leading to a net polarization , sketched in fig .
[ fig:3-lc](b ) .
the whole procedure is well - known as `` surface stabilized ferroelectric liquid crystal '' ( ssflc ) @xcite .
two energetically equal configurations exist with @xmath23 arranged in opposite directions , but perpendicular to the surface with the same tilting angle @xmath24 .
an applied electric field can now induce the transition between the two polarization states by changing the orientation of the molecule . ):
going from layer to layer the director @xmath21 successively helically precedes around the stacking normal vector @xmath22 in an angle @xmath25 .
the spontaneous polarization @xmath26 is perpendicular to the director and the normal vector .
the total polarization @xmath27 is zero .
( b ) oriented smectic c phase : all directors @xmath21 are aligned towards the same direction . by symmetry
the polarization @xmath28 can adopt two possible values .
( c ) side- and top view of the liquid crystal cell consisting of two caf@xmath29 plates ( light gray , transparent ) .
a thin film of ito was sputtered onto the plates ( bright brown ) .
polyimid ( green ) was deposited by the spin - coating process on top of the ito layer and serves as an orientation layer for the liquid crystals .
uv - glue in connection with small plastic spheres ( shaded area ) with a thickness of @xmath30 m keeps the two windows at a fixed distance .
( d ) transmission spectrum of caf@xmath29(black ) measured at room temperature in comparison to caf@xmath29 coated with ito ( blue ) and caf@xmath29 in combination with ito and polyimid ( red ) .
the dashed line indicates a spectral range where water absorption and ice on the detector window corrupt the data.,scaledwidth=80.0% ] while standard liquid - crystal cells consist of glass windows transparent in the visible , we have constructed a cell with two caf@xmath29 windows suitable for the infrared spectral range from 1000 to 50000 @xcite . in fig .
[ fig:3-lc](d ) the transmission spectrum of a caf@xmath29 window is plotted in the mid - infrared frequency range .
compared to alternative window materials , such as kbr , it is not hydroscopic , and with @xmath31 has a much lower refractive index than ge ( @xmath32 ) , for instance , reducing the reflection losses considerably .
the caf@xmath29 windows ( @xmath33 mm@xmath34 , thickness of @xmath35 mm , purchased from korth kristalle gmbh , kiel ) are polished optically to get the optimal surface quality for the optical experiments .
both caf@xmath29 windows are sputtered with a thin film of indium tin oxide ( ito ) commonly used as an electrode material for displays because it is lucent over a broad energy range and at the same time conductive . however , optical measurements plotted in fig .
[ fig:3-lc](d ) reveal that the transmission decreases rapidly towards low frequencies @xcite . in the range of interesting for the c = c double bond vibrations the transmission is between 5 and 20% , which is still sufficient for our experiment . in order to preorient the liquid - crystal molecules a very thin film of polyimide
was deposited on top of the ito layer by using the spin - coating technique which was rubbed afterwards unidirectional [ fig .
[ fig:3-lc](c ) ] .
as soon as the molecules were placed on the surface , they align themselves along the rubbing direction @xcite . as demonstrated in fig .
[ fig:3-lc](d ) the very thin layer of polyimide does not affect the overall light transmission of the system in the relevant frequency range in agreement with previous results @xcite .
both plates have been clued together with a uv glue mixed with small macroscopic plastic spheres with a diameter of @xmath30 m which serve as spacer to keep the windows on a constant distance , as sketched in fig .
[ fig:3-lc](c ) . after the glue was cured ,
the cell was heated up above @xmath36c so that the liquid - crystal mixture transformed in the isotropic phase with a reduced viscosity .
the liquid crystals were sucked in the chamber due to the capillary action .
afterwards , the temperature of the melt was slowly lowered so that a single domain was formed .
due to the large dimension of the cell the distance of the two plates was not constant and reduces to the center of the cell causing newton s rings .
it changes again , when placed in a vacumm spectrometer .
the contact wires were made out of copper and fixed on the exposed ito side areas with indium ; they are connected to a philips - pm 5768b pulse generator to switch the liquid crystals .
the generator sends during the measurement simultaneously to the switching voltage pulse a second ttl signal to the spectrometer to start the data acquiring process .
the cell is placed in a bruker vertex 66v / s fourier - transform infrared spectrometer together with a suitable infrared polarizer .
the absorption becomes maximal if the infrared light is polarized parallel to the electronic transition dipole moment @xmath37 of the molecular vibrational mode .
as soon as the external electric field @xmath38 is switched on , a force acts on the liquid crystal . in general , the switching process takes a few microseconds @xmath39 until most of the molecules are reoriented .
the new orientation of the molecules and its director @xmath21 change the direction of the transition dipole moment @xmath37 leading to an increase or decrease of the absorption signal . if the electric field is switched on , the directors shown in fig .
[ fig:3-lc](b ) spin collectively in one direction and therefore , the azimuthal angle changes .
the switching velocity and -time can be derived from the equation of motion @xcite : @xmath40 with @xmath41 the damping due the viscosity of the liquid , @xmath42 is the angle between the electric field and the polarization of the molecule , @xmath43 is the coordinate along the direction perpendicular to the window , and @xmath44 the elasticity constant . due to the large moment of inertia @xmath45 kg / m
the equation can be simplified to @xmath46 and solved for small angles @xmath42 @xmath47 with @xmath48 m as the spacing between the two cell windows .
we can extract two time constants , @xmath39 for rise with the field and @xmath49 for fall after the electric field @xmath38 is turned off : @xmath50 the switching time for the here examined material felix 017/100 can be estimated using @xmath51 ns/@xmath52 , @xmath53 nc/@xmath52 , @xmath54 n , @xmath55 kv / cm , and @xmath48 m to yield @xmath56s and @xmath57 ms
. the mid - infrared transmission spectrum of the liquid crystal cell filled with felix is displayed fig .
[ fig:4-transmission ] . the dips in the spectrum between 1000 and 1700 as well as
between 2800 and 3000 , enlarged in the insets ( b ) and(c ) , be assigned to the vibrational modes of the liquid - crystal molecules .
the features @xmath58 are stretching modes of ch@xmath29- and ch@xmath59 end group .
the modes @xmath60 are connected with the vibrations of the c = c double bonds belonging to the center part of the molecules .
the features at 1439 and 1395 can be referred to the wiggling and torsion of the ch@xmath59 group .
the low - lying resonances at 1269 and 1243 belong to an asymmetric stretch oscillation of the c - o - c bond .
the position of the detected features perfectly coincides with the resonance frequencies determined by huang and shih @xcite .
[ fig:5-angles ] shows a change of the infrared spectra when an electric field of @xmath61 v is applied .
all measurements were performed at @xmath62c . due to the rearrangement of the liquid - crystal molecules a strong modification of the intensity could be observed .
we can can deduce two different - oriented transition dipole moments : the first belongs to the vibrational mode of the rigid body of the molecule and the other points out from the molecule axis related to the end groups .
the modes containing the ch@xmath29 and ch@xmath59 bonds exhibit a minimum at @xmath63 and a maximum at @xmath64 , correspondingly .
the minimum of the modes belonging to the molecule body are slightly shifted and can be found at @xmath64 .
the green curve , for instance , belongs to a vibrational mode with an aligned dipole moment parallel to the central molecular frame . at @xmath63
it reaches a maximum for a voltage of 5 v , this means that the dipole moment rotates away from the electric field vector of the infrared light .
hence , the transmitted intensity increases at this specific wavelength .
in contrast , the intensity of the dipole moments related to the end groups of the molecule is reduced because the transition dipole moments are aligned parallel to the incident radiation . for various polarization states of the incident light measured at room temperature .
( b ) polarization - dependent measurement of the change of the transmitted intensity after creating 5 v on the two windows for seven different vibrational modes .
only the strongest modes are displayed .
the maximal change of the transmission appears at @xmath63 , followed by a minimum at @xmath64 .
the sinusoidal modification evidences that the liquid crystals react on the applied electric field .
, scaledwidth=100.0% ] in order to investigate the dynamics of the liquid - crystal switching process the polarizer was fixed at an angle of @xmath65 , and voltages between 4 and 6 v applied between the two plates .
the time - dependent infrared signal is presented in a contour plot in fig .
[ fig:6-contour ] .
the rectangular voltage pulse had a width of 2 ms with a repetition rate of 30 hz and the signal was recorded for a period of 4 ms . to improve the signal - to - noise ratio ,
five spectra were averaged .
the time resolution was set to 25 @xmath15s and the spectral resolution was 2 .
we can conclude that the liquid crystal molecules rotate and reorient like a stiff body under the influence of an external electric field .
kv / cm with a pulse width of 2 ms . in a spectral region from 1200 to 1800 several features can be recognized corresponding to the orientation of their transition dipole moment .
( b ) time - evolution of the spectra demonstrate for four selected times after the voltage pulse of 6 v was applied .
the spectrum ( blue ) , recorded 50 @xmath15s before the pulse is applied , exhibits almost no change and is flat .
several features appear at time 200 @xmath15s that can be ascribed to certain vibrational modes .
the signal saturates after 1150 @xmath15s and shows no further variation of the intensity.,scaledwidth=100.0% ] the extreme values of @xmath66 are displayed in blue or in red in the contour plot . the transition dipole moments @xmath37 are aligned parallel to @xmath67 lead to a reduced and , hence to a negative signal whereas in contrast for a positive @xmath37 is turned away from @xmath67 .
this becomes more obvious in fig .
[ fig:6-contour](b ) where spectra at different times are plotted . as a check
, data were taken @xmath68s before the pulse ; no variation of the vibrational modes and no shift is observed . after applying the voltage pulse , however , the spectrum dramatically alters and reveals the feature exactly as in fig .
[ fig:5-angles](a ) which marks the final state .
also the absolute values of the change are the same . in accord with fig .
[ fig:4-transmission](a ) and [ fig:5-angles](a ) no variation can be identified below 1000 because the ito layer absorbs most of the incident light .
the switching process is completed after about 1 ms and the signal stays constant over a period of 800 @xmath15s .
is displayed for the resonance frequency 1606 for three different voltages .
the rise of the signal as well as the drop can be fitted well by a single exponential function @xmath69
. the signal increases with increasing voltage . at the same time the switching speed increases .
in contrast to that the relaxation process retards with higher voltage .
( b ) voltage - dependence of the rise time @xmath39 for five different resonance frequencies .
with increasing voltage the switching time decreases continuously .
( c ) whereas the relaxation time @xmath49 deceases , but no direct dependence on the voltage can be discovered . for 6 v
@xmath49 reveals a larger variance.,scaledwidth=100.0% ] to determine the rise and fall time , the temporal evolution of @xmath70 of five vibrational modes were used .
the change of the intensity of the @xmath71 mode is plotted as a function of time in fig .
[ fig:7-time](a ) for three different voltages . with a rise time of less than 10 ns , the voltage pulse basically has a rectangular shape ; however , the infrared signal reacts slowly and requires several hundreds of microsecond until it saturates .
due to the viscosity of the liquid crystals the molecules react with a certain delay to the electric field .
the switching time @xmath39 diminishes with rising voltage whereas the amplitude increases slightly .
however , the signal overshoots at the beginning for 6 v and relapses back to the value of the 5 v pulse .
the rise time can be determined by fitting the experimental data with a single exponential function @xmath72 .
the extracted values of @xmath39 are summarized in fig .
[ fig:7-time](b ) and ( c ) as a function of the applied voltage . at the end of the pulse the signal relaxes slowly back to its initial values within 2 ms and is at least by a factor of four larger than the rise time @xmath39 .
the time for reorientation @xmath49 can be determined from the temporal evolution of the recovery process by a single exponential function . at a voltage of 4
v almost all vibrational modes reveal a switching time @xmath39 between 210 @xmath15s and 230 @xmath15s ; with increasing voltage this range enlarges to maximum 70 @xmath15s for 6 v whereas the values are between 190 @xmath15s and 120 @xmath15s . the various time
constants imply that constituents of the molecules react differently on the electric field .
the vibrational modes related to the molecule body ( @xmath71-mode ) rotate slower than the ones ( @xmath73-mode ) which are connected to the end groups . the theoretically calculated value for @xmath74s ( for 5 v ) derived from eq .
( [ eq : tau ] ) agrees very well with the experimentally determined @xmath39 which is of about 200 @xmath15s .
the reason for the small discrepancy may be due to uncertainties in the distance of the spacer plates or in the material specific parameters @xmath41 and @xmath26 ; they all depend strongly on temperature . from our limited data
set , we can not verify the predicted @xmath75 characteristic of the rise time ; our data indicate a constant time .
probably , @xmath39 exhibits the expected field dependence and the divergent behavior for smaller voltages . as it was expected from the theoretical calculations of @xmath49 the relaxation times in fig .
[ fig:7-time](c ) are larger than the switching time .
a time constant of 40 ms was predicted but not overserved .
furthermore , the relaxation rate is a function of the applied voltage but according to eq .
( [ eq : tau ] ) it should be independent of any external parameter . one possible explanation therefor is the stronger tilting of the molecules with increasing voltage as well as the total amount of realigned molecules increases as well .
thereby , the layer of oriented molecules increases which leads to a kind of self - stabilization effect resulting in the extended decay .
for this reason , the relaxation rate @xmath49 is not a function of the intrinsic elastic constant @xmath44 , but also of the external eclectic field .
over the last decades photo - induced phase transitions were investigated in several material classes , for instance , polymers @xcite , organic charge - transfer salts @xcite , transition - metal oxides like vanadium oxides @xcite , cuprates @xcite , and manganites @xcite . in the following we study the one - dimensional organic mixed - stack crystals tetrathiavulvalene - chloranil ( ttf - ca ) that undergoes a ionic - neutral phase transition at @xmath76 k that can also be induced by light .
initiated by the groundbreaking experiments by koshihara _
@xcite , ultrafast pump - probe experiments have been performed to examine the photo - induced phase transition pipt in the femto- and picoseconds time range @xcite .
the understanding of these phenomena was boosted by theories of nasu and yonemitsu @xcite . in the case of ttf - ca
open questions concern the relaxation of metastable domains , the related time scale , and the modification of the infrared spectrum due to photo - excitation ; some of these have recently be addressed by time - resolved ftir - spectroscopy @xcite .
cl@xmath77o@xmath29 ) and ( b ) tetrathiafulvalene ( ttf , c@xmath78s@xmath77h@xmath77 ) .
( c ) monoclinic unit cell of ttf - ca at room temperature .
the ttf and ca molecules are ordered along the crystallographic @xmath79-axis .
( d ) at @xmath80 k the space group of the unit cell is p2@xmath81/n and the ca and ttf molecules are stacked equally spaced along the @xmath79-axis .
a further stack is located at @xmath82 , respectively , at which the ttf - ca pairs are tilted opposite to the @xmath79-axis . in the ionic phase the ttf and ca molecules dimerize along the stack . by the charge transfer of about @xmath83 electric dipoles
are formed along the stacking direction .
, scaledwidth=100.0% ] at ambient conditions the planar molecules of ttf and ca are equidistantly arranged in alternating stacks along the @xmath79-direction ( fig .
[ fig:10-ttf - ca ] ) . since the charge transfer @xmath84 as determined from optical experiments is rather small @xcite , this state is referred to as the neutral phase ; upon cooling the charge transfer increases slightly . at @xmath76
k a phase transition occurs where the molecules dimerize along the stack with intermolecular distances 3.504 and 3.685 @xcite , and the ionicity @xmath85 increases from 0.3 @xmath86 to about 0.6 @xmath86 ; fig .
[ fig:10-ttf - ca](d ) displays a sketch of the arrangement .
accordingly the dielectric constant diverges at the transition with a pronounced frequency dependent response @xcite .
the application of 11 kbar pressure shifts the transition to room temperature @xcite .
( green ) of ttf - ca along the stacking direction . above the transition it behaves as a classic band insulator with an activation energy of @xmath87 ev.,scaledwidth=80.0% ] the resistivity of ttf - ca along the stacking direction increases upon cooling , following an activated behavior as displayed in the arrhenius plot of fig .
[ fig:11-ttf - ca - rho ] . at @xmath88
, @xmath89 drops by one order of magnitude before it increase again at lower temperatures .
the strong enhancement of the conductivity is attributed to the rising number of neutral - ionic domain walls in the neutral phase which also explains the dielectric behavior and non - linear transport @xcite .
alternatively it was suggested , that the multiphonon peierls coupling leads upon approaching the phase transition to a progressive shift of spectral weight and of the coupling strength toward the phonons at lower frequencies , ending in a soft - mode behavior only for the lowest - frequency phonon near the transition temperature @xcite . in the proximity of the phase transition ,
the lowest - frequency phonon becomes overdamped due to anharmonicity induced by its coupling to electrons . in fig .
[ fig:13-ttf - ca - refmodes ] the mid - infrared reflectivity and the optical conductivity of ttf - ca are presented for temperatures above and below the neutral ionic transition . in this energy range
the intramolecular modes of the ttf and ca molecules can be identified and assigned @xcite ; some of them are sketched in panels ( c ) and ( d ) . along the @xmath79-direction
the symmetric a@xmath90 as well as the infrared - active b@xmath91 modes of ca and ttf can be observed whereas the first one is only infrared active due to the emv - coupling . above the phase transition in the neutral phase the ttf and ca
molecules are not dimerized and thus the a@xmath92 modes are only weakly infrared - active and the optical conductivity is low .
k ( blue ) and 85 k ( red ) taken for the polarization along the stack .
the maximum is mainly caused by emv - coupled modes which gain intensity in the dimerized ionic phase .
( c ) moleuclar vibrations of chloranil ( ca ) and ( b ) tetrathiofulvalene ( ttf ) : the upper rows depict the gerade a@xmath90-modes and the lower rows the ungerade b@xmath93-modes.,scaledwidth=100.0% ] in the ionic phase the point inversion symmetry is lost and the molecules become dimerized : the intensity of a@xmath90 modes is enhanced enormously .
time - resolved measurements thus allow us to explore the dynamics at the neutral - ionic phase transition by looking at the change in dimerization of the ttf and ca molecules , the increase of the ionicity .
we can investigate the building of metastable domains until they eventually annihilate . for photo - induced experiments a nd : yag laser ( b.m industries / thales , yag-502dns - dps920 ) was operated in the second harmonics @xmath94 nm ( 2.35 ev ) of the fundamental wavelength ( 1064 nm ) .
the pulse length is 8 ns .
the repetition rate can be selected internally and externally between 1 hz and 20 hz . the laser intensity is adjusted continuously by a brewster plate in addition to neutral density filters between 0.1 and 0.5 optical density .
the laser intensity was checked by a power and energy meter in front of the sample .
the long term stability of the laser power is better than 5% .
a telescope arrangement reduces the diameter of the laser beam from 1 cm by a factor of 2 , illustrated in fig .
[ fig:8-laser ] .
the laser beam is directed from the optical bench to the infrared microscope via several mirrors ( see fig .
[ fig:7-setup ] ) .
there , it is deflected on the sample by a @xmath65 aluminum coated mirror mounted below the schmidt - cassegrain objective of the bruker hyperion infrared microscope .
a lens ( @xmath95 mm ) focuses the beam on the sample , by varying the focal length .
the light is circular polarized .
the laser system and optical setup are spatially decoupled from the spectrometer and mounted on an optical table to suppress possible external vibrations .
the laser pulse sequence is controlled externally to ensure the temporal synchronization between the laser , the pulse generators and the ftir - spectrometer , as depicted in fig .
[ fig:8-laser ] .
the charging of the flash lamps is triggered by an external signal as well as the pockels cell generating the laser pulse .
therefore , a hp pulse generator ( pm 5786 b ) sends the trigger signal x1 to the delay generator ( eg&g princeton applied research model 4144 ) .
one of the delay generator output signals x2 is forwarded to a second pulse generator ( hp 214b ) which releases a further delayed pulse x3 with a minimum length of 150 @xmath15s and minimum height of 5 v. it initializes the charging of the flash lamps of the laser system .
after signal x3 drops to zero the lamps are charged with a delay of 1.5 ms .
it also activates the discharge of the lamp with a delay of 15 @xmath15s .
about 30 @xmath15s after the end of the charging and discharge pulse x3 a further voltage signal x4 from the first hp pm 5768 b pulse generator ( length 6 @xmath15s , 6
v ) is sent to the pockels cell generating the laser pulse . in the case of the photoconductivity measurements
the delayed trigger signal x5 from the delay generator activates the pulse generator ( avtech electrosystems ltd . , ottawa ) to apply a voltage pulse x6 to the sample .
the synchronizing pulse x7 of the avtech device goes to the tektronix oscilloscope to start the acquisition of the photocurrent .
the third signal x8 from the delay generator is used to initialize the time - resolved infrared measurement of the ftir - spectrometer .
s. the pulse x6 is the voltage pulse which is applied to the sample for the photoconductivity measurements , for instance .
the oscilloscope records the variation of the sample current or resistance and is synchronized with the other instrument via the signal x7 .
, title="fig:",scaledwidth=57.0% ] s. the pulse x6 is the voltage pulse which is applied to the sample for the photoconductivity measurements , for instance .
the oscilloscope records the variation of the sample current or resistance and is synchronized with the other instrument via the signal x7 .
, title="fig:",scaledwidth=36.0% ] for the pipt measurements we excite only one of the first transitions of ttf@xmath96 by the second harmonic of the nd : yag laser ; all intermolecular transitions of ca@xmath97 lie above present photon energy . also excitations from lower lying bands into the valence band are possible .
_ compared the dependence of the conversion efficiency on the photon energy ; by excitation of intramolecular transitions no threshold intensity occurs to create neutral domains @xcite . and ca@xmath98 pairs in the ionic phase in ttf - ca .
a ttf@xmath98 molecules is excited with a laser pulse with the photon energy @xmath99 .
( b ) vertical excitation of the homos of ttf@xmath98 according to the frank - condon - principle in the lumo of ttf@xmath98 .
the excitation is strongly localized to the molecule .
( c ) creation of excitons , for instance frenkel - excitons , which are delocalized across the whole molecules .
( d ) via different relaxation processes and channels charge transfer excitons are created which triggers to the transition of the neutral phase .
( e ) the dimerization is suppressed and the charge between the molecules is redistributed .
a neutral domain is created in the ionic host matrix which is separated by neutral - ionic domain walls .
( f ) afterwards the neutral domains extend along the one - dimensional chain . by electron - phonon - coupling
also neighboring ionic chains are converted into neutral regions and a three - dimensional domain is established . ] in fig .
[ fig:14-pipt ] the optical generation of neutral domains is schematically illustrated .
the light pulse will create intramolecular frenckel - excitons , which are electron - hole pairs strongly localized on the excited molecules .
they decay via various channels into several charge - transfer excitons @xcite and by that lead to a phase transition . in the case of direct excitation only one exciton
is created , which is not enough to establish a macroscopic , metastable domain extended over several d@xmath100a@xmath100 pairs . with a sufficient number of photons a multiplicative
, non - linear effect can be established , formating metastable domains . by the creation of the one - dimensional , neutral , non - dimerized region
, neutral - ionic domain walls are formed between the neutral and ionic parts .
the excitation energy of these domain walls is about 0.1 ev @xcite and corresponds to the activation energy of 120 mev and 65 mev in the ionic phase @xcite .
this is in good agreement with predictions of an activation energy between 25 and 56 mev @xcite .
afterwards , the electronic system couples to the crystal lattice and exciting phonons @xcite via electron - phonon coupling and hence , creates shock - waves .
they can convert neighboring chains into neutral domains .
( solid lines ) after photo excitation is depicted for various times .
the behavior resembles the static reflectivity difference @xmath101 presented in fig .
[ fig:13-ttf - ca - refmodes](a ) .
( b ) and ( c ) normalized @xmath102 for various laser intensities recorded at 1390 for @xmath103 k and for different temperatures for 0.71 mj / cm@xmath34 .
the time profile can be successfully fitted by a stretched explonential function @xmath105 ( dashed linies).,scaledwidth=100.0% ] in fig .
[ fig:14-ttf - ca - transient](a ) the time - dependent behavior of the reflectivity change @xmath106 at @xmath103 k is plotted for a laser pulse intensity of 0.71 mj / cm@xmath34 .
it directly compares to the static reflectivity change @xmath107 plotted in fig .
[ fig:13-ttf - ca - refmodes](a ) . by photo excitation @xmath108 becomes negative within a short time which is below the experimental time resolution of 6 @xmath15s .
the direct comparison of the @xmath108 shape and the static reflectivity change @xmath109 reveal that the ionic phase was not only dissolved , but also a transition into a neutral state was induced . within several hundreds of microseconds
the signal @xmath108 relaxes back to zero which means that the ionic phase is reestablished .
moreover , no change of the spectral shape with the elapsed time and laser pulse intensity could be detected . by comparing the shape of @xmath108 with the corresponding difference of the reflectivity @xmath109 between 78 k and 85 k ,
we conclude that the vanishing of this features indicates the dissolving of the dimerization state between the ttf and ca molecules .
moreover , we suggest that metastable , non - dimerized neutral domains in the ionic matrix are created . to trace the temporal evolution of the pipt in dependence of the pump intensity and the sample temperature , we have chosen the very intense @xmath110 ( a@xmath111 ) mode of ttf residing at 1390 since we have asserted that the temporal evolution is the same for the whole spectra .
the normalized @xmath102 is represented in fig .
[ fig:14-pipt](b ) for different pulse intensities . at
the beginning the signal decays very fast and at the end it flattens out . at the vicinity of @xmath88
the first component decays faster with decreasing laser intensity .
a fit by a simple single or double exponential function @xcite does not yield satisfactory results .
however , we obtain an excellet agreement when using a stretched - exponential function , which is also called kohlrausch - william - watt function @xmath112 as depicted in fig .
[ fig:14-ttf - ca - transient](b ) and ( c ) .
the fitting parameters @xmath113 and @xmath114 are a function of the laser intensity and decreases from 0.35 to 0.42 and from @xmath115 to @xmath116 s with decreasing laser intensities . in fig .
[ fig:14-ttf - ca - transient](c ) @xmath102 is displayed for various temperatures @xmath117 , 73 , and 78 k. far below @xmath88 , the temporal dynamics of the reflectivity drops very fast within the first 20 @xmath15s and approaches asymptotically a constant value which is in contrast to the temporal profile at the vicinity ( @xmath118=78 k ) of @xmath88 which constantly diminishes .
similar to the dependence of the fitting parameters on the laser intensity the effective recombination time @xmath114 as well as the stretching exponent decrease from @xmath116 s to @xmath119 s and from 0.42 to 0.23 , respectively , with decreasing sample temperature .
the observed time profile can be explained by a random - walk annihilation process of the generated neutral - ionic domain walls .
our comprehensive time - resolved infrared study and the random - walk model @xcite allow us to conclude that close to the phase transition , large domains are formed due to the valence instability .
we find that the merger and interaction of the induced domains play an important role for the formation of the macroscopic domains and deduce from the model with decreasing laser intensity , the average domain size decreases . at lower temperatures the ionic phase is more robust ;
the average domain size is much smaller and changes less with laser intensity .
the random walk of the neutral - ionic domain walls is the dominant factor for the relaxation of the metastable domains in the temperature range considered .
we have presented two examples where time - resolved infrared investigations using step - scan fourier - transform spectroscopy provide insight in the molecular dynamics at the phase transition and molecular orientation . in the case of liquid crystals , the application of an electric field switches the orientation of the molecular dipoles changing the transmission of the cell for polarized infrared light .
watching the time evolution of the pronounced molecular vibrational modes allows us to extract the time constant of approximately 2 ms and its dependence on the applied voltage .
the neutral - ionic phase transition of ttf - ca can be photo - excited with a short laser pulse , creating domain walls that are mobile and eventually annihilate .
we follow the time - dependence of the vibrational modes activated by the changed dimerization and ionicity .
the relaxation back to the initial state strongly depends on the temperature and laser intensity , it extends from a 3 @xmath15s to almost 1 ms . for both examples we give details of the experimental setups and limitations .
we demonstrate the applicability of step - scan fourier - transform spectroscopy in a large dynamical range .
we thank f. schrg and f. giesselmann for providing the liquid crystals and g. untereiner for continuous help ; e. kurz and n. frhauf supported us when building the cell and use of their clean room .
funding by the deutsche forschungsgemeinschaft ( dfg ) is acknowledged .
thanks the carl - zeiss - stiftung for financial support .
m. mitrano , g. cotugno , clark , s.r .
, r. singla , s. kaiser , j. sthler , r. beyer , m. dressel , l. baldassarre , d. nicoletti , a. perucchi , t. hasegawa , h. okamoto , d. jaksch , a. cavalleri , phys .
lett . * 112 * , 117801 ( 2014 ) | the time - dependent optical properties of molecular systems are investigated by step - scan fourier - transform spectroscopy in order to explore the dynamics at phase transitions and molecular orientation in the milli- and microsecond range .
the electrical switching of liquid crystals traced by vibrational spectroscopy reveals a rotation of the molecules with a relaxation time of 2 ms .
the photo - induced neutral - ionic transition in ttf - ca takes place by a suppression of the dimerization in the ionic phase and creation of neutral domains .
the time - dependent infrared spectra depend on temperature and laser pulse intensity ; the relaxation of the spectra follows a stretched - exponential decay with relaxation times in the microsecond range strongly dependent on temperature and laser intensity .
we present all details of the experimental setups and thoroughly discuss the technical challenges .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
chirality in nuclear physics relates to features of nuclear hamiltonian in the context of time reversal symmetry .
the phenomenon of explicit time - reversal symmetry breaking comes up through already well known cabibbo - kobayashi - maskawa matrix of the standard model .
the ckm matrix , introduced in order to diagonalize the mass terms given by higgs mechanism with respect to quark flavor , contains t violating terms being associated with the heavy quark sector .
therefore no explicit t violation is expected to be present in low energy nuclear interactions giving t - symmetric nuclear hamiltonian .
what remains is the mechanism of spontaneous time reversal symmetry breaking and possible occurrence of spontaneous breakdown of the chiral symmetry .
this possibility has been studied for several years through the measurements of the energy levels and gamma transition probabilities @xcite,@xcite .
the latter ones have been intensively studied in the mass regions of a=130 and a=100 , while the specific gamma selection rules , i.e. staggering of the b(m1 ) values along the rotational bands have been observed only in cs isotopes @xcite .
absence of the b(m1 ) staggering in other nuclei is intriguing and needs clarification what is the main goal of the present paper .
although spontaneous time - reversal symmetry breaking is responsible for the phenomenon of chirality , there are no eigenstates of t operator and therefore the symmetry operator t alone is not used in the quantum - mechanical description of this problem .
however , one can introduce another symmetry operator @xmath0 being a combination of time - reversal and @xmath1-rotation , that allows proper description of t - symmetry breaking and simultaneously possesses its own eigenstates .
the @xmath0 is called t - signature @xcite or chiral symmetry @xcite operator .
as stated above , nuclear hamiltonian h is expected to be t symmetric . additionally , h is also rotationally invariant in the absence of external fields . as a result ,
the eigenstates of h are assigned to spin and spin projection quantum numbers @xmath2 , and the following commutation relation is valid @xmath3=0.\ ] ] the commutation relation ( [ commutation_1 ] ) shows that if @xmath4 is an eigenstate of h with the eigenvalue @xmath5 , then @xmath6 must also be the eigenstate of h with the same eigenvalue @xmath7 the @xmath0 operator is anti - unitary the action of which makes complex conjugation of all complex numbers .
therefore , it does matter which part of the eqn .
( [ eigenstates ] ) the operator acts on , and the appearance of the round bracket in the eqn .
( [ eigenstates ] ) is necessary .
the round bracket shows that @xmath8 acts only on the @xmath9 . skipping this bracket equals to the complex conjugation of the whole matrix element @xmath10 spin and its projection
are not changed by acting of @xmath0 operator @xcite , i.e. @xmath11 where @xmath12 is a complex number .
the above equation shows that eigenstates of the hamiltonian are the eigenstates of the symmetry operator @xmath0 as well and , therefore , they preserve the chiral symmetry . owing to the fact that eigenvalues of the hamiltonian are real values , one can show that @xmath12 must be phase only ( @xmath13 ) @xmath14 the above analysis relates to diagonal matrix elements of the hamiltonian but it can be generalized .
particularly , it is possible to make any matrix element of any operator @xmath15 commuting with @xmath0 to be real by choosing the phase @xmath16 @xmath17 the choice of this special phase @xmath16 introduces a phase convention @xcite , and this convention has to be kept for the further analysis .
in the most simple case of chirality , the nuclei are described as three - body objects being composed of two odd nucleons and an even - even core
. therefore presence of three angular momenta vectors forming a system with definite handedness is expected in tandem with specific conditions where one of the odd nucleons possesses particle - like character , the other one hole - like character and a core is triaxially deformed .
for that reason , apart of the eigenstate basis @xmath2 one can introduce equivalent basis set in which the handedness of the nucleus is specified . although the handedness , being dependent on the angular momenta vectors orientation , is a continuous parameter , it can be classified in two values @xmath18 and @xmath19 corresponding to the left- or right - handedness of the system .
the elements of the new basis set have several interesting features .
in contrast to the matrix elements calculated on the @xmath2 states being real , the matrix elements calculated on the @xmath20 , @xmath21 are complex since @xmath20 and @xmath21 are not the eigenstates of the hamiltonian .
this means also that they relate to dynamical features of the nucleus .
the states with specified handedness break the chiral symmetry since they are not the eigenstates of the @xmath0 symmetry operator i.e. @xmath22
the @xmath20 and @xmath21 not being the eigenstates of the hamiltonian posses an essential dynamical feature , i.e. the system can tunnel between left- and right - handed configurations @xmath23 one can estimate the tunneling frequency as @xmath24hz by taking the ratio @xmath25 , where @xmath26 kev is typical energy splitting observed in the chiral nuclei ( see following sections ) .
since the time of gamma quanta emission is larger than the tunneling period , it is not possible to observe such fast dynamical process experimentally through electromagnetic radiation . in this case
the time scale of the measurement is much larger than the characteristic time of the tunneling process . following the indeterminacy principle ,
the states with definite energy and undefined time are observed experimentally , i.e. only the eigenstates of the hamiltonian @xmath2 are visible .
therefore the @xmath20 and @xmath21 states have to be projected on @xmath2 basis elements in order to compare the theoretical predictions with the experimental observations .
the projection procedure is also called symmetry restoration . in the case of the chiral symmetry the rotational as well as the t - symmetry ( chirality ) has to be restored .
rotational symmetry can be restored in the spirit of generator coordinate method method ( see sect .
11.4.6 in ref .
@xcite ) by using projection operators @xmath27 @xmath28 where the handedness is not affected by spin projection since there is no @xmath0 in the operators @xmath27 .
indices @xmath29 in eqs .
( [ rzutowanko1 ] ) and ( [ rzutowanko2 ] ) stand for quantum numbers like spin of the core @xmath30 and of the odd nucleons @xmath31 , @xmath32 etc . related to structural composition of the partner band levels .
these quantum numbers will be omitted in order to examine the role of the handedness only .
the spin quantum number is not affected by action of @xmath0 operator since according to eq.(1 - 39 ) of ref.@xcite it fulfills the following commutation relations @xmath33 & = & 0 \\ \label{kolejnosc2 } \left [ r_y^t , i^2 \right ] & = & 0.\end{aligned}\ ] ] therefore the restoration of the chiral symmetry ( projection of chirality quantum number @xmath34 or @xmath35 ) have the following form @xmath36 being the normalization factors .
the @xmath37 states form chiral doubles that can be observed experimentally .
these doublets form two chiral partner bands .
phases of @xmath37 states have been chosen so as to fulfill the phase convention given by eq .
( 1 - 38 ) of ref .
@xcite . the formulae ( [ kolejnosc1],[kolejnosc2 ] ) show that the order of the projections can be reversed and one can project on the chirality first and then on the spin quantum number .
the probabilities of the electromagnetic transitions linking the excited states in the chiral bands relate to the matrix elements of the electromagnetic transition operator @xmath38 calculated on the laboratory states @xmath39 . as it was shown in ref .
@xcite , the @xmath38 fulfills the following commutation relation ( see also eqs .
( 1a-73 ) and ( 3c-10 ) in ref .
@xcite ) @xmath40 = 0\ ] ] @xmath41 being the m1 , e2 , m3 , e4 , ... transition type .
this commutation relation , together with the introduced phase convention makes all matrix elements of the electromagnetic transitions between the @xmath42 states real values and that for the inband electromagnetic transition @xmath43 , i.e. without chirality change , will assume the following form @xmath44 and , accordingly , for the @xmath45 interband transition that changes chirality @xmath46 one has to remember that matrix elements of @xmath47 and @xmath48 states take in general complex values .
relations ( [ creversed ] ) , ( [ chiral1 ] ) and ( [ chiral2 ] ) , ( [ emreversed ] ) lead to the following expressions @xmath49 the above relations show that summation and subtraction of complex conjugate values occurs in the matrix elements given by eqs.([inbandmt ] ) and ( [ interbandmt ] )
. therefore , the inband matrix element , @xmath50 , takes the final form @xmath51 and the interband matrix element , @xmath52 , transforms to @xmath53 no imaginary unit appeares in equations ( [ innband]),([innterband ] ) which is due to the the chosen phase convention .
these equations are valid when the chiral symmetry is close to the strong symmetry breaking limit . when this limit is attained , the tunneling effect between the left- and right - handed states disappears , which means that @xmath54 .
the left- and right - handed states are separated @xmath55 which means that normalization parameters @xmath56 equal to unity .
gamma transitions of m1 and e2 type can not change right - handed state to the left - handed one i.e. @xmath57 ( @xmath41= m1 , e2 ) since larger angular momentum change of the nuclear state would be required .
in this case eqs .
( [ innband ] , [ innterband ] ) reduce to @xmath58 @xmath15 being the m1 or e2 electromagnetic transition operator .
the above equations show that in the case of a strong chiral symmetry breaking limit , the b(m1 ) and b(e2 ) transition probabilities in both bands should be identical .
moreover , equations ( [ inband ] ) and ( [ interband ] ) show that if the dominating part of the @xmath59 matrix element is real then the inband electromagnetic transition should mainly be observed . on the contrary
, if the imaginary part of the @xmath59 matrix element dominates then the interband electromagnetic transition should be in favor of the inband one .
these features are observed in the majority of partner bands in which lifetimes of the excited states have been measured .
the role of the specified handedness of the intrinsic states is then explained .
staggering of the b(m1 ) value as a function of spin can not be attributed only to the existence of the @xmath20 and @xmath21 states in a quantum system .
this effect , as discussed in ref.@xcite , relates to properties ( here called structural composition ) that can occur in addition with the chiral symmetry breaking . in the quoted paper ,
the hamiltonian of the model consisting of a triaxially deformed core coupled to one proton particle and one neutron hole in the same single-@xmath60 shell was examined .
such a hamiltonian , appropriate for the description of chiral bands built on two quasi - particle @xmath61 configuration , possesses additional invariances and corresponding quantum numbers .
these quantum numbers have been used to derive the gamma selection rules which are based on the fact that the core angular momentum change for e2 and m1 transitions is @xmath62 .
this special feature of core rotation is due to its triaxial deformation .
staggering of the b(m1 ) transition probabilities relates therefore to the structural composition of the laboratory states and can be explained by examining the @xmath31 , @xmath32 , @xmath30 quantum numbers in matrix elements @xmath63 . if some of the conditions given in ref .
@xcite were not fulfilled then b(m1 ) staggering should vanish .
nonetheless , other properties related to the existence of the @xmath20 and @xmath21 states could still be observed .
such a situation takes place in @xmath64rh nuclei where partner bands are expected to be built on the configuration consisting of different @xmath60-shell levels @xcite @xcite , @xcite , as well as in the @xmath65nd isotope where partner bands are assumed to be built on the three ( instead of two ) quasiparticle configuration @xcite .
lifetimes of the excited states were measured in partner bands of @xmath66rh @xcite and @xmath65nd @xcite and , indeed , the staggering of b(m1 ) transition probability observed in those bands is very weak ( compared to that for cs isotopes ) or absent .
moreover , in the case of @xmath65nd the corresponding transition probabilities in the partner bands are almost equal .
this points to the strong chiral symmetry breaking limit in @xmath65nd and reflects different structural composition than in cs isotopes .
these examples indicate that b(m1 ) staggering should be a sensitive measure of triaxiality in two - quasiparticle single @xmath60-shell configurations .
the dependence of the b(m1 ) staggering on triaxial deformation has been studied in terms of the cqpc model @xcite where indeed the staggering effect vanishes rapidly when @xmath67-deformation deviates from 30@xmath68 .
this fact supports the role of the structural composition in description of the electromagnetic properties of the partner bands .
it follows from the above considerations that the chiral symmetry breaking in nuclear physics leads to the existence of two rotational bands , almost degenerated with similar electromagnetic properties . staggering of the b(m1 ) value quoted in many papers is an additional property originating from the structural composition of a given nucleus .
0 e. grodner , j. srebrny , a. a. pasternak , i. zalewska , t. morek , ch .
droste , j. mierzejewski , m. kowalczyk , j. kownacki , m. kisieliski , s. g. rohoziski , t. koike , k. starosta , a. kordyasz , p. j. napiorkowski , m. woliska - cichocka , e. ruchowska , w. pciennik , and j. perkowski _ phys .
lett . _ * 97 * , 172501 ( 2006 ) .
d. tonev , g. de angelis , p. petkov , a. dewald , s. brant , s. frauendorf , d. l. balabanski , p. pejovic , d. bazzacco , p. bednarczyk , f. camera , a. fitzler , a. gadea , s. lenzi , s. lunardi , n. marginean , o. mller , d. r. napoli , a. paleni , c. m. petrache , g. prete , k. o. zell , y. h. zhang , jing - ye zhang , q. zhong , and d. curien _ phys .
lett _ * 96 * , 052501 ( 2006 ) .
p. joshi , d.g .
jenkins , p.m. raddon , a.j .
simons , r. wadsworth , a.r .
wilkinson , d.b .
fossan , t. koike , k. starosta , c. vaman , j. timr , zs .
dombrdi , a. krasznahorkay , j. molnr , d. sohler , l. zolnai , a. algora , e.s .
paul , g. rainovski , a. gizon , j. gizon , p. bednarczyk , d. curien , g. duchne , j.n .
lett . _ * b595 * , 135 ( 2004 ) .
s. mukhopadhyay , d. almehed , u. garg , s. frauendorf , t. li , p.v .
madhusudhana rao , x.wang , s. s. ghugre , m. p. carpenter , s. gros , a. hecht , r.v .
f. janssens , f. g. kondev , t. lauritsen , d. seweryniak , and s. zhu _ phys .
lett _ * 99 * , 172501 ( 2007 ) .
t. suzuki , g. rainovski , t. koike , t. ahn , m. p. carpenter , a. costin , m. danchew , a. dewald , r. v. f. janssens , p. joshi , c. j. lister , o. mller , n. pietralla , t. shinozuka , j. timr , r. wadsworth , c. vaman , s. zhu _ phys .
rev . _ * c78 * , 031302(r ) ( 2008 ) . | nuclear chirality has been intensively studdied for the last several years in the context of experimental as well as theoretical approach .
characteristic gamma selection rules have been predicted for the strong chiral symmetry breaking limit that has been observed in cs isotopes .
the presented analysis shows that the gamma selection rules can not be attributed only to chiral symmetry breaking .
the selection rules relate to structural composition of the chiral rotational bands , i.e. to odd particle configuration and the deformation of the core . |
in the understanding of cancer development , mathematical modeling and numerical simulations have nowadays complemented experimental and clinical observations .
the field is now mature ; books and surveys are available , as for example @xcite .
a first class of models , initiated in the 70 s by greenspan @xcite , has considered that cancerous cells multiplication is limited by nutrients ( glucosis , oxygen ) brought by blood vessels .
models of this class rely on two kinds of descriptions ; either they describe the dynamics of cell population density @xcite or they consider the geometric motion of the tumor through a free boundary problem ; see and the references therein .
this stage lasts until the tumor reaches the size of @xmath0 mm ; then lack of nutrients leads to cell necrosis which triggers neovasculatures development @xcite that supply the tumor with enough nourishment .
this has motivated a new generation of models where growth is limited by the competition for space @xcite , turning the modeling effort towards mechanical concepts , considering tissues as multiphasic fluids ( the phases could be intersticial water , healthy and tumor cells , extra - cellular matrix ) @xcite .
this point of view is now sustained by experimental evidence @xcite .
the term homeostatic pressure , coined recently , denotes the lower pressure that prevents cell multiplication by contact inhibition .
the aim of this paper is to explain how asymptotic analysis can link the two main approaches , cell density models and free boundary models , in the context of fluid mechanics .
we depart from the simplest cell population density model , proposed in @xcite , in which the cell population density @xmath1 evolves under pressure forces and cell multiplication according to the equation @xmath2 where @xmath3 is the pressure field .
pressure - limited growth is described by the term @xmath4 , which typically satisfies @xmath5 for some @xmath6 ( the homeostatic pressure ) . the pressure is assumed to be a given increasing function of the density .
a representative example is @xmath7 with parameter @xmath8 . in the free boundary problem to be discussed later , the value @xmath9 represents the maximum packing density of cells , as discussed in @xcite . nutrients consumed by the tumor cells and brought by the capillary blood network are a usual additional ingredient to the modeling .
the situation is described in that case by the system @xmath10 \displaystyle \partial_t c - \delta c = - \varrho \ ; \psi(p , c ) , \\[2 mm ] c(x , t ) = c_b > 0 \qquad \text{as } \ ; |x| \to \infty , \end{cases}\ ] ] where @xmath11 denotes the density of nutrients , and @xmath12 the far field supply of nutrients ( from blood vessels ) . the coupling functions @xmath13
, @xmath14 are assumed to be smooth and to satisfy the natural hypotheses @xmath15 \displaystyle\partial_p\psi\leq 0,\qquad & \displaystyle\partial_c\psi \geq 0 , \qquad&\displaystyle \psi(p,0 ) = 0 .
\end{array}\ ] ] variants are possible ; for instance , we could assume that nutrients are released continuously from a vasculature , then leading to an equation as @xmath16 we will consider below the purely mechanical model under assumptions , and also the system , where nutrients are taken into account , under assumptions and .
in both cases we will show that the asymptotic limit @xmath17 yields a free boundary model of hele - shaw type , as we were looking for .
let us recall that the mathematical theory of the limit for an equation like in the absence of a growth term is well developed , the asymptotic limit being in this case the standard hele - shaw model for incompressible fluids with free boundaries .
early papers on the subject @xcite consider situations in which mass is conserved and the limit is stationary . in order to have a non - trivial limit evolution one needs some source , either in the equation or at the boundary of the domain . a first example , in which there is an inwards flux at infinity , is given in @xcite , where the authors study the limit @xmath17 of self - similar focusing ( hole - filling ) solutions to the porous medium equation @xmath18 .
a second example , in which there is a bounded boundary with a nontrivial boundary data , was first considered in @xcite , and later on in @xcite . to our knowledge
, the present paper is the first one in which the evolution in the limit is produced by a nonlinear source term in the equation . in order to pass to the limit
, three approaches have been used : weak solutions , variational formulations ( using the so - called baiocchi variable ) , and viscosity solutions ; see @xcite for this last case .
the weak formulation of hele - shaw was first introduced in @xcite , and the variational formulation in @xcite .
let us also mention that the hele - shaw graph , and hence the hele - shaw equation , can be approximated in other ways , for example by the stefan problem .
this situation has also been considered in the literature , even in situations where there is an evolution in the limit ; see for instance @xcite .
we would like to stress that including the growth term is not a simple change : several powerful but specific tools do not apply any longer .
more deeply , as we explain later , several approaches have no chance to work as they are .
this is the reason why we will work on an equation for the cell density itself rather than the pressure .
we also would like to point out that models of hele - shaw type are still an active field arising in several unexpected applications , see for instance @xcite , and that surface tension is not covered by the present work ; see @xcite .
_ organization of the paper .
_ we first study the simplest mechanical model in section [ sec : mechanical ] . the more complex model with nutrients
is studied in section [ sec : nutrients ] .
the uniqueness proofs for both cases are highly technical , and are performed separately , in sections [ sec : uniqueness ] and [ sec : nutrient_unique ] .
these two sections can safely be skipped by the readers who are mainly interested in the applications .
we finally include an appendix devoted to some interesting examples for the purely mechanical model , that illustrate several phenomena . _ some notations .
_ we will use several times the abridged notations @xmath19 , @xmath20 , meaning @xmath21 , @xmath22 .
given any @xmath23 , we denote @xmath24 , while @xmath25 .
we start our analysis by a detailed investigation of the purely mechanical model , which does not take into account the consumption of nutrients , with a pressure field given by .
a simple change of scale allows us to assume without loss of generality that @xmath26 .
we arrive to the porous medium type equation , set on @xmath27 , @xmath28 we assume that the initial data @xmath29 are such that , for some @xmath30 , @xmath31 \|\varrho_m^0- \varrho^0
\|_{l^1(\mathbb{r}^n)}\underset{m \to \infty } { \longrightarrow } 0,\quad & \|\partial_{x_i } \varrho^0_m\|_{l^1(\mathbb{r}^n)}\le c , \quad i=1,\dots , n . \end{array}\right.\ ] ] free boundary limit .
let @xmath32 be the unique bounded weak solution to .
we will prove that , along some subsequence , there is a limit as @xmath17 which turns out to be a solution to a free boundary problem of hele - shaw type .
[ th : fb ] let @xmath13 and @xmath33 satisfy and respectively . then ,
after extraction of subsequences , both the density @xmath34 and the pressure @xmath35 converge for all @xmath23 strongly in @xmath36 as @xmath17 to limits @xmath37 , @xmath38 , that satisfy @xmath39 , @xmath40 , and @xmath41 plus the relation @xmath42 , where @xmath43 is the hele - shaw monotone graph @xmath44 [ 0,\infty),\qquad & \varrho=1 .
\end{array } \right.\ ] ] note that means that a.e .
@xmath45 and @xmath46 a.e . in @xmath47 .
we will obtain also several important qualitative properties for the limit .
on the one hand , the tumor is growing and the pressure increases , that is , @xmath48 on the other hand , if the initial data @xmath33 have a common compact support , then the limit solution propagates with a finite speed : @xmath49 and @xmath50 are compactly supported for all @xmath51 .
finally , the constructed limit solution enjoys the monotonicity property , inherited from the case where @xmath52 is finite , @xmath53 p_\infty(t ) \geq
\bar p_\infty(t ) .
\end{array } \right.\ ] ] transport equation and complementarity formula .
we will also obtain @xmath54 estimates on the gradients @xmath55 that will show on the one hand that @xmath56 and on the other hand that @xmath57 solves a transport equation .
[ th : transport ] the limit solution @xmath57 obtained in theorem [ th : fb ] satisfies @xmath58 in a weak sense .
a direct calculation shows that the pressure @xmath35 satisfies @xmath59 hence , if we let @xmath17 , we formally obtain the _ complementarity formula _
@xmath60 this formal computation can be made rigorous .
[ thm : complementarity ] under the assumptions of theorem [ th : fb ] , the limit pressure @xmath61 satisfies the complementarity formula @xmath62 it is worth noticing that is equivalent to the strong convergence of @xmath55 in @xmath63 for all @xmath23 ; see lemma [ lemma : equivalence.complementarity.strong.convergence ] .
since the limit pressure @xmath61 is expected to be continuous in space for all positive times , the positivity set @xmath64 should be well defined for all @xmath51 .
notice that it coincides almost everywhere with the set where @xmath65 ; see figure [ fig : strict ] . indeed , on the one hand , by the definition of the graph @xmath43 we have @xmath66
; on the other hand , if we had @xmath67 and @xmath65 in some set with positive measure , then @xmath68 would continue to grow ( exponentially ) there , which is a contradiction .
therefore , @xmath69 may be regarded as the _ tumor _ , while the regions where @xmath70 ( mushy regions , in the literature of phase - changes ) correspond to _ precancer _ cells .
_ a solution to the mechanical model , in one dimension with @xmath71 .
left : @xmath72 .
right : @xmath73 .
the upper line is @xmath74 ; the bottom line is @xmath3 ( scale enlarged for visibility ) .
notice that the density scales are not the same in the two figures .
the initial data is taken with compact support and the solution is displayed for a time large enough ( see figure [ fig : strict+ ] below for and intermediate regime ) .
, title="fig:",width=264 ] large .
_ a solution to the mechanical model , in one dimension with @xmath71 .
left : @xmath72 .
right : @xmath73 .
the upper line is @xmath74 ; the bottom line is @xmath3 ( scale enlarged for visibility ) .
notice that the density scales are not the same in the two figures .
the initial data is taken with compact support and the solution is displayed for a time large enough ( see figure [ fig : strict+ ] below for and intermediate regime ) .
, title="fig:",width=264 ] + the complementarity formula indicates that the limit pressure at time @xmath75 , @xmath50 , should solve the elliptic equation @xmath76 a problem which is wellposed if @xmath69 is smooth enough , since @xmath77 is decreasing .
this implies in particular that in general regularity in time for the pressure is missing , since time discontinuities may show up when two tumors meet ; see subsection [ sec : timedisc ] in the appendix . geometric motion vs. equation on the cell number density .
to complete the description of the limit problem we should be able to trace @xmath69 starting from its initial position .
the pressure equation suggests that we should have @xmath78 which leads to a geometric motion with normal velocity @xmath79 at the boundary of @xmath69 given by @xmath80 thus we have arrived to a geometric hele - shaw type problem , which is the classical one when @xmath81 .
the above formal computation is expected to be true if we prescribe a fixed initial pressure @xmath82 ( which implies that the initial densities converge to the indicator function of the positivity set of @xmath83 ) . when @xmath81 , it was proved to be true in a viscosity sense in @xcite ; see also @xcite for an earlier result in this direction using a variational formulation of baiocchi type . however
, if the initial densities @xmath29 are such that @xmath84 is below 1 in a set with positive measure , the result is no longer true .
the main point is that the tumor may meet precancer zones . at a meeting point
@xmath85 , the example in subsection [ subsect : effect.of.eq.density ] in the appendix suggests that the tumor grows faster ( also for the case @xmath81 ) , with a normal velocity given by a rule of the form @xmath86 where @xmath87 is some limit of @xmath74 as @xmath88 and @xmath89 from the outside of @xmath69 .
we leave open this problem , which seems to be a challenging extension of the viscosity method in @xcite .
even if we consider the modified geometric motion law instead of , the geometric formulation does not carry all the information of the limit solution .
indeed , the density in precancer zones evolves , with an exponential growth , until it reaches the level @xmath90 , a fact that is not captured neither by , nor by ; see figure [ fig : strict+ ] and the examples in the appendix . when @xmath81 we also need to give a full description of the limit .
however , in this latter case the evolution in the mushy regions is simpler , since the density does not change until it interacts with the positivity set for the pressure . and the initial data @xmath74 is less than @xmath91 .
the solution is displayed at four different times .
it shows how the smooth part of @xmath74 strictly less than @xmath91 is growing with @xmath92 ( figure on the left ) . when @xmath74 reaches the value @xmath91 , the pressure becomes positive , increases and creates a moving front that delimitates the growing domain where @xmath93 .
thin line is @xmath74 and thick line is @xmath3 as functions of @xmath94 .
see also figure [ fig : tw ] in the appendix.,title="fig:",width=136 ] and the initial data @xmath74 is less than @xmath91 .
the solution is displayed at four different times .
it shows how the smooth part of @xmath74 strictly less than @xmath91 is growing with @xmath92 ( figure on the left ) .
when @xmath74 reaches the value @xmath91 , the pressure becomes positive , increases and creates a moving front that delimitates the growing domain where @xmath93 .
thin line is @xmath74 and thick line is @xmath3 as functions of @xmath94 .
see also figure [ fig : tw ] in the appendix.,title="fig:",width=136 ] and the initial data @xmath74 is less than @xmath91 .
the solution is displayed at four different times .
it shows how the smooth part of @xmath74 strictly less than @xmath91 is growing with @xmath92 ( figure on the left ) . when @xmath74 reaches the value @xmath91 , the pressure becomes positive , increases and creates a moving front that delimitates the growing domain where @xmath93 .
thin line is @xmath74 and thick line is @xmath3 as functions of @xmath94 .
see also figure [ fig : tw ] in the appendix.,title="fig:",width=136 ] and the initial data @xmath74 is less than @xmath91 .
the solution is displayed at four different times .
it shows how the smooth part of @xmath74 strictly less than @xmath91 is growing with @xmath92 ( figure on the left ) . when @xmath74 reaches the value @xmath91 , the pressure becomes positive , increases and creates a moving front that delimitates the growing domain where @xmath93 .
thin line is @xmath74 and thick line is @xmath3 as functions of @xmath94 .
see also figure [ fig : tw ] in the appendix.,title="fig:",width=136 ] + let us point out that , in the case of a compactly supported initial data , precancer zones disappear in finite time . uniqueness . to complete the description of the asymptotic limit , we conclude with a uniqueness result , relying on a hilbert duality method , for the free boundary problem .
[ th : unique ] there is a unique pair @xmath95 of functions in @xmath96 , @xmath97 , @xmath98 , satisfying and such that for all @xmath23 : * @xmath99 is uniformly compactly supported for @xmath100 $ ] ; * @xmath101 ; * @xmath102 , @xmath103 .
it is important to notice that if , in addition to , we assume that the initial data @xmath104 are uniformly compactly supported , then the limit solution @xmath57 to given by theorem [ th : fb ] falls within the uniqueness class . as a consequence , convergence is not restricted to a subsequence .
_ the compact support assumption in theorem [ th : unique ] can be removed , to the cost of using rather technical arguments , in the style of the ones employed in the uniqueness proof for the case with nutrients , theorem [ thm : uniqueness.system ] below .
we keep it for the sake of clarity .
the rest of this section is devoted to the proofs of the above mentioned results , except uniqueness , which is postponed to section [ sec : uniqueness ] .
we consider the solutions to and proceed to obtain the estimates that allow us to pass to the limit and prove theorem [ th : fb ] .
this is a standard porous media type equation and all the manipulations below can be justified , see the monograph @xcite .
the main observation here is that the lower estimate on @xmath105 that does the job when passing to the limit in porous media equations without a source , does not work in the present case .
however , a control can be obtained on the quantities @xmath106 , which will be enough for our purposes .
@xmath107 bounds for @xmath34 , @xmath35 .
standard comparison arguments yield
@xmath108 this allows in particular to avoid initial layers ( which are present when @xmath109 ; see @xcite ) .
@xmath110 bounds for @xmath34 , @xmath35 .
let @xmath34 , @xmath111 be two ( non - negative ) solutions to , and @xmath112 , @xmath113 the corresponding pressures .
we have @xmath114 from where we get @xmath115
this almost contraction property yields the uniform ( in @xmath52 ) bound @xmath116 on the other hand , @xmath117 .
thus , using we conclude @xmath118 a semiconvexity estimate for @xmath35 . according to the hypotheses on the growth function , @xmath119}\left(\phi ( p)- p\phi'(p)\right ) > 0
we will prove that @xmath120 as a consequence , in the limit we will have @xmath121 .
we borrow the idea to prove inequality from @xcite , where the case @xmath81 was considered .
we write the equation as @xmath122 let us denote @xmath123 .
since @xmath124 , we have @xmath125 \partial_t ( \phi(p_m))=\phi'(p_m ) \partial_t p_m=(m-1)p_m\phi'(p_m)w+ \phi'(p_m)|\nabla p_m|^2\\[10pt ] \qquad\qquad\ \ , \geq ( m-1)p_m\phi'(p_m)w+ 2 \nabla ( \phi(p_m ) ) \cdot \nabla p_m , \end{array}\ ] ] which gives @xmath126 the function @xmath127 is a subsolution to this equation , and follows . _ remarks . _
( i ) the right hand side of behaves as @xmath128 for @xmath129 .
\(ii ) for @xmath130 we have @xmath131 , and we recover the well - known result for the case @xmath81 , namely @xmath132 . bounds for @xmath133 , @xmath134 .
we combine with to obtain an estimate from below for the time derivative of the pressure , @xmath135 this in turn gives an estimate from below for the time derivative of the density , @xmath136 the monotonicity inequalities for the limit problem are then obtained just by letting @xmath17 in and .
we now use @xmath137 to obtain @xmath138 this , together with , leads to a uniform bound in space and time for time intervals of the form @xmath139 $ ] . an analogous computation shows that @xmath140 & & \displaystyle+(m-1)r_\phi \int_{\frac1{m-1}}^t \left(\frac{\textrm{e}^{-(m-1)r_\phi t } } { 1- \textrm{e}^{-(m-1)r_\phi t}}\int_{\mathbb{r}^n } p_m(t)\right)\,dt \le c(t ) . \end{array}\ ] ] @xmath110 bounds for @xmath141 , @xmath55 .
let @xmath142}|\phi'(p)|>0 $ ] .
we consider the equation for @xmath143 , multiply it by @xmath144 , and use kato s inequality ; thanks to the monotonicity of @xmath13 we obtain @xmath145 & \le & \phi(0)|\partial_{x_i}\varrho_m|-\alpha \varrho_m |\partial_{x_i } p_m| .
\end{array}\ ] ] integrating in @xmath146 , we get @xmath147 which yields , on the one hand , that @xmath148 and on the other hand that @xmath149 convergence and identification of the limit . since the families @xmath34 and @xmath35 are bounded in @xmath150 , we have strong convergence in @xmath151 both for @xmath34 and @xmath35 . to pass from local convergence to global convergence in @xmath36
, we need to prove that the mass in an initial strip @xmath152 $ ] and in the tails @xmath153 are uniformly ( in @xmath52 ) small if @xmath154 is large enough .
the control on the initial strip is immediate using our uniform , in @xmath52 and @xmath75 , bounds for @xmath155 and @xmath156 .
in order to control the tails , we consider @xmath157 such that @xmath158 , @xmath159 for @xmath160 and @xmath161 for @xmath153 , and define @xmath162 .
then , for any @xmath163 , @xmath164 & \le & \displaystyle cr^{-2}+\phi(0)\int_{\mathbb{r}^n}\varrho_m(t)\varphi_r .
\end{array}\ ] ] therefore , for any @xmath100 $ ] we have @xmath165 \displaystyle & \le&\displaystyle\textrm{e}^{\phi(0)t } \left(\|\varrho^0_m-\varrho^0\|_{l^1(\mathbb{r}^n)}+\int_{\mathbb{r}^n}\varrho^0\varphi_r+cr^{-2}t \right ) \le \displaystyle{\varepsilon}\end{array}\ ] ] for @xmath154 and @xmath52 are large enough . since @xmath166 for any @xmath167 , the tail control for the pressures follows easily .
after extraction of subsequences , we can pass to the a.e .- limit in the equation @xmath168 to obtain that @xmath169 .
passing to the limit in the estimates , we have also @xmath170 , @xmath171 and @xmath172 for all @xmath23 .
all the above is enough to prove that the pair @xmath57 solves the hele - shaw equation but for the question of the initial data .
time continuity and initial trace .
we need delicate arguments that are fortunately in the folklore of the topic of nonlinear diffusion equations , and borrow also the conclusion from @xcite . because @xmath68 is non - decreasing in time , we can write for a test function @xmath173 and @xmath174 , @xmath175 \zeta = \displaystyle\int_{t_1}^{t_2 } \int_{\mathbb{r}^n } \left [ p_\infty \delta \zeta + \varrho_\infty \phi(p_\infty ) \zeta\right ] \\[20pt ] & \leq c(t ) ( t_2-t_1)\left ( \| \delta \zeta \|_\infty + 1\right ) . \end{array}\ ] ] taking a sequence of such uniformly smooth functions that converge to @xmath91 , we find that @xmath176 ( in fact with a locally uniform lipschitz constant ) . in order to identify the initial trace , we observe that for any test function @xmath177 as above , @xmath178 letting @xmath17 , we have @xmath179 letting first @xmath180 and then @xmath181 , we conclude that @xmath182 in @xmath183 .
we now prove theorems [ th : transport ] and [ thm : complementarity ] , both of them connected with the fact that @xmath184 for all @xmath23 . since we have already proved that both @xmath34 and @xmath35 converge strongly in @xmath36 for all @xmath23 , the first of these theorems just depends on obtaining a uniform bound in @xmath63 for @xmath55 .
_ proof of theorem [ th : transport ] .
_ we rewrite the pressure equation as @xmath185 integrating in @xmath186 we obtain the required estimate , @xmath187 @xmath188 _ remark . _ combining with , we obtain @xmath189 for all @xmath163 .
this gives a uniform estimate for @xmath190 , @xmath191 $ ] , for any fixed values @xmath192 .
however , it does not allow to go down to @xmath193 . from the equation in and the @xmath110 bound on @xmath134
we conclude that @xmath194 this is an optimal bound since @xmath61 has corners on @xmath195 .
because this only gives space compactness , it is not enough to establish the complementarity formula on the pressure when passing to the limit in .
we will need to perform a time regularization argument _ la steklov_. _ proof of theorem [ thm : complementarity ] .
_ let @xmath196 be a regularizing kernel in time with support @xmath197 .
we use the notation @xmath198 .
the equation gives @xmath199 from here it follows that @xmath200 is uniformly ( in @xmath52 ) smooth in time and @xmath201 in space for @xmath202 fixed .
therefore @xmath203 converges strongly in @xmath151 .
hence , after multiplying by @xmath35 , we may pass to the limit to obtain @xmath204 in order to estimate the right hand side we make the following decomposition , @xmath205 \qquad\qquad\qquad + \underbrace{\frac{m}{m-1}\int_{\mathbb{r}}\left(\varrho_m^{m-1}(t ) -\varrho_m^{m-1}(s)\right ) \left(\partial_t \varrho_{m}(s ) + \frac cs\right ) \omega_{\varepsilon}(t - s)ds}_{{\text{ii } } _ m(t)}\\[5 mm ] \qquad\qquad\qquad\underbrace{-\frac{cm}{m-1}\int_{\mathbb{r } } \left(\varrho_m^{m-1}(t ) -\varrho_m^{m-1}(s)\right ) \frac{\omega_{\varepsilon}(t - s)}s\,ds}_{{\text{iii } } _ m(t ) } , \end{array}\ ] ] where @xmath206 is a constant such that @xmath207 ; see estimate . for the first term we have @xmath208 regarding the second term
, the estimate implies that @xmath209 .
therefore , @xmath210 . let @xmath211 , and @xmath212 the smallest time in its support .
then , @xmath213 the third term is easy to treat . since @xmath214 , for any test function @xmath177 as above , @xmath215 as @xmath216 .
these three calculations give , in the limit @xmath217 , @xmath218 to recover the desired information , @xmath219 it remains to pass to the limit as @xmath216 after noticing that @xmath220 and that the differential term can be treated through its weak formulation ; indeed , after testing it against a test function as before , it is written as @xmath221\ ] ] which passes to the limit @xmath222 because we already know that @xmath223 .
therefore , @xmath224=0.\ ] ] @xmath188 as we have already mentioned , the complementarity formula is equivalent to the strong convergence of the gradients .
[ lemma : equivalence.complementarity.strong.convergence ] the complementarity formula holds if and only if @xmath225 strongly in @xmath63 for all @xmath23 , _ proof .
_ we consider a test function @xmath226 and use and to obtain @xmath227 this implies , after an integration by parts , that @xmath228 \underset{m \to \infty}{\longrightarrow } 0.\ ] ] let @xmath229 .
thanks to fatou s lemma , @xmath230 . on the other hand ,
since @xmath35 converges strongly , after extraction of subsequences we have @xmath231 = r,\ ] ] and @xmath232 is equivalent to the strong convergence . @xmath188 if the initial data are compactly supported uniformly in @xmath52 , we can control the supports of @xmath233 and @xmath234 uniformly in @xmath52 and @xmath235 $ ] for all @xmath23 .
this implies that the speed of propagation is finite for the limit problem .
the control is performed comparing the pressures with functions of the form @xmath236 the key point is that these functions , which are viscosity solutions of the hamilton - jacobi equation @xmath237 , are supersolutions to the equation @xmath238 for some time interval which does not depend on @xmath52 .
this idea was already used for the case @xmath81 in @xcite . however , in the absence of reaction , @xmath239 is a supersolution for all times , and the proof is a bit easier . let @xmath33 satisfying and such that all their supports are contained in a common ball @xmath240 .
then , for every @xmath23 , there is a radius @xmath241 depending only on @xmath242 , @xmath154 and @xmath243 , such that the supports of @xmath244 are contained in the ball @xmath245 for all @xmath246 and @xmath100 $ ] .
_ we consider @xmath239 as in with @xmath247 and @xmath206 large enough so that @xmath248 for all @xmath249 and @xmath246 .
an easy computation shows that @xmath250 for all @xmath251 $ ] .
since the functions @xmath35 are subsolutions to , we conclude that the supports of the functions @xmath34 are contained in the ball of radius @xmath252 for all @xmath251 $ ] .
moreover , @xmath253 , and the argument can be iterated to reach the time @xmath243 in a finite number of steps
in this section we prove , under suitable assumptions , that the limit problem has a unique solution , theorem [ th : unique ] .
the main difficulty comes from the fact that @xmath3 is not a lipschitz , single - valued function of @xmath74 .
hence , we can not apply directly the ideas developed in @xcite to adapt hilbert s duality method to the porous medium equation .
the technique has to be modified , in the spirit of @xcite .
dual problem .
consider two solutions , @xmath254 , @xmath255 .
let @xmath256 be a bounded domain containing the supports of both solutions for all @xmath100 $ ] , and @xmath257 .
then we have @xmath258=0\ ] ] for all suitable test functions @xmath259 .
this can be rewritten as @xmath260=0,\ ] ] where , for some fixed @xmath261 , @xmath262 0 \leq \displaystyle b=\frac{p_1-p_2}{(\varrho_1-\varrho_2)+(p_1-p_2 ) } \leq 1,\\[10pt ] 0 \leq \displaystyle c= \ ; - \varrho_2\,\frac{\phi(p_1)-\phi(p_2)}{p_1-p_2}\ ; \leq \nu . \end{array}\ ] ] to arrive to these bounds on @xmath263 and @xmath264 , we define @xmath265 when @xmath266 , even when @xmath267 , and @xmath268 when @xmath267 , even when @xmath269 .
the idea of hilbert s duality method is to solve the _ dual problem _
@xmath270 \psi=0\quad\text{in } \partial\omega\times(0,t),\quad \psi(\cdot , t)=0\quad\text{in } \omega , \end{array } \right.\ ] ] for any smooth function @xmath271 , and use @xmath259 as test function .
this would yield @xmath272 from where uniqueness for the density is immediate .
uniqueness for the pressure would then follow from .
however , on the one hand the coefficients of the dual problem are not smooth , and , on the other , since @xmath263 and @xmath264 are not strictly positive , the dual equation is not uniformly parabolic . a smoothing argument is required . regularized dual problem
let @xmath273 , @xmath274 , @xmath275 , @xmath276 be sequences of smooth bounded functions such that @xmath277 \|b - b_n\|_{l^2(\omega_t)}<\beta / n,\qquad&1/n < b_n\le 1,\\[3 mm ] \|c - c_n\|_{l^2(\omega_t)}<\delta_1/n,\qquad&0 \leq c_n < m_1,\qquad&\| \partial_t c_n \|_{l^1(\omega_t ) } \leq k_1 \\[3 mm ] \|\phi(p_1)-\phi_{1,n}\|_{l^2(\omega_t)}<\delta_2/n,\qquad&|\phi_{1,n}|<m_2,\qquad & \| \nabla \phi_{1,n } \|_{l^2(\omega_t ) } \leq k_2 , \end{array } \right.\ ] ] for some constants @xmath278 . for any smooth function @xmath271
, we consider the solution @xmath279 of the backward heat equation @xmath280 \psi_n=0\quad\text{in } \partial\omega\times(0,t),\quad \psi_n(\cdot , t)=0\quad\text{in } \omega .
\end{array } \right.\ ] ] the coefficient @xmath281 is continuous , positive and bounded below away from zero .
thus , the equation satisfied by @xmath279 is uniformly parabolic in @xmath282 .
hence @xmath279 is smooth and can be used as a test function in . combining and , we have @xmath283 where @xmath284 i^2_{n}=\displaystyle\iint_{\omega_t }
\big((\varrho_1-\varrho_2)+(p_1-p_2)\big ) ( b - b_n ) \left(\delta\psi_n - c_n \psi_n\right ) , \\[10pt ] i^3_{n}=\displaystyle\iint_{\omega_t } ( \varrho_1-\varrho_2)(\phi(p_1)-\phi_{1,n})\psi_n,\\[10pt ] i^4_{n}=\displaystyle\iint_{\omega_t } \big((\varrho_1-\varrho_2)+(p_1-p_2)\big ) b(c - c_n)\psi_n . \end{array}\ ] ] limit @xmath285 .
uniform bounds and conclusion .
our aim is to prove that @xmath286 , @xmath287 , which implies that @xmath269 .
for this task we will use some bounds that we gather in the following lemma .
[ lem : bounds.uniqueness.scalar.equation ] there are constants @xmath288 , @xmath289 , depending on @xmath243 and @xmath271 , but not on @xmath290 , such that @xmath291 _ proof .
_ the first bound is just the maximum principle because @xmath292 is non - negative and @xmath293 uniformly bounded .
the second bound is obtained multiplying the equation in by @xmath294 .
after integration in @xmath295 , we get @xmath296 \displaystyle \quad + \int_t^t \int_\omega \left [ -\partial_t c_n \frac { \psi_n^2}{2 } - \phi_{1,n } |\nabla \psi_n|^2 - \psi_n \nabla \phi_{1,n } \cdot\nabla \psi_n+ c_n \phi_{1,n } \psi_n^2 + \psi_n \delta g - g c_n \psi_n \right]\\[4 mm ]
\displaystyle\ \leq k\left(1-t+ \int_t^t \|\nabla \psi_n ( s ) \|_{l^2(\omega)}^2\ , ds\right ) , \end{array}\ ] ] where we denote by @xmath297 various constants independent of @xmath290 .
we now use the grnwall lemma to conclude our second uniform bound .
then we can use again the inequality to conclude the third one .
@xmath298@xmath188 we now get @xmath299 & \le&\displaystyle kn^{1/2}\|a - a_n\|_{l^2(\omega_t)}\le k\alpha / n^{1/2};\\[10pt ] \displaystyle |i^2_{n}|&\le & \displaystyle k\iint_{\omega_t}|b - b_n|\ , k\|(a_nn^{1/2}/b_n)^{1/2}(b - b_n)\|_{l^2(\omega_t)}\\[10pt ] & \le&\displaystyle kn^{1/2}\|b - b_n\|_2\le k\beta / n^{1/2};\\[10pt ] \displaystyle |i^3_{n}|&\le & \displaystyle\iint_{\omega_t } k\|\phi(p_1)-\phi_{1,n}\|_{l^2(\omega_t ) } \le k\delta_2/n ; \\[10pt ] \displaystyle |i^4_{n}|&\le & \displaystyle k\iint_{\omega_t } b \end{array}\ ] ] once we have proved that @xmath269 , equation says that @xmath300 = 0,\ ] ] from where the result will follow by taking as test function @xmath301 , using the monotonicity of @xmath13 .
when nutrients are taken into account , we are led to problem .
we again assume that the pressure field is given by , with @xmath9 set equal to 1 .
our assumptions on the initial data are stronger than for the purely mechanical problem .
namely , in addition to , we assume that for some @xmath302 such that @xmath303 , @xmath304 \|{\rm div}(\varrho_m^0 \nabla p_m^0 ) + \varrho_m^0 \ ; \phi(p_m^0,c_m^0)\|_{l^1(\mathbb{r}^n)}\le c , \quad \|\delta c_m^0 - \varrho_m^0 \ ; \psi(p_m^0,c_m^0)\|_{l^1(\mathbb{r}^n)}\le c. \end{array } \right.\ ] ] an interesting difference with the purely mechanical model is that when nutrients are not enough the cells might die ( or become quiescent ) , which is represented by @xmath305 for @xmath11 small enough with @xmath3 given .
these cells , which are in principle in the center of the tumor , are replaced by cells from the boundary moving inwards by pressure forces .
in particular the growth inequalities can not hold true here .
a typical choice is @xmath306 , @xmath307 with @xmath308 two constants and @xmath309 a function having the same properties as in the purely fluid mechanical model .
as in the case without nutrients , in the limit we obtain a free boundary problem of hele - shaw type .
let @xmath13 , @xmath14 satisfy , and @xmath310 satisfy the hypotheses and . then ,
after extraction of subsequences , the density @xmath34 , the nutrient @xmath311 and the pressure @xmath35 converge for all @xmath23 strongly in @xmath36 as @xmath17 to limits @xmath312 that satisfy @xmath313 , @xmath314 and @xmath40 , and @xmath315 \displaystyle \partial_t c_\infty = \delta c_\infty - \varrho_\infty \psi(p_\infty , c_\infty)\quad&c_\infty(0)=c^0 , \end{array } \right.\ ] ] in a distributional sense , plus the relation @xmath316 , where @xmath43 is the hele - shaw graph .
[ th : fbn ] when we say that @xmath317 is a solution to in a distributional sense , we mean that @xmath318 \displaystyle \iint_{q } \big\{c_\infty \zeta_t+c_\infty\delta \zeta-\varrho_\infty\psi(p_\infty , c_\infty)\zeta\big\}=-\int_{\mathbb{r}^n}c^0\zeta(\cdot,0 ) , \end{array}\ ] ] for all test functions @xmath319 .
as mentioned above , for the system we lose the monotonicity properties , and we are not able to prove the expected continuity of @xmath68 and @xmath320 in @xmath183 .
however , we are still able to prove that @xmath321 in @xmath63 .
therefore , the equation on the cell density can also be written as a transport equation , @xmath322 if the initial data @xmath29 are compactly supported uniformly in @xmath52 , we can control the supports of @xmath234 uniformly in @xmath52 for each @xmath51 .
hence , in the limit problem tumors propagate with a finite speed ( this is not true for the nutrients ) , and a free boundary shows up .
this can be proved through a comparison argument with the same barrier functions as in the scalar case .
moreover , the constructed solution will fall within a class for which we prove uniqueness .
[ thm : uniqueness.system ] there is a unique triple @xmath323 , @xmath324 , @xmath325 , satisfying in the distributional sense and such that for all @xmath23 : * @xmath99 is uniformly compactly supported for @xmath100 $ ] ; * @xmath326 ; * @xmath327 , @xmath328 . as in the purely mechanical model
, we can write an equation for the pressure @xmath35 , @xmath329 this suggests that the complementarity formula @xmath330 holds .
however , we have not been able to establish it rigorously .
the reason is that , in contrast with the case without nutrients , we have failed to control @xmath331 from below by means of a comparison argument , or to prove the strong convergence of @xmath55 in @xmath63 for all @xmath23 .
we leave open the question to give conditions on @xmath332 allowing to prove formula .
the rest of this section is devoted to prove the first of these theorems .
the uniqueness result is postponed to a later section .
@xmath107 bounds for @xmath34 , @xmath35 , @xmath311 .
the assumptions on the growth functions imply , using standard comparison arguments , @xmath333 @xmath110 bounds for @xmath34 , @xmath35 , @xmath311 .
let @xmath334 and @xmath335 be two solutions of the system for a fixed @xmath52 , with corresponding pressures @xmath3 and @xmath336 .
we subtract the equation for @xmath337 from the equation for @xmath74 , and multiply the resulting equation by @xmath338 , and do an analogous manipulation for the @xmath11 variable .
we integrate by parts , and obtain @xmath339(t ) , \\[10pt ] \displaystyle\frac{d } { dt}\|(c-\hat c)(t)\|_{l^1(\mathbb{r}^n)}\le- \int_{\mathbb{r}^n}\big[\big(\varrho \psi(p , c)-\hat \varrho\psi(\hat p,\hat c)\big)\mathop{\rm sign}(c-\hat c)\big](t ) . \end{array } \right.\ ] ] let @xmath340 be a positive constant to be chosen later .
we have @xmath341 \quad\displaystyle \underbrace{\int_{\mathbb{r}^n}\big[\big(\varrho-\hat\varrho\big)\big(\phi(p , c)\mathop{\rm sign}(\varrho-\hat\varrho)-\mu\psi(p , c)\mathop{\rm
sign}(c-\hat c)\big)\big](t)}_{\mathcal{i}}\\[4 mm ] \quad+\underbrace{\int_{\mathbb{r}^n}\big[\hat\varrho\big((\phi(p , c)-\phi(p,\hat c))\mathop{\rm sign}(\varrho-\hat\varrho ) -\mu(\psi(p , c)-\psi(p,\hat c))\mathop{\rm sign}(c-\hat c)\big)\big](t)}_{\mathcal{j}}\\[4 mm ] \quad+\underbrace { \int_{\mathbb{r}^n } \big [ \hat\varrho \big ( ( \phi(p,\hat c)-\phi(\hat p,\hat c ) ) \mathop{\rm sign}(\varrho-\hat\varrho ) -\mu(\psi(p,\hat c)-\psi(\hat p,\hat c ) ) \mathop{\rm sign}(c-\hat c ) \big ) \big](t ) } _ { \mathcal{k}}. \end{array}\ ] ] let @xmath342 , @xmath343 , @xmath344 . since @xmath345 , if @xmath346 , there are constants @xmath206 independent of @xmath52 such that @xmath347(t)\leq c\|(\varrho-\hat\varrho)(t)\|_{l^1(\mathbb{r}^n ) } , \\[4 mm ] \displaystyle\mathcal{j } \le \| \hat \rho\|_{l^\infty(q ) } \left(\| \partial_c\phi\|_{l^\infty(a ) } + \mu \|
\partial_c\psi\|_{l^\infty(a ) } \right ) \|(c - \hat c)(t ) \|_{l^1(\mathbb{r}^n ) } \leq c \|(c - \hat c)(t)\|_{l^1(\mathbb{r}^n)},\\[4 mm ] \displaystyle\mathcal{k}\le\int_{\mathbb{r}^n } \big [ \hat \varrho \big ( - \big| \phi(p,\hat c)- \phi(\hat p,\hat c ) \big| + \mu \big| \psi(p,\hat c)- \psi(\hat p,\hat c ) \big| \big)\big](t)\le0 . \end{array}\ ] ] we conclude that @xmath348 therefore , @xmath349 this gives uniqueness , and choosing @xmath350 , and @xmath351 , we find the uniform estimates @xmath352 as in the purely fluid mechanical model , from the @xmath353 bounds and , we conclude that , for @xmath163 , @xmath354 @xmath110 bounds on the derivatives of @xmath34 and @xmath311 .
we differentiate the two equations of the system with respect to time , multiply the first one by @xmath355 and the second by @xmath356 and use kato s inequality to obtain @xmath357 \displaystyle \partial_t |\partial_t c_m|-\delta(|\partial_t c_m|)\le -\partial_t\varrho_m \psi\mathop{\rm sign}(\partial_t c_m)-\varrho_m\partial_p\psi \partial_t p_m\mathop{\rm sign}(\partial_t c_m)-\varrho_m\partial_c\psi|\partial_t c_m| . \end{array } \right.\ ] ] we integrate in space and add the two equations to obtain , using the monotonicity properties of the growth functions , @xmath358 \int_{\mathbb{r}^n } \left ( |\partial_t \varrho_m(t)| \big(\phi + \mu \psi\big)(0,c_b ) + \| \varrho_m \|_{l^\infty(q ) } \|\partial_c\phi\|_{l^\infty(a)}|\partial_t c_m(t)| \right ) , \end{gathered}\ ] ] and thus , thanks to assumption , @xmath359 we can estimate the space derivatives in the same way , and arrive to @xmath360 estimates on the derivatives of @xmath35 . from the first equation in , using the strict monotonicity of @xmath13 with respect to @xmath3 and the @xmath361 estimates that we have already proved , we get @xmath362 where , as above , @xmath343
. therefore , @xmath363 the control on the space derivatives follows by a similar argument , and we obtain @xmath364 convergence and identification of the limit .
the above estimates give strong convergence in @xmath151 for @xmath34 , @xmath365 and @xmath35 .
our uniform control of the @xmath110 norms of these quantities in @xmath52 and @xmath100 $ ] shows that the mass in thin initial strips is uniformly in @xmath52 small .
therefore , to get global convergence in @xmath36 , we just need to control the tails .
proceeding as in the case without nutrients , we get @xmath366 from where the tail control for @xmath34 , and then for @xmath35 , follows easily . as for the nutrients
, we have @xmath367 but we already have a tail control for @xmath34 . hence , for all @xmath154 and @xmath52 large enough , @xmath368 which immediately implies a tail control for @xmath365 .
the strong convergence in @xmath36 is enough to pass to the limit and recover the system in a distributional sense with the limiting pressure graph relation .
@xmath54 bounds for @xmath55 , @xmath369 .
the bound for the gradients of the pressures follows from equation written in the form @xmath370 indeed , integrating in @xmath186 we obtain @xmath371 this is enough to prove that the limit satisfies the transport equation .
the estimate for the gradients of the nutrients , which is needed to prove that the limit falls within the uniqueness class , is even easier .
we just have to multiply the equation for the nutrients by @xmath311 and integrate by parts in @xmath186 .
since @xmath372 , we obtain @xmath373
we meet two difficulties . on the one hand , as in the case without nutrients
, @xmath3 is not a lipschitz , single - valued function of @xmath74 . on the other hand ,
the density of nutrients is not compactly supported .
hence we can not restrict to the case of a bounded domain . to take care of the lack of compact support of @xmath11
, we will use an idea from @xcite . dual system .
let @xmath374 and @xmath375 be two solutions with the same initial data and fix a final time @xmath243 . since they satisfy in the sense of distributions , we have @xmath376 \displaystyle\iint_{q_t}\left\ { ( c_1-c_2 ) \partial_t
\zeta + ( c_1-c_2 ) \delta \zeta - ( \varrho_1 \psi(p_1,c_1 ) - \varrho_2 \psi(p_2,c_2 ) ) \zeta \right\ } = 0 , \end{cases}\ ] ] for any pair of test functions @xmath377 such that @xmath378 .
denoting @xmath379 and @xmath380 , we can rewrite the above equations as @xmath381 \displaystyle \iint_{q_t } \left\{(c_1-c_2 ) \left ( \partial_t \zeta + \delta \zeta - e \zeta \right ) - ( \varrho_1- \varrho_2+p_1-p_2 ) \left(a\psi_1 - fb \right ) \zeta \right\}= 0 , \end{cases}\ ] ] where @xmath382 0 \leq \displaystyle c= \ ; -\varrho_2\frac{\phi(p_1,c_1)-\phi(p_2,c_1)}{p_1- p_2 } \le \nu_1 , \ & \displaystyle 0\leq d= \varrho_2 \frac{\phi(p_2,c_1)-\phi(p_2,c_2)}{c_1- c_2}\le \nu_2 , \\[12pt ] \displaystyle0\leq e= \varrho_2 \frac{\psi(p_2,c_1)-\psi(p_2,c_2)}{c_1- c_2 } \le \nu_3,\ & 0 \leq \displaystyle f= \ ; -\varrho_2\frac{\psi(p_1,c_1)-\psi(p_2,c_1)}{p_1- p_2}\le \nu_4 . \end{array}\ ] ] adding these two equations we get @xmath383 where @xmath384 \displaystyle\mathcal{b}(\psi,\zeta)=\partial_t \zeta + \delta \zeta - e\zeta + d \psi .
\end{array } \right.\ ] ] let @xmath271 , @xmath385 be any non - negative functions in @xmath386 . if the _ dual system _ @xmath387 admits a smooth solution @xmath388 with a suitable decay , we may use @xmath259 and @xmath177 as test functions to obtain @xmath389 from where uniqueness would follow . unfortunately , the coefficients in the equations defining the dual system are not smooth .
even worse , @xmath263 an @xmath264 may vanish : the system is not uniformly parabolic , and a delicate regularization procedure is required to fulfill our plan .
regularized dual system .
given @xmath271 , @xmath385 as above , let @xmath390 be such that the supports of @xmath391 , @xmath392 , @xmath393 , @xmath394 , @xmath395 are contained in @xmath396 . given any @xmath397 , we introduce the dual system with regularized coefficients , posed in @xmath398 , @xmath399 \displaystyle \partial_t \zeta_{n , r } + \delta \zeta_{n , r } - e_n \zeta_{n , r } + d_n \psi_{n , r } = h , \end{array } \right.\ ] ] with final and boundary data given by @xmath400 where , thanks to the hypotheses on the solutions , the @xmath401 error in the approximation of each coefficient is @xmath402 , and @xmath403 \| \partial_t c_n \|_{l^1(q_{t})},\ \| \partial_t f_n \|_{l^1(q_{t})},\ \| \nabla \phi_{1,n } \|_{l^2(q_{t})},\ \| \nabla \psi_{1,n } \|_{l^2(q_{t } ) } \leq k. \end{array } \right.\ ] ] our aim is to use @xmath404 and @xmath405 , suitably extended by zero , as test functions .
since these functions have non - zero derivatives at the boundary , we find yet another difficulty : the lack of smoothness at the boundary of the ball @xmath406 , which requires an extra regularization .
avoiding the lack of smoothness at the boundary of the ball . for @xmath407
, we consider a family of cut - off functions @xmath408 such that @xmath409 , @xmath410 \|\nabla\eta_\varepsilon\|_\infty \le \mathcal{c}/\varepsilon,\qquad \|\delta\eta_\varepsilon\|_\infty \leq \mathcal{c}/\varepsilon^2 . \end{array}\ ] ] now put @xmath411 , @xmath412 in the weak formulation to get @xmath413 where @xmath414 \mathcal{k}_{n , r,\varepsilon}=\displaystyle\iint_{q_{r , t}}\left\{(c_1-c_2 ) \big(2\nabla\eta_\varepsilon\cdot\nabla\zeta_{n , r } + \zeta_{n , r}\delta\eta_\varepsilon\big)\right\},\\[10pt ] i^1_{n , r,\varepsilon}=\displaystyle\iint_{q_{r , t}}\left\{(\varrho_1-\varrho_2 + p_1-p_2)\eta_\varepsilon b_n\left(1-\frac{a}{a_n}\right ) \left(\delta\psi_{n , r}-c_n\psi_{n , r}+f_n\zeta_{n , r}\right)\right\},\\[10pt ] i^2_{n , r,\varepsilon}=\displaystyle\iint_{q_{r , t}}\left\{(\varrho_1-\varrho_2+p_1-p_2 ) \eta_\varepsilon ( b - b_n)\left(\delta\psi_{n , r}-c_n\psi_{n , r}+f_n\zeta_{n , r}\right)\right\},\\[10pt ] i^3_{n , r,\varepsilon}=-\displaystyle\iint_{q_{r , t}}\big\{(\varrho_1-\varrho_2+p_1-p_2 ) \eta_\varepsilon b [ \psi_{n , r}(c - c_n)+\zeta_{n , r}(f - f_n ) ] \big\},\\[10pt ] i^4_{n , r,\varepsilon}=\displaystyle\iint_{q_{r , t}}\big\{(c_1-c_2 ) \eta_\varepsilon\zeta_{n , r } [ ( e - e_n)+(d - d_n ) ] \big\},\\[10pt ] i^5_{n , r,\varepsilon}=\displaystyle\iint_{q_{r , t}}\left\{(\varrho_1-\varrho_2+p_1-p_2)\eta_\varepsilon a\left((\phi_1-\phi_{1,n})\psi_{n , r}- ( \psi_1-\psi_{1,n})\zeta_{n , r}\right)\right\}. \end{array}\ ] ] the term @xmath415 vanishes , because the supports of @xmath416 , @xmath417 , @xmath395 are contained in @xmath396 .
hence we can drop it .
we will now take the limit @xmath418 .
the only difficult term at this step is @xmath419 .
we write @xmath420 since @xmath421 on @xmath422 , we have @xmath423 and , if @xmath424 is the unit outwards normal to @xmath425 , then @xmath426 therefore , since @xmath427 and @xmath428 are bounded , @xmath429 limit @xmath285 .
uniform bounds .
proving that @xmath430 , @xmath431 , is easy , thanks to the following estimates .
there are constants @xmath432 , @xmath433 , depending on @xmath243 , @xmath271 and @xmath385 , but not on @xmath290 and @xmath154 , such that @xmath434 \left\| \left(b_n / a_n\right)^{1/2 } \left ( \delta \psi_{n , r } -c_n \psi_{n , r}+f_n\zeta_{n , r}\right)\right\|_{l^2(q_{r , t } ) } \leq \mathcal{c}_2 . \end{array}\ ] ] _ proof . _
_ @xmath107 bounds . _
an easy computation shows that the functions @xmath435 satisfy the differential inequalities @xmath436 - n'(t ) \leq \mathcal{c}(1 + m(t ) ) , \\[10pt ] m(t)=n(t)=0 , \end{array } \right.\ ] ] with @xmath437 , @xmath438 , where @xmath439 is the point where the maximum defining @xmath440 is achieved .
thanks to the assumptions on @xmath441 and @xmath442 , there is a constant @xmath443 such that @xmath444 .
let now @xmath445 as a combination of the equations on @xmath446 and @xmath447 , we find that @xmath448 indeed , assume that @xmath449 .
then , @xmath450 and the bad @xmath451 terms cancel .
if , on the contrary , @xmath452 , then the result follows by the harmless equation on @xmath453 . from
we have @xmath454 , hence the result . _ estimates on the laplacian . _
we multiply the second equation in by @xmath455 and integrate in space and time to obtain @xmath456 & \le&\displaystyle c\|\partial_t\zeta_{n , r}\|_{l^2(q_{r , t})}. \end{array}\ ] ] hence @xmath457 is uniformly bounded .
we now multiply the first equation in by @xmath458 and integrate in space and time .
we get @xmath459 \ \displaystyle=-\int_{b_r(0)}\left(\frac{c_n\psi_{n , r}^2}2+f_n\psi_{n , r}\zeta_{n , r}\right)(t)\\[4 mm ] \quad\displaystyle+\int_t^t\int_{b_r(0 ) } \left(-\frac{\psi_{n , r}^2\partial_t c_n}2 + \psi_{n , r}\zeta_{n , r}\partial_t f_n+f_n\psi_{n , r}\partial_t\zeta_{n , r}\right)\\[4 mm ] \quad \displaystyle + \int_t^t\int_{b_r(0)}\big(-\phi_{1,n}|\nabla\psi_{n , r}|^2+\psi_{n , r}\nabla\phi_{1,n}\cdot \nabla\psi_{n , r}+ c_n\phi_{1,n}\psi_{n , r}^2- \phi_{1,n}f_n\psi_{n , r}\zeta_{n , r}\big)\\[4 mm ] \quad \displaystyle + \int_t^t\int_{b_r(0)}\big ( -\psi_{1,n}\nabla\zeta_{n , r}\cdot\nabla\psi_{n , r}-\zeta_{n , r}\nabla\psi_{1,n}\cdot \nabla\psi_{n , r}-c_n\psi_{1,n}\psi_{n , r}\zeta_{n , r}+f_n\psi_{n , r}\zeta_{n , r}^2 \big)\\[4 mm ] \quad \displaystyle + \int_t^t\int_{b_r(0)}\big(\psi_{n , r}\delta g - c_ng\psi_{n , r}+f_ng\zeta_{n , r } \big)\\[4 mm ] \
\displaystyle\le k\left(1-t+\int_t^t\|\nabla\psi_{n , r}(s)\|_{l^2(b_r(0))}\,ds\right ) .
\end{array}\ ] ] using the grnwall lemma we get @xmath460 , and then the desired estimate @xmath188 in order to pass to the limit in the @xmath461 term we need a good enough estimate , uniform in @xmath290 , for the decay of the normal derivative . as @xmath462 , @xmath463 uniformly in @xmath290 .
_ as a first step we obtain a more precise estimate for the size of @xmath405 .
we first notice that , since @xmath464 , @xmath465 and @xmath385 are bounded and compactly supported in @xmath396 , and @xmath404 is bounded , then @xmath466 for some @xmath467 . on the other hand ,
given a fixed @xmath468 , if @xmath469 is large enough , the function @xmath470 satisfies @xmath471 & \le&\displaystyle-\mathcal{c } \mathds{1}_{b_{r_0}(0)}. \end{array}\ ] ] moreover , @xmath472 , and @xmath473 for @xmath474 and @xmath475 . therefore ,
comparison yields @xmath476 if @xmath477 , we consider a function @xmath478 defined by @xmath479 where @xmath480 , @xmath481 satisfy @xmath482 note that @xmath483 on @xmath484 .
moreover , @xmath485 g(x)=\zeta_{n , r}(x , t)=0,\qquad&|x|=r,\ 0<t < t,\\[4 mm ] g(x)\ge\zeta_{n , r}(x , t)=0,\qquad&r-1<|x|<r . \end{array}\ ] ] therefore , @xmath486 for all @xmath487 , @xmath475 . since @xmath488 for @xmath474 , @xmath475 , we conclude that @xmath489 and hence @xmath490 the result then follows from the following estimate , @xmath491 & = & \displaystyle o(1)\quad\text{as } r\to\infty . \end{array}\ ] ] an analogous computation with @xmath492 if @xmath493 , and @xmath494 if @xmath495 , lead to the same estimate .
@xmath188 limit @xmath462 and conclusion . passing to the limit in @xmath154
, we finally obtain that @xmath496 hence @xmath269 and @xmath497 . uniqueness for @xmath3
then follows just taking @xmath301 as test function in .
a typical application of the hele - shaw equations is to describe tumor spheroids @xcite . when nutrients are ignored , the tumor is assumed to fill a ball centered at @xmath499 , @xmath500 the radius @xmath501 of this ball is computed according to the geometric motion rules and ; that is , we consider the unique ( and thus radially symmetric ) solution to @xmath502 and evolve the radius according to @xmath503 then , we consider @xmath504 defined as @xmath505 this is indeed a correct solution to our model .
let @xmath506 be given . problem defines a unique dynamic @xmath501 , @xmath507 , @xmath508 , which turns out to be the unique solution to the hele - shaw limit problem with initial data @xmath509 . for long times
it approaches a traveling wave solution with a limiting speed independent of the dimension , @xmath510 [ th : ts ] the limit profile can also be calculated and is one - dimensional . _ proof . _ _ existence . _ a unique solution to can be defined for every fixed @xmath501 , because the elliptic problem comes with a decreasing nonlinearity . indeed , shooting or iterative methods apply and give a solution with smooth dependency on the parameter @xmath501 .
therefore the differential equation can be solved thanks to the cauchy - lipschitz theorem . _ equivalence with the hele - shaw problem . _ to show that this geometric method gives the solution to , we compute , in the distributional sense , that @xmath511 also , because @xmath61 satisfies @xmath512 in @xmath69 and since @xmath61 vanishes out of @xmath69 , we have , still in the distributional sense in @xmath513 , with @xmath424 the unit outward normal to @xmath514 , @xmath515 .
now @xmath516 is solved on the unit ball as @xmath517 by the comparison principle , @xmath516 is increasing in @xmath154 and @xmath518 . passing to the limit in the equation we find that @xmath519 as @xmath520 .
it is easy to see that @xmath521 , because we can build sub- and supersolutions of the form @xmath522 with @xmath523 for appropriate values of @xmath206 . by elliptic regularity
we conclude that @xmath524 is of order @xmath525 and , since @xmath526 , it is bounded in @xmath527 . as a consequence ,
@xmath528 is of order @xmath529 .
next we multiply equation by @xmath530 and integrate in @xmath531 to obtain @xmath532 from the above @xmath54 estimate on @xmath533 , we conclude that @xmath534 . going back to the dimensionalized variable , follows .
_ one - dimensional profile .
_ we consider moving coordinates , @xmath535 , @xmath536 .
then @xmath537 is positive if and only if @xmath538 .
hence , @xmath539 since @xmath540 as @xmath541 , we can pass to the limit uniformly on compact sets @xmath542 to obtain @xmath543 _ remark .
_ the asymptotic growth rate is generic for solutions to , since one can always sandwich , after a certain delay , any standard
initial data between two particular profiles of the form which asymptotically travel with the same speed .
the first and simplest example is constructed in radial symmetry .
it consists of an initial datum which is constant @xmath544 in a ball centered at @xmath499 and of radius 1 and zero otherwise .
then the solution will have 0 pressure for a certain time , @xmath545 , in which the density grows exponentially according to equation , so that @xmath546 at @xmath547 , @xmath68 reaches the threshold level @xmath90 in @xmath548 .
it can not happen that @xmath549 for @xmath550 , since otherwise @xmath68 would violate the bound @xmath551 . therefore , @xmath50 satisfies @xmath552 in some ball @xmath553 with @xmath554 .
an easy argument using the maximum principle shows that @xmath555 .
it means that @xmath50 is equal or larger than the solution @xmath556 of the problem @xmath557 , @xmath558 for @xmath559 , @xmath560 for @xmath561 .
taking the limit @xmath562 , we obtain a jump discontinuity in time at @xmath563 for every @xmath561 .
note that @xmath50 is a discontinuous function of time with values in any @xmath564 , @xmath565 .
the second example is best presented in one space dimension and consists of the evolution of an initial data consisting of two copies of the standard example presented in section [ sec : spheroids ] ( tumor spheroid ) , which have now an interval as support .
we locate the supports at a distance from each other of say 1 . for a time interval , @xmath545 they evolve independently . at @xmath563 ,
the supports meet and an easy application of the maximum principle , shows that the solution is strictly positive in the union of the two intervals , therefore , larger than the solution corresponding to the union of the two intervals at @xmath563 .
an easy inspection of the solutions at times @xmath566 and @xmath567 shows that there is a jump discontinuity in the space pressure profiles .
this example can be adapted to several space dimensions by replacing disjoint intervals by disjoint concentric annuli .
we consider the cell density ( defined for short enough times ) with two discontinuities @xmath568 we claim that the correct dynamics is defined by the speed and values of @xmath569 given through @xmath570 with @xmath61 vanishing outside the ball of radius @xmath571 and satisfying @xmath572 in @xmath573 , with zero dirichlet boundary condition .
these rules are simply derived from equation .
for @xmath574 we have @xmath67 and thus the equation is reduced to @xmath575 which gives us the evolution of @xmath569 . for @xmath576
, we have @xmath577 , while @xmath578 . using the equation
, we get the dynamics for @xmath571 . *
acknowledgments . *
authors fq and jlv partially supported by spanish project mtm2011 - 24696 .
the paper was started while they where visiting upmc .
they are indebted to this institution for the warm hospitality .
bp is supported by institut universitaire de france .
blank , i. a. ; korten , m. k. ; moore , c. n. _ the hele - shaw problem as a mesa limit of stefan problems : existence , uniqueness , and regularity of the free boundary_. trans .
( 2009 ) , no . 3 , 12411268 .
lowengrub , j. s. ; frieboes h. b. ; jin , f. ; chuang , y .- l . ; li , x. ; macklin , p. ; wise , s. m. ; cristini , v. _ nonlinear modelling of cancer : bridging the gap between cells and tumours_. nonlinearity 23 ( 2010 ) , no . 1 ,
r1r91 .
tang , m. ; vauchelet , n. ; cheddadi , i. ; vignon - clementel , i. ; drasdo , d. ; perthame , b. _ composite waves for a cell population system modelling tumor growth and invasion_. chin .
b 34 ( 2013 ) , no . 2 , 295318 . | models of tumor growth , now commonly used , present several levels of complexity , both in terms of the biomedical ingredients and the mathematical description .
the simplest ones contain competition for space using purely fluid mechanical concepts .
another possible ingredient is the supply of nutrients through vasculature .
the models can describe the tissue either at the level of cell densities , or at the scale of the solid tumor , in this latter case by means of a free boundary problem .
our first goal here is to formulate a free boundary model of hele - shaw type , a variant including growth terms , starting from the description at the cell level and passing to a certain limit .
a detailed mathematical analysis of this purely mechanical model is performed .
indeed , we are able to prove strong convergence in passing to the limit , with various uniform gradient estimates ; we also prove uniqueness for the asymptotic hele - shaw type problem .
the main tools are nonlinear regularizing effects for certain porous medium type equations , regularization techniques la steklov , and a hilbert duality method for uniqueness . at variance with the classical hele - shaw problem , here
the geometric motion governed by the pressure is not sufficient to completely describe the dynamics .
a complete description requires the equation on the cell number density . using this theory as a basis , we go on to consider the more complex model including nutrients .
we obtain the equation for the limit of the coupled system ; the method relies on some bv bounds and space / time a priori estimates
. here , new technical difficulties appear , and they reduce the generality of the results in terms of the initial data . finally , we prove uniqueness for the system , a main mathematical difficulty . *
key words * tumor growth ; hele - shaw equation ; free boundary problems ; porous media ; hilbert uniqueness method .
* mathematics subject classification * 35k55 ; 35b25 ; 76d27 ; 92c50 . |
the two magellanic clouds have larger number density of high mass x - ray binaries ( hmxb ) compared to our galaxy .
the small magellanic cloud ( smc ) with a mass of about 1% of our galaxy has a total of about 70 hmxbs and hmxb candidates , almost comparable to that in our galaxy ( haberl & sasaki 2000 ; yokogawa et al .
though the hmxb number density in the large magellanic cloud ( lmc ) is not as high as that in smc , it is still significantly higher than the galactic value ( sasaki , haberl & pietsch 2000 ) . in - spite of a large number of hmxb pulsar objects in the two magellanic clouds , individual objects have not been studied in great detail except for smc x-1 and lmc x-4
there are four hmxb pulsars known in the lmc , of which lmc x-4 is the best studied .
several of these sources show a very large x - ray luminosity in their high states , close to or exceeding the eddington limit for a 1 m@xmath0 object .
sources , in which a strong local absorption column is absent , nature of the soft spectral component can be studied in detail . due to a low galactic absorption column density towards the lmc and smc ,
these objects are also suitable for study of the local absorbing material
. a soft excess in the x - ray spectrum , now detected in several accreting pulsars can be a common feature of the hmxb pulsars .
but since most of the hmxb pulsars are in the galactic plane , large line of sight absorption usually makes it difficult to detect it .
it is therefore , important to study details of the individual hmxbs in the lmc and smc .
the transient be x - ray binary source exo 0531096609.2 was discovered with the exosat observatory in 1983 ( pietsch , rosso & dennerl 1989 ) .
hard x - ray emission from this source was detected with the sl2 xrt experiment ( hanson et al .
1989 ) and its intensity variations were studied in detail with the rosat ( haberl , dennerl , & pietsch 1995 ) .
x - ray pulsations were also discovered with the rosat ( dennerl , haberl & pietsch 1995 ) .
this pulsar is located at a distance of 17@xmath1 from the luminous binary x - ray pulsar lmc x-4 , and has a pulse period of 13.7 s , which is also close to the 13.5 s pulse period of lmc x-4 . during a beppo - sax observation of lmc x-4 ,
exo 0531096609.2 was in the field of view of the narrow field instruments of beppo - sax , and pulsations were detected along with a positive local period derivative ( burderi et al .
the source was also observed with the xmm in 2000 during which it was in a relatively bright state ( haberl , dennerl , & pietsch 2003 ) .
the optical counterpart is one of the two close double be - stars ( stevens , coe and buckley 1999 ) in the subsequent sections we present the archival observations that we have used to study this source ( 2 ) , results from temporal and spectral analysis of the data ( 3 ) and follow up this with a discussion on the nature of this pulsar ( 4 ) .
we have analysed archival data from four observations of the pulsar exo 0531096609.2 .
three of these observations were made with the beppo - sax observatory and one with the asca .
the field of exo 0531096609.2 was covered by asca - gis during several observations of nearby sources . in one observation of lmc
x-4 , made during 1996 may 2426 , we found exo 0531096609.2 to be bright enough for temporal and spectral investigation .
it was also present in the field of view of the beppo - sax narrow field instruments during two observations of lmc x-4 in 1997 ( march 1315 ) and 1998 ( october 2022 ) and in another one observation of the lmc in 2000 ( january 0102 ) .
exo 0531096609.2 was in a transient outburst during the 1997 observation and the pulsation properties during this period have been reported earlier ( burderi et .
al . 1998 ) .
we have found 36 ks of useful exposure with asca - gis , 117 , 82 and 32 ks with sax - mecs ( 1997 , 1998 and 2000 respectively ) , and 42 , 31 and 15 ks with sax - lecs . for more details of these instruments , refer to tanaka , inoue & holt ( 1994 , asca ) and boella et al .
( 1997 , sax ) .
we have used the standard data selection criteria of both the observatories .
the count rates and spectra of the source were obtained from circular regions of radius 6@xmath1 for asca - gis , 3@xmath1 for sax - mecs and 2@xmath1 for sax - lecs .
the corresponding background spectra were extracted from source free nearby regions in the respective fovs .
light curves from these observations were extracted with time resolution of 0.25 s for sax lecs and mecs and 0.50 s for asca - gis .
barycenter corrections were applied to the photon arrival times in the event files before extraction of light curves .
the pulse folding and @xmath2 maximisation search near the expected pulse period of 13.7 s clearly showed presence of pulsations in all the observations .
the new pulse period measurements with asca - gis and beppo - sax lecs / mecs are given in table 1 and plotted in figure 1 .
pulse profiles were created from background subtracted light curves in different energy bands .
the pulse profiles of all but the 1997 beppo - sax observation are shown in figure 2 along with the dates of each observation and the 210 kev luminosity of the source during the corresponding observation .
there is a hint of slight energy dependence of the pulse profile in the asca observation .
the pulse profile clearly shows a luminosity dependence ; a single peaked profile at low luminosity ( asca 1996 , sax 1998 , top panels of figure 2 . ) and a double peaked profile at high luminosity .
the pulse fraction is in the range of 3060% . the pulse period history of exo 0531096609.2 , obtained by combining the new observations and the earlier reported measurements with the rosat , beppo - sax and xmm observatories are plotted in figure 1 .
an overall spin - down trend during the last five years and some local variation from this trend can be seen in the figure .
the orbital elements of this binary system is not known , and we note that an orbital motion of the neutron star with a velocity of a few hundred km s@xmath3 can also result in the observed changes in the local pulse period measurements .
suitable binning was applied to all the energy spectra before spectral fitting .
we first fitted the asca - gis spectrum with a simple model consisting of a power - law component and line of sight absorption with solar abundance of metals .
the source was in a low intensity state during this asca observation .
we rebinned the spectrum to 12 spectral bins in the 0.79.0 kev range .
a photon index of 0.46 and equivalent hydrogen column density of @xmath4 atoms @xmath5 was obtained with a reduced @xmath2 of 0.23 for 9 degrees of freedom . assuming a distance of 55 kpc to the lmc , the x - ray luminosity during the asca observation was @xmath6 erg s@xmath3 .
no signature of iron fluorescence was detected and a broad binning scheme needed due to the poor statistics also does not allow us to put a strong upper limit on the equivalent width of a line feature at 6.4 kev . the spectrum measured with asca - gis
is shown in figure 3 along with the best fit model spectrum and the residuals .
in all the three beppo - sax observations , the source exo 0531096609.2 was at a distance of about 17@xmath1 away from the centre of the field of view .
therefore , it was necessary to consider the smaller collecting area of the mirrors at large off - axis angles . for spectral analysis of mecs spectra at an off - axis angle of 17@xmath1 , we have taken an average of the effective areas at 14@xmath1 and 20@xmath1 , which are available for the mecs .
this is likely to introduce a few percent uncertainty in the measurements of spectral parameters over the statistical errors . at low energies ,
the relative loss of effective area for different off - axis angles is not energy dependent .
therefore , the lecs spectra were fitted with the on - axis effective area curve , and a relative normalisation with respect to the mecs was used .
all the three beppo - sax observations were fitted with the same spectral model described above . in the 1997 and 2000 observations
, there was some signature of an iron emission line , and a separate narrow gaussian line at 6.4 kev was included in the model .
the equivalent width of the iron line during these two observations were 130 @xmath7 50 ev and 180 @xmath7 160 ev respectively .
the spectral parameters along with the reduced @xmath2 obtained are given in table 2 .
the x - ray luminosities in the 2.010.0 kev band during the three sax observations are @xmath8 , @xmath9 , and @xmath10 erg s@xmath3 during the 1997 , 1998 and 2000 observations respectively .
the spectra measured with beppo - sax lecs and mecs instruments are shown in figure 3 .
the pulsar exo 0531096609.2 , like many other hmxb pulsars with be star companions , is believed to be a transient source .
however , we have found this pulsar to be bright , at a luminosity level of @xmath11 erg s@xmath3 to @xmath12 erg s@xmath3 in most of the x - ray observations made since 1993 with the low background imaging instruments .
several rosat observations during march to may 1993 also found the source to have significant x - ray intensity ( haberl et al .
1995 ) . in this respect , this pulsar is more akin to 4u 1907 + 09 ( int zand 1999 , mukerjee et al .
2001 ) , which also shows intensity variations of similar scale but without any obvious transient phenomena .
its long term x - ray intensity behaviour is different from that of the be - star transients like exo 2030 + 375 ( parmar , white & stella 1989 ) .
the peak x - ray luminosity of @xmath12 erg s@xmath3 observed in 1997 is however , much smaller than that of the pulsars smc x-1 and lmc x-4 ( paul et al .
2002 ) , which have high mass companion stars in very close orbits .
this indicates a large size orbit and hence orbital period .
the archival x - ray observations are not sufficiently long to search for pulse arrival time delay that can help to determine the orbital parameters . analysing the rxte - asm light curve of exo
0531096609.2 we have found a peak in the power spectrum at 183 days and its integer multiples .
intensity variations of the bright nearby source lmc x-4 is unlikely to create this because the 30.5 day superorbital intensity variation of lmc x-4 is not reflected in the power spectrum of exo 0531096609.2 .
however , since this is very close to a half - year , yearly observational effects can not be ruled out . combining the present observations with the earlier measurements of pulse periods with the rosat and the beppo - sax , we have obtained a period history of the pulsar , shown in figure 1 .
it shows a clear monotonic spin - up trend since the asca observation in 1996 , though not with a constant period derivative . however , certainly there were spin - down episodes between the 1991 rosat observation and the 1996 asca observation .
a spin - down was also observed during the 1997 beppo - sax observation itself , which is a surprise considering the high intensity level during this observation .
the last 5 years of pulse period measurements presented here shows a spin - up of the neutron star .
this suggests that accretion onto the neutron star was in progress for at least a significant part of this period .
however , we can not rule out a significant binary motion related component in the observed pulse period variations
. the x - ray emission from the polar cap regions can have different geometric pattern at different mass accretion rates . in the transient x - ray pulsar exo 2030 + 375 ,
a change in beaming pattern , from a fan beam in high luminosity state to a pencil - beam in low luminosity state was observed with the exosat ( parmar et al .
the pulse profiles of exo 0531096609.2 at different luminosity levels presented here and elsewhere ( burderi et al . 1998 , haberl et al .
2003 ) clearly indicate a dependence on the mass accretion rate .
the pulse profile is single peaked and nearly sinusoidal in low intensity state ( beppo - sax : 1998 ) , and it has two distinct well separated peaks or an admixture of the above two during other observations .
it is possible that in exo 0531096609.2 also , the x - ray beam pattern changes from a pencil beam at low luminosity to a fan beam at high luminosity .
the change in beam pattern probably takes place at a luminosity level of @xmath13 erg s@xmath3 . during the asca observation in 1996 and the beppo - sax observation in 1998 , the source was in low intensity state and the spectral parameters
are not constrained well from these observations .
however , the spectral characteristics reported here from the other three observations indicate intrinsic spectral variability in the pulsar . the power - law component of the spectrum is softer at higher luminosity .
the xmm spectrum of this source unambiguously showed the presence of a soft spectral component in excess of a partially absorbed hard power - law extended to lower energies ( haberl et al . 2003 ) .
the same observation also showed that the pulse profile is different in the soft and the hard x - ray bands .
this indicates that the soft component may have a different origin , or at least the geometry of emission is different in different energy ranges .
presence of narrow features in the soft excess prompted haberl et al .
( 2003 ) to use a hot plasma model that yields a temperature of 0.1 kev and emission measure of 4 @xmath14 10@xmath15 @xmath16 .
such a plasma is unlikely to show pulsations due to its large physical size and large cooling time scale , contrary to what is seen in the 0.15 - 0.4 kev band .
a blackbody type component with separate low energy emission lines as seen in her x-1 ( endo et al . 2000 ) may be a better description of the soft excess of this pulsar .
several other accreting x - ray pulsars with relatively low line of sight absorption show soft excess ( paul et al .
2002 and references therein ) . in some sources
the soft excess probably originates in a part of the x - ray irradiated inner accretion disk ( endo et al .
2000 ; naik & paul 2004 ) . from independent analysis of the xmm spectra
we have found that in exo 0531096609.2 , the total luminosity of unabsorbed x - ray emission in the soft excess is @xmath17 erg s@xmath3 , which is only about 1% of the total emission in the 110 kev energy band .
the ratio of the soft excess flux to the total flux is smaller than that of smc x-1 ( 3.6% ) , lmc x-4 ( 6.4% ) and her x-1 ( 10% ) . in this pulsar , the size of the black body emission region
is required to be only about @xmath18 cm and , therefore , can easily be accommodated in inner part of the accretion disk heated by the hard x - rays .
we thank an anonymous referee for valuable suggestions .
this research has made use of data obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center .
snaj thanks the astrosat programme through which a visit to tifr was supported .
48560.16 & rosat - pspc & 13.67133 @xmath7 0.00005 & dennerl et al .
( 1995 ) + 50228.12 & asca - gis & 13.67875 @xmath7 0.00012 & present work + 50520.00 & sax - mecs & 13.67590 @xmath7 0.00008 & burderi et .
1998 ) + 51107.64 & sax - mecs & 13.670674 @xmath7 0.000026 & present work + 51600.72 & sax - mecs & 13.66884 @xmath7 0.00004 & present work + 51824.56 & xmm - mos+pn & 13.66817 @xmath7 0.00001 & haberl et al .
( 2003 ) + lccccc n@xmath19 & 11@xmath20 & 5.7@xmath21 & 29@xmath22 & 6.6@xmath23 + photon index ( @xmath24 ) & 0.46@xmath25 & 0.81 @xmath7 0.05 & 0.37@xmath26 & 0.63 @xmath7 0.17 + fe line flux & - & 3.0 @xmath7 1.2 & - & 2.8 @xmath7 2.5 + log l@xmath27 & 36.09 & 36.78 & 35.48 & 36.18 + reduced @xmath28/dof & 0.23/9 & 1.44/79 & 0.49/20 & 0.55/46 + | we report timing and spectral properties of the transient be x - ray pulsar exo 0531096609.2 studied using observations made with the asca and beppo - sax observatories . though there must have been at least one spin - down episode of the pulsar since its discovery
, the new pulse period measurements show a monotonic spin - up trend since 1996 . the pulse profile is found to have marginal energy dependence .
there is also evidence for strong luminosity dependence of the pulse profile , a single peaked profile at low luminosity that changes to a double peaked profile at high luminosity .
this suggests a change in the accretion pattern at certain luminosity level .
the x - ray spectrum is found to consist of a simple power - law with photon index in the range of 0.40.8 . at high intensity level
the spectrum also shows presence of weak iron emission line . |
nanoplasmonics continues to receive substantial interest from various fields of research including biology @xcite , chemistry @xcite and physics @xcite , with applications ranging from renewable energy technology @xcite to homeland security by the sensitive identification of explosive material @xcite . due to the collective excitations of electrons at the surface of a metallic resonator coupling to electromagnetic fields
, the local density of optical states ( ldos ) decreases rapidly for spatial positions away from the surface of the metal resonator ; so - called `` hot spots '' , where the local electric field can be enhanced by orders of magnitude in comparison to a bulk medium , are formed near the surface .
the unique properties of localized surface plasmon ( lsp ) resonances , manifest in a strong confinement of electric field far below the diffraction limit and result in an exotic electromagnetic response that underlines the wide application of surface plasmons , especially in nanoscience and nanophotonics ; however , the extreme spatial localization of lsps makes them experimentally challenging for direct detection of the spatial field distribution , e.g. , by employing conventional spectroscopy techniques , since the spatial resolution is constrained by the diffraction limit .
in addition , optically dark modes have a vanishing dipole moment , and traditional optical methods fail to excite such modes . due to the strong frequency dispersion and losses in metals , which satisfy the kramers - kronig relations , it is a very challenging problem to model the electromagnetic response of arbitrarily shaped metal resonators .
consequently , most optical studies of plasmonic structures rely heavily on brute force numerical simulations .
direct imaging of the lsp resonances is important both for revealing the exotic physics underlying these resonances and for verifying and testing numerical models used in theoretical studies .
there are presently several different experimental schemes developed to detect the electromagnetic component of lsps , including near - field scanning optical microscopy ( nsom ) @xcite , leakage radiation microscopy @xcite , and two - photon induced photoluminescence @xcite , and cathodoluminescence @xcite .
of particular interest , electron energy loss spectroscopy ( eels ) has been shown to be capable of accessing the subwavelength spatial variation of the surface plasmon modes of single metallic nanoresonator experimentally @xcite for almost a decade ; during this time , it has been recognized to be @xcite one of the most powerful approaches , in which high speed electron beams ( typically ranging from several tens to several hundreds of kev ) are injected and transmitted across an optically thin sample , with a spatial resolution of around 1 nm ; eels has been applied to various systems such as split ring resonator ( srr ) @xcite and single nanorod / antennas @xcite .
the eels method is also quite versatile , allowing one to probe both optically bright and dark resonances over broadband frequencies , and can be used for detecting both localized and extended excitations @xcite .
recently , vortex electron beams have also been experimentally demonstrated @xcite , which may find use for probing the magnetic component of lsps @xcite .
modelling eels is an extremely challenging and tedious numerical problem , and there has been different theoretical approaches developed to model the eels of plasmonic resonators , including boundary element method ( bem ) @xcite , discrete dipole approximation ( dda ) @xcite , discontinuous galerkin time - domain ( dgtd ) @xcite and finite - difference time - domain ( fdtd ) methods @xcite . with sufficient care and computational resources , all of these approaches can show good agreement with experimental results , since these are basically full dipole solutions to the classical maxwell equations ; typically , these approaches employ numerous dipole point calculations in a 3d spatial grid to obtain the photon green function , which allows one to obtain the eels profile in space and frequency ; non - local effect of the conductive electrons are typically negligible until spatial positions within a few nm from the metal surface @xcite , so most studies of eels in nanoplasmonics have worked with the classical maxwell equations without any non - local effects ; however , as shown very recently , non - local effects may become important for certain metallic resonators and geometries @xcite .
unfortunately , most eels calculation methods to date are computational expensive , hard to employ without parallel computers , limited to certain geometries , and offer little physical insight into the lsp resonance structures .
it is thus highly desired to find an approach that is both simpler and more intuitive in terms of explaining the features of the eels maps and frequency profiles , and applicable to arbitrarily shaped resonators .
recently , the physical meaning and applications of eels has been explored extensively .
for example , hrl _
et al_. @xcite show that eels is an efficient tomography probe of the surface plasmon modes , and they propose to get the 3d green function from the eels ; unlike eels , which is related to the projected full electromagnetic ldos ( eldos ) , losquin _
et al_. @xcite theoretically show that cathodoluminescence is related to the so - called projected radiative eldos ( i.e. , nonradiative coupling effects are not captured ) , which they illustrate with a quasistatic mode expansion technique based on the bem .
the quasistatic bem approach uses only geometry - defined modes , which gives a nonretarded modal solution with scale invariance @xcite ; in contrast , the mode expansion technique introduced below uses the rigorously defined qnms which are the true open system retarded eigenmodes , and we show only a few of them ( indeed usually only one ) will be needed around the frequency of interest . in this work , we introduce an accurate and physical intuitive method to model the eels map of the surface plasmon modes based on the quasinormal mode ( qnm ) expansion of the photon green function @xcite .
the qnms are the eigenfunction of the source - free maxwell equations with open boundary condition @xcite , and complex eigenfrequencies .
two key advantages of our qnm technique are as follows : ( i ) after obtaining the normalized qnm , the calculation of the eels is straightforward and essentially instantaneous in the frequency regime of interest ; ( ii ) our calculation includes the contribution of the lsp in a modal theory , and thus has intuitive and analytical insight . after introducing the basic theory of eels and connecting to the qnms , we present several example structures of interest including a gold srr , a single gold nanorod , and a dimer of gold nanorods as is shown schematically in fig .
we also use the same qnm green function to obtain the purcell factor ( and projected ldos ) from a coupled dipole emitter and we show how the spectral profile compares and contrasts with the eels as a function of frequency .
nm , @xmath0 nm , @xmath1 nm , @xmath2 nm , @xmath3 nm , @xmath4 nm , @xmath5 nm and @xmath6 nm , and we consider material properties for gold . ]
for the structures of interest , we consider a general shaped metallic nanoresonator inside a homogeneous background medium with refractive index @xmath7 . we assume the magnetic response is negligible , with permeability @xmath8 ; the electric response is described by the drude model , with permittivity @xmath9 with parameters similar to gold : @xmath10thz and @xmath11thz .
the electric - field green function , * g * , of the system is defined as @xmath12 where @xmath13 is the unit dyadic , and @xmath14 inside the metallic nanoresonator with @xmath15 elsewhere . considering the frequency regime of interest where there is only a single qnm , @xmath16 , which gives the mode profile of the lossy / dissipative mode of the source free maxwell equations with open boundary conditions , with complex eigenfrequency @xmath17
, then the contribution to the transverse green function in the near field of the nanoresonator , around the cavity resonance , is given by @xcite @xmath18 the qnm , @xmath19 is normalized here as @xmath20 with @xmath21 .
alternative qnm normalization schemes are discussed in @xcite .
when discussing eels , it is useful to also connect to common quantities for use in quantum plasmonics . for example , using the normalized qnm , the corresponding effective mode volume for use in purcell factor calculations is defined as @xmath22 , \label{eq : veee}\end{aligned}\ ] ] at some characteristic position @xmath23
the enhancement of spontaneous emission ( se ) , or the enhancement of the projected ldos , at this position is then obtained from @xmath24}{{\rm im}[\hat{n}_\alpha\cdot{\bf g}_{\rm b}({\bf r}_0,{\bf r}_0;\omega)\cdot\hat{n}_\alpha ] } , \label{eq : pf}\end{aligned}\ ] ] where @xmath25 = \frac{\omega^3n_{\rm b}}{6\pi c^3}{\bf i}$ ] is for a lossless homogeneous background with refractive index @xmath26 , and @xmath27 is a unit vector of the dipole emitter aligned along @xmath28 . using the qnm approach , then @xmath29 is simply obtained by using @xmath30 , and the accuracy of this approach can be checked by performing a full dipole calculation of @xmath31 at this position , which we will show later using accurate fdtd techniques @xcite .
we consider a high speed electron beam with initial kinetic energy , @xmath32 , e.g. , 50 kev@xmath33200 kev , which gives an electron speed , @xmath34 , 0.55@xmath35 0.70 @xmath36 with @xmath36 the velocity of light in vacuum ; specifically , @xmath37 with @xmath38 the electron rest mass ; the electron passes through the nanoresonator which is a few tens of nanometers thick .
generally , scanning transmission electron microscopes will be employed to obtain the eels map , as a function of frequency , for which the relevant length scale of the spatial path over which the electron beam is traveling is around a few hundred nanometers ; under this situation , the energy loss of the electron is negligible , which means that , to a very good approximation , we can take the velocity of the electron as a constant . as the electron comes near the surface , the electric quasistatic interaction can be described by the image charge @xcite , which is negligible until it comes to around a few nanometers from the surface ; at this scale , the local geometry details can be ignored and the surface can be approximated by a slab , and detailed analysis elsewhere shows that the electric quasistatic contribution is typically negligible @xcite for the eels calculation .
thus we assume the electron energy loss is primarily induced by the dominant qnm(s ) . in practice
there are also `` bulk losses''@xcite coming from other background modes such as evanescent modes in the metal , but these are regularized depending upon the finite size of the cross section of the electron beam and have little influence on a modal interpretation of the eels map .
in fact , in @xcite , in order to investigate the modal response of the lsp resonance of the nanoresonator , they eliminated the bulk contribution by subtracting the solution from a different fdtd simulation with a homogeneous metal calculation , thus eliminating fdtd grid - dependent effects . with our qnm approach , there is no need to subtract off such a term , and , moreover , this contribution can be obtained analytically @xcite , and can also formulated as a local field problem for emitters inside lossy resonators @xcite .
numerically , we inject a spatial plane wave modulated with a finite pulse length ( fwhm ) , @xmath39 , with a central frequency around the resonance of the qnm ; then a run - time fourier transform with a time window @xmath40 is employed @xcite to get the qnm numerically ; we also use a non - uniform conformal mesh scheme with a fine mesh of 1 - 2 nm around the metallic nanoresonator .
the energy loss is defined by @xmath41 where the electric field induced by the qnm is given by @xmath42d\omega\nonumber\\ = & -2\int_{0}^{\infty}d\omega { \rm im}[e^{-{\rm i}\omega t}\frac{1}{\varepsilon_0\omega}\int{\bf g}^{\rm c}({\bf r}_t,{\bf r}';\omega)\cdot{\bf j}({\bf r}',\omega)d{\bf r}']\end{aligned}\ ] ] with the effective current carried by the moving electron , @xmath43 , and @xmath44 is the absolute value of the charge of the electron .
in the calculations below , we will assume the electron moving along the -@xmath45-axis , so @xmath46 with @xmath47 the unit vector along @xmath45 .
under these assumption , the eels function , @xmath48 , due to the qnm for electrons injected along @xmath45-axis is simply given by @xcite @xmath49dzdz ' , \label{loss } \end{aligned}\ ] ] where @xmath50 and @xmath51 on a 2d spatial map of the image . as is shown in eq .
( [ loss ] ) , in order to calculate the eels for one particular @xmath52 in the plane , the green function along the electron beam should be calculated at various @xmath45 and @xmath53 ; the number of simulations required should be sufficiently large to model electric dipoles scanning over the trajectory of the electron beam , and this is the reason why thousands of dipole simulation are usually employed @xcite . in stark contrast , with the qnm technique , once the qnms are obtained numerically , the green function can be calculated with eq .
( [ gt ] ) analytically , with a computation that is basically instantaneous .
below we present a selection of example metal resonators using the drude model for the material properties of the metal .
we also show the purcell factor or enhanced se factor at selected positions as well as the full eels as a 2d image . for our first example of the qnm calculation of eels , we study the gold srr , which is the basic `` artificial atom '' unit cell of the negative index metamaterial , with a rich magnetic response to external electromagnetic fields @xcite . while the 2d metamaterial lattice is usually fabricated on a low index semiconductor , for the present study , we will ignore the effects of the substrate and assume the srr is in free space ( @xmath54 ) , though this is not a model restriction .
the srr with thickness , @xmath2 nm , is located in @xmath55-plane as is shown in fig .
[ f0](a ) with parameters @xmath56 nm , @xmath0 nm , @xmath1 nm and @xmath3 nm . a full dipole fdtd calculation @xcite shows that the dipole resonance of the qnm is around @xmath57thz with mode profile @xmath58 shown in fig .
[ f1](a ) .
we use an @xmath59-polarized spatial plane wave with central frequency around 216 thz and pulse width @xmath60 fs , which is injected along the @xmath45-axis ; a running fourier transform with a temporal bandwidth @xcite @xmath61fs is used to obtain the qnm . the corresponding effective mode volume ( which is complex in its generalized form @xcite )
is around @xmath62 at the chosen dipole position @xmath63 nm where @xmath64 ( the dipole is shown by the blue arrow in fig .
[ f1](a ) , and note we have set the center of the srr as the origin of the coordinate system ) . in order to first check the accuracy of the qnm calculation , we calculate the enhancement of the projected ldos ( or se enhancement of a dipole emitter ) @xmath65 from eq .
( [ eq : pf ] ) .
figure [ f1](b ) shows that the single qnm model calculation ( magenta solid , with eq .
( [ gt ] ) ) agrees very well with the full numerical dipole calculation using fdtd @xcite ( grey dashed ) at position @xmath66 .
as is shown above , due to the strong confinement of the lsp , extremely small mode volumes are obtained leading to a strong enhancement of se of an electric dipole ( or single photon emitter ) at this near field spatial position .
the broad bandwidth of the se enhancement is also a notable feature of metallic nanoresonators , making it much easier to spectrally couple to artificial atoms .
we further remark that the spatial dependent spectral function , @xmath67 $ ] , as is discussed in ref .
@xcite , usually has a non - lorentzian lineshape that in general changes as a function of position ; this effect is captured by the qnm technique @xcite through the spatial dependent phase factor of the qnm . due to the collective motion of the free electrons
, there is an oscillating electric current circling along the srr ; as a result , a temporarily non - zero magnetic dipole is created , which displays a strong magnetic response to external electromagnetic field around the resonance of the qnm and forms the basis of exciting optical properties of metamaterial such as negative index .
the working region of the srr could be controlled by changing the length of the srr which determines the qnm resonance .
the 2d eels image , @xmath68 for a high energy , @xmath69100 kev , ( @xmath70 ) electron beam injected along @xmath45-axis is shown in fig . [ f1](c ) , with @xmath71 , which is consistent with the calculations in refs .
@xcite using full fdtd , nodal dgtg , and bem , respectively .
the eels at the position of the blue circle ( near a maximum field position ) is obtained as shown in fig .
[ f1](d ) by the dark green solid line as a function of the frequency ; clearly the eels can be used to effectively explore the qnm response of the plasmonic resonator .
the magenta dashed line in fig .
[ f1](d ) shows that the se enhancement example almost has the same spectral lineshape as the spatially averaged eels calculation . for our second example
, we consider a single nanorod as is shown in fig .
[ f0](b ) , with radius @xmath72 nm and length @xmath73 nm .
frequently such nanorods are embedded inside liquids , so we assume the nanorod is located inside a homogeneous background medium with @xmath74 .
using fdtd simulations , the dipole resonance is found around @xmath75thz @xcite . in order to get the qnm , a @xmath76-polarized spatial plane wave with central frequency 325 thz and @xmath77fs
is injected along @xmath59-direction , and the running fourier transform with @xmath78fs is employed @xcite .
figure [ f2](a ) shows the mode profile of @xmath79 and the effective mode volume is found around @xmath80 ; the enhancement of the projected ldos at position @xmath81 nm is shown in fig .
[ f2](b ) ( see arrow for dipole position ) , and the qnm calculation ( magenta solid ) shows excellent agreement with the full numerical calculation ( grey dashed ) . the corresponding 2d eels for an injected electron beam with energy @xmath82kev is shown in fig .
[ f2](c ) at the resonance frequency , @xmath83thz . at the in - plane position @xmath84 nm , around which the maximum of the eels is obtained ( blue circle in fig .
[ f2](c ) ) , @xmath85 is shown by the blue solid line in fig . [
f2](d ) . it can be seen that the eels again picks up the correct resonant response of the qnm . as is shown in fig .
[ f4](d ) , by the magenta dashed line , the se enhancement now has a different lineshape than the eels calculation ; as discussed earlier , this is caused by the spatially varying nature of spectral lineshape . for our final resonator example
, we study a dimer composed of two identical gold nanorods in homogeneous background with @xmath86 as is shown in fig .
[ f0](c ) .
the eigenfrequency of the dipole mode is found at @xmath87thz @xcite , and the correspondent qnm , @xmath88 is shown in fig .
[ f4](a ) . here in order to obtain the qnm mode ,
a @xmath76-polarized plane wave with central frequency 291 thz and @xmath77fs is injected , and @xmath89fs is used for the running fourier transform @xcite . the effective mode volume at @xmath90 nm is found to be @xmath91 ; the enhancement of the ldos , @xmath92 at @xmath66 , using qnm and full numerical fdtd calculations are shown by the magenta solid and gray dashed in fig .
[ f4](b ) , respectively .
the 2d eels , @xmath93 is shown in fig .
[ f4](c ) at @xmath94thz , for injected electrons with energy @xmath95kev ; @xmath85 at @xmath96 nm ( blue circle in fig .
[ f4](a ) ) is shown by the dark green solid line in fig .
[ f4](d ) ; once again we see that the se enhancement ( magenta dashed line ) displays a rather different lineshape compared to eels calculation , similar to the case of the single nanorod .
the se enhancement is also much larger for the dimer which also has a larger output coupling efficiency @xcite .
as is shown in @xcite , due to the inherent loss of the plasmonic system , the normalized qnm is in general complex , @xmath97 , and one obtains a position dependent phase factor @xmath98 , which induces both the reshaping of the spectrum of ldos ( that is proportional to the imaginary part of the green function ) and the variation of the position of peak ; this causes the spectral lineshape between the eels and the enhancement of ldos to differ in general .
it is important to stress that for the qnm calculation of eels , one only needs two simulations to get the complex eigenfrequency and spatial distribution of the qnm ( or qnms ) , respectively .
the rest of the calculation can be done semi - analytically with eq .
( [ gt ] ) .
consequently , the qnm technique for eels is many orders of magnitude faster than full fdtd calculation using electric dipoles , and offers more insight .
furthermore , by obtaining the magnetic field component of the qnm , i.e. , @xmath99 , the magnetic green function , @xmath100 could be calculated just as easily as the electric green function ; this can be used to simulate vortex - eels as done in ref .
@xcite which used brute force fdtd simulations .
the qnm could also be applied to model the electromagnetic force on atoms and nanostructures for optical trapping , which as input , usually requires the maxwell stress tensor and/or the green function @xcite . recently , guillaume _
et al_. @xcite proposed an efficient modal expansion dda method to model eels .
they show that by choosing a small number , e.g. , 3 - 10 , of eigenvectors their eigenvector expansion technique could get good results with reasonable accuracy , and the number of usual dda operations are decreased considerably . for certain geometries this approach may outperform a qnm fdtd computation , though it is not clear how general the approach is for various shaped resonators .
the philosophy of the qnm approach is to use the source free eigenmode solutions , obtained here using fdtd with the aid of the modal response from a scattered plane wave @xcite . as a result ,
when there are a few dominant qnms around the frequency of interest , in principle a single fdtd simulation is enough to obtain the modes and green function as long as the overlap between the modes and injected field is sufficient .
moreover , we stress that the qnm computation is not limited to the fdtd method , e.g. , it can also be obtained using an efficient dipole excitation technique with comsol @xcite . the typical time needed for our qnm calculations with a fine mesh size as small as 1 nm around the metal resonator , and total simulation volume 1 - 2 micron cubed , takes around several days on a high performance workstation .
this is certainly not insignificant , however , having the full green function as a function of position and frequency can then solve numerous problems without any more numerical simulations for the electromagnetic response .
apart from an efficient calculation of eels , the qnm green functions that we use above can be immediately adopted to efficiently study quantum light - matter interactions and true regimes of quantum plasmonics with quantized fields .
for example , using the quantization scheme of the electromagnetic field in lossy structures @xcite , the interaction between a quantum dipole ( two level atom with frequency @xmath101 and dipole @xmath102 ) and electric field operator in the rotating wave approximation ( assuming no external field ) is given by the interaction hamiltonian @xmath103 $ ] ; where the electric field operator is @xmath104 , with @xmath105 the imaginary part of @xmath106 and @xmath107 is the collective excitation operator of the field and medium ; and @xmath108 is the pauli operator .
we stress that the electromagnetic response of the lossy structure is rigorously included by the classical green function ( obtained from eq .
( [ gt ] ) ) .
for example , in a born - markov approximation , the quantum dynamics of @xmath109 quantum emitters around a metal nanostructures can be described through a reduced density matrix whose coupling terms can be fully described , including emitter - emitter and emitter - lsp interactions , through the analytical properties of the qnm green function .
for example , a recent example study of the quantum dynamics between two plasmon - coupled quantum dots is shown in @xcite .
note that such an approach is ultimately more powerful than a standard jaynes - cummings model since it can include non - lorentzian decay processes and nonradiative coupling to the resonator in a self - consistent way , and it can also be used to improve the simpler jaynes - cummings models ( in a regime where they are deemed to be approximately valid ) with a rigorous definition of the various required coupling parameters @xcite .
we have introduced an efficient and semi - analytic calculation technique for modelling eels using a qnm expansion technique , and exemplified the approach for several different metallic nanostructures .
we first showed that the qnm technique works well for the srr , and demonstrated that the qnm could be used to obtain similar 2d eels maps to those shown in refs .
@xcite , but with orders of magnitude improvements in efficiency and deeper physical insight .
we then showed qnm calculations for a single gold nanorod and dimer of gold nanorods .
we also presented example purcell factor calculations and demonstrated how the spectral profiles may differ to eels .
z. guo , j. hwang , b. zhao , j. h. chung , s. g. cho , s .- j .
baek , and j. choo , `` ultrasensitive trace analysis for 2,4,6-trinitrotoluene using nano - dumbbell surface - enhanced raman scattering hot spots , '' analyst * 139 * , 807 ( 2014 ) . c. chicanne , t. david , r. quidant , j. c. weeber , y. lacroute , e. bourillot , a. dereus , g. cp;as des francs , and c. girard , `` imaging the local density of states of optical corrals '' phys .
lett . * 88 * , 097402 ( 2002 ) .
a. drezet , a. hohenau , d. koller , a. stepanov , h. ditlbacher , b. steinberger , f. r. aussenegg , a. leitner , and j. r. krenn , `` leakage radiation microscopy of surface plasmon polaritons , '' mater .
b * 149 * , 220 ( 2008 ) .
s. viarbitskaya , a. teulle , r. marty , j. sharma , c. girard , a. arbouet , and e. dujardin , `` tailoring and imaging the plasmonic local density of states in crystalline nanoprisms , '' nat .
* 12 * , 426 ( 2013 ) .
a. losquin , and m. kociak , `` link between cathodoluminescence and electron energy loss spectroscopy and the radiative and full electromagnetic local density of states , '' acs photonics * 2 * , 1619 ( 2015 ) .
j. nelayah , m. kociak , o. stphan , f. j. garca de abajo , m. tenc , l. henrard , d. taverna , i. pastoriza - santos , l. m. liz - marzn , and c. colliex , `` mapping surface plasmons on a single metallic nanoparticle , '' nat . phys . * 3 * , 348 ( 2007 ) .
f. j. garca de abajo , and m. kociak , `` probing the photonic local density of states with electron energy loss spectroscopy , '' phys .
* 100 * , 106804 ( 2008 ) .
u. hohenester , h. ditlbacher , and j. r. krenn , `` electron - energy - loss spectra of plasmonic nanoparticles , '' phys .
* 103 * , 106801 ( 2009 ) m. husnik , f. van cube , s. irsen , s. linden , j. niegemann , k. busch and m. wegener , `` comparison of electron energy - loss and quantitative optical spectroscopy on individual optical gold antennas , '' nanophotonics * 2 * , 241 ( 2013 ) .
o. nicoletti , m. wubs , n. a. mortensen , w. sigle , p. a. van aken , and p. a. midgley , `` surface plasmon modes of a single silver nanorod : an electron energy loss study , '' opt .
express * 19 * , 15371 ( 2011 ) .
d. rossouw , m. couillard , j. vickery , e. kumacheva , and g. a. botton , `` multipolar plasmonic resonances in silver nanowire antennas imaged with a subnanometer electron probe , '' nano lett .
* 11 * 1499 ( 2011 ) . f. von cube , s. irsen , j. niegemann , c. matyssek , w. hergert , k. busch , and s. linden , `` spatio - spectral characterization of photonic meta - atoms with electron energy - loss spectroscopy , '' opt . mater .
express * 1 * , 1009 ( 2011 ) .
sren raza1 , shima kadkhodazadeh , thomas christensen , marcel di vece , martijn wubs , n. asger mortensen , and nicolas stenger , `` multipole plasmons and their disappearance in few - nanometre silver nanoparticles , '' nature communications * 6 * , 1 ( 2015 ) .
a. hrl , a. trgler , and u. hohenester , `` full three - dimensonal reconstruction of the dyadic green tensor from electron energy loss spectroscopy of plasmonic nanoparticles , '' acs photonics * 2 * , 1429 ( 2015 ) .
g. boudarham , and m. kociak , `` modal decompositions of the local electromagnetic density of states and spatially resolved electron energy loss probability in terms of geometric modes , '' phys .
b * 85 * , 245447 ( 2012 ) .
ge , and s. hughes , `` design of an efficient single photon source from a metallic nanorod dimer : a quasi - normal mode finite - difference time - domain approach , '' opt .
lett . * 39 * , 4235 ( 2014 ) .
h. m. lai , p. t. leung , k. young , p. w. barber , and s. c. hill , `` time - independent perturbation for leaking electromagnetic modes in open systems with application to resonances in microdroplets , '' phys .
a * 41 * , 5187 ( 1990 ) .
the contribution of image charge when the electron coming to the slab is to accelerate the electron , while it will decelerate the electron when the electron is leaving the other side of the slab .
both process will be antisymmetric of each other since we have assume the speed change of the process is negligible with high energy electrons ; as a result , they would cancel each other . v. s. c. manga rao , and s. hughes , `` single quantum dot spontaneous emission in a finite - size photonic crystal waveguide : proposal for an efficient ' ' on chip single photon gun , phys .
lett . * 99 * , 193901 ( 2007 ) . c. van vlack , and s. hughes , `` finite - difference time - domain technique as an efficient tool for obtaining the regularized green function : applications to the local field problem in quantum optics for inhomogeneous lossy materials , '' opt
. lett . * 37 * , 2880 ( 2012 ) .
ge , and s. hughes , `` quantum dynamics of two quantum dots coupled through localized plasmons : an intuitive and accurate quantum optics approach using quasinormal modes , '' phys .
b * 92 * , 205420 ( 2015 )
. s .- t .
guillaume , f. j. garca de abajo , and l. henrard , `` efficient modal - expansion discrete - dipole approximation : application to the simulation of optical extinction and electron energy - loss spectroscopies , '' phys .
b * 88 * , 245439 ( 2013 ) .
rong - chun ge , c. van vlack , p. yao , jeff .
f. young , s. hughes , `` accessing quantum nanoplasmonics in a hybrid quantum - dot metal nanosystem : mollow triplet of a quantum dot near a metal nanoparticle , '' phys . rev .
b * 87 * , 205425 ( 2013 ) . | understanding light - matter interactions using localized surface plasmons ( lsps ) is of fundamental interest in classical and quantum plasmonics and has a wide range of applications . in order to understand the spatial properties of lsps , electron energy loss spectroscopy ( eels )
is a common and powerful method of spatially resolving the extreme localized fields that can be obtained with metal resonators .
however , modelling eels for general shaped resonators presents a major challenge in computational electrodynamics , requiring the full photon green function as a function of two space points and frequency . here
we present an intuitive and computationally simple method for computing eels maps of plasmonic resonators using a quasinormal mode ( qnm ) expansion technique . by separating the contribution of the qnm and the bulk material , we give closed - form analytical formulas for the plasmonic qnm contribution to the eels maps .
we exemplify our technique for a split ring resonator , a gold nanorod , and a nanorod dimer structure .
the method is accurate , intuitive , and gives orders of magnitude improvements over direct dipole simulations that numerically solve the full 3d maxwell equations .
we also show how the same qnm green function can be used to obtain the purcell factor ( and projected local density of optical states ) from quantum dipole emitters or two level atoms , and we demonstrate how the spectral features differ in general to the eels spectrum . |
a polygonal domains @xmath4 with @xmath1 corners and @xmath5 holes is a polygonal region of genus @xmath5 whose boundary consists of @xmath1 line segments . the holes and the outer boundary of @xmath4 are regarded as _
obstacles_. then , the geodesic distance between any two points @xmath6 in a given polygonal domain @xmath4 is defined to be the length of a shortest obstacle - avoiding path between @xmath7 and @xmath8 .
the euclidean shortest path problem in a polygonal domain has drawn much attention in the history of computational geometry @xcite . in the _ two - point shortest path query _ problem , we preprocess @xmath4 so that we can determine a shortest path ( or its length ) quickly for a given pair of query points @xmath9 . while we can compute a shortest path in @xmath10 time from scratch @xcite , known structures for logarithmic time query require significantly large storage @xcite .
chiang and mitchell @xcite developed several data structures that can answer a two - point query quickly with tradeoffs between storage usage and query time .
most notably , @xmath11 query time can be achieved by using @xmath12 space and preprocessing time ; sublinear query time by @xmath13 space and preprocessing time .
more recently , guo et al .
@xcite have shown that a data structure of size @xmath14 can be constructed in @xmath15 time to answer the query in @xmath16 time , where @xmath5 is the number of holes .
their results are summarized in [ table : summary ] . for more results on shortest paths in a polygonal domain
, we refer to survey articles by mitchell @xcite . in this paper , we focus on a variant of the problem , in which possible query points are restricted to a subset of @xmath4 ; the boundary @xmath17 of the domain @xmath4 or a set of line segments within @xmath4 .
in many applications , possible pairs of source and destination do not span the whole domain @xmath4 but a specified subset of @xmath4 .
for example , in an urban planning problem , the obstacles correspond to the residential areas and the free space corresponds to the walking corridors .
then , the query points are restricted to the spots where people depart and arrive , which are on the boundary of obstacles .
therefore , our goal is to design a data structure using much less resources than structures of chiang and mitchell @xcite when the query domain is restricted to the boundary of a given polygonal domain @xmath4 or to a set of segments in @xmath4 . to our best knowledge , no prior work seems to investigate this variation . as a main result , in section [ sec : log_query ]
, we present a data structure of size @xmath18 that can be constructed in @xmath19 time and can answer a @xmath17-restricted two - point shortest path query in @xmath11 time . here
, @xmath20 stands for the maximum length of a davenport - schinzel sequence of order @xmath21 on @xmath1 symbols @xcite .
it is good to note that @xmath22 for any constant @xmath21 as a convenient intuition , while tighter bounds are known @xcite .
we also provide a tradeoff between space and query time in section [ sec : tradeoff ] .
in particular , we show that one can achieve sublinear query time using @xmath3 space and preprocessing time .
new results in this paper are also summarized in [ table : summary ] .
[ table : summary ] .summary of new and known results on exact two - point shortest path queries , where @xmath23 is arbitrary and @xmath24 is a parameter .
[ new ] denotes our results . [ cols="^,^,^,^,^",options="header " , ] our data structure is a subdivision of two - dimensional domain parameterized in a certain way .
the domain is divided into a number of grid cells in which a set of constrained shortest paths between query points have the same structure .
each grid cell is divided according to the projection of the lower envelope of functions stemming from the constrained shortest paths . with careful investigation into this lower envelope ,
we show the claimed upper bounds .
also , our approach readily extends to the variant where query points are restricted to lie on a given segment or a given set of segments in @xmath4 .
we discuss this extension in section [ sec : extension ] . in the case where @xmath4 is a simple polygon ( @xmath25 ) , the two - point shortest path query can be answered in @xmath11 time after @xmath26 preprocessing time @xcite .
more references and results on shortest paths in simple polygons can be found in a survey article by orourke and suri @xcite before chaing and mitchell @xcite , fast two - point shortest path queries in polygonal domains were considered as a challenge . due to this difficulty , many researchers have focused on the _ approximate _ two - point shortest path query problem .
chen @xcite achieved an @xmath27-sized structure for @xmath28-approximate shortest path queries in @xmath11 time , and also pointed out that a method of clarkson @xcite can be applied to answer @xmath29-approximate shortest path queries in @xmath11 time using @xmath14 space and @xmath30 preprocessing time .
later , arikati et al .
@xcite have improved the above results based on planar spanners .
the problem on polyhedral surfaces also have been considered .
agarwal et al .
@xcite presented a data structure of size @xmath31 that answers a two - point shortest path query on a given _ convex _ polytope in @xmath32 time after @xmath31 preprocessing time for any fixed @xmath33 and any @xmath34 .
they also considered the problem where the query points are restricted to lie on the edges of the polytope , reducing the bounds by a factor of @xmath1 from the general case .
recently , cook iv and wenk @xcite presented an improved method using kinetic voronoi diagrams .
we are given as input a polygonal domain @xmath4 with @xmath5 holes and @xmath1 corners .
more precisely , @xmath4 consists of an outer simple polygon in the plane @xmath35 and a set of @xmath5 ( @xmath36 ) disjoint open simple polygons inside @xmath37 . as a set , @xmath4 is the region contained in its outer polygon _ excluding _ the holes , also called the _ free space_. the complement of @xmath4 in the plane is regarded as _
obstacles _ so that any feasible path does not cross the boundary @xmath17 and lies inside @xmath4 .
it is well known from earlier works that there exists a _
shortest ( obstacle - avoiding ) path _ between any two points @xmath38 @xcite .
let @xmath39 be the set of all corners of @xmath4 .
then any shortest path from @xmath40 to @xmath41 is a simple polygonal path and can be represented by a sequence of line segments connecting points in @xmath42 @xcite .
the _ length _ of a shortest path is the sum of the euclidean lengths of its segments .
the geodesic distance , denoted by @xmath43 , is the length of a shortest path between @xmath7 and @xmath8 .
also , we denote by @xmath44 the euclidean length of segment @xmath45 . a _ two - point shortest path query _ is given as a pair of points @xmath46 with @xmath47 and asks to find a shortest path between @xmath7 and @xmath8 . in this paper , we deal with a restriction where the queried points @xmath7 and @xmath8 lie on the boundary @xmath17 . a _ shortest path tree _ @xmath48 for a given source point @xmath40 is a spanning tree on the corners @xmath39 plus the source @xmath7 such that the unique path to any corner @xmath49 from the source @xmath7 in @xmath48 is a shortest path between @xmath7 and @xmath50 .
the combinatorial complexity of @xmath48 for any @xmath40 is at most linear in @xmath1 .
a shortest path map @xmath51 for the source @xmath7 is a decomposition of the free space @xmath4 into cells in which any point @xmath52 has a shortest path to @xmath7 through the same sequence of corners in @xmath39 .
once @xmath48 is obtained , @xmath51 can be computed as an additively weighted voronoi diagram of @xmath53 with weight assigned by the geodesic distance to @xmath7 @xcite ; thus , the combinatorial complexity of @xmath51 is linear .
a cell of @xmath51 containing a point @xmath54 has the common last corner @xmath55 along the shortest path from @xmath7 to @xmath8 ; we call such a corner @xmath50 the _ root _ of the cell or of @xmath8 with respect to @xmath7 . an @xmath27 time algorithm , using @xmath10 working space , to construct @xmath48 and @xmath51 is presented by hershberger and suri @xcite .
an _ spt - equivalence decomposition _
@xmath56 of @xmath4 is the subdivision of @xmath4 into cells in which every point has topologically equivalent shortest path tree .
an @xmath56 can be obtained by overlaying @xmath1 shortest path maps @xmath57 for every corner @xmath49
hence , the complexity of @xmath56 is @xmath58 .
note that @xmath59 consists of at most @xmath14 points ; they are intersection points between any edge of @xmath57 for any @xmath49 and the boundary @xmath17 .
we call those intersection points , including the corners @xmath39 , the _
breakpoints_. the breakpoints induce @xmath14 intervals along @xmath17 .
we shall say that a breakpoint is _ induced by @xmath57 _ if it is an intersection of an edge of @xmath57 and @xmath17 .
given a set @xmath60 of algebraic surfaces and surface patches in @xmath61 , the _ lower envelope _
@xmath62 of @xmath60 is the set of pointwise minima of all given surfaces or patches in the @xmath63-th coordinate .
the _ minimization diagram _
@xmath64 of @xmath60 is a decomposition of @xmath65 into faces , which are maximally connected region over which @xmath62 is attained by the same set of functions . in particular , when @xmath66 , the minimization diagram @xmath64 is simply a projection of the lower envelope onto the @xmath67-plane .
analogously , we can define the _ upper envelope _ and the _ maximization diagram_. as we intensively exploit known algorithms on algebraic surfaces or surface patches and their lower envelopes , we assume a model of computation in which several primitive operations dealing with a constant number of given surfaces can be performed in constant time : testing if a point lies above , on or below a given surface , computing the intersection of two or three given surfaces , projecting down a given surface , and so on .
such a model of computation has been adopted in many research papers ; see @xcite .
in this section , we present a data structure that answers a two - point query restricted on @xmath17 in @xmath11 time . to ease discussion , we parameterize the boundary @xmath17 . since @xmath17 is a union of @xmath68 closed curves , it can be done by parameterizing each curve by arc length and merging them into one .
thus , we have a bijection @xmath69 that maps a one - dimensional interval into @xmath17 , where @xmath70 denotes the total lengths of the @xmath68 closed curves forming @xmath17 .
conversely , the inverse of @xmath7 maps each interval along @xmath17 to an interval of @xmath71 .
a shortest path between two points @xmath72 is either the segment @xmath45 or a polygonal chain through corners in @xmath39 .
thus , unless @xmath73 , the geodesic distance is taken as the minimum of the following functions @xmath74 over all @xmath75 , which are defined as follows : @xmath76 where @xmath77 , for any point @xmath78 , denotes the _ visibility profile _ of @xmath52 , defined as the set of all points @xmath79 that are _ visible from _ @xmath52 ; that is , @xmath80 lies inside @xmath4 .
the symbol @xmath81 can be replaced by an upper bound of @xmath82 ; for example , the total length @xmath70 of the boundary of the polygonal domain @xmath4 . since the case where @xmath83 is visible from @xmath84 ,
so the shortest path between them is just the segment @xmath85 , can be checked in @xmath11 time using @xmath15 space @xcite , we assume from now on that @xmath86 . hence , our task is to efficiently compute the lower envelope of the @xmath14 functions @xmath87 on a 2-dimensional domain @xmath88 . using known results on the lower envelope of the algebraic surfaces in @xmath89-dimension , we can show that a data structure of size @xmath90 for @xmath11 query can be built in @xmath90 time as follows .
fix a pair of intervals @xmath91 and @xmath92 induced by the breakpoints . since @xmath91 belongs to a cell of an spt - equivalence decomposition , @xmath93 is independent of choices over all @xmath94 .
therefore , the set @xmath95 of corners visible from @xmath83 is also independent of the choice of @xmath96 and further , for a fixed @xmath97 , there exists a unique @xmath55 that minimizes @xmath98 for any @xmath99 over all @xmath49 @xcite .
this implies that for each such subdomain @xmath100 we extract at most @xmath1 functions , possibly appearing at the lower envelope .
moreover , in @xmath101 , such a function is represented explicitly ; for @xmath102 and @xmath103 , @xmath104 where @xmath105 and @xmath106 are the @xmath52- and the @xmath107-coordinates of @xmath83 , and @xmath108 and @xmath109 are the @xmath52- and the @xmath107-coordinates of a point @xmath110 .
note that @xmath105 and @xmath106 are linear functions in @xmath111 by our parametrization .
for each @xmath112 , there exists a corner @xmath113 that minimizes @xmath114 in @xmath101 over all @xmath115 .
therefore , the function @xmath116 on @xmath101 is well - defined .
observe that the graph of @xmath117 is an algebraic surface with degree at most @xmath118 in @xmath89-dimensional space . applying any efficient algorithm that computes the lower envelope of algebraic surfaces in @xmath119
, we can compute the lower envelope of the functions @xmath117 in @xmath120 time @xcite . repeating this for every such subdomain @xmath121 yields @xmath90 space and preprocessing time . since we would like to provide a point location structure in domain @xmath122 , we need to find the minimization diagram @xmath123 of the computed lower envelope .
fortunately , our domain is @xmath124-dimensional , so we can easily project it down on @xmath122 and build a point location structure with an additional logarithmic factor . in another way around , one could try to deal with _ surface patches _ on the whole domain @xmath122 .
consider a fixed corner @xmath125 and its shortest path map @xmath126 .
the number of breakpoints induced by @xmath126 is at most @xmath26 , including the corners @xmath39 themselves .
this implies at most an @xmath14 number of combinatorially different paths between any two boundary points @xmath83 and @xmath84 via @xmath127 .
that is , for a pair of intervals @xmath91 and
@xmath92 , we have a unique path via @xmath127 and its length is represented by a partial function of @xmath128 defined on a rectangular subdomain @xmath100 .
hence , we have @xmath14 such partial functions for each @xmath125 , and thus @xmath129 in total .
each of them defines an algebraic surface patch of constant degree on a rectangular subdomain .
consequently , we can apply the same algorithm as above to compute the lower envelope of those patches in @xmath130 time and space .
now , we present a way of proper grouping of subdomains to reduce the time / space bound by a factor of @xmath1 .
we call a subdomain @xmath100 , where both @xmath91 and @xmath92 are intervals induced by breakpoints , a _ grid cell_. thus , @xmath122 consists of @xmath58 grid cells
. we will decompose @xmath122 into @xmath129 _ blocks _ of @xmath26 grid cells .
consider a pair of boundary edges @xmath131 and let @xmath132 and @xmath133 be the number of breakpoints on @xmath134 and on @xmath135 , respectively .
let @xmath136 and @xmath137 be the breakpoints on @xmath134 and on @xmath135 , respectively , in order @xmath138 and @xmath139 .
take @xmath133 grid cells with @xmath140 and let @xmath141 be their union .
we redefine the functions @xmath87 on domain @xmath142 . as discussed above , for any @xmath143
, we have a common subset @xmath144 of corners visible from @xmath83 . for any @xmath97 , let @xmath145 be a function defined on @xmath142 and @xmath146 be the number of breakpoints on @xmath135 induced by @xmath126 .
the following is our key observation .
[ lemma : patch ] the graph of @xmath147 on @xmath142 consists of at most @xmath148 algebraic surface patches with constant maximum degree . if @xmath149 for any @xmath150 and some @xmath151 , then @xmath84 lies in a cell of @xmath126 with root @xmath50 ; by the definition of @xmath117 , the involved path goes directly from @xmath83 to @xmath127 and follows a shortest path from @xmath127 to @xmath84 .
on the other hand , when we walk along @xmath135 as @xmath152 increases from @xmath153 to @xmath154 , we encounter @xmath155 breakpoints induced by @xmath126 ; thus , @xmath156 cells of @xmath126 .
hence , the lemma is shown .
moreover , the partial function corresponding to each patch of @xmath157 is defined on a rectangular subdomain @xmath158 for some @xmath159 .
this implies that the lower envelope of @xmath117 on @xmath142 is represented by that of at most @xmath160 surface patches .
though this envelope can be computed in @xmath161 time , we do further decompose @xmath142 into @xmath162 blocks of at most @xmath1 grid cells .
this can be simply done by cutting @xmath142 at @xmath163 for each @xmath164 . for each such block of grid cells
, we have at most @xmath165 surface patches and thus their lower envelope can be computed in @xmath120 time .
hence , we obtain the following consequence
. one can preprocess a given polygonal domain @xmath4 in @xmath13 time into a data structure of size @xmath13 that answers the two - point shortest path query restricted to the boundary @xmath17 in @xmath11-time , where @xmath166 is an arbitrarily small positive number .
recall that @xmath167 . for a pair of boundary edges
@xmath134 and @xmath135 , we can compute the lower envelope of the functions @xmath87 in @xmath168 . summing this over every pair of boundary edges
, we have @xmath169 a point location structure on the minimization diagram can be built with additional logarithmic factor , which is subdued by @xmath170 .
the algorithms we described so far compute the lower envelope of surface patches in @xmath89-space in order to obtain the minimization diagram @xmath123 of the functions @xmath87 . in this subsection
, we introduce a way to compute @xmath123 rather directly on @xmath122 , based on more careful analysis . basically , we make use of the same scheme of partitioning the domain @xmath122 into blocks of ( at most ) @xmath1 grid cells as in section [ sec : log_query].2 .
let @xmath142 be such a block defined as @xmath171 such that @xmath172 is an interval induced by the breakpoints and @xmath173 is a union of @xmath174 consecutive intervals in which we have @xmath175 breakpoints @xmath176 . by lemma [ lemma : patch ]
, the functions @xmath177 restricted to @xmath142 can be split into at most @xmath165 partial functions @xmath178 with @xmath179 defined on a subdomain @xmath180 .
each @xmath181 is represented explicitly as @xmath182 in @xmath183 , so that we have @xmath184 for any @xmath185 and some @xmath186 .
note that it may happen that @xmath187 or @xmath188 for some @xmath189 and @xmath190 ; in particular , if @xmath187 , we have @xmath191 .
also , as discussed in section [ sec : log_query].2 , @xmath183 is represented as @xmath192 for some @xmath193 . in this section ,
we take the partial functions @xmath178 into account , and thus the goal is to compute the minimization diagram @xmath123 of surface patches defined by the @xmath178 .
we start with an ordering on the set @xmath144 of corners visible from @xmath83 for any @xmath194 based on the following observation .
[ lemma : ordering ] the angular order of corners in @xmath144 seen at @xmath111 is constant if @xmath111 varies within @xmath172 .
if this is not true , we have such an @xmath195 that @xmath196 and two corners @xmath197 in @xmath144 are collinear . since corners lie on the boundary of an obstacle , one of @xmath50 and @xmath198 is not visible from @xmath83 locally near @xmath199 ; that is , @xmath196 is a breakpoint , a contradiction
. increases , @xmath83 moves in direction where the obstacle is to the right . ] without loss of generality , we assume that as @xmath111 increases , @xmath83 moves along @xmath17 in direction that the obstacle lies to the _ right _ ; that is , @xmath83 moves _ clockwise _ around each hole and _ counter - clockwise _ around the outer boundary of @xmath4 .
( see figure [ fig : orientation ] . ) by lemma [ lemma : ordering ] , we order the corners in @xmath144 in counter - clockwise order at @xmath83 for any @xmath143 ; let @xmath200 be a total order on @xmath144 such that @xmath201 if and only if @xmath202 . from now on , we investigate the set @xmath203 which is a projection of the intersection of two surface patches defined by @xmath178 and @xmath204 .
one can easily check that @xmath205 is a subset of an algebraic curve of degree at most @xmath206 .
[ lemma : monotone ] the set @xmath205 is @xmath152-monotone .
that is , for fixed @xmath152 , there is at most one @xmath143 such that @xmath207 .
if @xmath207 , we have equation @xmath208 . from the equation ,
we get @xmath209 .
if we fix @xmath152 as constant , @xmath210 remains a constant even if @xmath111 varies within @xmath172 .
thus , @xmath83 is an intersection point with line segment @xmath211 and a branch @xmath212 of a hyperbola whose foci are @xmath213 and @xmath214 .
if @xmath215 consists of two points , then the line @xmath216 through @xmath213 and @xmath214 must cross @xmath211 at a point @xmath217 ; @xmath216 is the transverse axis of @xmath212 and both @xmath213 and @xmath214 lie in one side of the line supporting @xmath211 .
such a crossing point @xmath217 is a breakpoint by definition but we do not have any breakpoint within @xmath211 since @xmath172 is an interval induced by the breakpoints , a contradiction .
[ lemma : increasing ] the set @xmath205 is either an empty set or an open curve whose endpoints lie on the boundary of @xmath218 .
moreover , @xmath205 is either a linear segment parallel to the @xmath152-axis or @xmath111-monotone .
let @xmath91 and @xmath92 be intervals such that @xmath219 .
note that @xmath220 and @xmath221 for some @xmath222 .
first , note that if @xmath187 , @xmath223 and thus @xmath224 , so the lemma is true .
thus , we assume that @xmath225 .
regarding @xmath226 and @xmath227 , there are two cases : @xmath188 or @xmath228 . in the former case , we get @xmath229 from equation @xmath208 .
observe that variable @xmath152 is readily eliminated from the equation , and thus if there exists @xmath230 with @xmath231 , we have @xmath232 for every other @xmath233 .
hence , by lemma [ lemma : monotone ] , @xmath205 is empty or a straight line segment in @xmath218 which is parallel to @xmath152-axis , and thus the lemma is shown .
now , we consider the latter case where @xmath234 . without loss of generality
, we assume that @xmath235 .
recall that if @xmath236 , then @xmath237 for any @xmath111 in the interior of @xmath91 .
we denote @xmath238 and @xmath239 . on the other hand
, we also have a similar relation for @xmath226 and @xmath227 .
let @xmath240 and @xmath241 .
observe that @xmath242 and @xmath243 are continuous functions of @xmath152 , and if @xmath244 at @xmath245 , then @xmath246 is a breakpoint induced by @xmath247 or @xmath248 . since @xmath92 contains no such breakpoint induced by @xmath247 or @xmath248 in its interior , either @xmath249 or @xmath250 for all @xmath152 in the interior of @xmath92 ; that is , the sign of @xmath251 is constant .
since for any @xmath252 with @xmath253 we have @xmath254 by our parametrization , we can represent @xmath255 and @xmath256 , where @xmath257 , @xmath258 , @xmath259 and @xmath260 are constants depending on @xmath213 , @xmath226 , and parametrization @xmath7 .
more specifically , @xmath261 denotes a signed distance between @xmath83 and the perpendicular foot of @xmath213 onto the line supporting @xmath262 , and @xmath258 is the distance between @xmath213 and the line supporting @xmath262 .
see figure [ fig : bisector ] .
thus , @xmath181 can represented as @xmath263 . ]
now , we differentiate the both sides of equation @xmath264 by @xmath152 to obtain the derivative @xmath265 : @xmath266 rearranging this , we obtain @xmath267 since @xmath268 , we have @xmath269 . also , as discussed above , @xmath251 has a constant sign when @xmath152 varies within the interior of @xmath92 .
thus , @xmath270 has a constant sign , and @xmath271 also has a constant sign at any @xmath207 .
furthermore , @xmath271 is continuous and has no singularity in the interior of @xmath272 .
this , together with lemma [ lemma : monotone ] , proves the lemma .
now , we know that @xmath205 can be seen as the graph of a partial function @xmath273 . also , lemma [ lemma : increasing ] implies that @xmath205 _ bisects _ @xmath218 into two connected regions @xmath274 and @xmath275 , where @xmath276 and @xmath277 .
let @xmath278 be the set of points @xmath128 where the minimum of @xmath279 over all @xmath190 is attained by @xmath181 .
we then have @xmath280 for each @xmath189 . for easy explanation , from now on ,
we regard the @xmath111-axis as the _ vertical axis _ in @xmath122 so that we can say a point lies _ above _ or _ below _ a curve in this sense .
the idea of computing @xmath278 is to use the lower and the upper envelopes of the bisecting curves @xmath205 . in order to do so ,
we extend @xmath205 to cover the whole @xmath152-interval @xmath281 in @xmath282 by following operation : for each endpoint of @xmath205 , if it does not lie on the vertical line @xmath283 or @xmath284 , attach a horizontal segment to reach the vertical line as shown in figure [ fig : envelopes](a ) .
we denote the resulting curve by @xmath285 ; if @xmath286 , define @xmath285 as the horizontal segment connecting two points @xmath287 and @xmath288 in @xmath218 .
observe now that @xmath285 bisects @xmath218 into regions @xmath274 and @xmath275 , which lie _ above _ and _ below _
@xmath285 , respectively .
( dashed line ) to @xmath285 by attaching horizontal segments ( solid line ) ; ( b ) @xmath289 , ( c ) @xmath290 , and ( d ) the region @xmath278 between them . here
, the dotted boxes are grid cells whose union is @xmath183 and the @xmath111-axis appears vertical . ] let @xmath291 be the region above @xmath285 and @xmath292 be the region below @xmath285 . for a fixed @xmath189 with @xmath179 , we classify the @xmath285 into two sets @xmath293 and @xmath294 such that @xmath295 if @xmath296 or @xmath297 if @xmath298 . recall that @xmath299 .
thus , we have @xmath300 the boundary of @xmath301 in @xmath183 is the lower envelope @xmath289 of @xmath293 ; symmetrically , the boundary of @xmath302 in @xmath183 is the upper envelope @xmath290 of @xmath294 .
therefore , @xmath303 , the region below the lower envelope @xmath289 of @xmath293 and above the upper envelope @xmath290 of @xmath294 , and it can be obtained by computing the overlay of two envelopes @xmath289 and @xmath290 .
see figure [ fig : envelopes](b)(d ) .
we exploit known results on the davenport - schinzel sequences to obtain the following lemma @xcite .
[ lemma : ds ] the set @xmath278 is of combinatorial complexity @xmath304 and can be computed in @xmath305 time , where @xmath306 is the maximum length of a davenport - schinzel sequence of order @xmath21 on @xmath1 symbols .
@xmath285 consists of at most three arcs , at most one algebraic curve of degree @xmath206 and at most two straight segments .
thus , we have at most @xmath307 algebraic arcs of degree at most @xmath206 .
any two such arcs can intersect each other at most @xmath308 times by bzout s theorem @xcite .
thus , each of @xmath289 and @xmath290 has complexity @xmath304 and can be computed in @xmath309 time @xcite . after sorting the vertices on these envelopes in @xmath152-increasing order
, we can easily specify all intersections between @xmath289 and @xmath290 in the same bound .
it should be noted here that the exact constant @xmath310 is not relevant ; it only matters that this is some constant .
we can compute the minimization diagram @xmath123 by computing each @xmath278 in @xmath311 time . in the same time
bound , we can build a point location structure on @xmath123 .
finally , we conclude our main theorem .
[ thm : log - query ] one can preprocess a given polygonal domain @xmath4 in @xmath312 time into a data structure of size @xmath18 that answers the two - point shortest path query restricted on the boundary @xmath17 in @xmath11 time .
in this section , we provide a space / query - time tradeoff .
we use the technique of partitioning @xmath39 , which has been introduced in chiang and mitchell @xcite .
let @xmath313 be a positive number with @xmath24 .
we partition the corner set @xmath39 into @xmath314 subsets @xmath315 of near equal size @xmath316 . for each such subset @xmath317 of corners ,
we run the algorithm described above with little modification : we build the shortest path maps @xmath126 only for @xmath318 and care about only @xmath319 breakpoints induced by such @xmath126 .
thus , we consider only the paths from @xmath83 via @xmath320 and @xmath55 to @xmath84 , and thus @xmath319 functions @xmath87 for @xmath318 and @xmath49 . since we deal with less number of functions , the cost of preprocessing reduces from @xmath26 to @xmath316 at several places .
we take blocks of @xmath316 grid cells contained in @xmath122 and the number of such blocks is @xmath321 . for each such block
, we spend @xmath322 time to construct a point location structure for the minimization map @xmath323 of the functions . iterating all such blocks
, we get running time @xmath324 for a part @xmath317 of @xmath39 . repeating this for all such subsets @xmath317 yields @xmath325 construction time .
each query is processed by a series of @xmath21 point locations on every @xmath323 , taking @xmath326 time .
[ thm : tradeoff ] let @xmath313 be a fixed parameter with @xmath24 . using @xmath325 time and @xmath327 space
, one can compute a data structure for @xmath328-time two - point shortest path queries restricted on the boundary @xmath17 .
remark that when @xmath329 , we obtain theorem [ thm : log - query ] , and that @xmath3 time and space is enough for sublinear time query . note
that if @xmath26 time is allowed for processing each query , @xmath14 space and @xmath330 preprocessing time is sufficient .
our approach easily extends to the two - point queries in which queried points are restricted to be on a given segment or a given polygonal chain lying in the free space @xmath4 .
let @xmath331 and @xmath332 be two sets of @xmath333 and @xmath334 line segments , respectively , within @xmath4 . in this section ,
we restrict a query pair @xmath46 of points to lie on @xmath331 and @xmath332 each
. we will refer to this type of two - point query as a _
@xmath335-restricted two - point query_. as we did above , we take two segments @xmath336 and @xmath337 and let @xmath132 and @xmath133 be the number of breakpoints the intersection points with an edge of @xmath57 for some @xmath49 on @xmath134 and @xmath135 , respectively . also , parameterize @xmath134 and @xmath135 as above so that we have two bijections
@xmath338 \to s$ ] and @xmath339 \to t$ ] .
any path from a point on @xmath134 leaves to one of the two sides of @xmath134 .
thus , the idea of handling such a segment within the free space @xmath4 is to consider two cases separately . here , we regard @xmath134 and @xmath135 as directed segments in direction of movement of @xmath83 and @xmath340 as @xmath111 and @xmath152 increases , and consider only one case where paths leave @xmath134 to its left side and arrive at @xmath135 from its left side .
the other cases are analogous .
then , the situation is almost identical to that we considered in section [ sec : log_query].2 . for a pair of segments @xmath134 and @xmath135
, we can construct a query structure in @xmath341 time .
unfortunately , @xmath132 and @xmath133 can be as large as @xmath14 , yielding the same time bound for @xmath342-restricted two - point queries in the worst case .
thus , in the worst case , we need additional factor of @xmath343 as follows . [ thm : convexchain ] let @xmath331 and @xmath332 be two sets of @xmath333 and @xmath334 ( possibly crossing ) line segments , respectively , within @xmath4 , and @xmath313 be a fixed parameter with @xmath24 . then , using @xmath344 time and @xmath345 space , one can compute a data structure for @xmath346-time @xmath335-restricted two - point queries .
remark that in practice we expect that the number of breakpoints @xmath347 and @xmath348 is not so large as @xmath349 that the preprocessing and required storage would be much less than the worst case bounds .
in this paper , we posed the variation of the two - points query problem in polygonal domains where the query points are restricted in a specified subset of the free space .
and we obtained significantly better bounds for the boundary - restricted two - point queries than for the general queries . despite of its importance ,
the two - point shortest path query problem for the polygonal domains is not well understood .
there is a huge gap ( @xmath26 to @xmath12 ) about logarithmic query between the simple polygon case and the general case but still the reason why we need such a large storage is still unclear . on the other hand , restriction on the query domain provides another possibility of narrowing the gap with several new open problems : ( 1 ) what is the right upper bound on the complexity of the lower envelope defined by the functions @xmath87 on the parameterized query domain ? and what about any lower bound construction ? ( 2 ) if the query domain is a simple @xmath124-dimensional shape , such as a triangle , then can one achieve a better performance than the general results by chiang and mitchell ? we would carefully conjecture that our upper bound @xmath0 for logarithmic query could be improved to @xmath350 .
indeed , we have @xmath58 grid cells on the parameterized query domain and whenever we cross their boundaries , changes in the involved functions are usually bounded by a constant amount .
thus , if one could find a clever way of updating the functions and their lower envelope , it would be possible to achieve an improved bound .
the authors thank matias korman for fruitful discussion .
s. r. arikati , d. z. chen , l. p. chew , g. das , m. h. m. smid , and c. d. zaroliagis .
planar spanners and approximate shortest path queries among obstacles in the plane . in _ proc .
sympos . algo .
( esa96 ) _ , pages 514528 , 1996 . | we consider a variant of two - point euclidean shortest path query problem : given a polygonal domain , build a data structure for two - point shortest path query , provided that query points always lie on the boundary of the domain . as a main result
, we show that a logarithmic - time query for shortest paths between boundary points can be performed using @xmath0 preprocessing time and @xmath0 space where @xmath1 is the number of corners of the polygonal domain and the @xmath2-notation suppresses the polylogarithmic factor .
this is realized by observing a connection between davenport - schinzel sequences and our problem in the parameterized space .
we also provide a tradeoff between space and query time ; a sublinear time query is possible using @xmath3 space .
our approach also extends to the case where query points should lie on a given set of line segments . |
beautiful stellar disks only come to be with the accretion of gaseous clouds of star formation fuel .
there are several reasons why this fuel is thought to be gradually accreted from a galaxy s halo .
the first is that halo gas exists around our galaxy and other spiral galaxies ( e.g. , oort 1970 ; thilker et al .
2004 ; oosterloo , fraternali & sancisi 2007 ) .
halo gas therefore appears to be a relatively common phenomenon and the gas velocities indicate it will not escape from the galaxy , but rather eventually fall towards the disk .
the second reason is the metallicity distribution of the long - lived stars in the galactic disk , which indicates low metallicity fuel must be continually accreted , i.e. , the g - dwarf problem .
the metallicity distribution of the g and k dwarfs in the solar neighborhood can not be reproduced with simple closed box models , and the need for gaseous inflow for the majority of the life of the disk has persisted with further observations and increasingly complex chemical evolution models ( e.g. , larson 1972 ; fenner & gibson 2003 ; kotoneva et al . 2002 ; chiappini , matteucci & romano 2001 ; magrini , corbelli & galli 2007 ; worthey et al .
a third reason to bring in star formation fuel from the halo is the lack of radial gaseous inflow observed in the disk of spiral galaxies ( wong et al .
2004 ) . bringing the outer , relatively unenriched gas to the inner regions of the disk
would potentially provide a fresh source of star formation fuel .
there are two main potential sources of gaseous fuel in a galaxy halo : gas - rich satellites and condensed material from the hot diffuse halo
. the latter may be a combination of material left from the initial collapse of baryons into the dark matter halo , stripped material from the satellites that was integrated into this hot halo , and some level of galactic fountain material at low latitudes . in this review ,
we discuss these two reservoirs of disk star formation fuel with a focus on the milky way , given this is where we have the most information on the satellite distribution and the low column density halo gas .
we will also discuss the halo gas of m31 and m33 in the context of its origin and fate .
the number of satellites in the local group has increased substantially recently with the results from the sloan digital sky survey ( sdss ; e.g. , willman et al .
2005 ; zucker et al .
2006 ; belokurov et al .
2006 ) and surveys of the m31 environment ( e.g. , ibata et al .
the new satellites have typical v - band luminosities of @xmath0 l@xmath1 and total masses in the @xmath2 m@xmath1 range ( simon & geha 2007 ) .
they are found at a range of distances about both the galaxy and m31 .
gas from satellites is generally the first thing to be stripped as it passes through the diffuse halo medium of the larger galaxy .
the stripped gas will fuel , or has fueled , the galaxy and m31 .
figure [ hidist ] shows the distribution of galactic and m31 satellites in distance and hi mass ( grcevich & putman 2008 ) .
this plot is similar to blitz & robishaw ( 2000 ) , but is updated to incorporate new hi data , optical velocities , and all of the known local group dwarf galaxies .
the majority of the dwarf galaxies in the local group that still have neutral hydrogen are beyond 300 kpc from the galaxy and m31m@xmath1 . ] . in general those galaxies
still with gas lie towards the outskirts of the local group .
the gaseous detection of the new sdss galaxy leo t ( irwin et al .
2007 ) with the galactic arecibo l - band feed array ( galfa ) is shown in figure [ leot ] ( see also ryan - weber et al .
the majority of the galaxies that do not have neutral hydrogen are within @xmath3 kpc , with the exception of tucana and cetus
. there could be a sdss sensitivity effect for why there are not more gas - less galaxies at large radii ( koposov et al .
2007 ) , but the result of having all of the galaxies with hi beyond 300 kpc does not change .
the galaxies without gas have most likely been subjected to ram pressure stripping ( e.g. , mayer et al .
2006 ) , but halo densities high enough to effectively strip dwarf galaxies are not expected at radii of 150 - 300 kpc ( see grcevich & putman 2008 for details ) .
km s@xmath4 and the emission from the galaxy.[leot],height=220 ] the amount of gas brought into the milky way from the accretion of satellite galaxy material can be estimated by presuming that all of its currently gas - less satellites had their fuel assimilated into the galactic disk .
known stellar streams around the milky way can also be included , though obviously we can not account for the dwarfs that have also had their stellar component completely assimilated in the disk .
the gas - less dwarf galaxies in figure 1 most likely once had a similar gas mass to the dwarf galaxies with gas , or @xmath5 m@xmath1 .
the gas - less dwarf galaxies and 4 other destroyed dwarfs traced by stellar streams ( newberg et al . 2007 ;
belokurov et al .
2007 ) would therefore have contributed @xmath6
m@xmath1 of hi to fuel the galaxy .
this value would increase with the inclusion of the satellites molecular and ionized gas , but it is unlikely to increase by more than a factor of two ( e.g. , leroy et al .
chemical evolution models of the galactic disk typically require an average infall rate of 1 m@xmath1/yr over the past 5 - 7 gyr ( chiappini et al .
2001 ) , and therefore the dwarf galaxies appear to provide sufficient fuel for only @xmath7 years . it should be noted that the 1 m@xmath1/yr is an average rate of infall , and it should be closer to half that value today ( chiappini , this proceedings ) . in any case , to accomplish the average of 1 m@xmath1/yr from only the accretion of gas from small dwarf galaxies would require an average of 300 accretion events per gyr .
this is extremely high compared to the predictions of @xmath8cdm models ( zentner et al .
2005 ) , and the number of satellites we currently see in the halo .
there is a source of gaseous fuel associated with galactic satellites that has not yet been discussed , the magellanic system .
this system is currently at @xmath9 kpc and consists of the small and large magellanic clouds , and the magellanic stream and leading arm that are being stripped from the clouds ( putman et al . 2003 ) .
the magellanic system will eventually contribute @xmath10 m@xmath1 of fuel just in neutral hydrogen to the galaxy .
galaxies such as the magellanic clouds and larger bodies may have contributed to fueling the galaxy and m31 in the past . in this case , only an average of 1 accretion event per gigayear would be required ; however the stability of the disk would most likely be an issue ( stewart et al .
in the discussion of satellite galaxy material above , the focus was on the accretion of gas directly in the cold form .
another method a galaxy can obtain star formation fuel is by accreting clouds that have condensed within the hot halo medium .
this hot halo is predicted to originate from the initial collapse of baryons into the dark matter potential well , but currently it would be a mixture of this material , stripped satellite gas that was integrated into the halo before reaching the disk , and galactic fountain material , or hot gas rising into the halo from the cumulative effect of supernovae in the disk .
recent galaxy formation and evolution models have the hot halo gas gradually cool over time , with halo clouds forming from thermal instabilities ( maller & bullock 2004 ; lin & murray 2000 ) .
this is in contrast to the original models that have all of the hot halo gas cool and fall in at early times ( white & rees 1978 ; white & frenk 1991 ) .
the recent models predict at @xmath11 the existence of both a hot halo medium that hides a large percentage of a galaxy s baryons , and cool halo clouds that will gradually fuel the disk ( maller & bullock 2004 ; sommer - larsen 2006 ; kaufmann et al .
2006 ; fukugita & peebles 2006 ; connors et al .
2006 ) .
both cool halo clouds and a hot diffuse medium are found in the galactic halo .
the cool halo clouds are the high - velocity clouds ( hvcs )
neutral hydrogen clouds with velocities that are inconsistent with simple models of galactic disk rotation ( wakker & van woerden 1997 ; putman et al .
the hot diffuse halo medium is largely detected indirectly through the structure of halo clouds ( e.g. , peek et al .
2007 ) and the detection of high - velocity o vi absorption associated with the interaction of the halo clouds and this medium ( sembach et al . 2003 ) .
there are also constraints on the properties of the diffuse halo medium from x - ray absorption line detections ( fang et al .
2006 ; wang et al .
2005 ; williams et al .
figure [ ht ] shows examples of head - tail clouds formed in the galactic halo as hvcs move through the diffuse halo medium . with fabian heitsch
, we are simulating this process to reproduce the structure of the observed head - tail clouds and probe the properties of the diffuse halo medium . ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] ) , 1.6 , 2 , 4 , 8 , 12 , 16 , and 20 @xmath12 @xmath13.[ht],title="fig:",height=144 ] in peek , putman & sommer - larsen ( 2008 ) we examined the link between the observed hvcs and the clouds produced in a high - resolution sph cosmological simulation of galaxy formation .
the simulation includes chemical evolution , supernova feedback , and radiative cooling with an extragalactic uv field ( see sommer - larsen , gotz & portinari 2003 for more details ) .
a resolution of @xmath14 m@xmath1 for halo clouds was reached by finding a milky way ( v@xmath15 km s@xmath4 , m@xmath16 m@xmath1 ) in a lower resolution simulation and then tracing it back and replacing one particle with 12 at @xmath17 , and then one particle with 8 particles @xmath18 gyr later .
the simulation was then run for another 0.5 gyr and the halo cloud population examined at several timesteps .
the simulated and observed halo clouds are found throughout the galactic sky , at distances below 60 kpc .
there has been substantial recent progress on constraining the distances to the observed hvcs using direct and indirect methods ( e.g. , thom et al .
2007 ; westmeier et al .
the simulated clouds have an average galactic standard of rest velocity of v@xmath19 to @xmath20 km s@xmath4 ( depending on the location of the sun in the simulation ) , which is comparable to the observed hvc average velocity of v@xmath21 km s@xmath4 . the mass of the simulated clouds can be converted to an observed flux with the known distance and by assuming a sensible ionized component ( @xmath22% ionized ) .
the simulated cloud fluxes ( individual and total ) are also found to be comparable to the observed hvcs .
a large percentage of the flux from the simulated clouds comes from a few nearby complexes , which is also the case for the galactic hvcs ( i.e. complex c ) .
peek et al .
( 2008 ) present the figures that support the comparisons discussed in this paragraph .
the observed and simulated halo clouds show many similarities , indicating many hvcs may represent clouds condensing from a hot halo medium .
the accretion rate of these clouds can be determined from the simulation and is found to be @xmath18 m@xmath1/yr onto the galaxy s disk .
figure [ infall ] shows the infall rate of simulated clouds onto subsequent galactic radii .
the gradually increasing infall rate shows that clouds are created and/or grow at all radii .
the convergence of the infall rate at 0.2 m@xmath1/yr is at @xmath11 and this accretion rate is predicted to be much greater at higher redshifts consistent with the higher star formation rate ( sommer - larsen et al .
the previous sections focus on the accretion of gas onto the galactic disk .
up until recently the milky way was the only galaxy with a known halo cloud population , with the exception of some detections of large halo features similar to the magellanic stream ( @xmath23 m@xmath1 ) in nearby galaxies ( e.g. , ryder et al .
2001 ; hibbard et al . 2001 ) .
new receivers and larger allocations of telescope time have allowed deep hi surveys of nearby spiral galaxies to be completed , and these surveys are beginning to indicate cold halo clouds are a common property of spiral galaxies .
the local group spirals are discussed here , but see also the deep hi observations of other nearby spirals where additional halo clouds are observed ( e.g. fraternali et al .
2001 ; oosterloo et al . 2007 ; heald et al . ,
this proceedings ; battaglia et al 2006 ; boomsma et al .
2005 ) .
m31 is our closest spiral of similar size to the milky way , and was found to have numerous halo clouds by thilker et al .
the population of clouds detected is within 50 kpc of m31 and has a total hi mass of a few @xmath24 m@xmath1 .
westmeier et al .
( 2007 ) later showed there are no additional clouds at larger radii .
several of the m31 halo clouds are in the vicinity of the giant stellar stream ( to the south of m31 ) and ngc 205 , while others appear to be completely isolated .
the clouds in the vicinity of satellites may be remnants of their gaseous components , but they may also represent clouds formed in instabilities in the hot halo medium created by the passage of the satellite .
the isolated m31 halo clouds are most likely condensed clouds , although they could also represent the remnants of clouds stripped from satellites at earlier times .
the velocities of the clouds relative to m31 and the known satellites may cast some light on the most likely origin ( e.g. see figure 3 of peek et al .
the total mass in m31 s hi halo clouds is within a factor of three of the galaxy s population if you adopt the distance constraint of @xmath25 kpc for the galactic halo clouds and exclude the magellanic stream ( putman 2006 ) .
m33 also has clouds in its halo .
figure [ m33chans ] shows the preliminary galfa channel maps of m33 .
two main halo features can be noted in this figure , though additional clouds at lower contour levels are evident in the data cube .
the first feature is most predominant at @xmath26 and -240 to -260 km s@xmath4 , but continues beyond this velocity range and spatially wraps around to join the northern side of m33 s warped disk .
the second halo feature is a small cloud with a filament to the main galaxy at @xmath27 which is most evident at -148 km s@xmath4 ( see also westmeier et al . 2005 ) .
the first halo cloud feature potentially has an interesting link to what may be a stellar overdensity in the maps of ibata et al .
( 2007 ) ; but given the limited coverage of their current survey , it is unclear if this is actually a distinct stellar feature .
the potential continuation of this northern hi emission towards m31 ( braun & thilker 2004 ) appears to be due to contamination by a cloud that merges with galactic emission in the galfa data .
the galfa data extends up to @xmath28 , and with its sensitivity ( @xmath29 @xmath13 in the 11 km s@xmath4 channels shown in figure [ m33chans ] ) and resolution ( 3.5 and 0.2 km s@xmath4 ) reveals the detailed kinematic and spatial structure of numerous low velocity clouds that continue into the galactic disk .
this link to galactic emission would have been partially obscured in the braun & thilker data with the smoothing and exclusion of examining galactic emission .
the finding of halo clouds around m33 is somewhat surprising given m33 is only @xmath301/10 the mass of the milky way ( @xmath31 m@xmath1 ; corbelli 2003 ) .
models do not generally expect a galaxy of this size to have a large amount of mass locked up in its halo at @xmath11 ( maller & bullock 2004 ) .
the clouds in the halo of m33 may represent the cold accretion mode that is expected to dominate for galaxies of this mass ( keres et al . 2005 ) , or the structures may represent the gradual destruction of m33 by m31 which is now @xmath32 kpc away .
the recent proper motion measurements of m33 by brunthaler et al .
( 2005 ) have been used by several authors to put constraints on the motion of m31 and its impact on m33 .
loeb et al .
( 2005 ) rule out a large region in parameter space because of the lack of disruption evident in the stellar component of m33 .
van der marel & guhathakurta ( 2007 ) use the motion of m33 , along with the motions of other local group satellites , to calculate the most likely motion of m31 .
the most likely transverse velocity for m31 has m33 on a tightly bound orbit and signatures of disruption are therefore likely .
muratov and gnedin at the university of michigan are completing more detailed orbit calculations and finding similar results .
the hi features discussed here are most likely the beginning of this disruption process , and the disruption is currently only evident in the extended gaseous component , similar to the magellanic clouds .
it is possible m33 will eventually form giant hi features akin to those found in the magellanic system as it approaches perigalacticon .
both of these systems have a similar total hi mass ( @xmath33 m@xmath1 ) and will provide a substantial amount of fuel to the galaxies .
there is growing evidence that gas clouds in galaxy halos are a common phenomenon and a natural part of the galaxy evolution process .
this halo gas is part of accreting satellites in some cases , but some of the clouds are consistent with a condensing cloud model .
the hot material the clouds condense out of at @xmath11 most likely consists of a mixture of materials that has been gradually enriched over time . since the medium is thought to be responsible for much of the stripping of dwarf galaxies ( mayer et al .
2006 ) , and extends out to at least 55 kpc as evident from the observations of clouds in the magellanic stream ( stanimirovic et al .
2002 ; putman et al .
2003 ; sembach et al .
2003 ) , it is unlikely to all originate from feedback from the galaxy itself . in fountain scenarios the hot gas generally rises to only @xmath34 kpc ( booth & theuns 2007 ; de avillez 2000 ) .
one plausible scenario is the hot halo is created with the initial collapse of baryons into the dark matter potential well and is supplemented through time with fountain material and stripped gas from satellite galaxies . in terms of the local group galaxies ,
the three largest galaxies all show evidence for hi halo clouds with a total mass in clouds of @xmath35 m@xmath1 .
this is not a large reservoir of gas in the context of the chemical evolution models , but with a continual new supply through incoming gas - rich satellites and condensing clouds it may be sufficient at @xmath11 . presumably both of these sources were more abundant at higher redshift .
the future may be bright for both the milky way and m31 in the context of future star formation fuel with the accretion of gas from the magellanic system for the milky way and possibly m33 s gas for m31 .
there may also be additional leo t s sitting at the outskirts of the local group that will serve as future galactic and m31 morsels .
the role of the diffuse halo of these galaxies will be further probed with observations and simulations of the head - tail clouds and new results from the cosmic origins spectrograph .
thanks f. fraternali , r. sancisi , and j. ostriker for useful discussions .
many thanks to the turn on galfa survey team , snezana stanimirovic , kevin douglas , steven gibson , carl heiles , and eric korpela for their help with the galfa data presented here . battaglia , g. , fraternali , f. , oosterloo , t. , sancisi , r. 2006 , a&a , 447 , 49 belokurov , v. et al .
2007 , apj , 658 , 337 belokurov , v. et al . 2006 , apj , 642 , l137 blitz , l. & robishaw , t. 2000 , apj , 541 , 675 boomsma , r. , oosterloo , t.a . ,
fraternali , f. , van der hulst , j.m . ,
sancisi , r. 2005 , a&a , 431 , 65 booth , c.m . & theuns , t. 2007 , mnras , 381 , 89 braun , r. & thilker , d. 2004 , a&a , 417 , 421 chiappini , c. , matteucci , f. , & romano , d. 2001 , apj , 554 , 1044 connors , t. , kawata , d. , bailin , j. , tumlinson , j. & gibson , b. 2006 , apj , 646 , l53 corbelli , e. 2003 , mnras , 342 , 199 de avillez , m.a .
2000 , mnras , 315 , 479 fang , t. , mckee , c.f . , canizares , c.r . , & wolfire , m. 2006 , apj , 644 , 174 fenner , y. & gibson , b.k .
2003 , pasa , 20 , 189 fraternali , f. , oosterloo , t. , sancisi , r. , van moorsel , g. 2001 , apj , 562 , 47 fukugita , m. & peebles , p.j.e .
2006 , apj , 639 , 590 grcevich , j. & putman , m.e .
2008 , apj , in prep .
hibbard , j.e .
, van gorkom , j.h . ,
rupen , m.p . ,
schiminovich , d. 2001 , aspc , 240 , 657 ibata , r. et al .
2007 , apj , 671 , 1591 irwin , m.j .
2007 , apj , 656 , l13 kaufmann , t. , mayer , l. , wadsley , j. , stadel , j. , & moore , b. 2006 , mnras , 370 , 1612 keres , d. , katz , n. , weinberg , d. , dave , r. 2005 , mnras , 363 , 2 koposov , s. et al .
2007 , apj , submitted ( arxiv0706.2687 ) kotoneva , e. , flynn , c. , chiappini , c. & matteucci , f. 2002 , mnras , 336 , 879 larson , r.b .
1972 , nature , 236 , 21 leroy , a. , cannon , j. , walter , f. , bolatto , a. & weiss , a. 2007 , apj , 663 , 990 lin , d.n.c .
& murray , s.d . 2000 , apj , 540 , 170 magrini , l. , corbelli , e. & galli , d. 2007 , a&a , 470 , 843 maller , a. & bullock , j. 2004 , apj , 355 , 694 mayer , l. , mastropietro , c. , wadsley , j. , stadel , j. , & moore , b. 2006 , mnras , 369 , 1021 newberg , h.j .
, yanny , b. , cole , n. , beers , t. , re fiorentin , p. , schneider , d. , & wilhelm , r. 2007 , apj , 668 , 221 oort , j.h .
1970 , a&a , 7 , 381 oosterloo , t. , fraternali , f. , sancisi , r. 2007 , aj , 134 , 1019 peek , j.e.g . ,
putman , m.e . ,
mckee , c.f . , heiles , c. & stanimirovic , s. 2007 , apj , 656 , 907 peek , j.e.g . ,
putman , m.e . ,
sommer - larsen , j. 2008 , apj , in press putman , m.e .
2006 , apj , 645 , 1164 putman , m.e . , et al . 2002 ,
aj , 123 , 873 putman , m.e . , staveley - smith , l. , freeman , k.c . ,
gibson , b.k .
, & barnes , d.g .
2003 , apj , 586 , 170 ryder , s.d .
2001 , apj , 555 , 232 ryan - weber , e.v .
et al . 2008 ,
mnras , 384 , 535 sembach , k.r .
2003 , apjs , 146 , 165 simon , j.d . & geha , m. 2007 , apj , 670 , 313 sommer - larsen , j. 2006 , apj , 644 , l1 sommer - larsen , j. , gotz , m. , portinari , l. 2003 , apj , 596 , 47 stanimirovic , s. , dickey , j.m .
, krco , m. , brooks , a.m. 2002 , apj , 576 , 773 stewart , k. , bullock , j. , wechsler , r. , maller , a. , zentner , a. 2007 , apj , submitted ( astro - ph/0711.5027 ) thilker , d. et al .
2004 , apj , 601 , l39 thom , c. , peek , j.e.g . ,
putman , m.e . ,
heiles , c. , peek , k.m.g . , wilhelm , r. 2008 , apj , in press wakker , b.p . &
van woerden , h. 1997 , ara&a , 35 , 217 wang , q.d .
2005 , apj , 635 , 386 westmeier , t. , braun , r. , thilker , d. 2005 , a&a , 436 , 101 westmeier , t. , braun , r. , bruens , c. , kerp , j. , thilker , d. 2007 , newar , 51 , 108 white , s. & frenk , c. 1991 , apj , 379 , 52 white , s. & rees , m. 1978 , mnras , 183 , 341 williams , r.j .
2005 , apj , 631 , 856 willman , b. et al .
2005 , apj , 626 , l85 wong , t. , blitz , l. & bosma , a. 2004 , apj , 605 , 183 worthey , g. , espana , a. , macarthur , l. & courteau , s. 2005 , apj , 631 , 820 zentner , a. , berlind , a. , bullock , j. , kravtsov , a. , & wechsler , r. 2005 , apj , 624 , 505 zucker , d. et al .
2006 , apj , 643 , l103 | halo clouds have been found about the three largest galaxies of the local group and in the halos of nearby spirals .
this suggests they are a relatively generic feature of the galaxy evolution process and a source of fuel for galaxy disks . in this review ,
two main sources of disk star formation fuel , satellite material and clouds condensing from the hot halo medium , are discussed and their contribution to fueling the galaxy quantified .
the origin of the halo gas of m31 and m33 is also discussed . |
powerful classical double - lobed radio galaxies are ( at low redshifts ) hosted almost exclusively by luminous early - type galaxies .
these galaxies are usually elliptical galaxies . given that normal elliptical galaxies usually exhibit undetectable quantities of cool ( neutral ) gas , what fuels the central supermassive black holes in elliptical radio galaxies ?
our @xmath4co survey of a complete sample of 3c galaxies at low redshift ( lim et al .
, this volume ) reveals that the vast majority have 5@xmath5 upper limits in molecular gas masses of 2 - 5 x 10@xmath6 ( for a linewidth of 500 km / s and a galactic co - to - h@xmath7 conversion factor ) . to better understand the molecular gas content of classical radio galaxies
, we are undertaking a survey of nearby early - type powerful radio galaxies .
these galaxies were originally selected for study by verdoes kleijn et al .
( 1999 ) with the hst , and were drawn from a catalog of one - hundred seventy - six radio - loud galaxies constructed by condon & broderick ( 1988 ) .
of the eighteen galaxies where sensitive 12co measurements have now been made , we detected eight ( i.e. , a detection rate of 45% ) .
the galaxies have molecular gas masses in the range 10@xmath8 10@xmath9 .
all the galaxies detected have double - horned 12co line profiles characteristic of a rotating disk or torus , commensurate with the disk - like dust features seen in these galaxies . in the case of 3c 31 , our follow - up imaging observations with the iram pdbi confirm that the molecular gas is distributed in a rotating disk / torus with size similar to its dust disk .
these results , summarized in table 1 , reinforce our argument ( lim et al .
2000 ) that , in the vast majority of cases , the molecular gas in powerful classical radio galaxies most likely originates from a merger between a pre - existing elliptical galaxy and a relatively less massive gas - rich ( disk ) galaxy .
radio activity is triggered only when sufficient gas reaches the nucleus of the elliptical galaxy , thereby explaining the relatively relaxed nature of both the accreted molecular gas and parent elliptical galaxy .
lcccc name & redshift & iras & optical dust & @xmath10 + & ( z ) & & & @xmath11 + ngc 193 & 0.014 & & @xmath12 & @xmath13 + ngc 315 & 0.016 & @xmath12 & @xmath12 & @xmath14 + ngc 383 ( 3c 31 ) & 0.017 & @xmath12 & @xmath12 & @xmath15 + ngc 541 & 0.018 & & @xmath12 & @xmath16 + ugc 1841 ( 3c 66b ) & 0.021 & & & @xmath17 + ngc 2329 & 0.019 & @xmath12 & & @xmath18 + ngc 2892 & 0.023 & & & @xmath19 + ngc 3801 & 0.011 & & @xmath12 & @xmath20 + ngc 3862 ( 3c 264 ) & 0.022 & @xmath12 & @xmath12 & @xmath21 + ngc 4261 ( 3c 270 ) & 0.007 & @xmath12 & @xmath12 & @xmath22 + ngc 4335 & 0.015 & & @xmath12 & @xmath23 + ngc 4374 ( 3c 272.1 ) & 0.003 & @xmath12 & @xmath12 & @xmath24 + ngc 4486 ( 3c 274 ) & 0.007 & @xmath12 & & standing wave ripples + ngc 5127 & 0.016 & & @xmath12 & @xmath25 + ngc 5141 & 0.018 & & @xmath26 & @xmath27 + ngc 7052 & 0.016 & @xmath12 & & @xmath28 + ugc 12064 ( 3c 449 ) & 0.018 & tentative & @xmath12 & @xmath29 + ngc 7626 & 0.011 & & @xmath12 & @xmath30 + | we report a survey for molecular gas in nearby powerful radio galaxies .
eight of the eighteen radio galaxies observed were detected with molecular masses in the range @xmath0@xmath1 , similar to the same survey we performed towards 3c radio galaxies .
the upper limits of molecular gas in the remainder are typically of @xmath2 , indicating that very few radiogalaxies have molecular gas reservoir with more than 10@xmath3 . |
one of the possible solutions to the hierarchy problem , is a class of models in which extra spatial dimensions @xcite give rise to strong gravity due to the fact that in @xmath0 dimensions the fundamental planck mass ( md ) can be as low as few tev .
the most significant consequence of the low - scale gravity , is the possibility of production of the tev - scale black holes in particle colliers , such as large hadron collier ( lhc ) @xcite and the available energy must be well above the fundamental planck scale @xcite . in string theory , in addition to einstein - hilbert term , there are higher curvature corrections @xcite . on the other hand ,
higher dimensional black holes have been used to analyzed strongly coupled finite temperature field theories .
so , extension of einstein gravitational theory into those with higher power of curvature in higher dimensions is natural . in this
regards , the lovelock theory of gravity is a candidate belonging to such class of theories @xcite which had received a lot of attention , in particular , in the brane world scenario , low energy string theory , and black hole thermodynamics . since the thermodynamic property of black hole is essentially a quantum feature of gravity , it is expected to gain some insights into quantum gravity in the field of black hole thermodynamics . in this case , the effective action proposed by lovelock @xcite contains the higher powers of curvature in particular combinations and the only produced field equations are the second order one which consequently arise no ghosts @xcite . in this case , the action is precisely of the form @xmath1 where @xmath2 denotes the corresponding lagrangian . + here ,
we consider the first four terms of the lovelock gravity which is up to the third order .
the combination of these terms in @xmath3 is the most general lagrangian producing second order field equations .
it is very difficult to find out nontrivial exact analytical solutions of einstein s equation with the higher curvature terms due to the non linearity of the field equations and mostly need to adopt some approximation methods or find numerical solutions .
wheeler has found @xcite exact static spherically symmetric black hole solutions of the gauss - bonnet ( the second order lovelock ) gravity .
also , one can find related solution of the einstein - maxwell - gauss - bonnet and einstein - born - infeld - gauss - bonnet models in refs @xcite .
the thermodynamics of the charged static spherically black hole solutions has been considered in @xcite and of charged solutions in @xcite . in this paper
we want to find new static solutions of third order lovelock gravity by using algebraic equation ( see eq.(9 ) ) which are asymptotically flat and contain second and third order coupling constant of lovelock gravitational theory and investigate their thermodynamics . in this case
, we will investigate the effect of the coupling constants on hawking temperature and bekenstein - hawking entropy as well , specially in 7 and 8 dimensional spacetime .
+ the outline of the paper is as follows . in section 2 ,
we briefly review lovelock theory and we obtain general algebraic equation . in section 3 ,
we present the static solution of the third order lovelock theory and investigate the thermodynamic parameters for seven dimensional spacetime .
we obtain the solution for eight dimensional spacetime and investigate its thermodynamical properties in section 4 .
the final section 5 is devoted to conclusion and discussion .
one of the natural generalization of einstein theory in higher dimensional spacetime is lovelock gravity @xcite which maintains the properties of einstein equation . thus , for eq .
( 1 ) , the effective action , we consider third order lovelock black hole with a cosmological constant term , where the corresponding lagrangian is given by @xmath4 where @xmath5 is an arbitrary constant and @xmath6 is the euler density of @xmath7-dimensional manifolds which is defined by @xmath8 in eq . ( 3 ) , @xmath9 and @xmath10 are the generalized totally antisymmetric kronecker delta and the reimann tensor in @xmath11 dimension respectively .
we would like to draw attention that in @xmath12 dimensions , all terms for which @xmath13 are identically equal to zero , and the term @xmath14 is a topological term .
so , the only term for which @xmath15 are contributing to the field equations .
hereafter , we set @xmath16 , @xmath17 , @xmath18 , @xmath19 , @xmath20 ( for @xmath21 ) , and @xmath22 .
so , the second and third order of the lagrangian terms are given by @xmath23 and @xmath24 one can then derive the lovelock equation with respect to the metric as follows @xmath25 where @xmath26 is the einstein tensor and @xmath27 @xmath28 @xmath29 are the second and the third order of lovelock tensors respectively .
+ it was shown @xcite that there exist static exact black hole solutions of the equation ( 6 ) . in this case
, the line element takes the following form @xmath30 where @xmath31 is the metric of @xmath32 dimensional constant curvature space . in order to determine the function @xmath33 by eq .
( 6 ) , it is convenient to define a new variable @xmath34 as @xmath35 with the constant curvature @xmath36 or @xmath37 . then , after calculating riemann tensor components and substituting into ( 6 ) , one can obtain an algebraic equation as follows @xmath38 } + \psi - \frac{{2\lambda } } { { n\left ( { n + 1 } \right ) } } = \frac{\mu } { { { r^{n + 1}}}}\ ] ] where @xmath39 is cosmological constant , @xmath40 , and @xmath41 is an integration constant related to @xmath42 mass as @xmath43 @xcite , where we used a unit @xmath44 . in the following section we obtain the exact solution for static black holes and its thermodynamical properties for 7 dimensional spacetime .
the most general second order differential equation which presents the solutions of gravity is the gravitational field equation of third order lovelock gravity in seven dimensions . in order to obtain seven dimensional static asymptotically solution of third order of the lovelock gravity , we set @xmath45 and @xmath46 and we consider only positive mass black holes @xmath47 . + in the case of @xmath48 , the theory is reduced to einstein - gauss - bonnet theory and its related @xmath33 reads @xmath49 so , for black hole solution ( 7 ) considering ( 10 )
, one can find the hawking temperature as the inverse of the special value of the period of the euclidean time by using @xmath50 as follow @xmath51 therefore , the behavior of the hawking temperature crucially depends on the second and third order coupling constants of the lovelock gravity theory .
the hawking temperature always increases monotonically from @xmath52 at @xmath53 for small horizon radius ( see figure 1 ) .
figure 2 shows the effect of the @xmath54 variations on the temperature . by the way ,
maximum of the temperature decreases by increasing of the second order coupling constant of the lovelocke gravity in 5 dimensional asymptotically flat spacetime .
we see that , as expected , ( 11 ) behaves like the hawking temperature of a gr solution if the black hole is large enough . on the other hand
, temperature tends to zero for small values .
it is easy to verify that the hawking temperature associated to the solution in @xmath55 ( @xmath56 dimension ) is given by @xmath57 which implies that in @xmath56 dimensional case , the hawking temperature is related just to the second order of the lovelock coupling constant same as the @xmath58 dimensional case .
it is worth pointing out that for dimension 6 the functional form of the temperature is substantially different from the case @xmath56 , as it includes an additional term which is actually proportional to @xmath54 . in this case ,
figure 3 shows that the temperature decreases by increasing the @xmath54 . in order to obtain the temperature for @xmath59 dimension , substituting @xmath60 into eq .
( 9 ) , we obtain a simple equation as @xmath61 then , one can easily find @xmath62 and by using eq .
( 8) obtain the function @xmath33 as follows @xmath63 where @xmath64 and solution ( 10 ) has only one branch unlike the solutions in gauss - bonnet gravity which have two branches .
indeed , eq . ( 10 ) with above @xmath54s and @xmath65s has one real solution and two complex solutions .
+ the hawking temperature of the black holes can be obtained by requiring the absence of conical singularity at the event horizon in the euclidean sector of the black hole solution which can be obtained by analytical continuation of the metric .
we continue analytically the metric by @xmath66 which yields the euclidean section .
the regularity at @xmath67 requires that we should identify @xmath68 where @xmath69 and @xmath70 is the inverse hawking temperature of the event horizon given as @xmath71 where @xmath72 clearly here , the temperature is related to the second order coupling constant as the 5 and 6 dimensional case . also , it depends on the third order constant of lovelock as well as second order .
figure 4 shows the @xmath54 effects on the hawking temperature of fixed @xmath65 .
we also show the effect of @xmath65 of fixed @xmath54 .
it is seen that , the hawking temperature increases when the second order constant of the lovelock increases but the hawking temperature decreases when the third order constant of the lovelock increases ( see figure 4 and 5 ) . we have shown that in the absence of the second order constant , @xmath54 , the third order constant of the lovelock , @xmath65 , still decreases the hawking temperature ( figure 6 ) . [ 4 ] [ 5 ] + in general , the bekenstein - hawking entropy of black holes satisfies the so - called area law of entropy which states that the black hole entropy equals one - quarter of the horizon area , @xmath73 @xcite .
however , in higher derivative gravity the area law of entropy is not satisfied mostly @xcite.the entropy per unit volume obeys the law of the entropy of asymptotically flat black holes of pth - order lovelock gravity @xcite @xmath74 where the integration is done on the @xmath75-dimensional space like hypersurface of killing horizon , @xmath76is the kth order lovelock lagrangian of @xmath77 which is the induced metric on it and @xmath78 is the determinant of @xmath77 . by the way , a simple method of finding the entropy is through the use of first law of thermodynamics , @xmath79 @xcite .
the horizon radius s tarts from zero when the mass parameter is nonnegative .
so , by integrating the first law , the entropy reads @xmath80 in this case , one can obtain the entropy for @xmath48 as follows @xmath81 which is related to the @xmath54 as second order constant of the lovelock theory .
figure 7 shows the entropy deviation by variation of the @xmath54 and the entropy decreases when the @xmath54 increases . in the same way
, one can compute the entropy for 6 dimensional asymptotically flat black hole in lovelock back ground as follows @xmath82 where we neglect the terms related to @xmath83 as it is so small . in this case
, it is interesting that increasing of the @xmath54 caused increase of the entropy ( see figure 8) .
if we set @xmath60 by using equations ( 15 ) and ( 16 ) , the entropy of the 7 dimensional spacetime , @xmath84 , reads @xmath85{{54}}{{\left ( { \alpha { \mkern 1mu } \beta { \mkern 1mu } { r^6 } } \right)}^{2/3}}\beta { \mkern 1mu } { r^{10 } } - 27{\mkern 1mu } \sqrt[3]{{54}}{{\left ( { \alpha { \mkern 1mu }
\beta { \mkern 1mu } { r^6 } } \right)}^{2/3}}\alpha { \mkern 1mu } \beta { \mkern 1mu } \mu { \mkern 1mu } { r^4 } } } + o\left ( { { \alpha ^{n \succ 3}},{\beta ^{n \succ 3 } } } \right ) } \right]{\mkern 1mu } { \rm{d}}r}\ ] ] solving ( 21 ) analytically is not straightforward but one can use some numerical methods in order to compute the entropy . however , we can neglect @xmath54 and @xmath65 higher than order of 3 . obviously
, the entropy for 7 dimensional asymptotically flat black hole in lovelock back ground is related to the second and third order of the coupling constant of the lovelock gravity theory .
in this paper , we have calculated the hawking temperature ( equations ( 11 ) , ( 12 ) , ( 15 ) ) and the bekenestein - hawking entropy ( equations ( 19 ) , ( 20 ) , ( 21 ) ) of the asymptotically flat black hole in lovelock background for 5,6 , and 7 dimensional spacetime by using a general algebraic equation ( 9 ) . in this case , the coupling constant of the lovelock gravity theory take into account up to the third order .
it is evident that the thermodynamic parameters , like hawking temperature and the entropy , are related to the second order of the coupling constant of the lovelock for @xmath86 . in this manner , for 7 dimensional case , they are related to the second and third order of the constant of the lovelock .
+ we have found , in case of @xmath48 the maximum of the temperature decreases when the @xmath54 increase as well as the entropy in a fixed mass .
for 6 dimensional case , the temperature decreases with increasing @xmath54 .
we have investigated the effects of the coupling constants of the lovelock theory on the temperature and entropy in @xmath59 as well .
we have calculated the temperature analytically but we found that the entropy equation ( 21 ) needs to be solved numerically . in this case , both the temperature and the entropy are related to the second and third order of the coupling constant of the lovelock gravity theory .
it is evident the hawking temperature increases when the @xmath54 increases but it decreases with increasing @xmath65 ( see figure 4 and 5 ) .
we have shown that variation of the @xmath54 is more effective on the thermodynamics of the black holes rather than @xmath65 .
the paper is supported by universiti kebangsaan malaysia ( grant no .
frgs/2/2013/st02/ukm/02/2 ) and university of malaya ( grant no . ru-023 - 2014 ) . s. giddings and s. thomas , _ phys .
* 65 * ( 2002 ) 056010 .
d. g. boulware , and s. deser , _ phys .
* 55 * ( 1985 ) 2656 .
d. lovelock , _ j. math
* 12 * ( 1971 ) 498 .
c. charmousis,_lect .
notes phys . _
* 769 * ( 2009 ) 299 .
arxiv:0805.0568n[gr - qc ] b. zwiebach , _ phys .
b _ * 156 * ( 1985 ) 315 ; b. zumino , _ phys . rep _
* 137 * ( 1986 ) 109 .
j. t. wheeler , _ nucl .
b _ * 273 * ( 1986 ) 732 .
d. l. wiltshire , _ phys .
b _ * 169 * ( 1986 ) 36 .
d. l. wiltshire , _ phys . rev .
* 38 * ( 1988 ) 2445 .
j. e. lidsey , s. nojiri and s. d. odintsov , _ jhep _ * 06 * ( 2002 ) 026 .
r. c. myers and j. s. simon , _ phys .
* 38 * ( 1988 ) 2434 .
r. g. cai , _ phys .
* 65 * ( 2002 ) 084014 ; r. g. cai , _ phys .
b _ * 582 * ( 2004 ) 237
. j. d. beckenstein , _ phys .
d _ * 7 * ( 1973 ) 2333
. m. lu , and m. b. wise , _ phys .
d _ * 47 * ( 1993 ) r3095 .
t. jacobson and r. c. myers , _ phys .
* 70 * ( 1993 ) 3684 .
r. m. wald , _ phys .
* 48 * ( 1993 ) r3427 . | we study the thermodynamics of the asymptotically flat static black hole in lovelock back ground where the coupling constants of the lovelock theory effects are taken into account .
we consider the effects of the second order of the coupling constant , and third order of the lovelock constant coefficient on the thermodynamics of asymptotically flat static black holes . in this case
the effect of the coupling constants on the thermodynamics of the black hole are discussed for 5 , 6 , and 7 dimensional spacetime .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
unconventional superconductivity in heavy fermion metals often develops in the vicinity of antiferromagnetic ( af ) order@xcite .
its understanding is intimately connected with that of the af quantum critical points ( qcps ) @xcite .
traditional descriptions of heavy fermion quantum criticality are based on those for purely itinerant magnetism , in terms of the fluctuations of the spin - density - wave ( sdw ) order parameter @xcite .
studies in the recent past have emphasized the beyond landau " physics of kondo destruction @xcite . in these studies ,
considerable progress has been made based on the extended dynamical mean field theory ( edmft ) solution of kondo lattice models @xcite .
one of the important questions along this direction concerns the implications of these theoretical studies on the understanding of superconductivity .
the edmft approach builds on the dynamical mean field theory ( dmft ) @xcite , which was developed through the infinite dimensional limit of the hubbard model @xcite .
dmft maps an interacting lattice problem to a single quantum impurity model coupled to a self consistent weiss field .
the latter captures the environment as seen at the single - particle level , and is represented by a fermionic bath .
dmft has made significant contributions to a variety of strongly correlated problems @xcite , and has in particular provided significant new insights on the mott transition .
the edmft approach treats inter - site density - density or spin - spin interactions , leading to a single impurity model coupled to self consistent fermionic and bosonic baths@xcite .
it has been extensively applied to the study of af quantum critical heavy fermion metals @xcite . for a kondo lattice ,
the edmft approach yields a bose - fermi kondo model ( bfkm ) with self consistent bosonic and fermionic baths .
kondo destruction arises from this approach .
this may be seen already at the level of the bfkm in the absence of self consistency .
the coupling of the local moment to the bosonic bath competes with the kondo effect , i.e. the tendency of singlet formation due to the af exchange coupling between the local moment and the fermionic bath .
when the spectrum of the bosonic bath is sufficiently soft , corresponding to the spectral function being sub - ohmic " , this competition gives rise to the destruction of the kondo effect , in a way that is associated with the criticality of the bfkm @xcite .
studying this type of criticality in a variety of quantum impurity models has led to a number of new insights regarding kondo destruction qcps . for the lattice case , through the edmft equations ,
the bosonic spectrum is particularly soft near the af qcp due to the critical slowing down , and one consistent solution is that the kondo destruction occurs at the af qcp .
there is by now extensive experimental evidence for this type of beyond - landau qcp from experiments in heavy fermion metals , both in terms of a unusual scaling of the dynamical spin susceptibility in the quantum critical regime and a sudden jump of the fermi surface across the qcp @xcite .
however , in order to study the important problem of the interplay between this unconventional quantum criticality and superconductivity , a cluster generalization of the edmft is called for . in this manuscript
, we develop such a formalism . in dmft based approaches , incorporating real space correlations beyond a single site
have naturally been done with the development of quantum cluster theories @xcite . in this case , strongly correlated problems can be mapped to a quantum cluster model with self consistent fermionic baths and the interactions within the cluster are treated exactly .
importantly , dynamical cluster theories incorporate non - perturbative corrections to dmft without introducing a non - causal self energy @xcite .
this can be formulated in real space which leads to cluster dmft ( cdmft ) @xcite , or in momentum space which is known as the dynamical cluster approximation ( dca)@xcite ; there are other cluster embedding schemes possible such as the variational cluster approximation ( vca)@xcite . when the weiss fields are neglected these cluster schemes are no longer self consistent and reduce to cluster perturbation theory ( cpt ) , which approximates lattice quantities by expanding about the isolated cluster limit @xcite .
a main advantage of dynamical cluster theories is the ability to account for various types of order not possible within dmft @xcite .
for example , a four site cluster can treat a d - wave superconducting order parameter as well as stripe charge or spin order .
such a pairing mechanism is expected to be appropriate for , _
e.g. _ , the cuprates and heavy fermion materials . in this case
, any superconducting ground state will have cooper pairs formed between sites which can lead to a variety of different pairing symmetries , such as extended s - wave , p - wave or d - wave . in this manuscript
, we present a cluster extended dynamical mean field theory scheme that we dub c - edmft .
we derive the equations by generalizing ref . to the cluster case using a locator expansion about a dressed cluster limit .
we formulate the equations in both real and momentum space .
we introduce magnetic order in the same fashion as edmft in refs .
distinct from dmft , and then generalize this approach to include superconductivity as well .
we also construct the pairing correlation functions induced by magnetic interactions in the normal state within this approach .
lastly , we use the formalism to derive effective impurity models associated with strongly correlated problems of central interest .
we note that cluster generalizations of the edmft have been carried out in various forms in the past @xcite .
where there is overlap , our approach is consistent with these formulations .
we will make the specific comparisons as we go through the derivation of our approach . in short , the c - edmft formalism developed here
brings out two new aspects ( for definiteness , we will describe these with eq .
( [ eqn : ham ] ) in mind ) .
first , it is the inter - site @xmath0 interactions which underlie both the magnetic and superconducting orders .
such orders develop through decoupling the @xmath0 interaction term into the appropriate channels .
second , the approach avoids double - counting the inter - site interaction by suppressing the induced inter - site interactions associated with the polarization of the fermonic bath by the order parameter . as a result
, the @xmath1-dependence of the dynamical spin susceptibility arises through the @xmath0 interaction term , instead of the bare particle - hole bubble at the special-@xmath1 " ( as opposed to the generic-@xmath1 " , see section iii ) .
these two aspects are in the same spirit as discussed for the case of the edmft @xcite . for illustration purposes ,
we consider a one band hubbard model with two body inter - site interactions on a generic lattice . @xmath2 where @xmath3 destroys an electron of spin @xmath4 at site @xmath5 , @xmath6 , and @xmath7 .
the index @xmath8 runs over @xmath9 , where for @xmath10 ( or for @xmath11 ) the operator @xmath12 , is the spin operator , where @xmath13 is the @xmath8-pauli matrix for @xmath11 .
in addition we consider the charge channel with @xmath14 , where the operator @xmath15 denotes the normal ordered density @xmath16 .
we denote nearest neighbors by @xmath17 and only consider nearest neighbor hopping @xmath18 and two body exchange interaction @xmath19 . for @xmath20 ,
the model reduces to the standard hubbard model with an onsite coulomb repulsion of strength @xmath21 .
it is natural to extend these techniques to multi - band models and longer range interactions .
a main focus of this work is a self consistent solution of the single particle greens function @xmath22 as well as the spin and charge susceptibilities @xmath23 . in general , for the single particle greens function , a perturbative expansion about the non - interacting limit yields the dyson equation @xmath24 where @xmath25 is the fourier transform of @xmath18@xmath26 , @xmath27 is the chemical potential , @xmath28 is the single particle self energy , and we denote fermionic matsubara frequencies as @xmath29 .
analogous to the single particle greens function , we introduce a spin and charge self energy @xmath30 which is defined in terms of each susceptibility as @xmath31 where @xmath32 is the fourier transform of @xmath19 and @xmath33 is a bosonic matsubara frequency .
the spin / charge self energies specify how much their corresponding susceptibilities differ from a gaussian model where @xmath34 ( ref . ) . in the following
we will derive a self consistent c - edmft approach to approximate the lattice quantities @xmath28 and @xmath30 and in turn the single particle greens function and spin / charge susceptibilities . the remainder of the paper is organized as follows .
we focus on the c - edmft equations in the absence of any order in section [ sec : ect ] , and those in the presence of magnetic order in section [ sec : mag - order ] .
we then apply the approach to superconducting order and correlations in section [ sec : sup ] .
we use the formalism to derive an effective cluster model in section [ sec : ecm ] .
finally , we outline the relevant solution methods in section [ sec : sol_methods ] , discuss several pertinent points in section [ sec : dis ] , before concluding the paper in section [ sec : con ] .
we begin by dividing the lattice of @xmath35 sites into clusters of size @xmath36 , where each lattice site is now labeled by @xmath37 , where @xmath38 labels the cluster and @xmath39 labels the sites within the cluster ( see figure [ fig : cluster ] ) . in the following
we will use upper case latin letters to denote cluster indices ( i.e. indices within the cluster ) and lower case letters to label each cluster .
this is then fourier transformed to @xmath40 , where @xmath41 is the intra cluster momentum and @xmath42 the inter cluster momentum . with this notation @xmath18 and @xmath19 can be written as @xmath43 , where the bold @xmath44 denotes a matrix in cluster indices .
we then separate @xmath18 and @xmath19 into intra and inter cluster parts @xmath45 where @xmath46 and @xmath47 are the interactions within the cluster , whereas @xmath48 and @xmath49 are the interactions between clusters , note that by construction @xmath50 and @xmath51 vanish . clusters , of size @xmath52 , where @xmath53 is the dimensionality of the cluster .
the vector @xmath38 labels each cluster , while sites within the cluster are labelled by @xmath39 .
interactions are divided into within the cluster @xmath54 and between clusters @xmath55 , ( which is also done for the hopping elements @xmath56 and @xmath57 . )
we have omitted the channel index @xmath8 for clarity .
, width=216 ] we will first derive the equations in real space .
we first focus on a ground state with no broken symmetry , and will then generalize the equations to the case of magnetic order and superconductivity in sections [ sec : mag - order ] and [ sec : sup ] .
we perform a locator expansion in @xmath57 and @xmath55 about the cluster limit @xcite . the isolated cluster single particle greens function and susceptibilities
are defined by @xmath58 and @xmath59 respectively , where @xmath60 is the _ isolated _ cluster hamiltonian at cluster @xmath38 . in the
following we consider problems that have translational invariance between clusters which implies each cluster correlation function is identical and we can drop the label @xmath38 .
we now generalize the effective cumulant expansion of metzner @xcite for the greens function and smith and si @xcite for the susceptibilities from a single site to a cluster , which leads to matrix quantities . along these lines , we introduce the _ effective _ cluster greens function @xmath61 and spin / charge susceptibilities @xmath62 which are defined as the isolated cluster greens function and susceptibility ( in the @xmath8 channel ) with all local decorations that are irreducible by cutting a single @xmath63 and @xmath64 line respectively ( see figure [ fig : dyson ] ) .
( solid square ) , as an expansion in bare cumulants @xmath65 ( empty square ) with all local decorations that are completely irreducible by cutting a single @xmath55 line .
( b ) diagrammatic representation of the spin susceptibility in equation ( [ eqn : c - chi ] ) , expanding about the effective cluster limit .
each term is for a specific two body interaction channel index @xmath8 , ( with the index omitted for clarity).,width=307 ] the effective cluster correlation functions can be regarded as `` dressed '' cluster correlation function , generalizing the dressed atom picture @xcite to the cluster case . retaining this class of diagrams can be formally justified in the large dimensional limit after rescaling @xmath63 and @xmath64 by the square root of the coordination raised to the manhattan distance between clusters while keeping the dimension and number of sites in the cluster fixed@xcite . performing the locator expansion about the effective cluster correlation functions we arrive at the following dyson like equations @xmath66
we note that these are matrix equations and we are using a bold notation to denote matrices in cluster indices .
fourier transforming equations ( [ eqn : c - g ] ) and ( [ eqn : c - chi ] ) to inter - cluster momentum @xmath42 and @xmath67 we arrive at for the greens function @xmath68^{-1}\end{aligned}\ ] ] and for the susceptibility @xmath69^{-1}.\end{aligned}\ ] ] rewriting the dyson equations for @xmath70 and @xmath71 in equations ( [ eqn : g ] ) and ( [ eqn : chi ] ) in real space cluster indices we find the self energy and spin / charge self energies are @xmath42 and @xmath67 independent respectively , and only depend on cluster indices .
we arrive at the following equations for the effective cluster correlation functions @xmath72 where the free isolated cluster greens function is @xmath73 and @xmath74 is the identity matrix in cluster indices . the fact that both self energies are @xmath42 and @xmath67 independent implies they can be calculated by an effective cluster model .
the effective cluster model can be obtained through a generalized cavity method @xcite , by expanding the partition function in terms of @xmath63 and @xmath64 about a particular cluster @xmath75 and effectively integrating out all other degrees of freedom .
this leads to the cluster action @xmath76 we have dropped the cluster label @xmath75 , defined the isolated cluster action as @xmath77 and introduced the effective weiss fields @xmath78 and @xmath79 that are related to lattice quantities by @xmath80 where @xmath81 and @xmath82 are the greens function and spin susceptibility of the lattice with the cluster @xmath75 removed , and we have taken the cluster @xmath75 to be at the origin . generalizing the arguments of ref . to the case of matrix cluster quantities
we can relate @xmath81 and @xmath82 to the full greens function and spin / charge susceptibilities to obtain ( omitting the frequency labels ) @xmath83 with these relations , equations ( [ eqn : c - g ] ) and ( [ eqn : c - chi ] ) , as well as the self consistency conditions @xmath84 the weiss fields are completely determined by the self energies and the local correlation functions where @xmath85 we have used equations ( [ eqn : c - g ] ) and ( [ eqn : c - chi ] ) to eliminate the dependence on the effective cumulants and use the subscript `` @xmath86 '' denote averages calculated with the effective cluster action in equation ( [ eqn : c - action ] ) , which also corresponds to the lattice quantities within the cluster as enforced via the self consistent equations ( [ eqn : self - g ] ) and ( [ eqn : self - chi ] ) .
one can generalize the arguments of refs . to prove that the approach here has manifestly causal self energies for both @xmath87 and @xmath88 .
it is useful to consider a few limiting cases of the above equations .
first , we note that setting @xmath89 and @xmath90 implies the effective cluster correlation functions reduce to the isolated cluster quantities , and the self energies are then completely determined by solving the isolated cluster problem .
therefore , in the absence of weiss fields this approach reduces to cpt for both @xmath91 and @xmath92 and is no longer self consistent .
this clarifies the meaning of keeping all local decorations for @xmath93 and @xmath94 and is necessary to properly introduce the dynamical weiss fields . in the limit of one site in the cluster @xmath95
, the equations reduce to edmft ; this underscores the fact that the cluster theories incorporate spatial fluctuations beyond standard dynamical mean field theories .
lastly , in the limit of large cluster sizes , @xmath96 the theory becomes exact . in this sense , extended dynamical cluster theories interpolates between the edmft and the exact answer as the cluster size is increased .
after self consistency has been reached it is possible to restore translational invariance to the self energies and thereby the correlation functions by interpolating the cluster quantities .
since the self energies are only defined for sites within the cluster ( or cluster momentum ) the interpolation scheme must respect the symmetry of the original lattice .
following ref .
, after the self consistent solution has been reached we interpolate the cluster self energies to obtain the lattice quantities with the estimation @xmath97 we then use these to determine the lattice greens function and spin susceptibility in equations ( [ eqn : g ] ) and ( [ eqn : chi ] ) .
other interpolation schemes are possible as described in ref .
, where each scheme preserves the symmetry of the lattice . our equations without restoring the translational invariance have some similarities with those of ref . , which invoked a different procedure .
namely , a hubbard - stratonovich transformation decouples the inter site two body interaction term , and the hubbard - stratonovich field becomes a self consistent dynamic bosonic weiss field in the cluster limit .
we also note that , in the absence of conduction electrons similar equations for the self consistent bosonic bath appeared in the construction for spin only models in ref . .
the momentum space construction parallels the dca formulation , which restores translation symmetry by giving the cluster periodic boundary conditions@xcite .
this is achieved by modifying the cluster fourier transform to @xmath98_{{\bf x}_i,{\bf x}_j}&\equiv&j^{\alpha}(\tilde{{\bf q}})_{{\bf x}_i,{\bf x}_j}e^{-i\tilde{{\bf q}}\cdot({\bf x}_i-{\bf x}_j ) } \label{eqn : dcaft } \\
& = & \frac{1}{n_c}\sum_{{\bf q}}e^{i{\bf q}\cdot({\bf x}_i-{\bf x}_j)}j^{\alpha}_{{\bf q}+\tilde{{\bf q}}}\end{aligned}\ ] ] in the following we will refer to this as the dca fourier transform .
this leads to periodic boundary conditions in the cluster and a coarse graining of the cluster quantities @xmath99 and @xmath100 and in turn , the inter - cluster quantities become @xmath101 and @xmath102 . applying the dca fourier
transform to equations ( [ eqn : c - g ] ) and ( [ eqn : c - chi ] ) leads to matrix equations that are diagonal in momentum space . here
we specify the equations for the spin susceptibility , with the equations for the greens function being identical in form to those of dca .
we have a dyson like equation @xmath103 with an effective spin cumulant @xmath104 the weiss field is specified by @xmath105 with the self consistent equation @xmath106
lastly , the cluster action is now diagonal in cluster momentum , which leads to @xmath107 where the isolated cluster action [ in equation ( [ eqn : isolated - action ] ) ] is written in cluster momentum @xmath108 $ ] .
these equations are similar to those of ref . , where the edmft approach was adopted through a hubbard - stratonovich transformation that decouples the inter - site interaction terms , and then the dca approach is applied .
in both the real space and momentum space formulations , it is seen that the momentum dependence of the dynamical lattice spin susceptibility reflects the momentum dependence associated with the inter - site interaction @xmath0 .
this feature is similar to what happens in the edmft @xcite , with the advantage that the inter - site interactions that give rise to the momentum dependence of the dynamical susceptibility also appear in the self - consistent dynamical equations .
it is useful to stress that the momentum dependence of the lattice susceptibility in the paramagnetic phase does not reflect that of the bare particle - hole bubble @xcite .
this is to be contrasted with the standard dmft and cluster generalizations , where the bare particle - hole bubble is responsible for the momentum dependence of the lattice susceptibility , which we discuss in detail in the following section .
before we discuss the cluster generalization of symmetry broken phases within edmft , we find it very useful to review the different schemes used to introduce magnetic order in the single site case . in the context of the edmft ,
there are two ways to introduce magnetic order into the system , namely whether or not the magnetic order parameter polarizes the single particle weiss field @xmath109 ( see ref . for details ) .
it has been shown @xcite that allowing @xmath109 to polarize amounts to keeping the particle hole bubble contribution , @xmath110^{-1}-[\chi_{\mathrm{ph},{\mathrm{loc}}}(\omega)]^{-1}$ ] , to the spin susceptibility @xcite ( where the particle hole bubbles are constructed using the full lattice and local greens function respectively obtained with dmft and the brackets @xmath111 $ ] denote a matrix form @xcite ) . within the context of dmft , such a term will exist due to the distinction between `` normal '' and `` special '' @xmath1 s ; _ cf . _
however , due to promoting @xmath0 to the same level as @xmath18 within the edmft , there are no special @xmath1 s allowed , since this would make @xmath112 which would diverge in the large @xmath113 limit .
the absence of any special @xmath1 s implies @xmath110^{-1}=[\chi_{\mathrm{ph},{\mathrm{loc}}}(\omega)]^{-1}$ ] , and the particle hole bubble contribution vanishes @xcite . keeping the particle hole bubble within the edmft amounts to double counting contributions from the spin - spin interaction @xcite . in the following section , we focus on the cluster edmft case .
this is to be contrasted with dca , which parallels dmft and therefore retains the distinction between special and generic @xmath114 s and a similar particle hole contribution to the spin susceptibility , namely @xmath115^{-1}-[\bm{\chi}_{\mathrm{ph},{\mathrm{loc}}}(\omega)]^{-1}$ ] ( here the particle hole bubble is constructed with the full single particle lattice and local cluster greens function respectively , obtained within dca and the bold denotes matrices in cluster momentum ) @xcite . as in edmft ,
the cluster edmft promotes @xmath55 to the same level as @xmath57 .
generalizing the edmft argument @xcite to the cluster case we conclude that there should only be generic @xmath116 s .
this amounts to not allowing the single particle weiss field @xmath117 to polarize ( i.e. is @xmath4 independent ) , and is equivalent to the suppression of the particle hole bubble contribution .
we now consider magnetic order with an ordering wave vector @xmath118 within the channel @xmath119 with @xmath120 .
the cluster chosen must be large enough to accommodate the type of magnetic order under consideration , for example a four site cluster can describe the collinear af order with @xmath121 whereas a two site cluster can only treat either ferro- or af order .
this then implies that the order pattern within each cluster must be the same and therefore @xmath122 .
we will consider the equations in real space and momentum space consecutively .
after separating @xmath123 into inter- and intra - cluster parts we treat the cluster interactions exactly and normal order the interaction between clusters , via @xmath124 .
all of the previous steps apply , but now we perform the locator expansion in the normal ordered interaction between clusters @xmath125 .
this corresponds to adding an additional term to the cluster action @xmath126 and the local magnetic field is determined self consistently from @xmath127m_{y},\ ] ] where @xmath128 and the average is over the cluster action . in momentum space , this approach amounts to adding to the action @xmath129 now the local field is given by @xmath130 m \label{eqn : hloc - q}\ ] ] where @xmath131 and in this case the inter - cluster interaction is @xmath132 where we have coarse grained the cluster interaction as described previously .
it is useful to note that , in the limit of no dynamical weiss field , the mean field equations for the self consistent field @xmath133 reduce to that of cluster weiss mean field theory . here
, the dynamical weiss field renormalizes the static field due to the dynamical interactions mediated by @xmath134 . in the following section
, we use the self consistent equations that incorporate magnetic order to provide an alternative way of deriving the lattice spin susceptibility at an ordering wave vector @xmath135 , which we define as @xmath136 .
this approach will also provide insight into the way a superconducting long range order can be incorporated in c - edmft . in order to treat an ordering wave vector more naturally
, we will consider the momentum space formalism . in order to ease the notation in the following subsection we use a subscript to denote each matsubara frequency , e.g. @xmath137 .
first we observe that if we include a dynamic source field for each matsubara frequency @xmath138 , we must normal order every frequency modes instead of the static mode only , by writing @xmath139 , as a result equation ( [ eqn : dcaordered ] ) will become @xmath140 with @xmath141 \langle s^{\lambda}_{{\bf q}_{\mathrm{or}},n } \rangle $ ] .
we introduce an additional term , @xmath142 , to the lattice action which couples the @xmath143 component of the spin operators at wave vector @xmath144 to a dynamic source field @xmath145 mapping this into the cluster action , and assuming that the weiss fields can not be polarized ( due to the absence of any special @xmath67 s ) , we end up adding _ only _ one extra term @xmath146 to equation ( [ eqn : dcaordered ] ) , namely @xmath147 here @xmath148 , which is defined in the cluster , need to be distinguished from @xmath149 , which is defined on the lattice , since we have @xmath150 and @xmath151 respectively .
the self consistency between the lattice and the cluster quantities implies the following .
@xmath152 here @xmath153 and @xmath154 denote averaging over cluster action of equation ( [ eqn : newsc ] ) and the lattice action with the additional source field term @xmath142 and @xmath146 respectively . to stress that the expectation value @xmath155 is calculated self
consistently we define @xmath156 . differentiating both sides of equation ( [ eqn : self_con_for_s ] ) with respect to @xmath157 at @xmath158 ,
the right - hand side gives the lattice correlation function , @xmath159 whereas the left - hand side gives the cluster correlation function , using @xmath160 and , @xmath161 \chi_{\mathrm{loc , or}}(i\nu_{n } ) \nonumber \label{eqn : pfpm}\end{aligned}\ ] ] where we have defined @xmath162 as the local spin susceptibility in the ordering channel @xmath163 .
we also have @xmath164 because of eq .
( [ eqn : self_con_for_s ] ) and eq .
( [ eqn : dfdh ] ) .
finally , we can solve for @xmath165 by combining eq.([eqn : self_con_for_s ] ) through eq.([eqn : pfpm ] ) @xmath166 we immediately recognize this result as the c - edmft self - consistent equation for the spin susceptibility at the ordering wave vector @xmath135 and spin component z. most importantly , this result indicates that a non - polarized weiss field is consistent with the c - edmft treatment .
we will show in the appendix how the additional particle hole bubble contribution to the expression for @xmath167 , forbidden by the absence of special @xmath116 s , is introduced if we allow the weiss field to be polarized .
these considerations have important implications for the incorporation of superconductivity into the formalism , which we turn to in the following section .
we now apply the approach to the study of superconductivity , with the pairing driven by the spin - spin interaction in equation ( [ eqn : ham ] ) .
we first focus on an af ising interaction with @xmath168 and set the other @xmath123 to zero .
the af interaction @xmath0 favors pairing of electrons with opposite spins ; a parallel consideration can be made for the case of ferromagnetic interactions , which favors pairing between electrons of the same spins .
we then discuss the case of a full af heisenberg interaction , where @xmath169 are the same for @xmath170 .
following the discussion of magnetic order in section [ sec : mag - order ] , in the context of pairing , the absence of special @xmath116 s implies we should not include a contribution from the particle particle bubble in the susceptibility .
this amounts to not allowing the conduction electron band to become `` polarized '' by a finite superconducting order parameter . in the appendix
, we also discuss what happens when the conduction electrons are allowed to be pair - polarized .
conceptually , we would like to keep the strong inter site interactions ( that give rise to the dynamic bosonic bath ) while promoting a single mode in the static pairing channel in order to give it the chance to condense .
we do so , by singling out the static , attractive pairing interaction between the paring operators @xmath171 and @xmath172 [ defined as @xmath173 .
we rewrite the spin - spin interaction as @xmath174 where the repeated indices @xmath175 are summed over . here in the first step we separated out the pairing interaction , and
then further separated out the static mode of the pairing interaction in the second step .
we then introduce a hubbard - stratonovich transformation to decouple the attractive interaction in the pairing channel .
@xmath176\exp\bigg(-\sum_{\langle i , j \rangle,\sigma } \big [ \-\frac { 4\beta|\delta_{i\sigma , j\bar{\sigma}}|^{2}}{j_{ij } } \nonumber \\ & + & \int_{0}^{\beta}d\tau \delta_{i\sigma , j\bar{\sigma}}\hat{\delta}^{\dagger}_{i\sigma j\bar{\sigma}}(\tau ) + \bar{\delta}_{i\sigma , j\bar{\sigma}}\hat{\delta}_{i\sigma j\bar{\sigma}}(\tau ) \big ] \bigg ) .
\label{eqn : hs}\end{aligned}\ ] ] we note that if we choose to decouple the first term of equation ( [ eqn : pair_sep ] ) instead of that of equation ( [ eqn : freq_sep ] ) and then make the approximation that the hubbard - stratonovich field is constant as a function of @xmath177 ( i.e. keeping only the static mode of the hubbard - stratonovich field ) , we would arrive at exactly the rhs of equation ( [ eqn : hs ] ) .
thus we see that by separating out the static mode before rather than after the hubbard - stratonovich decoupling , the non - static modes of the pairing interaction will be absorbed in the second term in equation ( [ eqn : freq_sep ] ) .
this term is interpreted as the remaining spin - spin interaction .
now we take the saddle point approximation of @xmath178 and follow the steps in section [ subsec : ectr ] to carry out a generalized cavity construction . up to additive constants
, we obtain the effective cluster action , @xmath179 the saddle point equation leads to an additional self consistent equation for the superconducting order parameter @xmath180 where the average is taken with respect to the effective cluster model in equation ( [ eqn : c - scaction ] ) . the additional term in the action
, @xmath181 represents all the modifications in the effective action caused by separating the zero frequency pairing interaction .
the exact expression of @xmath181 can be found in the appendix b. we see from the expression there that all terms in @xmath181 are suppressed by factors of @xmath182 ; therefore , at sufficiently low temperatures including the quantum critical regime , @xmath181 can be safely neglected . in the following we only consider the low temperature limit and make the approximation that @xmath183 .
working in this approximation , the self consistence conditions for both @xmath78 and @xmath184 remain the same as in the previous sections .
it is useful to note that this approach can also be formulated in momentum space .
using the dca fourier transform defined in equation ( [ eqn : dcaft ] ) amounts to replacing @xmath0 by @xmath185_{{\bf x}_i,{\bf x}_j}$ ] in the spin channel . in order to treat the pairing channel on the same footing
, we use @xmath186 starting in equation ( [ eqn : freq_sep ] ) .
this then amounts to replacing @xmath187 in equation ( [ eqn - gapsc ] ) with @xmath188_{x_i x_j}$ ] , which is the coarse grained interaction in cluster momentum @xmath189 ) that is fourier transformed back to real space cluster variables ( see ref . )
@xmath188_{x_i x_j } = 1/n_c\sum_{{\bf q}}e^{i{\bf q}\cdot ( { \bf x}_i - { \bf x}_j)}\bar{j}_{\bf q } $ ] .
in addition , this leads to periodic boundary condition on the cluster in real space .
we now consider the case of an af heisenberg spin spin interaction @xmath190 .
the derivation proceeds in parallel with the previous ising case . here ,
we separate out the singlet term with an attractive interaction @xcite @xmath191 we have defined the singlet creation operator between sites @xmath5 and @xmath192 as @xmath193 and its hermitian conjugate @xmath194 . as before , we then introduce a static hubbard - stratonovich field to decouple the singlet interaction @xmath195 , the saddle point equation for @xmath196 now becomes @xmath197 with an effective action , @xmath198 again , we have not allowed the single particle weiss field to become polarized from the finite superconducting order parameter .
the heisenberg isolated cluster action @xmath199 , is equation ( [ eqn : isolated - action ] ) with @xmath200 for @xmath14 and @xmath201 for @xmath10 .
we have also ignored the additional part of the action @xmath202 that is suppressed by a factor of @xmath203 ( see the appendix for a discussion of the ising case ) .
we remark that it is possible to use this formalism to describe states that have both magnetic order and superconductivity by including a finite @xmath204 as described in section [ sec : mag - order ] . in this section
we derive the zero momentum lattice pairing susceptibility @xmath205 defined as @xmath206 where @xmath207 is the number of bonds in the lattice , with @xmath208 being the number of nearest neighbors , and @xmath209 , @xmath210 are the pairing symmetry factors in real and spin space respectively .
we will focus on the case of an ising spin interaction @xmath0 but this can be easily generalized to the case of a heisenberg interaction .
we first project the superconducting gap onto a particular symmetry channel assuming the gap amplitude @xmath211 is uniform across each bond , i.e. @xmath212 ( and complex conjugate @xmath213 ) , where the phase factor in real space is given by @xmath214 and that in spin space is @xmath215 , and they have the property @xmath216 ( see ref . ) .
the procedure is similar to that described in section [ sec : lt ] for magnetism , although here we have to project into a particular symmetry channel ( through @xmath217 ) .
we arrive at the following expression for the zero momentum lattice pairing susceptibility .
@xmath218 where we have defined the effective pairing interaction @xmath219 , and the cluster pairing susceptibility @xmath220 , where @xmath221 where @xmath222 is the number of individual bonds in the cluster .
previous treatments of two particle response functions in various cluster theories demands much more computational effort because it involves the inversion of the bethe - salpeter equation , which is in principle a matrix equation of infinite dimension in the space of three wave vectors and frequencies . in our approach ,
the cluster susceptibilities completely determine the corresponding lattice quantities .
we note that this can also be formulated in momentum space after the symmetry factors are first coarse grained in momentum space @xmath223 ( where @xmath224 is the fourier transform of @xmath225 ) and then fourier transformed to real cluster space @xmath226_{x_i x_j}=1/n_c\sum_{{\bf k}}e^{i{\bf k}\cdot ( { \bf x}_i - { \bf x}_j)}\bar{f}_{\bf k } $ ] . in this case
, the coarse grained symmetry factors no longer have to satisfy @xmath227 , and as a result the effective pairing interaction becomes @xmath228 . in the appendix
, we contrast this expression with the result of allowing the single particle weiss field to acquire anomalous terms . in this case
, we find additional contributions , corresponding to the particle particle bubble which only contributes if special @xmath229 s exist .
the formalism we have discussed so far also applies to a variety of strongly correlated problems aside from the hamiltonian we have been considering in equation ( [ eqn : ham ] ) .
one such example is a two band model , namely the anderson lattice hamiltonian appropriate for the description of heavy fermion materials @xcite .
the model describes a band of conduction electrons hybridized with a band of localized , highly correlated @xmath230-electrons and is defined as @xmath231 as usual , we have explicitly included an ising rkky interaction between the @xmath230-electron spins .
the rkky and kondo interactions compete , and tuning their ratio can lead to a quantum phase transition between a heavy fermi liquid and an antiferromagnet . in certain cases ,
the qcp is of the sdw type , where the heavy quasiparticles remain intact across the transition and undergo a sdw transition
. in other cases , the sdw description fails , and the physics of critical kondo destruction comes into play . focusing on the normal state properties , applying the extended dynamical cluster theory of section [ subsec : ectm ] in momentum space we arrive at the effective cluster action @xmath232 for the case of 2-d af exchange fluctuations , the divergence of the spin susceptibility at the ordering wave vector implies [ through the self consistent equation ( [ eqn : self - chi - q ] ) ] that the local spin susceptibility with cluster momentum @xmath135 is also ( logarithmically ) divergent @xcite .
this leads to a spin weiss field associated with the critical momentum channel @xmath233 that develops a sub - ohmic spectral density @xmath234 .
based on universality , we can regard the effective cluster model in equation ( [ eqn : c - alaction ] ) with a sub - ohmic density of states for the ordered channel , as an effective model that contains both kondo destruction and pairing correlations induced from af exchange interactions .
for the simplest case of @xmath235 , and keeping only the critical degrees of freedom [ i.e. only the weiss field in the ordered channel @xmath233 ] we arrive at a simplified model to study pairing correlations near a kondo destroyed qcp .
this model was proposed and solved in ref . using a combination of continuous time quantum monte carlo and the numerical renormalization group .
it was shown that the cluster pairing susceptibility @xmath237 is enhanced at the kondo destruction qcp .
it will be important to consider the full self consistent solution and determine if local quantum criticality with kondo destruction survives finite size cluster corrections and if so , how large @xmath238 is near the local qcp .
another strongly correlated problem of central interest is the extended hubbard model , which adds to the standard hubbard model an inter - site density - density interaction .
we note in passing that incorporating a spin orbit coupling can lead to topologically non - trivial ground states in the presence of interactions @xcite .
the extended hubbard model is thought to be the appropriate model to describe certain types of organic superconductors , stripe charge order in the cuprates and different types of mott transitions .
in addition , when the inter - site interaction is attractive ( @xmath239 ) , the model may also feature inter - site pairing and , in the presence of a spin - orbit coupling , topological superconductivity @xcite .
it is defined as @xmath240 where we have generalized the tight - binding hopping terms to a spin dependent hopping matrix @xmath241 ; this form is sufficiently general to allow for a spin orbit coupling term and keep @xmath242 that are not necessarily nearest neighbors .
tuning the ratio of @xmath243 can lead to a variety of quantum phase transitions between a fermi liquid , a band insulator , and a mott insulator . whereas tuning the strength of the spin orbit coupling can lead to topological transitions between a band and topological insulator . applying
the formalism of section [ subsec : ectr ] leads to the effective cluster action @xmath244 studying the different types of quantum phase transitions in the effective cluster model alone [ i.e. , without implementing the self - consistency ] can lead to significant new insights regarding behavior of the qcps in the lattice problem .
in other words , the effective cluster model of equation ( [ eqn : c - action - eh ] ) serves as a simplified model that can provide insights into different types of mott transitions , superconductivity and even interacting topological phase transitions .
the self consistent cluster dynamical mean field equations form a set of highly non - linear equations .
their solution requires an accurate and reliable solution of the cluster impurity model , which is iteratively solved .
there are analytical tools and computational methods that are suitable for studying the cluster impurity models . from a computational perspective , solving the cluster impurity models represents the most extensive efforts of solving the self consistent equations . in the presence of a phase transition , the number of iterations necessary to solve the equations can become quite large due to a `` critical slowing down '' . in this case
, it is very useful to use mixing techniques that are well known in the context of density functional theory , in order to reduce the number of iterations that are needed to achieve self consistency @xcite .
the cluster model can be solved using , for example the exact diagonalization ( ed ) , the numerical renormalization group ( nrg ) , the density matrix renormalization group ( dmrg ) and quantum monte carlo methods . including bosonic baths in diagonalization based techniques can be done , but it requires a truncation of the infinite bosonic hilbert space .
in addition , the presence of numerous bosonic baths can make such an approach quite computationally demanding .
however , this does not rule out ed and nrg techniques , provided there is some physical intuition of which bosonic bath is going to drive the system through a quantum phase transition . then retaining this single bosonic bath is necessary to capture the critical universal properties , while the other baths serve to renormalize the effective model parameters .
the recently developed continuous time quantum monte carlo ( ct - qmc ) has the advantage that the bosons are traced out and the algorithm is numerically exact .
these methods have been adapted to treat scalar bosonic baths that interact with the impurities charge @xcite or spin @xcite degrees of freedom .
the case of a vector bosonic bath @xcite has also been studied .
recently , the ct - qmc has been generalized to a two impurity @xcite model in the presence of a single bosonic bath . as the additional bosonic baths that arise in the c - edmft treatment commute with each other , the method in ref .
can naturally be used to solve the c - edmft equations to high accuracy .
generalizations to larger cluster is still possible within such a framework , provided the algorithm does nt suffer from a sign problem arising from the fermionic degrees of freedom . as described in section
v , for the minimal case of the two - impurity c - edmft approach to the periodic anderson model , in the vicinity of an antiferromagnetic qcp , it would be adequate to keep only a single bosonic bath that is coupled to the staggered combination of the local - moment spins in the cluster .
the method of ref . then makes it feasible to study the quantum critical behavior of the periodic anderson model .
in this work we have derived a cluster edmft formulation by a locator expansion about a dressed cluster limit .
an alternative derivation can also be done using a baym - kadanoff functional @xcite . in this case
the generating functional of the grand potential is ( focusing on one of the two particle channels ) @xmath245 & = & \mathrm{tr}\left[\log ( { \bf g})\right]- \mathrm{tr}\left[{\bf g}(\bf{g}_0^{-1}-{\bf g}^{-1})\right ] \nonumber \\ & -&\frac{1}{2}\mathrm{tr}\left[\log ( \bm{\chi})\right]+\mathrm{tr}\left[\bm{\chi}(\bm{\chi}_0^{-1}-\bm{\chi}^{-1})\right ] \nonumber \\ & + & \phi[{\bf g},\bm{\chi}],\end{aligned}\ ] ] with the stationary conditions @xmath246 .
the self energies are then given by @xmath247/\delta { \bf g}=\bm{\sigma}$ ] , @xmath247/\delta \bm{\chi}={\bf m}$ ] .
within the cluster approximation , the self energies are calculated from an effective cluster model .
therefore , the approximation that the functional @xmath248 is only a functional of @xmath249 and @xmath250 leads to the cluster edmft equations .
therefore , analogous to the edmft approach @xcite , the c - edmft approach is also conserving @xcite .
as we have discussed in the introduction , a main success of edmft compared to dmft has been in its solution to the kondo lattice hamiltonian and the theory of local quantum criticality @xcite .
this rests on treating the rkky interactions between the local moments in a dynamical fashion , through the self - consistent bosonic bath .
the self - consistent solution yields an unconventional qcp with a dynamical spin susceptibility satisfying @xmath251 scaling and anomalous dynamical critical exponent , similar to what has been observed in quantum - critical heavy - fermion metals @xcite . for this result , it is important that the mapped bose - fermi kondo model itself contains a kondo - destruction qcp @xcite .
studies of the two impurity bose - fermi anderson model in ref . have shown that a kondo - destruction qcp persists , with similar scaling properties for the staggered dynamical spin susceptibility .
this raises the prospect that the c - edmft approach discussed here will be able to study the interplay between the unconventional quantum critical normal state and superconductivity . as discussed earlier in the paper ,
such an interplay is important to the understanding of heavy fermion superconductivity .
we note that the dynamical treatment of the rkky interactions differentiates edmft and c - edmft from dmft , which does not treat the rkky interactions beyond a static mean field approximation , as well as the cluster - dmft @xcite , which treats the rkky interactions within the cluster in a dynamical fashion , but does not do so for those interactions outside of a chosen cluster .
in this manuscript we have presented a new cluster extended dynamical mean field approach .
we have developed the equations in both real and momentum space , incorporating magnetic order and superconductivity .
we have also determined the form of the superconducting correlation functions in the normal state .
we have then used the formalism to derive effective cluster models that are relevant to heavy fermion metals and mott - hubbard systems . in particular
, this formulation for unconventional superconductivity is expected to play a central role to the study of quantum critical heavy fermion metals .
_ acknowledgements .
_ we would like to thank kevin ingersent and lili deng for useful discussions .
this work has been supported by the nsf grant no .
dmr-1309531 , the robert a. welch foundation grant no .
c-1411 ( j.h.p . , a.c . &
q.s . ) , the alexander von humboldt foundation ( q.s . ) , and the east - demarco fellowship ( j.h.p . ) .
one of us ( q. s. ) acknowledges the hospitality of the aspen center for physics ( nsf grant no .
1066293 ) , the institute of physics of chinese academy of sciences , and the karlsruhe institute of technology .
in this appendix we derive the expression for @xmath181 , defined in equation ( [ eqn : c - scaction ] ) .
we use the generalized cavity approach to derive the effective action @xmath252 , @xmath253 we have defined @xmath254 as the part of the action that connects cluster @xmath75 to its neighbors .
the expectation value @xmath255 is taken with respect to the action @xmath256 , which is defined as the action with cluster @xmath75 removed , and @xmath257 corresponds to the action of the isolated cluster @xmath75 .
similar to dmft , after rescaling @xmath57 and @xmath55 , in the limit of large coordination , only the @xmath258 and @xmath259 terms contribute . in the remainder of this appendix
, we omit the `` connected '' label and take cluster @xmath75 to be equal to @xmath260 .
we separate @xmath254 into three pieces @xmath261 since there is no interference between the one particle sector and two particle sector in the expansion , we can ignore @xmath262 in the calculation of @xmath181 .
we first note that @xmath263 vanishes , since we are considering a case with no magnetic order .
the expectation value @xmath264 gives the standard expression for the spin weiss field ( see equation ( [ eqn : cav - chi ] ) of the main text ) .
the rest of the terms combined give @xmath181 .
@xmath265 the expression for each term is listed below , @xmath266 @xmath267 @xmath268 the effect of @xmath269 is to modify the one particle weiss field .
the other two terms modify the interactions by generating a general two particle interaction which is nonlocal in time .
in contrast to the standard weiss fields [ as in equation ( [ eqn : c - action ] ) ] , each term contains an additional factor of @xmath270 , and the last term carries a factor of @xmath271 .
this implies each term is suppressed by at least a factor of @xmath182 relative to the standard weiss fields , and @xmath182 can serve as a small parameter for sufficiently low temperatures .
in the zero temperature limit , all three terms vanish and do not affect any of the quantum critical properties .
here we discuss when happens when the single particle weiss fields become polarized by a finite magnetic or superconducting order parameter . in the following ,
we focus on the momentum space formulation .
we show that the additional terms ( particle hole or particle particle bubble contributions ) come from a polarized single particle weiss field .
our considerations here parallel those for edmft discussed in ref . .
we first discuss the case of magnetic order , with an order parameter finite in the @xmath208-direction . allowing the magnetic order parameter to polarize the single particle weiss field we obtain an effective cluster action @xmath272 where now the single particle weiss field @xmath273 is different for different spin components and @xmath204 is given by equation ( [ eqn : hloc - q ] ) . as a result ,
the expression for the lattice spin susceptibility has changed , focusing on the static spin susceptibility we obtain @xmath274 where @xmath275 is @xmath276^{-1}.\end{aligned}\ ] ] in addition , we have defined @xmath277 , which is equivalent to @xmath275 , with @xmath278 replaced with @xmath279 and the derivative is evaluated at @xmath280 .
we now consider allowing the finite superconducting order parameter to `` polarize '' the single particle weiss fields , i.e. introduce anomalous terms .
we take the saddle point approximation of @xmath178 and carry out a generalized cavity construction .
up to additional constants , we obtain the effective cluster action @xmath282 where the ising isolated cluster action @xmath283 is defined in equation ( [ eqn : isolated - action ] ) with @xmath200 for @xmath284 .
we have defined the nambu spinor @xmath285 and we adopt the nambu - gorkov formalism : @xmath286 .
we use the subscript @xmath287 to label a @xmath288 matrix in nambu space where each element is a matrix in cluster indices .
now the single particle weiss field has additional anomalous terms @xmath289 the spin weiss field @xmath290 assumes the same form as equation ( [ eqn : cav - chi ] ) when expressed in terms of the cavity correlation function . as we have discussed in the main text , the additional term in the action , @xmath181 represents all the modifications in the effective action caused by separating the zero frequency pairing interaction .
again , in the following , we only consider the low temperature limit and make the approximation that @xmath183 . to determine the one particle weiss field @xmath291
, we perform a cumulant expansion in inter cluster interactions .
similar to the inter - site interactions , we begin by separating @xmath292 also into intra and inter cluster parts @xmath293 .
this naturally arises from a locator expansion in @xmath294 after we rescale @xmath295 by @xmath296 , which leads to a single particle greens function in nambu space @xmath297^{-1}.\end{aligned}\ ] ] the single particle cumulant is now @xmath298 where @xmath299 is the @xmath208-pauli matrix in nambu space , together with the following definition for the generalized intra cluster hopping matrix @xmath300 and the inter cluster hopping matrix @xmath301 now the generalized one particle weiss field takes the form @xmath302 and the self consistency condition becomes @xmath303 the superconducting order parameter is then determined self consistently from the saddle point value : @xmath304 in principle , for a real space cluster scheme such as cdmft , the translation invariance inside the cluster is broken since the couplings on the boundary are now treated different than those inside the cluster .
thus , the order parameter @xmath305 may in principle take different value on different bonds .
we could obtain an estimate of the pairing amplitude @xmath306 by averaging @xmath305 over each bond ( note that the @xmath307 is a special case where the four sites are indeed equivalent , and such a procedure is unnecessary ) .
after that @xmath295 is then constructed using the translation invariance of the lattice .
two major differences between these self consistent equations and those in cdmft @xcite are the explicit appearance of the order parameter and the fact that the inter - site magnetic interaction is driving the superconducting pairing . considering the case of an ising interaction and following section [ sec : normal - state ] , we focus on the momentum construction , but we write all of the terms in real cluster space to make the connection to our previous discussions more explicit . focusing on the static pairing susceptibility , allowing the single particle weiss field to acquire anomalous terms , we find an additional contribution to the paring susceptibility , @xmath308 here , the additional contribution is @xmath309^{-1}\end{aligned}\ ] ] where again @xmath310 , is @xmath311 , with @xmath211 replaced with @xmath312 , and the derivative is evaluated at @xmath313 , where @xmath312 is a source field that couples to the pairing operators
. we can relate our result to that using the bethe - salpeter equation within dca and conclude that this additional contribution is related to the particle particle bubble contribution which only arises if there are both special and normal @xmath116 s . | the extended dynamical mean field theory has played an important role in the study of quantum phase transitions in heavy fermion systems . in order to incorporate the physics of unconventional superconductivity
, we develop a cluster version of the extended dynamical mean field theory . in this approach ,
we show how magnetic order and superconductivity develop as a result of inter - site spin exchange interactions , and analyze in some detail the form of correlation functions .
we also discuss the methods that can be used to solve the dynamical equations associated with this approach . finally , we consider different settings in which our approach can be applied , including the periodic anderson model for heavy fermion systems . |
pairs of closely stacked quantum dots ( qds ) coupled via inter - band dipole moments @xcite ( double quantum dots , dqds ) or by tunneling resulting from carrier wave function overlap and coulomb correlations @xcite ( quantum dot molecules , qdms ) attract much attention due to the richness of their physical properties which show huge technological promise for nanoelectronics , spintronics and quantum information processing applications .
the unique features of these systems , as compared to individual qds , can be used as the basis for long - time storage of quantum information @xcite and conditional optical control of carrier states @xcite which pave the way to an implementation of a two - qubit quantum gate @xcite .
double dot structures enable also coherent optical spin control and entangling @xcite or may act as sources of entangled photons @xcite .
it has been shown that the exciton spectrum of a qdm can be used to define an excitonic qubit with an extended life time @xcite , that information can be written on the spin state of the dopant mn ion located in one of the dots forming a dqd @xcite , and that a photon emitted by a nearby quantum point contact may induce carrier transfer in a dqd system @xcite .
the richness and complexity of the physical properties of these systems have been manifested in many optical experiments @xcite .
the spacing between the dots forming intentionally manufactured dqds and qdms is typically on the order of nanometers which is two to three orders of magnitude smaller than the wavelength of a resonantly coupled photon . under such conditions , in atomic samples the effect of superradiant emission
is observed @xcite .
a similar effect of coupling to common radiative modes appears also in the emission from qd systems . here also
superradiance - like phenomena occur in the evolution of the exciton occupation and polarization .
apart from the modified evolution for spontaneous emission , collective coupling to the radiation field can lead to other physical effects , one of which is the vacuum induced coherence ( vic ) @xcite .
this effect consists in spontaneous , partial coherent excitation transfer from an initially occupied qd to initially empty one which results in exciton occupation trapping in a decoherence - resistant state . like the superradiance phenomena
, the vic effect can be expected to occur also in qd systems . on the other hand , although natural atoms and qds have many properties in common , the characteristic feature of the latter is the inhomogeneity of transition energy which , for technologically realistic structures , is on the order of milli - electron - volts .
moreover , small spatial separation of the dots leads to coupling between them @xcite . in our previous works we studied the impact of the transition energy mismatch and the coupling between the two emitters on the stability of collective effects @xcite and their role in the linear @xcite and nonlinear optical response of the dqd systems @xcite .
we showed that , in the absence of coupling the between the qds , the appearance of an optically inactive subradiant state and a rapidly decaying superradiant one are extremely sensitive to the fundamental transition energy mismatch and already for a mismatch on the order of micro - electron - volts the collective character of the evolution is replaced by oscillation around the average exponential decay .
the destructive effect of the system inhomogeneity may be , to some extent , overcome by sufficiently strong coupling between the dots which rebuilds collective behavior even in structures with technologically realistic values of energy mismatch @xcite .
we have also pointed out that phonon - induced dynamics can slow down the decay of a superradiant state or speed up the emission from the superradiant one @xcite . based on these previous investigations of superradiance phenomena in qds
, one can expect that the vic effect should , in principle , also be observable in these systems , at least in the presence of sufficiently strong coupling @xcite .
however , to our knowledge , its stability against various inhomogeneities and perturbations typical for the solid state environmnet ( in particular non - parallel orientation of the interband dipoles in the two dots as well as phonon perturbations ) have not been studied . in this paper
, we study the necessary conditions for the vic effect to appear in a system of two vertically stacked semiconductor qds .
we investigate the role of the fundamental transition energy mismatch , coupling between the dots , and phonon - induced kinetics in this process .
we also pay particular attention to the difference of the magnitude of the interband dipole moments as well as to their non - parallel alignment ( due to subband mixing ) .
we show that in spite of all these inhomogeneities that are typical to qd structures ( as opposed to natural atoms ) the vacuum - induced coherence can be almost fully stabilized in realistic pairs of non - identical qds in a certain range of parameters .
in particular , different interband dipole moments for the two dots lead to the appearance of a state which is perfectly immune to radiative decay for a particular choice of the system parameters .
the paper is organized as follows . in sec .
[ sec : system ] , we describe the system under investigation and define its model . in sec .
[ sec : evolution ] , the method for describing the evolution is described .
section [ sec : results ] contains the discussion of the results . concluding remarks
are contained in section [ sec : concl ] .
the investigated system is composed of two vertically stacked semiconductor qds interacting with quantum electromagnetic field and lattice vibrations .
we restrict the discussion to the ground - level transitions with fixed polarization and spin orientations .
we take into consideration only ` spatially direct ' states in which electron - hole pairs reside in the same qd . due to the strong coulomb coupling these states
have a much lower energy than the ` dissociated ' states ( the external electric fields which would change this picture @xcite are not considered in our discussion ) . in this manner
, the dqd or qdm may be described as a four - level system , with the state @xmath0 denoting empty dots , states @xmath1 and @xmath2 representing single - exciton states with electron - hole pairs residing in the lower or higher qd , and the @xmath3 corresponding to the biexciton state , that is , to both qds occupied with an exciton . as it is currently impossible to produce on demand pairs of qds with identical fundamental transition energies , we assume that the exciton transition energies for the two dots are different , @xmath4 where @xmath5 is the average transition energy and @xmath6 is the energy mismatch . as in ref .
, we describe the evolution in the ` rotating basis ' defined by the unitary transformation @xmath7t/\hbar},\ ] ] where @xmath8 and @xmath9 are the standard free photon and phonon hamiltonians , respectively .
the hamiltonian of the system is then @xmath10 the first term describes exciton states in a dqd structure .
@xmath11 where @xmath12 is a biexciton shift due to the interaction of static dipole moments and @xmath13 is the amplitude of the coupling between the single - exciton states of the dots .
the qds are separated by a distance on the order of a few nm which is much smaller than the relevant wave length of the electromagnetic field with which the dots interact .
this allows us to neglect the space dependence of the electromagnetic field and describe the coupling of excitons to the photon modes in the dicke limit @xcite .
the relevant hamiltonian in the dipole and rotating wave approximations is then @xmath14 where @xmath15 are the creation and annihilation operators for the exciton in the @xmath16th qd , @xmath17 is the creation operator of the photon mode with the wave vector @xmath18 , and @xmath19 is a coupling constant for the @xmath16th qd , where @xmath20 is the inter - band dipole moment for the @xmath16th qd , @xmath21 is the unit polarization vector of the photon mode with polarization @xmath22 , @xmath23 is the corresponding frequency , @xmath24 is the vacuum dielectric constant , @xmath25 is the relative dielectric constant of the semiconductor and @xmath26 is the normalization volume .
we investigate only wide - gap semiconductors with fundamental transition energies on the order of 1 ev for which zero - temperature approximation for the electromagnetic modes may be used at any reasonable temperature .
interaction of the carriers confined in the two dots with phonon modes is modeled by the hamiltonian @xmath27 where @xmath28 and @xmath29 are creation and annihilation operators of the phonon mode with the wave vector @xmath30 and @xmath31 are the system reservoir coupling constants for the first and second qd , respectively .
we model the electron and hole wave functions by identical gaussians with extensions @xmath32 in the @xmath33 plane and @xmath34 along the growth direction , @xmath35 for such wave functions and for the deformation potential couplings between the confined carriers and longitudinal phonon modes , the coupling constants have the form@xcite @xmath36,\ ] ] where @xmath37 is the distance between the dots and @xmath38.\ ] ] here @xmath39 are deformation potential constants for electrons / holes , @xmath40 is the crystal density , @xmath41 is the speed of longitudinal sound ( linear phonon dispersion is assumed ) and @xmath42 are momentum components in the @xmath33 plane and along the @xmath43 axis .
analytical formulas describing the evolution of a pair of qds are available for uncoupled systems ( @xmath44 ) interacting only with phonon modes @xcite and , in the markov limit , if only radiative decay is included @xcite . in this paper
we use the previously proposed method @xcite that allows us to deal with the simultaneous action of both the phonon and photon surroundings .
our approach is based on the equation of motion for the reduced density matrix of the carrier subsystem in the interaction picture , @xmath45+\mathcal{l}_{\mathrm{ph}}[\rho].\ ] ] here the first term accounts for the effects induced by the radiative environment , which is described in the markov limit by the lindblad dissipator @xmath46= } \nonumber \\ & & \sum_{\alpha\beta=1}^{2}\gamma_{\alpha\beta } \left[\sigma^{\alpha}_{-}(t)\rho\sigma^{\beta}_{+}(t ) -\frac{1}{2}\left\{\sigma^{\beta}_{+}(t)\sigma^{\alpha}_{-}(t),\rho\right\}\right ] , \label{superlindblad}\end{aligned}\ ] ] where @xmath47^{\dagger}= \exp\left[\frac{ih_{\mathrm{dqd}}t}{\hbar}\right]\sigma^{\alpha}_{- } \exp\left[-\frac{ih_{\mathrm{dqd}}t}{\hbar}\right]\ ] ] and @xmath48 if we assume that the spontaneous decay rates for the two qds are @xmath49 and @xmath50 then it follows directly from eq .
( [ gamma ] ) that @xmath51 where @xmath52 . by redefining the relative phase of the exciton states in the two dots one can assume without any loss of generality that @xmath53 and @xmath54 are real . the second term in eq .
( [ evolution ] ) accounts for the effects due to interaction with phonon surrounding , allowing for non - markovian dynamics .
to describe these effects we use the time - convolutionless equation @xmath55=}\\ & & -\int\limits^{t}_{0}d\tau\mathrm{tr_{ph}}\bigl[h_{\mathrm{dqd - ph}}(t ) , \left[h_{\mathrm{dqd - ph}}(\tau),\rho(t)\otimes\rho_{\mathrm{ph}}\right]\bigr],\end{aligned}\ ] ] where @xmath56h_{\mathrm{dqd - ph}}\exp\left[-\frac{ih_{\mathrm{dqd}}t}{\hbar}\right],\ ] ] @xmath57 is the phonon density matrix at thermal equilibrium , and @xmath58 denotes partial trace with respect to the phonon degrees of freedom .
.parameters used in numerical simulations .
the values correspond to a self - assembled inas / gaas system . [ cols="<,^,^",options="header " , ]
below , we present our results of simulations of the vic process in pairs of qds . in all the analyzed cases , we assume that the system is prepared initially in a localized state @xmath1 . in sec .
refsubsec : identical we explain the effect in a system of identical , uncoupled dots .
then , in sec . [ subsec : dipole ]
, we analyze the role of the relative magnitude and orientation dipole moments and in sec .
[ subsec : dv ] the effect of the energy mismatch and coupling between the dots . in sec .
[ subsec : all ] we analyze the interplay of all the parameters that distiguish qds from natural atoms in the evolution of qds interacting only with radiation reservoir . the phonon impact on the vic
is discussed in sec .
[ subsec : fonony ] . , @xmath59 ns@xmath60 ) two - level systems .
( a ) exciton occupation of the system and occupations of the states @xmath1 ( @xmath61 ) and @xmath2 ( @xmath62 ) .
( b ) the off - diagonal density matrix element @xmath63 . ]
collective coupling of two identical qds ( systems with identical fundamental transition energies and parallel dipole moments of equal magnitudes ) to the quantum electromagnetic vacuum leads to the appearance of a short - living ( bright ) superradiant state , @xmath64 , and an optically inactive ( dark ) subradiant state @xmath65 .
the initial state of the analyzed system is a localized single - exciton state ( @xmath1 or @xmath2 ) which can appear naturally , e.g. , as an effect of incoherent trapping or controlled tunnel injection of carriers in an injection structure similar to that studied in ref . .
such a state of a system of two identical and uncoupled qds may be expressed as an equal combination of the sub- and superradiant states , @xmath66 and @xmath67 , @xmath68a ) .
while this process is taking place , coherence builds up spontaneously in the system ( fig .
[ fig : vic_delta=0]b ) . as a consequence
, the pair of identical qds is trapped in a delocalized and decoherence - resistant state with the exciton occupation number equal @xmath69 and real off - diagonal matrix element equal @xmath70 .
this effect is referred to as vic . )
qds with identical transition energies ( @xmath71 ) but with non - identical values of the spontaneous recombination rates ( @xmath72 ) .
( b - d ) the total exciton occupation ( red solid line ) and the occupations of the individual dots ( green and blue dashed lines ) for the three cases shown in ( a ) .
the value of @xmath73 ns@xmath60 is the same for all the graphs in this figure . ]
vertically stacked semiconductor qds differ slightly in size and shape .
if the system was formed in a self - assembled two - layer process then the upper qd is usually bigger than the lower one @xcite .
the geometry and the structure of the qds is reflected in the carrier wave functions and thus in the interband dipole moments . according to eq .
( [ gamma ] ) , this leads to different spontaneous decay rates for the two qds forming the investigated system .
the dipoles corresponding to the upper and lower qd may differ in amplitude and , if the hole states in the two structures have different light - hole admixtures , also in orientation ( see appendix ) .
the equation of motion describing the evolution of a system of two energetically identical ( @xmath71 ) and uncoupled ( @xmath44 ) qds interacting only with the radiation reservoir is described by eq . with @xmath74 and @xmath75 given by eq .
( [ superlindblad ] ) with @xmath76 .
it is known that in three - level open systems of this kind , non - radiating superposition states occur under certain conditions @xcite . in our case ,
a non - trivial ( different than the ground state @xmath77 ) stationary solution to the open system evolution equation , corresponding to the spontaneously formed , stable , delocalized dark state discussed above , exists only for @xmath78 , i.e. , for parallel dipole moments [ eq . ( [ gamma ] ) ] ( such that @xmath79 up to a phase factor ) . for dipoles that are parallel but have different amplitudes ,
the evolution is similar to the case of identical systems , i.e. , coupling to photon surrounding leads to excitation transfer and occupation trapping . as can be seen in fig .
[ fig : delta=0_rozne_dipole ] , the fraction of trapped exciton occupation depends on the values of the single dot decay rates and stabilizes at the level @xmath80 . as in the case of identical qds ,
the suppression of the exciton decay is due to the existence of a dark state which , for @xmath72 , is not strictly anti - symmetric , @xmath81 due to unequal contribution from the single - exciton states to the dark state ( [ darkg11g22 ] ) the final occupation of the dots is non - symmetric , either , and equals @xmath82 respectively .
different dipole moments allow one to achieve different final situations . in the trapped state ,
the occupation of the initially excited dot may be lower [ figs .
[ fig : delta=0_rozne_dipole](b ) and [ fig : delta=0_rozne_dipole](c ) ] or higher [ fig .
[ fig : delta=0_rozne_dipole](d ) ] than that of the initially empty dot . ) qds with identical transition energies ( @xmath71 ) and different spontaneous recombination rates ( @xmath72 ) for a few values of the angle @xmath83 between the dipole moments .
the values of @xmath49 and @xmath50 shown in fig .
( a ) are valid for both figures . ]
another factor that influences the interband dipole moments is the light - hole admixture . if this admixture is different for the two dots then the dipole moments corresponding to the qds forming the qdm or dqd become non - parallel ( see appendix )
this means that the off - diagonal decay rate @xmath84 , where @xmath83 is the angle between the interband dipole moments of the two dots [ see eq . ] .
consequently , the stationary solution to eq . does not exist and quenching of the final exciton occupation is observed .
the values of @xmath83 are determined by the light hole admixture and typically are on the order of @xmath85 @xcite . however , as can be seen in fig .
[ fig : teta_rozne_dipole ] , even for much larger values of the angle between the dipoles , quenching of the occupation is weak .
although the system looses its coherence at long times , the character of the evolution remains the same as in the case of parallel dipoles on time scales much longer than the nominal exciton lifetime , i.e. , initially excitation transfer between the dots takes place until the occupation rate @xmath86 close to that defined by the decay rates @xmath49 and @xmath50 is reached and , then the impact of of hole subband mixing is manifested as a slow and equal decay of occupations [ fig .
[ fig : teta_rozne_dipole](b ) ] . in spite of rapid technological progress , manufacturing of dqds or qdms with identical fundamental transition energies is still not feasible . as can be seen in fig .
[ fig : vic_d_v](a ) , for the realistic case of non - zero energy mismatch , the effect of trapping the system in an occupied and optically inactive state is destroyed . already for the energy splitting on the order of tens of @xmath87ev quenching of the final occupation
is observed ( the relevant energy scale is the transition energy line width @xmath88ev ) .
qds constituting such an inhomogeneous double dot systems interact with disjoint energy ranges of the electromagnetic field which destroys the collective character of coupling to photon modes .
the energy mismatch of the dots slows the decay of the superradiant state and induces emission from the subradiant one @xcite . due to the lack of a stable exciton state in which the system might be trapped the vic effect in inhomogeneous qdms
is destroyed .
ns@xmath60 ) .
] initially , the character of the evolution of an exciton occupation of the inhomogeneous pairs of qds does not differ considerably from the corresponding case of a pair of identical dots . until @xmath89 ,
the coupling to photon modes maintains its collective character and excitation transfer from the initially occupied qd to the initially empty one takes place .
later , due to the emission from the subradiant as well as from the superradiant state , occupations of both dots decay @xcite .
if the distance between the qds is sufficiently small coupling between the systems ( frster or tunneling ) becomes effective and affects the evolution of carriers confined in the structure .
since sub- and superradiant states are eigenstates of the coupling part of the hamiltonian [ eg.([hamiltoniandqd ] ) ] separated by the energy @xmath90 , sufficiently strong interaction between the dots rebuilds the collective character of the interaction even in structures with technologically realistic energy mismatches on the order of 1 mev @xcite .
this also enables the vic effect in inhomogeneous dqds to be rebuilt . as can be seen in fig .
[ fig : vic_d_v](b ) , for @xmath91 , a transition from a localized initial single - exciton state to a nearly stable state is possible for a system with a technologically achievable energy mismatch .
although full stabilization of the vic effect is impossible for non - identical dots and quenching of the exciton occupation always takes place sufficiently strong coupling between the dots considerably reduces the decay of the trapped state . for a weaker coupling between the dots ,
the exciton occupation decay is faster but still reduced compared to the system of uncoupled qds .
in contrast to identical dots , the two localized states have different contributions from the from sub- and superradiant states .
this results in a different degree of trapping depending on the choice of the initially occupied state @xcite .
mev , @xmath92 mev , @xmath73 ns@xmath60 , and for a few values of @xmath50 .
( b ) the total occupation and the occupations of the individual dots for the last case shown in ( a ) .
( c ) exciton occupation for three sets of parameters stabilizing the occupation trapping .
red solid line : @xmath93 mev , @xmath94 mev , @xmath95 ns@xmath60 , @xmath96 ns@xmath60 . green dashed line : @xmath93 mev , @xmath97 mev , @xmath98 ns@xmath60 , @xmath99 ns@xmath60 .
blue dotted line : @xmath100 mev , @xmath92 mev , @xmath73 ns@xmath60 , @xmath101 ns@xmath60 .
( d ) the total occupation and the occupations of the individual dots corresponding to the red line in ( c ) . ] for technologically realistic dqds and qdms , the fundamental energy mismatch of the two qds forming the system is on the order of milli - electron - volts , the dipole moments differ slightly between the dots and the systems are coupled with one another via frster or tunneling coupling .
the evolution of such systems , for parallel dipole moments , is shown in fig .
[ fig : realistic ] . in fig .
[ fig : realistic](a ) , the dynamics of a system strongly stabilized by a coupling between the dots is presented for a few values of the decay rates . as can be seen , the rate of occupation decay depends on the choice of the two single - dot decay rate parameters .
in such a system , the coherent excitation transfer between the dots takes place on time scales similar to the previously discussed cases .
however , the initial state is now a superposition of non - degenerate system eigenstates , which leads to fast oscillations of the exciton occupation between the two dots [ fig .
[ fig : realistic](b ) ] . in the present general case , the spontaneously formed coherence and the trapped occupation
can again be fully stabilized by an appropriate choice of parameters in a system with parallel dipoles .
the single - exciton eigenstates of the system hamiltonian ( [ hamiltonianh ] ) are @xmath102 where @xmath103 and @xmath104 .
the corresponding eigenenergies will be denoted by @xmath105 .
if the coupling between the single exciton states satisfies the relation @xmath106 ) and ( [ wlasne2 ] ) ] corresponds to the dark state ( [ darkg11g22 ] ) . in such case
, the evolution of a realistic system also leads to occupation trapping in a decoherence - resistant state [ fig .
[ fig : realistic](c ) ] at the occupation @xmath107 .
as can be seen from eq .
( [ vvvv ] ) , non - equal values of the spontaneous decay rate for the two dots open a possibility to stabilize the evolution of the system for wide range of parameters . as for a system of identical dots ,
the occupations of single dots also stabilizes , but again initial fast oscillations due to contribution from two energy eigenstates are observed [ fig .
[ fig : realistic](d ) ] .
mev , @xmath72 ) and coupled ( @xmath92 mev ) qds for a few values of @xmath83 ( a ) and for two values of @xmath50 ( b ) . ] for the non - parallel dipole moments , i.e. , for @xmath108 , the effect of occupation trapping is in principle destroyed but for strongly coupled dots , the decay is very weak . for the interesting time scale on the order of ns
the impact of non - parallel dipoles becomes visible only for the values of @xmath83 that exceed the realistic ones by an order of magnitude @xcite ( fig .
[ fig : realisticqds_teta ] ) .
( color online ) the phonon spectral density for @xmath109 nm.,width=226 ] spontaneous emission from a system of qds is affected by phonon dynamics @xcite .
coupling between the qds and lattice vibrations induces excitation transfer between the two single exciton eigenstates [ eq . and ] . for a pair of identical qds ( @xmath110 )
, these eigenstates exactly coincide with the bright ( superradiant ) and dark ( subradiant ) states @xmath67 and @xmath66 defined in sec .
[ subsec : identical ] .
if the dots are non - identical , there is no perfect correspondence between these two pairs of states but , for non - zero coupling , still one of them is brighter and the other is darker .
as we have shown previously @xcite , the phonon - induced redistribution of single exciton occupations among these two states strongly affects spontaneous emission in a temperature - dependent way . for the system of vertically stacked dots analyzed in this paper , the amplitudes of both tunneling and frster couplings are negative , so that @xmath111 . in this case , the darker eigenstate has a higher energy and will be affected by relaxation even at low temperatures . as we show below , phonon dynamics
indeed breaks the relative stability of the darker state and leads to considerably accelerated decay of the excitonic occupation , except for special parameter choices . while , in general , non - markovian effects may be important in the carrier - phonon dynamics that affects spontaneous emission @xcite , the phonon - related effects to be discussed below can be understood within a markovian picture of transitions between energy eigenstates . the corresponding rate for a transition between the eigenstates @xmath112 and @xmath113 is @xmath114 , where @xmath115 , @xmath116 and the spectral density is defined as @xmath117 the spectral density for the inter - dot spacing @xmath109 nm is plotted in fig .
[ fig : spdens ] .
apart from the usual cutoff at frequencies larger than @xmath118 due to the restriction on the momentum non - conservation @xmath119 , where @xmath120 is the average dot size , it is characterized by the oscillations with the period @xmath121 which result from the double dot structure of the system @xcite . ratio ( corresponding to @xmath122 ) : the comparison of the total exciton occupation in a system coupled to phonons and without phonons .
( a ) @xmath123 mev ; ( b ) short - time section of ( a ) with solid ( red ) line representing the total exciton occupation , dotted ( magenta ) line corresponding to the total occupation without phonons and the dashed ( green ) and dash - dotted ( blue ) lines showing the occupations of the two dots ; ( c ) @xmath124 mev ( corresponding to the 3rd minimum of the spectral density ) ; ( d ) @xmath125 mev ( 5th minimum of the spectral density ) .
] the phonon - related effects on the spontaneous buildup of coherence is shown in fig .
[ fig : vic_ph ] . here
we keep the mixing angle @xmath126 constant ( there is , a constant ratio @xmath127 ) and use the energy splitting @xmath128 as a parameter . in fig .
[ fig : vic_ph](a ) we plot the evolution of the total exciton occupation for @xmath129 mev , where the spectral density is relatively large , corresponding to phonon transition rates on the order of 1 ps@xmath60 .
one can see here that indeed the phonon - induced relaxation suppresses the vacuum - induced coherence , which is manifested by the rapid decay of the excitonic occupation .
the role of phonons in this strong change of the occupation dynamics is clear in fig .
[ fig : vic_ph](b ) , where we show the first 100 ps of the same evolution . a very fast phonon - induced redistribution of occupations of the two dots ( shown in green dashed and blue dash - dotted lines ) takes place on a picosecond time scale after which the radiative effects are completely dominated by the phonon - induced thermalization which , at low temperatures , forces the system to stay in the bright state @xmath130 .
the detrimental effect of carrier - phonon coupling can only be avoided if the phonon - induced occupation dynamics is made slow compared to the spontaneous emission .
this can be achieved by using the oscillating form of the phonon spectral density ( see fig . [
fig : spdens ] ) . as can be seen in fig .
[ fig : vic_ph](c , d ) , if choosing the energy splitting @xmath131 such that it corresponds to the 3rd minimum of the spectral density has little positive effect but for @xmath131 in the 5th minimum , the phonon effects become very weak and the long living tail of the excitonic occupation is restored .
an interesting additional effect can be seen if one compares figs .
[ fig : vic_ph](a ) and [ fig : vic_ph](d ) : if the phonon effects are strong then increasing the temperature slows down the occupation decay .
however , for weak phonon influence the temperature dependence is opposite .
this can be explained by noting that in the presence of fast phonon - induced redistribution of single exciton occupations , the occupation of these two states remains in quasi - equilibrium , which means that the occupation of the higher energy , darker state increases as the temperature grows . on the contrary , if the phonon - induced dynamics is slow the system state is almost unperturbed and close to the spontaneously formed dark superposition .
phonons lead to transitions out of this stable state with intensity growing with temperature .
we have studied the formation of vacuum - induced coherence and the associated long - living trapped excitonic population in a pair of vertically stacked semiconductor qds .
we focused on the features that distinguish qd systems from natural atoms : the mismatch of transition energies , coupling , possibly non - identical dipole moments for the optical transition , and strong interactions with the phonon environment .
we have shown that the vic effect is very sensitive to the inhomogeneity of the qds .
already for the fundamental transition energy mismatch on the order of the emission line width , the exciton occupation is quenched .
however , the destructive effect of the energy inhomogeneity can be strongly reduced by coupling between the dots . while for dots with identical magnitudes of the interband dipole moments full stabilization of the spontaneously formed coherence
can only be achieved in the limit of infinite coupling , in pairs of qds with different interband dipoles it is possible to adjust the energy mismatch , the coupling between the qds , and the dipole moments in such a way that a perfectly stable state forms in the spontaneous emission process and a fraction of the initial excitonic population attains a formally infinite life time .
the non - parallel orientation of the interband dipoles , that may be caused by heavy - light hole mixing leads to negligibly weak effects in realistic structures .
while carrier - phonon coupling typically destroys the vacuum induced coherence on picosecond time scales , it can be overcome by appropriately selecting the energy splitting between the single exciton states .
these results show that the vic effect can be observed in realistic systems with energy splitting on the order of milli - electron - volts provided that the system parameters ( interband dipoles , coupling and energy mismatch ) can be controlled with sufficient flexibility .
this seems to be possible by appropriately designing the system on the manufacturing stage and then employing the dependence of various parameters on external fields ( e.g , via stark effect or modification of electron - hole wave function overlap ) .
let us note , finally , that the major experimentally detectable consequence of the appearance of vacuum - induced coherence in the double dot system is the long - living tail in the exciton occupation ( or in luminescence intensity ) .
since this effect is of considerable amplitude and evolves on long , nanosecond time scales it should be relatively easily detectable with time - resolved luminescence or pump probe spectroscopy .
a. s. acknowledges support within a `` moda kadra 2015 plus '' project , co - financed by the polish ministry of science and higher education ( mnisw ) and the european union within the european social fund , and within a scholarship for outstanding young scientists granted by the polish mnisw .
in this appendix , we discuss the effect of light hole admixture on the non - parallel alignment of the interband dipoles for ( nominally ) heavy hole transitions . within the 4-band luttinger model @xcite , the electron and hole wave functions are @xmath132 where @xmath133 refers to a higher or lower qd
, respectively , @xmath134 is a bloch function for the valence subband @xmath22 ( heavy holes : @xmath135 , light holes : @xmath136 ) , @xmath137 is the space coordinate , @xmath138 denotes spin , @xmath139 is an electron bloch function with a fixed spin @xmath140 ( for definiteness ) and @xmath141 refer to electron and hole envelope functions , respectively . the matrix element of the interband dipole moment for the @xmath16th qd is @xmath142 to calculate the above integral we sum over unit cells ( labeled by @xmath143 ) and integrate over one unit cell ( @xmath144 labels the position within the cell ) .
the envelope functions vary slowly which allows us to assume that they are constant over one unit cell , @xmath145 .
the bloch functions are periodic , @xmath146 , and orthogonal for different bands . as a result , we can write @xcite @xmath147 here @xmath148 is the bulk interband dipole moment , @xmath149 @xmath26 is the unit cell volume , @xmath150 is the envelop function overlap integral , and we have replaced the summation over unit cells with integration over @xmath143 .
we are investigating bright heavy hole excitons , hence @xmath151 and the other coefficients @xmath152 are much smaller .
the non - vanishing bulk dipole moment matrix elements involving the spin-@xmath153 electron state are @xcite @xmath154 because the wave functions differ for the two qds , the values of the heavy hole overlap integrals @xmath155 may vary . since , to the leading order , @xmath156 the difference of the overlap integrals leads to different magnitudes of the dipole moments @xmath157 , @xmath158 .
moreover , one has @xmath159 hence , @xmath160 thus , if the light hole admixture is different for the two dots , @xmath161 then @xmath162 , that is , the dipoles are non - parallel .
as the subband mixing is typically small in self - assembled structures the angle @xmath83 between the dipole moments is small .
therefore , one can write @xmath163 where @xmath164 is an irrelevant phase . comparing eqs .
( [ dipole1 ] ) and ( [ dipole2 ] ) one gets @xmath165 | we present a theoretical study of vacuum - induced coherence in a pair of vertically stacked semiconductor quantum dots .
the process consists in a coherent excitation transfer from a single - exciton state localized in one dot to a delocalized state in which the exciton occupation gets trapped .
we study the influence of the factors characteristic of quantum dot systems ( as opposed to natural atoms ) : energy mismatch , coupling between the single exciton states localized in different dots , different and non - parallel dipoles due to subband mixing , as well as coupling to phonons .
we show that the destructive effect of the energy mismatch can be overcome by an appropriate interplay of the dipole moments and coupling between the dots which allows one to observe the trapping effect even in a structure with technologically realistic energy splitting on the order of milli - electron - volts .
we also analyze the impact of phonon dynamics on the occupation trapping and show that phonon effects are suppressed in a certain range of system parameters .
this analysis shows that the vacuum induced coherence effect and the associated long - living trapped excitonic population can be achieved in quantum dots . |
this paper gives an overview of the neutrino physics possibilities at a future muon storage ring , which can be either a muon collider ring or a ring dedicated to neutrino physics that uses muon collider technology to store large muon currents .
it summarizes a previous more detailed description of these topics by this author @xcite . after a general characterization of the neutrino beam and its interactions ,
some crude quantitative estimates are given for the physics performance of a muon ring neutrino experiment ( murine ) consisting of a high rate , high performance neutrino detector at a 250 gev muon collider storage ring .
neutrinos are emitted from the decay of muons in the collider ring : @xmath0 the thin pencil beams of neutrinos for experiments will be produced from long straight sections in either the collider ring or a ring dedicated to neutrino physics . from relativistic kinematics , the forward hemisphere in the muon rest frame will be boosted , in the lab frame , into a narrow cone with a characteristic opening half - angle , @xmath1 , given in obvious notation by @xmath2 the large muon currents and tight collimation of the neutrinos results in extremely intense beams intense enough even to constitute a potential off - site radiation hazard @xcite .
for the example of 250 gev muons , the neutrino beam will have an opening half - angle of approximately 0.4 mrad .
the final focus regions around collider experiments are important exceptions to equation [ eq : thetanu ] since the muon beam itself will have an angular divergence in these regions that is large enough to significantly spread out the neutrino beam . for tev - scale neutrinos ,
the neutrino cross - section is approximately proportional to the neutrino energy , @xmath3 .
the charged current ( cc ) and neutral current ( nc ) interaction cross sections for neutrinos and antineutrinos have numerical values of @xcite : @xmath4 these cross sections are easily converted into approximate experimental event rates for the example used in reference @xcite of a 250 + 250 gev collider with a 200 meter straight section . for a general purpose detector subtending the boosted forward hemisphere of the neutrino beam : @xmath5 , \label{eq : genevents}\ ] ] where @xmath6 is the detector length . for a long baseline detector in the center of the neutrino beam : @xmath7}{(l[km])^2 } } , \label{eq : longevents}\ ] ] where @xmath8 is the detector mass and @xmath9 the distance from the neutrino source .
these event rates are several orders of magnitude higher than in today s neutrino beams from accelerators .
figure [ detector_fig ] is an example of the sort of high rate general purpose neutrino detector that would be well matched to the intense neutrino beams .
note the contrast with the kilotonne - scale calorimetric targets used in today s high rate neutrino experiments .
the neutrino target is a 1 meter long stack of ccd tracking planes with a radius of 10 cm chosen to match the beam radius at approximately 200 meters from production for a 250 gev muon beam .
it contains 750 planes of 300 micron thick silicon ccd s , corresponding to a mass per unit area of approximately 50 @xmath10 , about 2.5 radiation lengths and 0.5 interaction lengths .
equation [ eq : genevents ] predicts a very healthy @xmath11 cc interactions per year for this target . besides providing the mass for neutrino interactions , the tracking target allows precise reconstruction of the event topologies from charged tracks , including event - by - event vertex tagging of those events containing charm or beauty hadrons or tau leptons .
given the favorable vertexing geometry and the few - micron typical ccd hit resolutions , it is reasonable to expect almost 100 percent efficiency for b tagging , perhaps 70 to 90 percent efficiency for charm tagging and excellent discrimination between b and c decays . the target in figure [ detector_fig ] is surrounded by a time projection chamber ( tpc ) tracker in a vertical dipole magnetic field .
the characteristic de / dx signatures from the tracks would identify each charged particle . further particle i d is provided by the cherenkov photons that are produced in the tpc gas then reflected by a spherical mirror at the downstream end of the tracker and focused onto a read - out plane at the upstream end of the target .
the mirror is backed by electromagnetic and hadronic calorimeters and , lastly , by iron - core toroidal magnets for muon i d .
the dominant interaction of tev - scale neutrinos is deep inelastic scattering ( dis ) off nucleons ( i.e. protons and neutrons ) .
there are 2 types of dis : neutral current ( nc ) and charged current ( cc ) scattering . in neutral current ( nc ) scattering , the neutrino is deflected by a nucleon ( @xmath12 ) and loses energy with the production of several hadrons ( @xmath13 ) : @xmath14 this comprises about 25 percent of the total cross section and is interpreted as elastic scattering off one of the many quarks inside the nucleon through the exchange of a virtual neutral z boson : @xmath15 charged current ( cc ) scattering is similar to nc scattering except that the neutrino turns into its corresponding charged lepton : @xmath16 where @xmath6 is an electron / muon for electron / muon neutrinos . at the more fundamental quark level
a charged w boson is exchanged with a quark ( @xmath17 ) , which is turned into another quark species ( @xmath18 ) whose charge differs by one unit .
@xmath19 the relativistically invariant quantities that are routinely extracted in dis experiments are 1 ) feynman @xmath20 , the fraction of the nucleon momentum carried by the struck quark , 2 ) the inelasticity , @xmath21 , which is related to the scattering angle of the neutrino in the neutrino - quark com frame , and 3 ) the momentum - transfer - squared , @xmath22 .
murine s will have the further capability of reconstructing the hadronic 4-vector , resulting in a much better characterization of each interaction .
the final state quark always `` hadronizes '' at the nuclear distance scale , combining with quark - antiquark pairs to produce the several hadrons seen in the detector .
final state c and b quarks can be identified by vertex tagging of the decaying charm or beauty hadrons that contain them , and some statistically based flavor tagging will also be available for u , d or s final state quarks , using so - called `` leading particle effect '' @xcite .
neutrino interactions are interesting both in their own right and as probes of the quark content of nucleons , so a murine has wide - ranging potential to make advances in many areas of elementary particle physics .
this section gives an overview for measurements involving the ckm quark mixing matrix , nucleon structure and qcd , electroweak measurements , neutrino oscillations and , finally , studies of charmed hadrons .
there is considerable theoretical interest in the mixture of final state quarks produced in cc interactions .
the struck quark can be converted into any of the three final state quarks that differ by one unit of charge : a down ( d ) , strange ( s ) , or bottom(b ) quark can be converted into an up ( u ) , charmed ( c ) , or top ( t ) quark and vice versa . in practice ,
production of the heavy top quark is kinematically forbidden at these energies and the production of other quark flavors is influenced by their mass . beyond this , the standard model of elementary particle physics ( sm ) predicts the probability for the interaction to be proportional to the absolute square of the appropriate element in the so - called cabbibo - kobayashi - maskawa ( ckm ) quark mixing matrix , a unitary matrix with 4 free parameters whose values are not predicted by the sm .
improved measurements involving the ckm matrix will test the sm hypothesis .
it is of particular interest that one of the 4 parameters is a complex phase that is postulated as an explanation for cp violation the intriguing experimental phenomenon that particles may have tiny deviations from the properties that mirror those of their antiparticles .
the experimentally determined values for the 9 mixing probabilities are given in table [ ckm_table ] @xcite , along with their current percentage uncertainties and speculative projections @xcite for how the uncertainties could be reduced by a murine . from the large improvements in 4 of the 9 uncertainties
it is clear that a murine has potential for tremendous improvements in measuring the quark mixing matrix , and more detailed studies are clearly desirable . .
quark mixing probabilities . threshold suppression due to quark masses has been neglected . in practice
, this will reduce the mixing probabilities to the heavier c and b quarks to below the values given in the table and will prevent any mixing to the top quark . the second row for each quark
gives current percentage uncertainties in quark mixing probabilities and speculative projections of the uncertainties after analyses from a murine .
the two uncertainties in brackets have not been measured directly from tree level processes .
the uncertainties assume that no unitarity constraints have been used .
[ cols="^ , < , < , < " , ] another major motivation for murine s is the potential for greatly improved measurements of nucleon structure functions the momentum distributions of quarks inside the nucleon .
this provides @xcite important tests of quantum chromodynamics ( qcd ) the theory of the strong interaction that is widely accepted for its elegance and simplicity but which has not been experimentally verified at the level of the electroweak theory .
a murine might well be the best single experiment of any sort for the examination of perturbative qcd @xcite .
neutrino physics has also had an important historical role in measuring the electroweak mixing angle , which is simply related to the mass ratio of the w and z intermediate vector bosons : @xmath23 now that @xmath24 has been precisely measured at lep , measurements of @xmath25 in neutrino physics can be directly converted to predictions for the w mass .
the comparison of this prediction with direct @xmath26 measurements in collider experiments constitutes a precise prediction of the sm and a sensitive test for exotic physics modifications to the sm @xcite .
reference @xcite estimates that the predicted uncertainty in @xmath26 from a murine analysis might be of order 10 mev , which improves by more than an order of magnitude on today s neutrino experiments @xcite and is comparable with the projected best direct measurements from future collider experiments .
a neutrino property that is currently drawing much interest is the question of whether neutrinos have a non - zero mass .
if they do then it is possible for the 3 neutrino flavors to mix , perhaps producing neutrino oscillations that can be observed using a neutrino beam .
the probability for an oscillation between two of the flavors is given by@xcite : @xmath27.l[km ] } { e_\nu [ gev ] } \right ) , \label{eq : oscprob}\ ] ] where the first term gives the mixing strength and the second term gives the distance dependence .
reference @xcite obtains the following order - of - magnitude mass limit for an assumed long - baseline detector with reasonable parameters and with full mixing : @xmath28 independent of the distance to the detector .
similarly , a mixing probability sensitivity for @xmath29 events in a short - baseline detector is found to be as low as @xmath30 for the most favorable value of @xmath31 .
both of these estimates apply generically to all 3 possible mixings between 2 flavors : @xmath32 , @xmath33 and @xmath34 .
( see also reference @xcite for another discussion of neutrino oscillations at a murine . )
the @xmath31 estimate is more than an order of magnitude better than any proposed accelerator or reactor experiments for @xmath34 and @xmath33 , and is competitive with the best such proposed experiments for @xmath32 .
the estimated value for @xmath35 is even more impressive
orders of magnitude better than in any other current or proposed experiment for each of the three possible oscillation . as an interesting final topic
, murine s should be rather impressive factories for the study of charm with a clean , well reconstructed sample of several times @xmath36 charmed hadrons produced in @xmath29 neutrino interactions .
there are several interesting physics motivations for charm studies at a murine @xcite . as an example
, particle - antiparticle mixing has yet to be observed in the charm sector @xcite , and it is quite plausible @xcite that a murine would provide the first observation of @xmath37 mixing .
the intense neutrino beams at muon collider complexes should usher in an exciting new era of neutrino physics experiments , with great advances expected in both traditional and new areas of neutrino physics .
king , _ neutrino physics at a muon collider _ , proc .
workshop on physics at the first muon collider and front end of a muon collider , fermilab , november 6 - 9 , 1997 .
king , _ assessment of the prospects for muon colliders _
, paper submitted in partial fulfillment of requirements for ph.d .
, columbia university , new york ( 1994 ) ; b.j .
king , _ a characterization of the neutrino - induced radiation hazard at tev - scale muon colliders _ , bnl center for accelerator physics internal report 162-muon-97r , to be submitted for publication .
see , for example , chris quigg , _ neutrino interaction cross sections _
, fermilab - conf-97/158-t .
values extracted from andrzej j. buras , _ ckm matrix : present and future _ , tum - hep-299/97 .
janet m. conrad , michael h. shaevitz and tim bolton , _ precision measurements with high energy neutrino beams _
, hep - ex/9707015 , submitted to rev . mod
( 1997 ) k.s .
et al . _ ( ccfr / nutev collaboration ) _ a precision measurement of electroweak parameters in neutrino - nucleon scattering _ , fnal - pub-97/001-e .
b.j . king , columbia university ph.d .
thesis , 1994 ; nevis report : nevis-283 , cu-390 , nevis preprint r-1500 ( 1994 ) .
barnett et al . , physical review d54 , 1 ( 1996 ) and 1997 off - year partial update for the 1998 edition available on the pdg www pages ( url : http://pdg.lbl.gov/ ) . s. geer , _ the physics potential of neutrino beams from muon storage rings _ , proc . workshop on physics at the first muon collider and front end of a muon collider , fermilab , november 6 - 9 , 1997 .
i.i bigi , _ open questions in charm decays deserving an answer _ , cern - th.7370/94 , und - hep-94-big08 ( 1994 ) .
i.i bigi , _ the expected , the promised and the conceivable - on cp violation in beauty and charm decays .
_ , und - hep-94-big11 ( 1994 ) .
tiehui ( ted ) liu , _ the d0-dobar mixing search current status and future prospects _ , hutp-94/e021 ( 1994 ) .
gustavo burdman , _ charm mixing and cp violation in the standard model _ , fermilab - conf-94/200 ( 1994 ) . | an overview is given of the neutrino physics potential of future muon storage rings that use muon collider technology to produce , accelerate and store large currents of muons . |
let @xmath0 be a closed surface of genus @xmath1 and let @xmath2 be the riemannian metric on @xmath0 . denote by @xmath3 the laplace - beltrami operator on @xmath0 , and by @xmath4 the smallest positive eigenvalue of the laplacian . let the area @xmath5 be fixed .
how large can @xmath4 be on such a surface ?
sharp bounds for the first eigenvalue are known only for the sphere ( @xcite , see also @xcite ) , the projective plane ( @xcite ) , the torus ( @xcite , @xcite ) , and the klein bottle ( @xcite , @xcite ) .
the present paper is concerned with the surface of genus @xmath6 .
let @xmath0 be orientable and let @xmath7 be a non - constant holomorphic map ( or , conformal branched covering ) of degree @xmath8 .
it was proved in @xcite that @xmath9 any riemann surface of genus @xmath1 can be represented as a branched cover over @xmath10 of degree @xmath11 $ ] , where @xmath12 $ ] denotes the integer part ( see @xcite , @xcite ) .
therefore , @xmath13.\ ] ] in general , is not sharp , for example for @xmath14 ( @xcite ) .
let @xmath15 be a surface of genus @xmath16
. then implies @xmath17 the aim of this paper is to show , using a mixture of analytic and numerical tools , that ( [ sharp ] ) is sharp .
main results of this paper were announced ( without proofs ) in @xcite .
let @xmath18 be a branched covering of degree @xmath19 .
the riemann - hurwitz formula ( see @xcite ) implies that this cover is ramified at @xmath20 points .
we choose these points to be the intersections of the round sphere @xmath21 centered at the origin with the coordinate axes in @xmath22 .
the surface @xmath23 can be realized as @xmath24 this surface has the conformal structure of the bolza surface .
it has an octahedral group of holomorphic automorphisms and its symmetry group is the largest among surfaces of genus two @xcite .
interestingly enough , the bolza surface appears in some other extremal problems , in particular for systoles ( see @xcite ) . to simplify calculations
it is convenient to rotate the equatorial plane by @xmath25 .
the equation of @xmath23 becomes @xmath26 the projection @xmath27 is defined by @xmath28 .
the set of ramification points in the complex @xmath29 plane is @xmath30 .
the spherical and complex models are related by the stereographic projection ; the induced metric in the complex plane ( which we assume coincides with the equatorial plane of @xmath21 ) is @xmath31 let @xmath32 be the metric on @xmath23 which is the pullback of the round metric on @xmath21 .
one can see that the metric @xmath32 has conical singularities at the points of ramification .
it has curvature @xmath33 everywhere except the branching points . because of the presence of singularities we have to specify what we mean by the first positive eigenvalue of the laplacian on @xmath34 .
we set @xmath35 where the scalar product @xmath36 and the norm @xmath37 are taken in the space @xmath38 .
the sobolev space @xmath39 of functions supported away from the singularities is obtained by the closure of @xmath40 with respect to the norm @xmath41 .
we start with the following [ conj1 ] the equality in is attained for the metric @xmath32 on @xmath23 , i.e. @xmath42 since @xmath34 is a double cover of the standard @xmath21 , we have @xmath43 therefore , in order to prove conjecture [ conj1 ] it suffices to show that @xmath44 unfortunately , we are unable to prove , and therefore establish conjecture [ conj1 ] .
we can however reduce the conjecture to the following spectral problem on a quarter - sphere @xmath45 that can be treated using numerical methods .
namely , let , in usual spherical coordinates @xmath46 , @xmath47 we split the boundary @xmath48 into two parts : @xmath49 , where @xmath50 @xmath51 and consider the spectral boundary value problem for the laplace - beltrami operator on @xmath52 : @xmath53 let @xmath54 denote the first eigenvalue of the problem ( which we understand as usual in the variational sense ) .
[ conj2 ] @xmath55 our main result is [ conj2to1 ] conjecture [ conj2 ] implies conjecture [ conj1 ] .
theorem [ conj2to1 ] is proved in sections 2 and 3 .
extensive numerical calculations ( see section [ sec : num ] ) show that @xmath56 which implies conjecture [ conj1 ] .
the best lower bound we are able to prove is just @xmath57 , which follows from dirichlet - neumann bracketing ( replace the dirichlet condition by the neumann one on the arc @xmath58 , @xmath59 , see @xcite ) .
finally , we note that the spectral problem easily reduces via the stereographic projection to the following mixed dirichlet - neumann problem on a half - disk @xmath60 ( here @xmath61 are usual planar polar coordinates ) : @xmath62 here @xmath63 and @xmath64 .
problems and are quite remarkable in their own right each of them is an example of a mixed dirichlet - neumann boundary value problem whose spectrum is invariant under a swap of dirichlet and neumann boundary conditions .
namely , the spectrum of coincides with the spectrum of @xmath65 we refer to @xcite for a further discussion on dirichlet - neumann swap isospectrality .
one can check that a surface with a finite number of conical singularities can be approximated by a sequence of smooth surfaces of the same genus and area in such a way that the corresponding sequence of the first non - zero eigenvalues converges to @xmath4 on the original surface .
thus , conjecture [ conj1 ] means that is sharp in the class of smooth metrics , although the equality is not necessarily attained . for a general result about the convergence of the whole spectrum see @xcite
let @xmath66 , @xmath67 be a map that intertwines the preimages of points of @xmath21 under a two - sheeted covering @xmath18 .
clearly , the laplace operator @xmath3 commutes with @xmath68 . by the spectral theorem
, we can consider separately the restrictions of the laplacian onto the spaces of functions which are either even or odd with respect to @xmath68 .
the even functions on @xmath23 can be identified with the functions on @xmath21 .
therefore , as @xmath69 , we have @xmath70 , and the equality in will be achieved if and only if the first eigenvalue @xmath71 of the laplacian acting on the odd subspace satisfies @xmath72 .
consider the following isometries of @xmath21 ( as usual , we identify @xmath21 and @xmath73 by stereographic projection ) : @xmath74 here @xmath75 is a point in the equatorial plane upon which a point @xmath76 is stereographically projected .
the hyperelliptic involution @xmath68 is given by @xmath77 .
for @xmath78 , a symmetry @xmath79 of @xmath21 has two corresponding symmetries @xmath80 and @xmath81 satisfying @xmath82 those symmetries , with account of are given by the explicit formulae @xmath83 as an illustration , we demonstrate how the last of these formulae is obtained : if @xmath84 , then by , @xmath85 thus giving the expression for @xmath86 .
it easily seen that all @xmath80 commute with @xmath68 and satisfy @xmath87 in the proof of theorem [ conj2to1 ] we will use only the symmetries @xmath88 and @xmath86 .
calculations for @xmath89 are presented for the sake of completeness ( see remark [ rems2 ] ) .
let @xmath90 denote a fixed point set of a mapping @xmath91 .
as easily seen from , the sets @xmath92 , for @xmath93 , lie in the union of the coordinate lines and a unit circle of @xmath73 , and we introduce the following notation for future reference .
the coordinate lines are divided into two rays each by the ramification point @xmath94 , and we denote @xmath95 @xmath96 the circle is divided into four arcs by the ramification points @xmath97 , @xmath98 , @xmath99 , and @xmath100 , and we denote the arcs by @xmath101 so that the arc @xmath102 goes from @xmath103 to @xmath104 , the arc @xmath105 goes from @xmath104 to @xmath106 , the arc @xmath107 goes from @xmath106 to @xmath108 , and finally @xmath109 goes from @xmath108 to @xmath103 . in this notation , the fixed point sets @xmath92 are written as @xmath110 note that each of the rays @xmath111 ( @xmath112 ) intersects an arc @xmath113 at a single point which we denote @xmath114 : @xmath115 see figure [ fig : points ] .
each of the points @xmath114 has exactly two pre - images @xmath117 , @xmath118 , where @xmath119 are the solutions of the equation @xmath120 , with @xmath121 given in .
these solutions are easily found from ; we are of course at liberty to choose which of the two solutions is denoted @xmath122 and which is denoted @xmath123 .
for definiteness we set @xmath124 for future use , we need to know the images of points @xmath125 under the symmetries @xmath126 , @xmath127 . these are easily calculated from ; it turns out that @xmath128 with some indices @xmath129 , @xmath130 .
the results of the calculations are summarized in the following table [ tab : ws_and_ss ] .
.the points @xmath125 and their images @xmath131 under symmetries @xmath126 .
the table lists the pairs of indices @xmath132 and the resulting pairs @xmath133 for @xmath127 .
note also that @xmath68 acts by interchanging the second indices @xmath134.[tab : ws_and_ss ] [ cols="^,^,^,^ " , ] in each of the experiments above , we found that the computed eigenvalues appeared to converge to a value greater than 2.27 .
the associated eigenfunctions also appear to converge to a function whose contour lines are shown in figure [ fig : contour ] .
the authors would like to thank herschel farkas , eyal goren , daniel grieser , jacques hurtubise , and rafe mazzeo for useful discussions and suggestions .
part of this work was completed when i.p .
was visiting heriot - watt university and the mathematical research institute at oberwolfach , m.l .
was visiting universit de montral , and d.j .
was visiting max planck institute for mathematics in bonn .
hospitality of these institutions is greatly appreciated .
, m.l . and i.p .
also wish to thank the organizers of the lms durham symposium `` operator theory and spectral analysis '' for providing excellent conditions for collaboration . | sharp upper bounds for the first eigenvalue of the laplacian on a surface of a fixed area are known only in genera zero and one .
we investigate the genus two case and conjecture that the first eigenvalue is maximized on a singular surface which is realized as a double branched covering over a sphere .
the six ramification points are chosen in such a way that this surface has a complex structure of the bolza surface .
we prove that our conjecture follows from a lower bound on the first eigenvalue of a certain mixed dirichlet - neumann boundary value problem on a half - disk .
the latter can be studied numerically , and we present conclusive evidence supporting the conjecture .
laplacian , first eigenvalue , surface of genus two , mixed boundary value problem . |
agn are observed to be copious x - ray emitters .
these x - rays are thought to originate from the innermost regions of an accretion disk around a central supermassive black hole .
since the accretion disk itself is expected to be an optical / uv emitter , the most likely mechanism producing the x - rays is inverse compton scattering of these soft photons in a hot and tenuous corona that sandwiches the accretion disk .
thus , in principle , the study of these x - rays should allow the immediate environment of the accreting black hole as well as the exotic physics , including strong - field general relativity , that operates in this environment to be probed . in the past decade
, x - ray astronomy has begun to fulfill that promise .
guided by observations with the _ ginga _ , asca , rxte and bepposax satellites , there is a broad consensus that x - ray irradiation of the surface layers of the accretion disk in a class of agn known as seyfert 1 galaxies gives rise to fluorescent k@xmath2 emission line of cold iron via the process of `` x - ray reflection '' .
accompanying the iron line is a reflected continuum possessing a characteristic shape that is determined by the competing effects of photoelectric absorption and compton scattering . in this contribution
, i will review the basic physical processes leading to the observed x - ray reflection signatures , and the evidence leading us to believe that they do indeed originate from the innermost regions of the accretion disk .
we will then discuss three recent observational campaigns on ngc 4258 ( m106 ) , ngc 5548 and mcg6 - 30 - 15 .
the purpose of presenting results from these campaigns is to describe the successes , and mysteries , resulting from recent studies of x - ray reflection signatures .
the basic physics of x - ray ` reflection ' can be understood by considering a hard x - ray ( power - law ) continuum illuminating a semi - infinite slab of cold gas . in this context , ` cold ' is taken to mean that metal atoms are essentially neutral , but h and he are mostly ionized .
when a hard x - ray photon enters the slab , it is subject to a number of possible interactions : compton scattering by free or bound electrons , photoelectric absorption followed by fluorescent line emission , or photoelectric absorption followed by auger de - excitation .
a given incident photon is either destroyed by auger de - excitation , scattered out of the slab , or reprocessed into a fluorescent line photon which escapes the slab .
figure 1 shows the results of a monte carlo calculation which includes all of the above processes ( reynolds 1996 ; based on similar calculations by george & fabian 1991 ) . due to the energy dependence of photoelectric absorption , incident soft x - rays
are mostly absorbed , whereas hard photons are rarely absorbed and tend to compton scatter back out of the slab .
this gives the reflection spectrum a broad hump - like shape .
in addition , there is an emission line spectrum resulting primarily from fluorescent k@xmath2 lines of the most abundant metals .
the iron k@xmath2 line at 6.4kev is the strongest of these lines . for most geometries relevant to this discussion , the observer will see this reflection component superposed on the direct ( power - law ) primary continuum . under such circumstances ,
the main observables of the reflection are a flattening of the spectrum above approximately 10kev ( as the reflection hump starts to emerge ) and an iron line at 6.4kev . for solar or cosmic abundances and a plane - parallel slab geometry ,
the expected equivalent width of the iron line is 150200ev ( george & fabian 1991 ; reynolds , fabian & inoue 1995 ) .
a major advance came with the launch of _ asca _ and its medium - resolution ccd spectrometers . a long ( 4.5 day )
observation of the bright seyfert galaxy mcg@xmath36 - 30 - 15 in july-1994 allowed the iron line profile to be determined with some accuracy .
the resulting line profile is shown in fig . 2 ( published by tanaka et al . 1995 ) .
the line was found to be _ extremely _ broad ( almost @xmath4 fwzi ) and asymmetric in so far that it possesses an extensive red - wing .
such a broad and asymmetric line is expected if the x - ray reflection is occurring in the inner regions of an accretion disk strong line - of - sight doppler shifts , transverse doppler shifts and gravitational redshifts combine to produce an extensive low - energy wing and a sharp truncation of the line at high - energies ( fabian et al .
tanaka et al .
( 1995 ) showed that the mcg@xmath36 - 30 - 15 data is in good agreement with the disk model provided the inclination of the disk is @xmath5 and line fluorescence occurs down to @xmath6 , the innermost stable orbit around a schwarzschild black hole .
_ this result is the best evidence to date for a radiatively - efficient accretion disk around a black hole in any object .
_ subsequent studies of large samples of objects by nandra et al .
( 1997 ) confirmed the presence of these features in many other seyfert 1 galaxies , and show that there is a tendency for the iron lines to indicate face - on accretion disks ( as expected for seyfert 1 nuclei from the unified seyfert scheme ) .
the claim that iron line studies are probing the region within a few gravitational radii of the black hole is a bold one , and should be tested against other models at every opportunity . given the quality of data , the july-1994 mcg@xmath36 - 30 - 15 line profile has become a test bed for such comparisons .
fabian et al . ( 1995 ) examined many alternative models including lines from mildly relativistic outflows , the effect of absorption edges on the observed spectrum , and broadening of the line via comptonization .
fabian et al . found that none of these models were viable alternatives for the mcg@xmath36 - 30 - 15 line profile .
the idea of producing the broad line via comptonization has been revived recently by misra & kembhavi ( 1997 ) and misra & sutaria ( 1999 ) .
they argue that the spectrum initially consists of a narrow iron line superposed on a power - law continuum and that comptonization in a surrounding cloud with optical depth @xmath7 produces the broad line .
the comptonizing cloud must be both cold ( @xmath8kev in order to predominately downscatter rather than upscatter the line photons ) , and fully - ionized ( since no strong iron absorption edges are seen in the continuum spectrum ) .
the cloud is kept fully ionized and yet cool by postulating that the continuum source has a very luminous optical / uv component .
there are strong arguments against such a model .
since the power - law continuum emission also passes through any such comptonizing cloud , one would observe a break in the continuum spectrum at @xmath9kev .
such a break is not observed in the _ bepposax _ ( guainazzi et al .
1999 ) or _ rxte _ data ( lee et al . 1999 ) for mcg6 - 30 - 15 ( see misra 1999 ) . also , both continuum variability ( which is seen on timescales as short as 100s ) and ionization arguments limit the size of the comptonizing cloud in mcg6 - 30 - 15 to @xmath10 cm .
the essence of this ionization argument is that the ionization parameter at the outer edge of the cloud ( which , for a fixed cloud optical depth , scales with cloud size as @xmath11 ) must be sufficiently high that all abundant metals , including iron , are fully ionized ( fabian et al .
1995 ; reynolds & wilms 2000 ) . in the case of mcg6 - 30 - 15
, these constraints on the cloud size turn out to so tight that the postulated optical / uv component required to compton cool the cloud would violate the black body limit ( reynolds & wilms 2000 ) .
comptonization moreover provides a poor fit ( ruszkowski & fabian 2000 ) .
hence , we consider the comptonization model for the broad iron line not to be viable . in another alternative model ,
skibo ( 1997 ) has proposed that energetic protons transform iron in the surface of the disk into chromium and lower @xmath12 metals via spallation which then enhances their fluorescent emission . with limited spectral resolution
, such a line blend might appear as a broad skewed iron line .
this model suffers both theoretical and observational difficulties . on the theoretical side
, high - energy protons have to be produced and slam into the inner accretion disk with a very high efficiency ( skibo assumes @xmath13 for this process alone ) . on the observational side , it should be noted that the broad line in mcg6 - 30 - 15 ( tanaka et al 1995 ) is well resolved by the _ asca _ sis ( the instrumental resolution is about 150 ev at these energies ) and it would be obvious if it were due to several separate and well - spaced lines spread over 2 kev .
low luminosity agn ( llagn ) are almost certainly systems which are accreting at very low eddington fractions .
this places them in a regime of accretion which has been the subject of active theoretical work for many years .
it was realized by several authors that when the accretion rate is low ( relative to the eddington rate ) , an accretion disk may switch into a hot , radiatively - inefficient mode ( ichimaru 1977 ; rees 1982 ; narayan & yi 1994 ; narayan & yi 1995 ) .
in essence , the plasma becomes so tenuous that the timescale for energy transfer from the protons to the electrons ( via coulomb interactions ) becomes longer than the inflow timescale .
the energy remains as thermal energy in the protons ( which are very poor radiators ) and gets advected through the event horizon of the black hole .
these are the so - called advection dominated accretion flows ( adafs ) .
adafs are to be contrasted with ` standard ' radiatively - efficient accretion disks in which the disk remains cool and geometrically thin all of the way down to the black hole ( shakura & sunyaev 1973 ; novikov & thorne 1974 ) .
llagn provide an ideal laboratory in which to study adafs .
if the central regions of llagn accretion disks are operating in an adaf mode , we would not expect to observe broad iron lines since the adaf is far too hot to contain combined iron ions .
on the other hand , if we see broad iron lines from a llagn , this would be strong evidence for radiatively efficient disks even at these low accretion rates .
however , the fact that llagn are faint has hampered such detailed x - ray studies of these sources . in an attempt to study one llagn in some detail
, we performed a 200ksec joint _
asca_/_rxte _ observation of the nearby llagn ngc 4258
. a full discussion of this observation is presented in reynolds , nowak & maloney ( 2000 ) . here , we summarize the iron line properties of this object . as can be seen in fig . 3
, we clearly detect a iron line with an equivalent width of @xmath14ev and a velocity width of @xmath15 ( fwhm ) which is much narrower than those seen in higher luminosity seyfert galaxies .
if it is assumed that the observed iron line originates from the accretion disk , then we can use relativistic `` disk - line '' models to place constraints on the location of the line emission . using such models
it is found that the narrowness of the line implies that the bulk of the line emission originates from a distance @xmath16 from the black hole .
such a large line emitting radius is consistent with the adaf model for ngc 4258 ( lasota et al .
1996 ) in which the thin fluorescing accretion disk goes through an transition into an adaf at about @xmath1
. however , the data are also consistent with a radiatively efficient fluorescing disk extending down to the radius of marginal stability and a large ( @xmath0 ) x - ray corona . in this latter case , there would be faint broad wings to the line profile which could be found with a high signal - to - noise _ xmm newton _ spectrum .
one way or another , the iron line properties of this llagn are different to those of many higher - luminosity seyfert galaxies which display a broad iron line indicating radiative disks and small ( @xmath17 ) x - ray emitting regions .
we can not rule out the possibility that most ( or all ) of the observed iron line originates from matter that is not associated with the accretion disk .
however , simple arguments lead us to disfavor such a scenario . consider iron line emission in a geometrically - thick torus surrounding the accretion disk of ngc 4258 . an upper limit to the column density of this structure along our line of sight to the agn is given by the observed column density of @xmath18 .
if we suppose that this torus is in the same plane as the accretion disk ( so that we are also viewing it edge - on ) , it is plausible to assume that we are looking through the optically - thickest part of the torus .
we can then show that the maximum equivalent width produced by fluorescence of this material is @xmath19ev ( reynolds , nowak & maloney 2000 ) .
however , there are several reasons why this upper limit would almost certainly not be achieved .
firstly , modeling of the accretion disk warp strongly suggests that our line of sight intercepts the disk and that the bulk of the column density which obscures the agn originates in the disk ( herrnstein , priv .
communication ) .
therefore , we might expect significantly smaller column densities along lines of sight that have smaller inclinations angles relative to the accretion disk . secondly , the accretion disk may well occult half of this fluorescing cloud , thereby reducing this prediction further . thus
, the true iron line from surrounding non - disk material may well be reduced from our naive prediction by a factor of several . despite these arguments ,
let us now suppose that the observed iron line is _ not _ associated with the accretion disk .
our data then permit the presence of a `` seyfert - like '' broad iron line which is buried in the noisy continuum data .
if we suppose ngc 4258 to possess a normal `` seyfert - like '' broad iron line ( with an intrinsic equivalent width in the range 200400ev , modified by the light bending and limb - darkening effects that are important in edge - on systems such as ngc 4258 ) , we can set some interesting constraints ( see fig .
4 ) . in particular
, we can rule out models in which the limb - darkening is very weak ( as would be the case if , for example , the @xmath20 surface of the accretion disk is very filamentary rather than planar ) .
of course , higher signal - to - noise data from _ xmm newton _ are required if one is to constrain the nature of the disk in this object any further .
in the rest of this contribution , we turn to the properties of higher luminosity systems .
firstly , we shall discuss the classical seyfert galaxy ngc 5548 .
we observed ngc 5548 with _
euve _ , _ asca _ , and _ rxte _ in june / july 1998 . the main part of the campaign consisted of four observations ( with all of these satellites simultaneously ) separated by approximately one week .
one of these observations ( 18-june-1998 ) was substantially longer than the others ( @xmath21ksec of good data , as compared with an average of @xmath22ksec for the other observations ) .
this long observation provided us with intensive monitoring over a period of 3 days from the euv to hard x - rays .
the various aspects of this campaign are discussed in detail in chiang et al .
( 2000 ) .
although not strictly related to x - ray reflection studies , it is worth mentioning the continuum variability properties found during this campaign .
figure 5 shows the _ euve _ , _ asca _ and _ rxte _ light curves for our long observation .
the most distinct feature is a step at @xmath23s . by computing the z - transformed discrete correlation function ( zdcf ; alexander 1997 ) between these various data
, we find clear evidence that the euv emission leads the _ rxte_-band emission by @xmath22ksec , and the _ asca_-band emission by @xmath24ksec .
if confirmed , this results has at least two important implications .
firstly , the euv emission can not simply be reprocessed hard x - ray emission since these scenario would predict the euv to lag the hard x - rays . secondly ,
if we make the assumption that the observed lags are due to the compton upscattering of seed photons , we can constrain the size of the comptonizing region to be about @xmath25 for a @xmath26 black hole . returning to the issue of x - ray reflection
, this campaign produced a major surprise . before discussing the data
, it is worth recapping our theoretical expectations from simple x - ray reflection models . for a fixed disk / corona geometry and ionization state
, we expect the iron line equivalent width to remain constant provided one is probing timescales longer than the reverberation timescale of the system ( which is always the case for spectral studies with _ asca _ and _ rxte _ ) .
if the iron line equivalent width does change due to , for example , a change in the geometry of the system , then the relative strength of the reflection continuum would be expected to vary in step . in other words ,
we expect a strict proportionality between the iron line equivalent width and relative strength of the reflection continuum .
figure 6 shows the various spectral parameters , measured from the _ rxte _ data , for ngc 5548 as a function of the 210kev continuum flux of the source .
we found that the equivalent width of the iron line declined as the continuum source flux increased in fact , the iron line _ flux _ was consistent with being constant .
simultaneous spectroscopy with _ asca
_ showed the iron line to be broadened .
hence , the line does appear to be originating from the central regions ( @xmath27 ) of the accretion disk and , for any reasonable size black hole mass , the constancy of the line flux can not be due to light travel time effects .
moreover , the relative reflection fraction ( @xmath28 ) fails to track the changes in the iron line equivalent width .
indeed , these quantities appear to be weakly _ anti - correlated _ ( see fig .
7 ) , contrary to the simple reflection models .
a very long , simultaneous _ asca_/_rxte _ observation of the seyfert 1 galaxy mcg6 - 30 - 15 by an ( almost ) independent team of researchers revealed very similar spectral results to those obtained for ngc 5548 .
figure 8 shows the clear anti - correlation between iron line equivalent width and the relative strength of the reflection continuum found by lee et al .
( 2000 ) . to produce this figure , almost nine days of _ rxte _ data
have been binned into four flux - sorted spectra . unlike with ngc 5548
, these data strongly require a change in the relative strength of the reflection continuum as well as the iron line equivalent width .
the details of this observation are discussed by lee et al .
( 1999 , 2000 ) .
this particular _ rxte _ campaign yielded data of an almost unprecedented nature the observation lasted over 100 consecutive orbits of _ rxte _ , giving us dense coverage ( @xmath29 ) for almost 9 days .
reynolds ( 2000 ) took advantage of this unique dataset and performed a detailed search for delays between a continuum band ( 24kev ) and the iron line band ( 57kev ) in an attempt to search for iron line reverberation effects .
in essence , the method of press , rybicki & hewitt ( 1992 ) was used to interpolate across the gaps in the continuum light curve ( caused primarily by earth occultation ) and estimate the errors in the interpolation .
figure 9a shows a small portion of the light curve , together with the press et al .
( 1992 ) reconstruction .
the continuous ( interpolated ) continuum light curve can then be folded through a trial transfer function and compared with the line - band light curve in a @xmath30 sense ( e.g. , fig .
9b shows the @xmath30 surface that results from such a procedure ) . by varying the parameters of the trial transfer function , we can place constraints on any time delays between bands .
no time delays were found on timescales down to 5001000s .
instead , it was found that the line flux remained constant , thereby extending the result of lee et al .
( 2000 ) down to these timescales .
to summarize the mysteries resulting from the ngc 5548 and mcg6 - 30 - 15 campaigns , it is found that : 1 .
the iron line _ flux _ , rather than the equivalent width , appears to remain approximately constant as the continuum x - ray source undergoes rapid variability . in ngc 5548 , this result is derived by direct x - ray spectral fitting and so applies on timescales of @xmath31ksec ( which is approximately timescale that can be probed via such methods ) . in mcg6 - 30 - 15 ,
the spectral fitting of lee et al .
( 2000 ) combined with the analysis of reynolds ( 2000 ) leads us to believe that such behavior occurs on all timescale from 500ksec down to 500s .
this can not be due to light travel delays .
the iron line equivalent width and relative strength of the reflection continuum are _ anti - correlated _ , in contradiction to simple x - ray reflection models .
changing the ionization state of the surface of the accretion disk is one of the few ways of breaking the expected proportionality between the iron line equivalent and the ( inferred ) relative strength of the reflection continuum .
reynolds ( 2000 ) proposed a toy model to explain these _ rxte _ results ( fig .
10 ) . in this model
, there is a radial ionization gradient on the disk surface due to the radial dependence of both the ionizing flux and the density of the surface layers .
the surface of the inner disk , which is strongly irradiated by the primary x - ray flux , is highly ionized and acts as a pure compton mirror .
it thereby fails to imprint distinct ( atomic ) features in the observable spectrum .
by contrast , the outer disk surface ( starting at @xmath32 ) possesses a relatively low ionization parameter and produces ` neutral - like ' x - ray reflection signatures such as those calculated by george & fabian ( 1991 ) .
the interesting and uncertain processes occur in the partially ionized intermediate zone of the accretion disk . in this region
, there are two possible physical mechanisms by which the iron line can be strongly suppressed relative to the reflected continuum .
these mechanisms are described in the next two paragraphs .
reynolds ( 2000 ) noted that , if the boundaries between these zones are defined by some given set of ionization parameters , the zones will move as the primary continuum flux varies .
if the primary continuum flux increases , the fluorescing outer zone will move outwards and so subtends a smaller solid angle as seen by the primary x - ray source .
the observed iron line equivalent width will then be seen to decrease .
however , the measured relative strength of the reflection continuum may remain roughly constant if the intermediate zone still produces bound - free absorption features in the reflection continuum . as mentioned above
, there are at least two mechanisms which may operate in the intermediately ionized zone of the accretion disk to suppress iron line emission relative to the reflection continuum .
nayakshin , kazanas & kallman ( 2000 ) showed that the well - known thermal instability of photoionized plasma could lead to the formation of a highly ionized skin on the surface of the accretion disk .
compton scattering in this skin would smear out sharp spectral features such as the iron line , possibly to the level where line photons from this region would no longer be identified with the iron line
. however , very broad spectral features such as the iron edge and broad hump that observationally define the reflection continuum , would be much less affected by compton scattering in the skin .
thus , from the observational point of view , the presence of such a skin might reduce the iron line strength while leaving the strength of the reflection continuum unaffected .
the second line destruction mechanism is that of resonant trapping followed by auger destruction ( ross & fabian 1993 ) .
a fluorescent iron line photon is typically created at an optical depth of @xmath33 . however , when the bulk of the ions are at least as ionized as fexvii , the presence of a vacancy in the l - shell of the iron ions leads to resonant scattering of the line photons with a rather small path length .
if , furthermore , the bulk of the ions possess at least two l - shell electrons ( i.e. fexxiii or less ) , then the auger effect acts as a photon destruction mechanism . the combination of resonant ` trapping ' and auger ` destruction ' might reduce the observed iron line to a very small strength .
clearly , more work is needed on the physics and modeling of ionization accretion disks in order to assess whether such a toy model can explain the mysterious x - ray variability that we see in current _ rxte _ data .
the rapid x - ray variability of many seyfert galaxies leads us to believe that the primary x - rays are emitted during dramatic flare - like events in the accretion disk corona . when a new flare becomes active
, the hard x - rays from the flare will propagate down to the cold disk and excite iron fluorescence . due to the finite speed of light , the illumination from the flare sweeps across the disk , and the reflected x - rays act as an ` echo ' of this flare .
such flaring will cause temporal changes in the iron line profile and strength due to the changing illumination pattern of the disk and , more interestingly , time delays between the observed flare and the its fluorescent echo .
this latter effect is directly analogous to the optical broad line reverberation that is a major theme of this meeting .
the observational study of iron line reverberation is beyond the capabilities of current instruments ( even , probably , _
xmm newton _ ) but should be within reach of future high - throughput x - ray observatories such as _ constellation - x _ and _
xeus_. in recent work , it has been shown that interesting diagnostics of the black hole spin and x - ray source geometry are encoded within the iron line reverberation pattern ( reynolds et al .
1999 ; young & reynolds 2000 ; ruzskowski 2000 ) . however , there are many complications and issues that will affect our ability to extract these reverberation signatures from real data .
firstly , there are likely to be several x - ray flares active at any one time which will give overlapping reverberation signatures . moreover , due to the fact that the x - ray emitting corona is probably extended on the same spatial scales as the inner disk , each x - ray emitting flare will have its own distinct transfer function .
this breaks the linearity of the transfer problem .
young & reynolds ( 2000 ) showed that if there are only a small number of powerful flares , the individual transfer functions can be identified and separated in _ constellation - x data_. however , we must examine more powerful statistical methods for extracting useful information from this system .
possible reverberation signatures will be affected by short timescale ionization changes .
the models of reynolds et al .
( 1999 ) and young & reynolds ( 2000 ) assume that the disk has a fixed , fairly cold , ionization structure .
however , as discussed extensively above , _ rxte _ argues for ionization changes in the disk surface that are correlated with the continuum flux . on a local scale , this effect may be driven by ionization fronts that sweep across the disk when an x - ray flare erupts . in future work , these effects must be assessed so that instruments and observing strategies can be planned to study reverberation effects .
it must be noted that , even though _ rxte _ data argue for almost constant iron line flux due to the ionization changes , we would still expect to see variability of the line strength and/or profile on either the recombination timescale in the disk surface or the light - crossing timescale from the flare to the disk , whichever is longest .
given all of these complications and difficulties , why bother to pursue the idea of iron line reverberation ?
the answer to this question is straightforward until we have the technology to directly image the immediate environment of a black hole ( and this is at least 20 years away ) , x - ray iron line reverberation is simply our best hope of mapping mapping out the detailed astrophysical environment of supermassive black holes , including the effects of the black hole s spin .
in this contribution , we present three case studies in order to describe the current state of x - ray reflection studies in agn . for the low - luminosity agn ngc 4258 , we find that the iron line is much narrower than is typically found in higher luminosity agn .
we argue that this is evidence for either a truncated cold accretion disk ( possibly due to a transition to an adaf at @xmath0 ) or a large ( @xmath1 ) x - ray emitting corona surrounding the accretion disk .
we also present results for the higher luminosity seyfert nuclei in ngc 5548 and mcg6 - 30 - 15 . in both of these sources ,
_ rxte _ shows that the iron line equivalent width decreases with increasing luminosity .
furthermore , the iron line equivalent width is found to be _ anticorrelated _ with the relative strength of the reflection continuum , contrary to all simple reflection models .
it is proposed that continuum - flux correlated changes in the ionization of the accretion disk surface can explain this spectral variability .
finally , we address the issue of x - ray iron line reverberation in the light of these complicating factors .
the author appreciates support from hubble fellowship grant hf-01113.01 - 98a .
this grant was awarded by the space telescope institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa under contract nas 5 - 26555
. he also appreciates support from nasa under ltsa grant nag5 - 6337 , and the national science foundation under grants ast-9529170 and ast-9876887 .
alexander t. , 1997 , in astronomical time series , ed .
d. maoz et al . , ( dordrecht : kluwer ) , 163 chiang j. et al . , 2000 , apj , 528 , 292 fabian a. c. , rees m. j. , stellar l. , white n. e. , 1989 , mnras , 238 , 729 fabian a. c. et al .
1995 , mnras , 277 , l11 george i. m. , fabian a. c. , 1991 , mnras , 249 , 352 guainazzi m. et al . , 1999 ,
a&a , 341 , l27 ichimaru s. , 1977 , apj , 214 , 840 lasota j .-
p . , abramowicz m. a. , chen x. , krolik j. , narayan r. 1996 , apj , 462 , 142 lee j. , fabian a. c. , reynolds c. s. , brandt w. n. , iwasawa k. , 2000 , mnras , in press lee j. , fabian a. c. , brandt w. n. , reynolds c. s. , iwasawa k. , 1999 , mnras , 310 , 973 misra r. , 1999 , iucaa preprint 32/99 misra r. , kembhavi a. k. , 1998 , apj , 499 , 205 misra r. , sutaria f. k. , 1999 , apj , 517 , 661 nandra k. , george i. m. , mushotzky r. f. , turner t. j. , yaqoob t. , 1997 , apj , 477 , 602 nayakshin s. , kazanas d. , kallman t. r. , 2000 , apj , submitted ( astro - ph/9909359 ) narayan r. , yi i. , 1995 , apj , 452 , 710 narayan r. , yi i. , 1994 , apj , 428 , l13 novikov i. d. , thorne k. s. , 1973 , in _ black holes _ , eds c. dewitte & b. s. dewitte ( gordon and breach science publicaters , new york ) , p.344 press w. h. , rybicki g. b. , hewitt j. n. , 1992 , apj , 385 , 404 rees m. j. , 1982 , in riegler g. , blandford r. d. , eds , the galactic center , am .
, new york , p.166 .
reynolds c. s. , 1996 , phd thesis , university of cambridge reynolds c. s. , 2000 , apj , 533 , 811 reynolds c. s. , wilms j. , 2000 , apj , 533 , 821 reynolds c. s. , fabian a. c. , inoue h. , 1995 , mnras , 276 , 1311 reynolds c. s. , nowak m. a. , maloney p. r. , 2000 , apj , in press ( astro - ph/0004068 ) reynolds c. s. , young a. j. , begelman m. c. , fabian a. c. , 1999 , apj , 514 , 164 ross r. r. , fabian a. c. , 1993 , mnras , 261 , 74 ruszkowski m. , 2000 , mnras , 315 , 1 ruszkowski m. , fabian a. c. , 2000 , mnras , submitted shakura n. l. , sunyaev r. a. , 1973 , a&a , 24 , 337 skibo , j.g . , 1997 , apj , 478 , 522 tanaka y. et al . , 1995 , nat , 375 , 659 young a. j. , reynolds c. s. , 2000 , apj , 529 , 101 | after reviewing the basic physics of x - ray reflection in agn , we present three case studies which illustrate the current state of x - ray reflection studies . for the low - luminosity agn ngc 4258 , we find that the iron line is much narrower than is typically found in higher luminosity agn .
we argue that this is evidence for either a truncated cold accretion disk ( possibly due to a transition to an advection dominate accretion flow at @xmath0 ) or a large ( @xmath1 ) x - ray emitting corona surrounding the accretion disk .
we also present results for the higher luminosity seyfert nuclei in ngc 5548 and mcg6 - 30 - 15 . in both of these sources ,
_ rxte _ shows that the iron line equivalent width decreases with increasing luminosity .
furthermore , the iron line equivalent width is found to be _ anticorrelated _ with the relative strength of the reflection continuum , contrary to all simple reflection models .
it is proposed that continuum - flux correlated changes in the ionization of the accretion disk surface can explain this spectral variability .
finally , we address the issue of x - ray iron line reverberation in the light of these complicating factors .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
dwarf elliptical galaxies ( des ) are the most common galaxy type by number in the local universe , dominating the galaxy luminosity function of nearby clusters . yet
these galaxies remain among the most poorly studied galaxies due to their faint luminosities , @xmath5 , and characteristic low effective surface brightness @xmath6 mag arcsec@xmath7 @xcite . unlike brighter , classical elliptical galaxies whose surface brightness profiles are well fit by the de vaucouleurs @xmath8 law @xcite , des have brightness profiles that are characterized by sersic profiles @xcite with indices ranging between @xmath93 ( where @xmath9 corresponds to an exponential law and @xmath10 to an @xmath8 law ) making them appear more diffuse than classical ellipticals of the same total magnitude @xcite . in the virgo cluster ,
the majority of des brighter than @xmath11 contain compact central nuclei ; fainter than @xmath12 most des show no sign of a nucleus @xcite .
nuclei typically contain 5% to 20% of the total galaxy light and are slightly resolved at the distance of virgo by _ hubble space telescope _
( _ hst _ ) imaging @xcite .
in hierarchical models of galaxy formation , dwarf galaxies form out of small density fluctuations in the early universe and are predicted to be less spatially clustered than normal elliptical or spiral galaxies @xcite .
however , dwarf elliptical galaxies are preferentially found in dense cluster environments , more so than either ellipticals or spirals @xcite ; there are few , if any , examples of isolated des .
thus , current models favor de formation from a progenitor galaxy population . the proposed progenitors of des are spiral or irregular galaxies which are morphologically transformed into des through the processes of galaxy harassment and interaction @xcite .
detailed internal kinematics of des are a powerful observational tool with which to test these scenarios . until recently ,
radial velocity dispersion profiles were only available for the local group des and two of the brightest des in the virgo cluster @xcite .
in addition , a handful of global velocity dispersion measurements existed for des in various environments outside the local group @xcite .
these observations suggest that des have lower mass - to - light ratios than local group dwarf spheroidals ( e.g. , draco , fornax ) and are flattened by velocity anisotropy rather than by rotation .
however , @xcite and @xcite recently presented kinematic profiles for a few des in the virgo and fornax clusters , respectively , with rotation velocities comparable to that expected for a rotationally - flattened spheroid .
these rotating des are more luminous on average than the non - rotating des we have observed , and hint at a possible association between de luminosity and the presence of rotation . the question of whether des have significant rotation compared to their velocity dispersion is particularly important in the context of de formation scenarios .
@xcite have demonstrated that the process of galaxy harassment in cluster environments can morphologically transform a spiral galaxy into a de .
although this process tends to increase the velocity dispersion in a system , it is less efficient at disrupting rotational motions and a significant fraction of the progenitor s rotation is preserved .
thus , measuring the amount of internal angular momentum in des can constrain the progenitor galaxy type and/or the amount of disruption required .
we present internal kinematics as a function of radius for a sample of six de galaxies in the virgo cluster based on keck observations .
these data are interpreted in conjunction with archival _ hst _ imaging .
preliminary results from this study were presented in @xcite .
this paper is organized as follows : in [ data_sec ] we present the keck spectroscopic and _ hst _ imaging observations of our target galaxies , along with an outline of data reduction procedures ; in [ res_sec ] we present velocity and velocity dispersion profiles and describe the dynamical models that are applied to the data to derive mass - to - light ratios , constraints on orbital anisotropy , and limits on the central black hole mass ; the broader implications of the results are discussed in [ disc_sec ] .
the virgo cluster is the closest large reservoir of de galaxies beyond the local group and presents a significantly different environment in which to study such galaxies .
the six de galaxies presented below were drawn from the bright end of the virgo de luminosity function and have been imaged with _
these objects lie at a variety of distances from the center of the virgo cluster as shown in figure [ vcc ] .
the positions and photometric properties of the observed galaxies are listed in table 1 .
foreground reddening values are taken from @xcite assuming a standard galactic extinction law with @xmath13 . throughout this paper a virgo cluster true distance modulus of @xmath14 is adopted , i.e. , a distance of 15.3 mpc , as determined by the _ hst _ key project on the extragalactic distance scale @xcite .
six de galaxies were observed on 2001 march 2021 using the keck ii 10-m telescope and the echelle spectrograph and imager ( esi ; * ? ? ? * ) .
observations were made in the echellette mode with continuous wavelength coverage over the range @xmath15@xmath16 across 10 echelle orders with a spectral dispersion of 11.4 km s@xmath17 pixel@xmath17 .
the spectra were obtained through a @xmath18 slit , resulting in an instrumental resolution of 23 km s@xmath17 ( gaussian sigma ) over the entire spectrum , or @xmath19 .
the slit was positioned on the major axis of each galaxy , such that the galaxy s center was displaced @xmath20 from the center of the slit along its @xmath21 length .
three consecutive exposures of 20 minutes were obtained for each galaxy except vcc 1577 , for which @xmath22 minutes were obtained .
a summary of the observing parameters is given in table 2 .
high signal - to - noise ratio spectra of giant stars covering the range of spectral types g8iii to m0iii were taken with the same instrumental setup for use as templates in the kinematic profile fitting described in [ vp ] .
standards stars were observed both centered on and trailed across the slit width in order to recover an accurate estimate of the instrumental broadening for point and extended sources , respectively .
the esi data were reduced using a combination of iraf echelle and long - slit spectral reduction tasks .
first , the overscan and a dark frame , scaled by the exposure time , were subtracted from the data .
the sum of a bright calibration star spectral exposure and a flat field spectral exposure were used to trace the ends of the slit as well as a fiducial spatial point ( the location of the bright star ) for each of the 10 curved echelle orders .
this process yielded an empirical measurement of the spatial pixel scale for each order : it varies from @xmath23 in the bluest order to @xmath24 in the reddest order .
scattered light was subtracted from individual frames by fitting a smooth function to the areas outside these apertures ( spaces between echelle orders ) using the apscatter task . to preserve spatial information ,
the apall task was used in `` strip '' mode to extract and rectify two - dimensional rectangular strips for each echelle order by shifting and aligning each spatial column based on the aperture trace information .
the rectified orders were then interpolated to a common spatial pixel scale of @xmath24 per pixel .
calibration frames such as flat field , arc lamp , and template star exposures were also extracted into rectified , aligned and spatially - corrected strips using the same procedure .
data reduction was carried out on these rectified strips using procedures similar to those routinely used on long - slit spectra .
each strip was divided by its corresponding normalized flat - field image ( from the same echelle order ) .
cosmic rays on individual exposures , identified on the basis of object sharpness and peak pixel brightness , were masked and the exposures were then combined .
each order was logarithmically binned in wavelength using a two - dimensional wavelength calibration ( i.e. , as a function of spatial position along the slit ) determined from a combined cu / ar / hg / ne arc lamp spectrum .
the rms of the residuals in the wavelength solution is @xmath25 or less in each order .
the sky spectrum was determined for each combined frame from a section near the end of the slit farthest from the galaxy center ( @xmath26 ) , and subtracted from the rest of the two - dimensional spectrum .
we recognize that this `` sky '' spectrum is contaminated by light from the outer parts of the target de galaxy but this is unlikely to have a significant effect on our results .
we estimate that the `` sky '' spectrum is contaminated by galaxy light at the level of @xmath27 and @xmath28 relative to the de s center , with and without the nuclear contribution , respectively . in the outer parts of the de profile ( @xmath29 ) , the contamination level is higher ( 20%30% ) , but even this should not be a problem since there are no strong radial gradients in the mean velocity and velocity dispersion in the outer parts of the galaxies .
] bright , poorly - subtracted sky lines were masked out during the kinematic fitting discussed below .
the galaxy continuum flux in each order was then individually normalized to unity using the iraf continuum task .
a noise frame was created for each galaxy spectrum which kept track of uncertainties in every pixel due to read noise and poisson noise taking into account the ccd gain and number of readouts .
this noise frame is used as input into the kinematic analysis of [ vp ] as part of the formal error calculations on the velocity profile parameters .
finally , the strips from the different echelle orders were combined , weighted by the noise frame , to create a single two - dimensional long - slit spectrum . due to low signal - to - noise in the reddest and bluest echelle orders
, we do not include them in the analysis ; the final combined spectrum covers @xmath30 .
the spectra show no evidence for gaseous emission lines at any radii .
representative combined galaxy spectra are shown in figure [ spec ] .
the seeing fwhm during each spectroscopic observation was determined by comparing the observed intensity profile along the esi slit to high spatial resolution @xmath31-band surface brightness profiles derived from _
hst _ images ( as discussed in [ sb ] ) .
these brightness profiles were convolved with a gaussian seeing point spread function ( psf ) , integrated over the @xmath32 slit width and binned into @xmath24 pixels to match the esi pixel scale in the spatial direction .
this was then compared to the observed intensity profile along the esi slit in the matching spectral region .
the best - fit gaussian fwhm seeing estimates are given in table 2 for each galaxy .
these are consistent with the less accurate estimates of the seeing fwhm determined by coadding short ( few second ) exposures taken with the esi guider camera at intervals of approximately 5 minutes during the spectroscopic observations .
the mean line - of - sight velocity and velocity dispersion as a function of radius were determined using a pixel - fitting method first described in @xcite .
these quantities were determined by comparing the observed galaxy spectrum to a stellar template convolved with a series of gaussian line profiles .
the best - fitting gaussian profile was determined by @xmath33 minimization in pixel space .
the free parameters in this analysis are : mean line - of - sight velocity @xmath34 , velocity dispersion @xmath35 , and a line - strength parameter @xmath36 , which measures the ratio of equivalent width in the galaxy to that of the template star and accounts for template mismatch ( e.g. , due to differences in effective temperature or metallicity ) .
night sky absorption features ( a and b bands ) and strong sky emission lines were masked out in the fitting procedure ; masked pixels were not included in the calculation of @xmath33 .
deviations from gaussian profiles @xcite were not fit as this requires higher signal - to - noise spectra than presented in this paper .
in addition , an arbitrary continuum term was simultaneously fit to the data approximated by the sum of legendre polynomials . in analyzing the combined esi spectra ( @xmath37 pixels ) ,
the continuum was fit with a 20th order polynomial .
we have tested this method on broadened template stars to determine the minimum signal - to - noise required to accurately recover velocity dispersions and to estimate our sensitivity to template mismatch .
stellar templates were broadened with gaussian kernels of varying @xmath35 and poisson noise was added .
these tests suggest that for a signal - to - noise level @xmath38 per spectral pixel , a galaxy s internal velocity dispersions can be measured down to the instrumental resolution of 23 km s@xmath17 with an accuracy of 1% and down to 18.5 km s@xmath17 with an accuracy of 10% .
we have spatially rebinned our galaxy data to achieve a signal - to - noise level of @xmath39 per pixel at all radii , while ensuring that the spatial bin size is at least as large as the fwhm of seeing during the observations ( @xmath40 ) .
velocity profiles were recovered using template stars ranging in spectral type g8iii to m0iii .
the best fitting template , the k1iii star hd 40460 ( [ fe / h ] = @xmath41 ) , was used to recover all profiles presented here .
the choice of template star did not affect the recovered profiles at a level in excess of the formal error bars .
we also find no significant difference between the profiles presented here to those determined by separately recovering profiles for the eight esi echelle orders and computing a profile based on the weighted mean . we have previously attempted to observe virgo de galaxies with keck / hires in march 1998 .
the significantly higher spectral resolution of hires ( 2.1 km s@xmath17 pixel@xmath17 ) and its lower throughput as compared to esi made these observations prohibitively difficult .
the de galaxies vcc 1254 , vcc 1073 , vcc 452 and vcc 1876 were observed for 5 , 3 , 2 and 1.8 hours , respectively , through a custom - made @xmath42 slit .
we were able to determine the kinematics for vcc 1254 inside @xmath43 ; however , data for the last three galaxies did not have sufficient signal - to - noise to recover velocity profiles .
although this hires spectrum does not provide additional information on vcc 1254 , it does provide an excellent reliability check on our esi observations .
the esi and hires data complement each other in that the esi data have relatively high s / n but an instrumental resolution approaching the intrinsic dispersion of our target galaxies , whereas the instrumental resolution of hires is significantly higher at the price of low signal - to - noise . as shown in figure [ hires ] , the line profile shapes and velocity profiles determined with hires match that measured by esi .
the velocity profiles were calculated in the wavelength region @xmath44@xmath45 , however , we have no reason to believe that this argreement would be any less in other spectral regions .
_ hst _ wide field planetary camera 2 ( wfpc2 ) imaging is available for each of our target galaxies .
these images provide high spatial resolution surface brightness profiles needed for the dynamical modeling discussed in [ models ] , and allowed us to measure photometric properties for our target galaxies .
the data , first presented in @xcite and @xcite , consist of @xmath46-s wfpc2 images in the f555w bandpass and a single 300-s exposure in f814w .
the galaxies are centered on the wf3 ccd in the _ hst _ pointings , and we use only the f555w wf3 ccd image to determine surface brightness profiles .
the images were cleaned of cosmic rays and combined .
the instrumental f555w magnitudes were calibrated into @xmath31-band using the transformations of @xcite , assuming @xmath47 @xcite .
surface brightness profiles were determined for each galaxy using the iraf ellipse isophotal fitting routine down to a surface brightness of @xmath48 .
the average ellipticity , @xmath49 , determined between @xmath50 and the total integrated apparent magnitude , determined by integrating the total flux inside a @xmath51 aperture , are given in table 1 .
our apparent magnitudes agree with those determined by @xcite . unlike @xcite , we consider vcc 1577 to be a nucleated de galaxy since its bright , central star cluster is within a few tenths of an arcsecond of the galaxy s isophotal centroid position .
the observed surface brightness profiles are shown in figure [ sbprof ] . in subsequent analysis
, we differentiate between light from the central de nucleus and the underlying galaxy .
we therefore fit separate analytic profiles to the inner and outer surface brightness profiles . for the underlying galaxy
, a sersic profile is fit of the form : @xmath52 $ ] , where a sersic index @xmath9 represents an exponential profile and @xmath10 is a de vaucouleurs law . the best - fit sersic profile is determined by non - linear least - squares fitting to the region @xmath50 ; in this region , contributions from the nucleus and effects of the _ hst _ wfpc2 psf should be negligible .
the resulting profiles are shown in figure [ sbprof ] , and sersic indices and half - light effective radii are listed in table 3 .
best - fitting sersic indices range between @xmath53 .
there is a slight trend towards smaller @xmath54 values ( more closely exponential ) at fainter magnitudes , consistent with that seen in the much larger sample of virgo de surface brightness profiles analyzed by @xcite . at the distance of the virgo cluster , the nuclei of de galaxies
are slightly extended compared to the wfpc2 psf .
we have used the ishape software developed by @xcite to derive shape parameters for these nuclei .
the intrinsic shape of the nuclei was modeled as a plummer profile whose projected intensity scales as @xmath55 ^ 2 $ ] , where @xmath56 is the scale radius of the nucleus which , for this profile , is also the half - light radius .
the ishape software convolves the analytic profile with the wpfc2 f555w psf and a diffusion kernel generated by the tinytim software @xcite , and determines the best - fitting model parameters by minimizing residuals between the model and original two - dimensional image .
nuclear profiles were fit inside the central @xmath57 ( 10 pixels ) and were assumed to be circularly symmetric .
the free model parameters are the effective radius , @xmath56 , the profile normalization , @xmath58 , and a constant background level .
although a constant background level is a good approximation for the non - nuclear component inside the fitting radius , it is more appropriate to subtract the underlying galaxy sersic profile .
therefore , we use the ishape software to fit only the profile shape , @xmath56 , and determine the profile normalization , @xmath59 , such that the total nuclear magnitude equals the luminosity leftover after subtraction of the galaxy sersic profile from the total observed surface brightness profile .
for the five nucleated des , the resulting nuclear profiles are plotted as dotted lines in figure [ sbprof ] ; effective half - light radii and total nuclear magnitudes are given in table 3 .
results of the kinematic analysis of [ vp ] are shown for all galaxies in figure [ vp_fig ] ; the derived kinematic profiles are summarized in table 4 .
the mean line - of - sight velocity and velocity dispersion are plotted as a function of major axis radius in arcseconds ; the radius was measured relative to the peak position of the intensity profile along the esi slit .
the systemic radial velocity of each de was determined from the mean of the velocity data points and subtracted from the velocity profile .
the corrected heliocentric velocities are listed in table 4 and agree , within measurement errors , with previously published radial velocity measurements for vcc 917 , vcc 1073 , vcc 1254 and vcc 1876 @xcite .
the velocity profiles in figure [ vp_fig ] show no evidence for substantial rotation along the major axis of any of the six de systems observed . to quantify the maximum rotation velocity allowed by the data
, we have differenced the average velocities on either side of the major axis of the galaxy and divided by two ( @xmath60 in table 4 ) .
this quantity is not particularly meaningful for galaxies which do not show a coherent rotation curve ; it merely represents an upper limit on rotational motion .
error bars on rotational motion were determined by adding in quadrature the error of the mean velocity on either side of the major axis .
if the observed flattening of these galaxies were determined by rotational motion alone , the expected rotation velocity can be calculated directly from the tensor virial theorem , given the observed velocity dispersion @xcite .
the ratio of the maximum rotational velocity to the average velocity dispersion ( @xmath61 ) is plotted versus ellipticity in figure [ rotation ] and compared to the ratio expected from an isotropic , rotationally - flattened body .
ellipticity and average velocity dispersion are determined outside @xmath62 in order to exclude any contributions from a central nucleus and are listed in tables 2 and 3 , respectively .
the upper limits on @xmath61 determined for these galaxies are significantly smaller than expected if the observed flattenings were due to rotation .
thus , from figure [ rotation ] we conclude that these des are primarily flattened by anisotropic velocity dispersions .
the low rotation velocities of this study are consistent with previous measurements of de velocity profiles in the local group @xcite .
however , recent studies of de kinematics suggest that a fraction of de galaxies are rotationally supported . a rotational velocity of 15 km s@xmath17 was measured by @xcite for fs 76 , a de in the fornax cluster , slightly below the value expected if this system is rotationally supported .
@xcite found that five of their six virgo cluster des rotated with maximum velocities between 15 to 30 km s@xmath17 , placing them on or above the relation for rotational support .
the sixth galaxy presented by @xcite , ic 794 or vcc 1073 , is also in our sample .
their measurement , @xmath63 km s@xmath17 , is consistent with our observation of low rotation velocity , @xmath64 km s@xmath17 .
thus , of the 11 virgo des with measured velocity profiles , 5 have significant rotation velocities and 6 are non - rotating .
we note that the average brightness of the rotating de sample ( @xmath65 , assuming ( b - v ) = 0.8 @xcite ) is slightly brighter than that of our non - rotating des ( @xmath66 ) . in [ fp ] we find that these two populations are also slightly separated in the fundamental plane . we have recently observed a larger sample of virgo de galaxies , some of which have significant rotation velocities and some which do not
. we will explore in more depth the differences between these two classes in a forthcoming paper .
the mean velocity dispersions of the six de galaxies presented in figure [ vp_fig ] lie between 20 and 55 km s@xmath17 , and show a wide range of profile shapes .
although virgo de nuclei are unresolved from the ground , the kinematics of the central nucleus appear to be distinguished from the underlying galaxy .
surprisingly , the central velocity dispersion can be either larger ( vcc 1254 ) or smaller ( vcc 1073 , vcc 452 ) than the surrounding galaxy .
below , we construct dynamical models for each of the observed galaxies in order to explore the range of mass - to - light ratios , velocity dispersion anisotropy and central black hole masses allowed by the observed profiles .
high spatial resolution wfpc2 imaging available for all of the observed des ( [ sb ] ) allows dynamical modeling of the velocity dispersion profiles through the assumption that the stellar mass density is proportional to the luminosity density times some mass - to - light ratio at all radii . solving the spherically symmetric jeans equation ,
the predicted kinematics are convolved through the observational setup , allowing a direct comparison to the observations .
the simplifying assumption of spherical symmetry is justified as more generalized models can not be discriminated against without additional information such as minor axis kinematics or higher order velocity profile moments .
we produce spherical models in which the radius is related to the observed semimajor / semiminor axes and ellipticity of the corresponding de galaxy via the relation : @xmath67 .
the square root of the product of the semi - major and minor axis is a more appropriate quantity than the semi - major axis for these elliptical systems .
the models are based on dynamical software described in more detail by @xcite .
the luminosity density , @xmath68 , is determined for each galaxy by abel transformation of the projected wfpc2 @xmath31-band surface brightness profile measured in [ sb ] . to avoid noise amplification ,
the observed surface brightness profiles were first fit to an arbitrary function , a generalization of the `` nuker law '' @xcite , as shown in figure [ sbprof ] .
the total luminosity density is assumed to be composed of two parts , @xmath69 , representing the central nucleus and underlying galaxy , respectively .
we model the nuclear component as a plummer model with fixed scale length , @xmath56 , and total nuclear luminosity , @xmath70 in the @xmath31 band , as determined for each galaxy from the projected luminosity density in [ sbfits ] .
the luminosity density of a plummer model , taken from @xcite , is : @xmath71}^{-5/2}\ ] ] the luminosity density of the underlying galaxy , @xmath72 , is the total luminosity density minus the contribution from the nucleus .
the mass density of stars , @xmath73 , is modeled as the luminosity density times a mass - to - light ratio . in our models
, we allow two distinct mass - to - light ratios as free parameters , @xmath74 and @xmath75 , for the underlying galaxy and nuclear component .
the mass density distribution of the galaxy component is then : @xmath76 .
the mass density distribution of the nucleus is similarly defined as : @xmath77 .
the total mass density of the system is the sum of these densities : @xmath78 . rearranging terms ,
the total density is modeled as : @xmath79\ ] ] where the total luminosity density @xmath68 is inferred from the _ hst _ wfpc2 surface brightness profile , @xmath80 is given in eqn .
( [ jn ] ) , and the mass - to - light ratios @xmath75 and @xmath74 are free parameters . the case where @xmath81 is equivalent to the mass density of the galaxy being equal to the luminosity density times a constant mass - to - light ratio .
the total gravitational potential , @xmath82 , is obtained by integrating over the mass density determined by eqn .
( [ rho ] ) plus an added possible contribution , @xmath83 , from a central black hole .
the velocity dispersion as a function of radius , @xmath84 , is then obtained by solving the spherically - symmetric jeans equation : @xmath85 where @xmath86 describes the velocity dispersion anisotropy .
models with @xmath87 are isotropic , @xmath88 are tangentially anisotropic , and @xmath89 are radially anisotropic .
the radial velocity dispersion @xmath84 is numerically evaluated for any combination of @xmath90 , @xmath91 , @xmath74 , @xmath75 , and surface brightness profile . in order to compare to observations ,
the velocity dispersions are projected along the line - of - sight .
the projected dispersions are convolved using monte carlo integration with a gaussian kernel to take into account the seeing fwhm ( as determined in [ datared_sec ] ) .
the dispersions are then sampled using the slit width , pixel size , and rebinning scheme specific to each observation .
the predicted dispersion @xmath92 can be compared directly to the observed velocity dispersions @xmath93 over all radial bins @xmath94 by the defined quantity : @xmath95}^2\ ] ] the best - fitting model is determined by minimization of @xmath96 . given the well known degeneracy between the mass profile and velocity anisotropy , and given the quality of our data , we choose not to explore the full range of allowed parameter space .
instead , we consider three limiting cases : ( 1 ) constant mass - to - light ratio and velocity anisotropy , ( 2 ) the same model plus a central black hole , and ( 3 ) models with separate mass - to - light ratios for the nucleus and underlying galaxy and no central black hole .
we first consider models without a central black hole for which the free parameters are a single mass - to - light ratio @xmath97 ( i.e. @xmath98 ) and velocity dispersion anisotropy , @xmath90 , both independent of radius .
for each galaxy , models were run for values of the velocity dispersion anisotropy ranging between @xmath99 .
best fitting values of @xmath90 and @xmath97 were determined by overall minimization of @xmath96 and are listed in table 4 .
formal @xmath100 ( 68% confidence ) error bars are calculated for each individual free model parameter by the variation needed to increase @xmath96 by 1 with respect to its minimum value @xcite .
best - fit models , as well as several representative @xmath90-value models are plotted over the observed data points in figure [ aniso_models ] .
although isotropic models ( @xmath101 ) do not fit the observed profiles in detail , such models do in general reproduce the dip or rise in the central dispersion observed in all six de galaxies .
most of the galaxies are best fit with tangential anisotropy . in some cases , the amount of anisotropy needed to fit the profiles is unphysically large , motivating the models described in
[ nuc_vs_gal_sec ] .
the @xmath31-band mass - to - light ratios determined for our six virgo des range between @xmath102 .
these mass - to - light ratios are plotted against the absolute total magnitude of each galaxy in figure [ fig_ml ] and compared to higher luminosity classical elliptical galaxies of @xcite .
the @xcite galaxies show a clear trend of decreasing mass - to - light ratio towards fainter magnitudes .
the observed des tend to have larger mass - to - light ratios at a given absolute magnitude than expected by extrapolation of this relationship . combining the wfpc2 colors of these de galaxies with the dynamically
determined mass - to - light ratios , it is possible to roughly determine ages and metallicities for these galaxies .
the colors of the six des lie in the range @xmath103 , as measured by @xcite from wfpc2 data averaged inside @xmath104 . approximating
the stellar populations of these des galaxies by a single - burst population @xcite , the ages and metallicities implied by the above combined constraints lie between 5 to 12 gyr and @xmath105 \le 0 $ ] dex , respectively .
this rough calculation suggests that the mass determined from the observed kinematics can be accounted for by stellar populations alone without the need for a significant dark matter component , at least inside the radius of our observations ( @xmath106 effective radius ) .
this does not rule out a significant dark matter component at larger radii .
accurate determination of the ages and metallicities of these galaxies requires a rigorous analysis of their line strengths and will be presented in a forthcoming paper .
we next allow an additional free model parameter in the form of a central black hole with the goal of placing upper limits on the black hole mass allowed by our kinematic data . if , for example , des are the morphologically - transformed remnants of larger progenitor galaxies , limits on the central black hole mass can place potentially interesting constraints on such a progenitor population .
models were run for a two - dimensional grid of velocity anisotropy versus black hole mass . for each grid point , the best fitting mass - to - light ratio is determined .
contours of constant @xmath107 are shown for these two parameters in figure [ bh_models ] .
confidence intervals were assigned to @xmath107 values in a two dimensional parameter space , as discussed in @xcite .
black hole masses greater than @xmath108 can be ruled out at the 99.9% confidence level . for most objects ,
a zero mass black hole model is either the best fitting model , or statistically similar to the best fit at the 90% confidence level ( @xmath109 ) .
the galaxy vcc 1254 is the only de in which a non - zero black hole mass , @xmath110 , is a significantly better fit to the data than models without a central black hole .
this does not necessarily imply the presence of a black hole , as we will show below that this profile is equally well fit by a model in which the mass - to - light ratio of the nucleus is larger than the underlying galaxy .
in addition , although the upper limits on black hole mass determined in this section are robust , actual black hole mass determinations would require more complicated models which allow @xmath90 variations with radius .
upper limits on central de black hole mass are compared , in figure [ fig_sigbh ] , to the black hole mass - bulge velocity dispersion relationship , @xmath111 , derived for bulge - dominated galaxies @xcite .
although this relationship may not be applicable to de galaxies , which lack a bulge component , our upper limits are still consistent with the relationship .
the implication of these upper limits on possible de progenitor galaxies is discussed in [ disc_sec ] .
the kinematic profiles of the five nucleated des in figure [ vp_fig ] show that the nuclei tend to have velocity dispersions distinct from the surrounding galaxy .
in addition , photometric studies suggest that de nuclei tend to have different colors than the underlying light of the host de galaxy @xcite . motivated by these observations , we consider models allowing two distinct mass - to - light ratios , one for the nucleus , @xmath112 , and another for the underlying galaxy , @xmath74 .
we explore only isotropic models ( @xmath113 ) and search for combinations of @xmath74 , @xmath75 which minimize @xmath96 . for vcc 1254 ,
in which the nuclear dispersion is larger than the surrounding galaxy , the best fit nuclear mass - to - light ratio is twice that for the surrounding galaxy ( @xmath114 ) .
this model fits the data equally well as the models presented in the previous two sections .
however , a larger nuclear mass - to - light ratio implies an older , and therefore redder , stellar population , contrary to observations that the nucleus of vcc 1254 is bluer than the surrounding galaxy @xcite .
the true dynamical state of vcc 1254 is likely to lie between the three extreme models presented in this and previous sections .
for the four nucleated galaxies in which the velocity dispersion dips in the central regions ( vcc 452 , vcc 1073 , vcc 1577 , and vcc 1876 ) , a smaller nuclear mass - to - light ratio ( @xmath115 ) is only a marginally better fit to the data . in these four systems , we are not able to directly constrain the nuclear mass - to - light ratio .
even the unphysical case in which the nucleus contributes no mass , isotropic models are inadequate fits to the profiles of vcc 452 and vcc 1073 , implying that these galaxies must have some degree of tangential velocity anisotropy .
isotropic , single mass - to - light ratio models are adequate fits to the profiles of vcc 1577 and vcc 1876 , and variations in @xmath75 do not significantly improve the fit .
this can be understood because the nuclear component does not dominate the observed spectroscopic light of these galaxies , not even in the central data point . inside @xmath116
, the nucleus contributes between 5% and 25% of the total light in this region , as compared to 60% for vcc 1254 .
thus , the measured central velocity dispersion is not a good estimate of the velocity dispersion of the nucleus . in order to place these nuclei on the fundamental plane ( see [ fp_nuc ] ) , we do need an estimate of the nuclear velocity dispersions .
for this , we assume a nuclear mass - to - light ratio equal to the galaxy ( @xmath117 ) and calculate the central projected velocity dispersion for a plummer model @xcite with total luminosity and scale radius as determined in
[ sbfits ] .
the resulting nuclear velocity dispersions are given in table 4 .
in the multivariate space defined by central velocity dispersion , @xmath118 , effective surface brightness , @xmath119 , and effective radius , @xmath120 , de galaxies occupy a region of the so - called fundamental plane distinct from classical elliptical galaxies .
the separation is best demonstrated by the @xmath121-space projection of this parameter space defined by @xcite as : @xmath122 + \log r_{\rm eff } ) / \sqrt2\ ] ] @xmath123 + 2 \log i_{\rm eff } - \log r_{\rm eff})/\sqrt6\ ] ] @xmath124 - \log i_{\rm eff } - \log r_{\rm eff})/\sqrt3\ ] ] where @xmath125 is defined as @xmath126 and is the mean intensity inside the radius @xmath120 .
these coordinates are related to physical quantities as follows : @xmath127 is proportional to the logarithm of mass , @xmath128 is proportional to the effective surface brightness times mass - to - light ratio and @xmath129 is proportional to the logarithm of mass - to - light ratio . to compare the location of our des in the fundamental plane to other galaxy types , we plot data compiled by @xcite for classical ellipticals , spiral bulges , previously observed des , dwarf spheroidals and globular clusters .
these data have been compiled in the @xmath130-band . for comparison ,
we transform our de data to the @xmath130-band assuming @xmath131 @xcite .
in addition , we add four of the five rotating de galaxies presented by @xcite for which photometric data is available .
photometric properties for ugc 7436/vcc 543 were determined from wfpc2 imaging as in [ sbfits ] .
the properties of the remaining objects were taken from @xcite . as seen in the edge - on , @xmath127 vs. @xmath129 , view of the fundamental plane ( lower left panel , fig .
[ fp_fig ] ) dwarf ellipticals appear to lie in a plane parallel to , but offset from classical ellipticals . in the face - on ,
@xmath127 vs. @xmath128 , view ( upper left panel , fig .
[ fp_fig ] ) , des lie in a very different region of this plane , on a sequence perpendicular to the locus of classical ellipticals .
the offset in @xmath129 was first noted by @xcite and can be interpreted as either non - homology between dwarf and classical ellipticals or as a difference in mass - to - light ratios .
we have shown in figure [ fig_ml ] that des tend to have larger mass - to - light ratios at a given absolute magnitude compared to classical ellipticals , favoring the latter interpretation of the @xmath129 offset . comparing our non - rotating des to the rotating des of @xcite
, these two groups lie in slightly different regions of the fundamental plane .
the rotating des lie at larger @xmath127 and @xmath129 than the non - rotating sample , suggesting that they have both higher masses and mass - to - light ratios . from the location of the rotating des in the fundamental plane , these galaxies are not part of the low luminosity extension of classical ellipticals known to be rotationally - supported @xcite .
thus , des appear to have a wider range of rotational properties than previously assumed .
the separation in both luminosity and fundamental plane space between these two samples suggests a correlation between rotation and another physical quantity , possibly mass . however , more data is required to establish such a correlation . in the right panels of figure [ fp_fig ] ,
the nuclei of the five observed nucleated des are plotted . unlike the underlying de galaxies in the fundamental plane , de nuclei lie nearest to the region occupied by globular clusters .
these @xmath121-space parameters were calculated using central velocity dispersions determined directly from a plummer model fit to the nucleus .
this is a more accurate estimate of the nuclear velocity dispersion than the measured central velocity dispersion , but requires the assumption that the nuclear mass - to - light ratio equals that of the galaxy , as discussed in [ nuc_vs_gal_sec ] .
this assumption most strongly affects values of @xmath129 .
however , for any reasonable assumed nuclear mass - to - light ratio , the de nuclei lie closest to globular clusters in the all three @xmath121 indices .
the absolute luminosities of these de nuclei ( @xmath132 ) are as bright or brighter than the most luminous galactic globular clusters @xcite .
the nuclear effective radii are also larger than an average galactic globular cluster , but are smaller than the largest known globulars .
the resulting central luminosity densities of de nuclei , determined from the plummer models , are comparable to the average globular cluster central luminosity density .
the position of the best studied de nucleus , that of the local group de ngc 205 , a well - resolved supermassive star cluster of intermediate age and absolute magnitude @xmath133 @xcite , is also plotted on the fundamental plane .
this nucleus lies squarely in the region occupied by globular clusters .
the offset position of the virgo de nuclei , particularly in the face - on view of the fundamental plane ( top right panel fig .
[ fp_fig ] ) , relative to galactic globular clusters is most likely due to larger nuclear masses .
velocity and velocity dispersion profiles are presented for the major axes of six de galaxies in the virgo cluster .
these galaxies do not show evidence for substantial rotation ; upper limits on rotation velocities are well below that expected if these objects were rotationally flattened .
dynamical models for these galaxies suggest mass - to - light ratios in the range @xmath134 .
we argue that such ratios are expected for intermediate to old stellar populations and thus these des do not require significant dark matter inside an effective radius .
our observations do not rule out significant dark matter in des at larger radii as demonstrated by giant elliptical galaxies which exist in massive dark halos , but are not necessarily dark matter dominated at small radii @xcite . in fundamental plane space ,
we find that the virgo de galaxies , similar to previously observed des , lie in a plane parallel to , but offset from , that occupied by normal elliptical galaxies . in this space ,
de nuclei lie near the region occupied by galactic globular clusters .
the origin of nuclei in de galaxies remains an open question . in the present sample
, there is no obvious difference between the single non - nucleated de galaxy ( vcc 917 ) and the underlying galaxies of the observed nucleated des .
the mass - to - light ratio , anisotropy , and photometric parameters measured for vcc 917 are indistinguishable from those determined outside the nucleus of the other five des . however , as a population , non - nucleated de galaxies in the virgo cluster do have different properties .
they are observed to be less spatially concentrated , have lower specific globular cluster frequencies , and , on average , have flatter shapes as compared to nucleated de galaxies @xcite .
proposed scenarios for the origin of de nuclei include the remnant cores of larger stripped galaxies @xcite , the results of gas infall and star formation or the coalescence of several globular clusters whose orbits have decayed to the de center @xcite .
we have shown that the observed de nuclei share many properties with globular clusters , suggesting similar formation processes .
since de galaxies are preferentially found in dense environments , it is likely that galaxy interactions play a large role in their formation and evolution .
the models of @xcite suggest that galaxy harassment in clusters can morphologically transform a spiral galaxy into a dwarf elliptical .
harassment tends to increase internal velocity dispersions , but is less efficient in disrupting rotational motion and is not obviously reconciled with the low rotational velocities observed in the present de sample . if des are the morphologically - transformed remnants of larger progenitor galaxies , a constraint on such a progenitor population is provided by the central black hole mass limits determined in [ bh_sec ] .
the upper limit of @xmath4 for the observed de galaxies implies that any de progenitor must have had a bulge dispersion less then 100 km s@xmath17 , assuming the @xmath111 relation @xcite .
although this is not a stringent constraint on de galaxy formation models , higher spatial resolution kinematics , and therefore more stringent mass limits , could be a significant constraint on such models . as the number of de galaxies with measured internal kinematics increases , their position in the fundamental plane strengthens the conclusion that dwarf and classical elliptical galaxies evolve via very different physical processes . to determine whether dwarf ellipticals
as a galaxy class evolve under homogeneous conditions requires more observations
. a critical question is understanding the apparent dichotomy between the anisotropy - supported des presented in this paper and the rotationally - supported des presented by @xcite and @xcite .
the fact that rotating and non - rotating des appear to form a `` sequence '' in fundamental plane space , with the latter having somewhat lower mean luminosity , mass , and mass - to - light ratio , suggests that these are not two distinct types of de galaxies but rather are part of a continuous family .
larger samples are required to establish what , if any , physical property correlates with the observed rotational velocities and what this implies for de galaxy formation .
we would like to thank dennis zaritsky , ruth peterson and doug lin for help with the keck / hires data discussed in [ esi_hires_comp_sec ] .
we are grateful to bryan miller for making his reduced _ hst _
wfpc2 images available to us and to soeren larsen for help with the ishape software . m.g . acknowledges support from the stsci director s discretionary research fund .
freedman , w. l. , madore , b. f. , gibson , b. k. , ferrarese , l. , kelson , d. d. , sakai , s. , mould , j. r. , kennicutt , r. c. , jr . , ford , h. c. , graham , j. a. , huchra , j. p. , hughes , s. m. g. , illingworth , g. d. , macri , l. m. , & stetson , p. b. 2001 , , 553 , 47 gebhardt , k. , bender , r. , bower , g. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , grillmair , c. , ho , l. c. , kormendy , j. , lauer , t. , magorrian , j. , pinkney , j. , richstone , d. , & tremaine , s. 2000 , , 539 , l13 jones , d. h. , mould , j. r. , watson , a. m. , grillmair , c. , gallagher , j. s. , ballester , g. e. , burrows , c. j. , casertano , s. , clarke , j. t. , crisp , d. , griffiths , r. e. , hester , j. j. , hoessel , j. g. , holtzman , j. a. , scowen , p. , stapelfeldt , k. r. , trauger , j. t. , & westphal , j. a. 1996 , , 466 , 742 tremaine , s. , gebhardt , k. , bender , r. , bower , g. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , grillmair , c. , ho , l. c. , kormendy , j. , lauer , t. r. , magorrian , j. , pinkney , j. , & richstone , d. 2002 , , 574 , 740 lccccccc vcc 452 & 12:21:04.7 & 11:45:18 & de4,n & 0.08 & 15.34 & 0.09 & @xmath135 + vcc 917/ic 3344 & 12:26:32.4 & 13:34:43 & de6 & 0.45 & 13.93 & 0.11 & @xmath136 + vcc 1073/ic 794 & 12:28:08.6 & 12:05:36 & de3,n & 0.30 & 13.82 & 0.09 & @xmath137 + vcc 1254 & 12:30:05.3 & 08:04:29 & de0,n & 0.05 & 14.58 & 0.07 & @xmath138 + vcc 1577 & 12:34:38.4 & 15:36:10 & de4,n & 0.17 & 15.24 & 0.09 & @xmath139 + vcc 1876/ic 3658 & 12:41:20.4 & 14:42:02 & de5,n & 0.23 & 14.65 & 0.10 & @xmath140 + lcrc vcc 452 & 3600 & @xmath141 & 1.1 + vcc 917/ic 3344 & 3600 & 52 & 0.9 + vcc 1073/ic 794 & 3600 & @xmath142 & 0.8 + vcc 1254 & 3600 & 0 & 0.8 + vcc 1577 & 6000 & 20 & 1.0 + vcc 1876/ic 3658 & 3600 & 68 & 1.0 + lccccccc vcc 452 & @xmath143 9.6 ( 0.71 ) & 22.3 & 1.6 & 0.15 ( 0.011 ) & 20.7 & 22.54 & @xmath144 + vcc 917/ic 3344 & 12.2 ( 0.90 ) & 21.4 & 2.9 & ... & ... & ... & ... + vcc 1073/ic 794 & 11.1 ( 0.82 ) & 21.1 & 1.9 & 0.13 ( 0.010 ) & 17.4 & 19.86 & @xmath145 + vcc 1254 & 14.4 ( 1.07 ) & 22.4 & 2.9 & 0.17 ( 0.013 ) & 16.7 & 18.67 & @xmath146 + vcc 1577 & 10.5 ( 0.78 ) & 22.4 & 1.1 & 0.16 ( 0.011 ) & 20.1 & 22.23 & @xmath147 + vcc 1876/ic 3658 & 10.5 ( 0.78 ) & 21.8 & 0.8 & 0.11 ( 0.008 ) & 17.9 & 20.92 & @xmath148 + lcccccc vcc 452 & 1380 & @xmath149 & 23.8 & @xmath150 7.7 & @xmath151 & @xmath152 + vcc 917/ic 3344 & 1186 & @xmath153 & 31.1 & ... & @xmath154 & @xmath155 + vcc 1073/ic 794 & 1862 & @xmath156 & 44.6 & 31.7 & @xmath157 & @xmath158 + vcc 1254 & 1220 & @xmath159 & 31.0 & 48.6 & @xmath160 & @xmath161 + vcc 1577 & @xmath150 361 & @xmath162 & 26.8 & @xmath150 9.8 & @xmath163 & @xmath164 + vcc 1876/ic 3658&@xmath150 95 & @xmath165 & 25.7 & 16.5 & @xmath166 & @xmath167 + lcccccc vcc 452 & 1.84 & 2.45 & 0.78 & @xmath168 & 3.33 & 0.87 + vcc 917/ic 3344 & 2.07 & 2.80 & 0.63 & ... & ... & ... + vcc 1073/ic 794 & 2.29 & 3.05 & 0.79 & 0.71 & 4.93 & 0.85 + vcc
1254 & 2.13 & 2.45 & 0.83 & 1.04 & 5.26 & 0.86 + vcc 1577 & 1.93 & 2.45 & 0.82 & 0.03 & 3.61 & 0.85 + vcc 1876/ic 3658 & 1.94 & 2.64 & 0.70 & 0.24 & 4.53 & 0.71 + | spectroscopy with the keck ii 10-meter telescope and echelle spectrograph and imager is presented for six virgo cluster dwarf elliptical ( de ) galaxies in the absolute magnitude range @xmath0 .
the mean line - of - sight velocity and velocity dispersion are resolved as a function of radius along the major axis of each galaxy , nearly doubling the total number of des with spatially - resolved stellar kinematics .
none of the observed objects shows evidence of strong rotation : upper limits on @xmath1 , the ratio of the maximum rotational velocity to the mean velocity dispersion , are well below those expected for rotationally - flattened objects .
such limits place strong constraints on de galaxy formation models .
although these galaxies continue the trend of low rotation velocities observed in local group des , they are in contrast to recent observations of large rotation velocities in slightly brighter cluster des . using surface photometry from _ hubble space telescope _ wide field planetary camera 2 images and spherically - symmetric dynamical models , we determine global mass - to - light ratios @xmath2 .
these ratios are comparable to those expected for an old to intermediate - age stellar population and are broadly consistent with the observed @xmath3 colors of the galaxies .
these de galaxies therefore do not require a significant dark matter component inside an effective radius .
we are able to rule out central black holes more massive than @xmath4 .
for the five nucleated des in our sample , kinematic and photometric properties were determined for the central nucleus separately from the underlying host de galaxy .
these nuclei are as bright or brighter than the most luminous galactic globular clusters and lie near the region of fundamental plane space occupied by globular clusters . in this space ,
the virgo de galaxies lie in the same general region as local group and other nearby des , although non - rotating des appear to have a slightly higher mean mass and mass - to - light ratio than rotating des ; the de galaxies occupy a plane parallel to , but offset from , that occupied by normal elliptical galaxies . |
we study the dynamics of two - color beams in a nonlinear isotropic medium .
we take the initial conditions to be a one - color beam with gaussian profile . with a cubic ( @xmath0 ) nonlinearity , which is the lowest order possible in an isotropic medium , the possible nonlinearities are thg , kerr , and raman .
any other nonlinearity requires either a different susceptibility ( second - harmonics , high - harmonics ) , or more than two slowly - varying envelopes ( sum and difference frequency generation ) .
this is one of the simplest nonlinear optics problems , and is for that reason important ; it has an application to third - harmonic generation ( thg ) microscopy @xcite .
the usual model of thg takes the fundamental beam to be gaussian , and ( inconsistently , or as an approximation ) has thg as the sole nonlinearity @xcite .
this model has exact analytic solution in which the third - harmonic ( th ) beam takes a gaussian profile . for a phase - mismatch that is zero or negative
, the energy in the th peaks at the focus of the fundamental beam , and , after the focus , destructive interference causes _ all _ the th to be re - absorbed by the fundamental . the latter is often said to be due to the guoy shift @xcite , the phase - shift of @xmath1 that a gaussian beam experiences in going from a far - field , through a focus , to a far - field .
but , in fact , the details show it to depend on the beam s phase and shape everywhere along its path . however , the fast ( electronic ) nonlinear response of optical materials that yields thg also generates kerr effects ( self- and cross - phase modulation ) .
therefore , one should include kerr and perhaps also raman effects when modeling thg .
these can have quite drastic effects on a beam @xcite .
although it was recognized as early as 1973 that kerr effects could influence thg @xcite , and although kerr , thg , and dispersion have been examined in studies _ without _ transverse spatial dynamics @xcite , the results of transverse nonlinear dynamics on thg have not heretofore been studied in quantitative detail . we have derived a set of coupled nonlinear schrdinger ( nls ) equations for two slowly - varying envelopes , with the relevant nonlinearities treated rigorously and consistently @xcite . because of the microscopy application , we are interested in very tightly focused beams and pulses , with large momentum and frequency spreads .
the nls equations are thus given to all orders in dispersion and diffraction ; numerical simulations are carried out with a method that is accurate to all orders relevant for the grid @xcite .
we express the field as a fundamental , @xmath2 , centered about a carrier wave at frequency @xmath3 , and a th , @xmath4 , centered about a carrier wave at frequency @xmath5 , [ nls : fund_and_th ] @xmath6 a_{\omega } \nonumber \\ & & + \sum_{p , m=0}^{\infty } 2\pi \frac{i^p \ , ( 2 m - 1)!!}{(-2)^m \ , m ! \ , p ! } \left ( \frac{\partial^p}{\partial\omega^p } \frac{(\omega / c)^2}{\beta(\omega)^{2 m+1 } } \right)_{\omega_0 } \!\!\ !
\triangle_\perp^m \partial_t^p \ , p_{\omega } ( { { \bf x}},t ) \ ; , \label{nls : fundamental } \\
-i \partial_z a_{3\omega}({{\bf x}},t ) & = & \left [ \sum_{p=1}^{\infty } \frac{i^p}{p ! } \left ( \frac{\partial^p}{\partial\omega^p } \beta(\omega ) \right)_{\!\!3\omega_0 } \!\!\!\!\ ! \partial_t^p - \sum_{m=1}^{\infty } \frac{(2 m - 3)!!}{(-2)^{m } \ , m ! } \frac{\triangle_\perp^m}{\beta(3\omega_0)^{2 m-1 } } - \!\ ! \sum_{p , m=1}^{\infty } \frac{i^p \ , ( 2 m - 3)!!}{(-2)^{m } \ , m ! \ , p ! } \left ( \frac{\partial^p}{\partial\omega^p } \frac{1}{\beta(\omega)^{2 m-1 } } \right)_{\!\!3\omega_0 } \!\!\!\!\ ! \triangle_\perp^m
\partial_t^p \right ] a_{3\omega } \nonumber \\ & & + \sum_{p , m=0}^{\infty } 2\pi \frac{i^p \ , ( 2 m - 1)!!}{(-2)^m \ , m ! \ , p ! } \left ( \frac{\partial^p}{\partial\omega^p } \frac{(\omega / c)^2}{\beta(\omega)^{2 m+1 } } \right)_{\!\!3\omega_0 } \!\!\!\!\ !
\triangle_\perp^m \partial_t^p \ ,
p_{3\omega } ( { { \bf x}},t ) \ ; .
\label{nls : th}\end{aligned}\ ] ] in the nls equations ( [ nls : fund_and_th ] ) , @xmath7 is the transverse laplacian , @xmath8 is the index of refraction at frequency @xmath9 , and @xmath10 is the wavenumber .
the first summation terms on the right - hand sides ( rhs ) of eqs .
( [ nls : fund_and_th ] ) are dispersion , the second summation is diffraction , the third summation is cross - dispersion / diffraction .
the nonlinear polarization includes thg , kerr , and raman effects @xcite .
self - steepening terms @xcite , of the form @xmath11 and self - frequency shifting terms of the form @xmath12 are contained in the first - order time derivatives of the nonlinear polarizations @xmath13 and @xmath14 in eqs .
( [ nls : fund_and_th ] ) .
the nls equations ( [ nls : fund_and_th ] ) show that self - steepening terms and self - frequency shifting terms are just the first terms of a family of higher order nonlinear terms .
the nonlinear polarization is taken to be of the form [ nonlinear_polarization : yehuda ] @xmath15 z \ } \ ; a_{\omega}({{\bf x}},t)^ { * 2 } a_{3\omega}({{\bf x}},t ) \nonumber \\ & + & 3\chi^{\rm elec}(-\omega_0;\omega_0,-\omega_0,\omega_0 ) \ ; |a_{\omega}({{\bf x}},t)|^2 a_{\omega}({{\bf x}},t ) + 6\chi^{\rm elec}(-\omega_0;3\omega_0,-3\omega_0,\omega_0 ) \ ; |a_{3\omega}({{\bf x}},t)|^2 a_{\omega}({{\bf x}},t ) \nonumber \\ & + & a_{\omega}({{\bf x}},t ) \int_0^\infty \!\!\ !
[ 3\chi^{\rm nucl } ( -\omega_0;\omega_0,-\omega_0,\omega_0{\bf ; } \ ; s ) \ ; |a_{\omega}({{\bf x}},t - s)|^2 + 3\chi^{\rm nucl } ( -\omega_0;\omega_0,-3\omega_0,3\omega_0{\bf ; } \ ; s ) |a_{3\omega}({{\bf x}},t - s)|^2 ] ds \nonumber \\ & + & a_{3\omega}({{\bf x}},t ) \int_0^\infty \!\!\ !
\exp(-2 i \omega_0 s ) 3\chi^{\rm nucl } ( -\omega_0;3\omega_0,-\omega_0,-3\omega_0{\bf ; } \ ; s ) a_{\omega}({{\bf x}},t - s ) a_{3\omega}({{\bf
x}},t - s)^ * ds \\ p_{3\omega}({{\bf
x}},t ) & = & \chi^{\rm elec}(-3\omega_0;\omega_0,\omega_0,\omega_0 ) \ ; \exp\{i [ 3\beta(\omega_0 ) - \beta(3\omega_0 ) ] z \ } \ ; a_{\omega}({{\bf x}},t)^3 \nonumber \\ & + & 6\chi^{\rm elec}(-3\omega_0;\omega_0,-\omega_0,3\omega_0 ) \ ; |a_{\omega}({{\bf x}},t)|^2 a_{3\omega}({{\bf x}},t ) + 3\chi^{\rm elec}(-3\omega_0;3\omega_0,-3\omega_0,3\omega_0 ) \ ; |a_{3\omega}({{\bf x}},t)|^2 a_{3\omega}({{\bf x}},t ) \nonumber \\ & + & a_{3\omega}({{\bf x}},t ) \int_0^\infty \!\!\ !
[ 3\chi^{\rm nucl } ( -3\omega_0;\omega_0,-\omega_0,3\omega_0{\bf ; } \ ; s ) \ ; |a_{\omega}({{\bf x}},t - s)|^2 + 3\chi^{\rm nucl } ( -3\omega_0;3\omega_0,-3\omega_0,3\omega_0{\bf ; } \ ; s ) \ ; |a_{3\omega}({{\bf x}},t - s)|^2 ] ds \nonumber \\ & + & a_{\omega}({{\bf x}},t ) \int_0^\infty \!\!\ !
\exp(2 i \omega_0 s ) \ , 3\chi^{\rm nucl } ( -3\omega_0;\omega_0,-\omega_0,3\omega_0{\bf ; } \ ; s ) \ ; a_{\omega}({{\bf x}},t - s)^ * a_{3\omega}({{\bf x}},t - s ) ds \ ; .\end{aligned}\ ] ] this breaks up the nonlinear response into an electronic ( fast ) part and a nuclear ( slow ) part .
it also assumes that the electronic part of the response may be considered frequency - independent on the scale of the pulse bandwidths ; it does not make this assumption for the nuclear part of the response .
for the calculations , we take a more specific nonlinear polarization , essentially the standard model of the third - order susceptibility of fused silica @xcite plus a generalization , [ nonlinear_polarization : kerr_raman_thg ] @xmath16 z \ } \ ; \chi^{\rm thg } a_{\omega}({{\bf x}},t)^ { * 2 } a_{3\omega}({{\bf x}},t ) + \chi^{\rm elec } ( |a_{\omega}({{\bf x}},t)|^2 + 2 |a_{3\omega}({{\bf x}},t)|^2 ) a_{\omega}({{\bf x}},t ) \nonumber \\ & & + a_{\omega}({{\bf x}},t ) \int_0^\infty \!\!\ !
\chi^{\rm nucl}(s ) ( |a_{\omega}({{\bf x}},t - s)|^2 + |a_{3\omega}({{\bf x}},t - s)|^2 ) ds \nonumber \\ & & + a_{3\omega}({{\bf
x}},t ) \int_0^\infty \!\!\ ! \exp(-2
i \omega_0 s ) \chi^{\rm nucl}(s ) a_{\omega}({{\bf x}},t - s ) a_{3\omega}({{\bf x}},t - s)^ * ds \ , , \\
p_{3\omega}({{\bf x}},t ) & = & \exp\{i [ 3\beta(\omega_0 ) - \beta(3\omega_0 ) ] z \ } \ ; \frac{1}{3 } \chi^{\rm thg }
a_{\omega}({{\bf x}},t)^3 + \chi^{\rm elec } ( 2 |a_{\omega}({{\bf
x}},t)|^2 + |a_{3\omega}({{\bf x}},t)|^2 ) a_{3\omega}({{\bf x}},t ) \nonumber \\ & & + a_{3\omega}({{\bf x}},t ) \int_0^\infty \!\!\ ! \chi^{\rm nucl}(s ) ( |a_{\omega}({{\bf x}},t - s)|^2 + |a_{3\omega}({{\bf x}},t - s)|^2 ) ds \nonumber \\ & & + a_{\omega}({{\bf x}},t ) \int_0^\infty \!\!\ ! \exp(2 i \omega_0 s ) \chi^{\rm nucl}(s ) a_{\omega}({{\bf
x}},t - s)^ * a_{3\omega}({{\bf x}},t - s ) ds \ , .\end{aligned}\ ] ] here the thg coefficient is decoupled from the other electronic susceptibilities ; this is outside the usual model for pure fused silica in which the electronic contribution is considered instantaneous and the nuclear contrubution takes the form arising from a single damped harmonic oscillator , @xmath17 , with @xmath18 , @xmath19 , @xmath20 , @xmath21 m@xmath22/w , @xmath23 , @xmath24 fs , and @xmath25 fs . if the electronic thg susceptibility differs from the other electronic susceptibilities , the electronic contribution to the susceptibility is not instantaneous compared to all scales .
direct experimental measurements of the thg susceptibilities are available @xcite .
the more recent measurement of the thg coefficient @xmath26 @xcite is smaller than the electronic contribution to self - phase modulation @xcite @xmath27 by a factor of almost four ; older measurements @xcite give a thg coefficient smaller than the electronic part of the self - phase modulation by a factor of about @xmath28 . in the absense of direct experimental measurements of all the nonlinear polarization coefficients for doped silica , we use the simplest case @xmath29 in numerical simulations , but discuss how the results scale for different values of the thg coefficient .
vector effects are neglected .
inter - band raman scattering is negligible _ at the carrier frequencies _ because of the fast relative phase oscillation ; but in the pulse simulations , inter - band raman scattering is possible between the lower frequencies within the higher frequency band , and the higher frequencies within the lower frequency band .
a numerical nls propagation scheme may be said to be accurate to all orders of dispersion and diffraction if it is accurate to as many orders as there are grid points .
accuracy up to all available orders requires the index of refraction over the entire numerically represented frequency range . in the split - step
fast fourier transform scheme , linear propagation is carried out in momentum space ; the algorithm may be made accurate to all orders by putting the frequency dependent index of refraction directly into the formulas for the propagators @xcite , i.e. , @xmath30 \ , { \cal f } \ { a_\omega({{\bf x}},t ) \ } \right\ } + { \cal f}^{-1 } \left\ { \frac{2\pi \ , ( \omega_0+\omega)^2 / c^2 } { \sqrt{\beta(\omega_0+\omega)^2 - k_\perp^2 } } { \cal f }
\ { p_\omega ( { { \bf x}},t ) \ } \right\ } \ ; , \\
-i \partial_z a_{3\omega}({{\bf x}},t ) & = & { \cal f}^{-1}\!\ ! \left\ { \left [ \sqrt{\beta(3\omega_0+\omega)^2 - k_\perp^2 } - \beta(3\omega_0 ) \right ] { \cal f } \ { a_{3\omega}({{\bf x}},t ) \ } \right\ } + { \cal f}^{-1}\!\ ! \left\ { \frac{2\pi \ , ( 3\omega_0+\omega)^2 / c^2 } { \sqrt{\beta(3\omega_0+\omega)^2 - k_\perp^2 } } { \cal f } \ { p_{3\omega } ( { { \bf x}},t ) \ } \right\ } \ , .\end{aligned}\ ] ] here @xmath31 and @xmath32 are fourier and inverse fourier transforms in @xmath33 , @xmath34 is the transverse momentum , and @xmath9 is frequency .
the linear dispersion effects are contained in @xmath35 ; diffraction is contained in @xmath36 .
the offsets by @xmath3 and @xmath5 are due to the fact that the fields are slowly - varying envelopes about carrier waves at those frequencies ; when these appear , @xmath9 is the frequency relative to the offset . there is no need to compute the coefficients for dispersion , diffraction , self - steepening , _
etc_. , explicitly because they are contained implicitly in the linear dispersion @xmath37 .
since we simulate focusing and collapse , during which diffractive , dispersive , and nonlinear length scales can easily change by factors of a hundred or more @xcite , we allow the propagation step to vary such that it remains an order of magnitude less that the smallest relevent scale .
the effects that we are interested in are larger and clearer when the fundamental and th are not too far from being phase - matched , @xmath38 , or @xmath39 .
optical materials may be doped to obtain desired properties @xcite .
we consider silica doped with neodynium to obtain approximate phase - matching between the fundamental @xmath40 m and th @xmath41 m .
to model the frequency dependent index of refraction , we use the sellmeier relation for fused silica @xcite , but add one additional resonance at @xmath42 m , which is neodynium s largest resonance in the vicinity of our th . for this , phase - matching is achieved at the sellmeier coefficient @xmath43 , which corresponds to a few percent doping of the material . phase - mismatch is varied by changing the dopant concentration ( in this model , the sellmeier coefficient @xmath44 ) .
we take the conventional nonlinear coefficients for pure silica @xcite , as given above .
we simulated the propagation of both pulses and continuous beams , over a range of light intensities , phase - mismatch values , and focusing strengths .
the effects of the nonlinearity are clearly visible in fig .
[ fig1 ] , which shows the evolution of beams with varying intensity . for all simulations
, we took initial conditions with zero power in the th , and the fundamental in the form of a gaussian with some radial phase factor @xmath45 $ ] .
the radial phase factor @xmath46 can be related to the wavefront radius of curvature @xmath47 via the magnitude of the wavevector @xmath48 according to the formula @xmath49 . because the fundamental focuses very intensely , while only a relatively small part of it is converted to th , we illustrate the _ peak intensity _ of the fundamental beam and the _ power _ of the th beam . in fig .
[ fig1 ] , the input beam has initial width fwhm = @xmath50 m , a radial phase factor @xmath51 that brings the beam to a focus in 0.6 mm , and the material is phase - matched ( the reciprocal of the phase - mismatch is more than an order of magnitude greater than the simulation distance ) . the fact that the _ normalized _ th power varies with fundamental input intensity shows that the beams experience nonlinear dynamics , changing shape as the intensity changes . in fig .
[ fig1 ] , the initial wave front radii of curvature are small and their effect on the position of the focus overshadow the intensity dependence .
one can see that the beams approach the linear limit at low intensities , as these curves start to almost overlap .
[ fig2 ] shows the normalized _ far - field _ th power as a function of input power , for the data in fig . [ fig1 ] and for another set of runs with weaker focusing .
for a qualitative explanation of these figures , we first contrast our results with the model in which thg is the only nonlinearity @xcite . here ,
in the case with phase - matching , the th power @xmath52 $ ] reaches a maximum at the focus of the fundamental and then drops off as the inverse square of @xmath53 , where @xmath54 is the rayleigh range of the fundamental _ and _ the th beam , and @xmath55 is the width of the fundamental at its focus .
our simulations show that at high intensities , the kerr effect causes the beams to lose reflection symmetry about the focal plane . with this asymmetry ,
thg from the incoming and outgoing beams does not fully interfere destructively , and allows some th to propagate to the far - field .
the lower - intensity curves in fig .
[ fig1 ] do not quite drop off as the inverse square of distance because the numerical simulations did not start out from minus infinity , but began with merely a large ( finite ) beam width .
the reflection asymmetry in fig .
[ fig1 ] is due partly to the nonlinearities and partly to starting with a finite initial beam width .
as long as the amount of energy in the th is relatively small , the nonlinear dynamic effects will remain in the fundamental beam , and the th peak intensities and beam powers may scale up or down by a uniform factor , but will be otherwise unaffected .
clearly , there is significant variation with the fundamental input power .
moreover , this variation is quite different for different focusing conditions . in another series of simulations ,
phase - mismatch is varied .
[ fig3 ] shows the peak intensity of the fundamental and the th beam power with initial fwhm = @xmath56 m , radial phase - factor @xmath57 , and phase - mismatch from -12.9 to 51.8 mm@xmath58 . because the th power first increases and then decreases with phase - mismatch , we show the former range on one plot and the latter on another .
clearly , there is a residual far - field th and its power depends on phase - mismatch .
[ fig4 ] shows the far - field th conversion efficiency as a function of phase mismatch , for the runs in fig .
[ fig3 ] and another set of runs with weaker focusing , @xmath59 . as in the linear model @xcite , thg by a focusing beam
is maximized around a certain phase - mismatch ; but the nonlinear dynamics complicate the results considerably . in a further series of simulations , the intensity and phase - mismatch were held constant , and the radial phase factor varied .
this mixes up a few physical effects , since initial conditions with small radial phase factors can not be considered as starting from the far - field .
[ fig5 ] has extremely small phase - mismatch ( @xmath60 mm@xmath58 ) , and radial phase factors which vary from zero ( i.e. , starting at a focus ) up to @xmath61 .
the horizontal axis is on a log scale to help visually distinguish the superimposed sumulations .
the peak fundamental intensity varies , but over the range studied , the peak th power is rather insensitive to the wavefront radius of curvature of the fundamental , but the far - field th power does vary considerably .
[ fig6 ] shows the far - field th power as a function of radial phase factor , for the runs in fig .
[ fig5 ] , and also for a series of runs with significant positive and negative phase - mismatch . in increasing ( decreasing ) the tightness of the focusing , the nonlinear and diffraction lengths both decrease ( increase )
; the complexity of the results for far - field thg reflect the complexity of the nonlinear dynamics of the beam .
[ fig7 ] shows the dependence of the third - harmonic power on the when the thg suseptibility @xmath26 is varied .
the curve labeled @xmath62 corresponds to the same conditions used in fig .
[ fig3 ] but with @xmath60 mm@xmath58 ( i.e. , almost phase - matched ) .
the third harmonic power scales with @xmath63 when the third - harmonic intensity is small since then the nonlinear dynamics of the fundamental is unaffected by th and the generation of th field is proportional to @xmath26 .
the values of @xmath26 used in the calculations shown in fig .
[ fig7 ] are a factor of @xmath64 , @xmath65 , and @xmath66 times the value used in the previous figures , where the latter two factors correspond to the measured values of @xmath67 reported in refs .
@xcite numerical simulations showed the dynamics of pulses to differ from those of continuous waves in essentially two ways .
first , the group velocities of the fundamental and th will , except for special cases , not be the same .
a th pulse will thus generally walk off from a fundamental pulse .
this tends to reduce destructive interference ; it also limits the effective distance over which the pulses interact . compared to beams ,
th pulses tend to carry off a larger part of the fundamental pulse energy .
secondly , the dynamics of pulses ( 3 + 1dimensional ) are qualitatively different to those of continuous beams ( 2 + 1dimensional ) . for strongly focused but not very short pulses , these differences tend to be minor .
a thorough analysis of focused pulses in this system is quite involved , and will be pursued elsewhere .
thus , the detailed results for continuous - wave beams apply to pulses for which group - velocity differences are relatively small or for pulses that are relatively long .
in a nonlinear medium , intense beams or pulses of finite diameter which converge to a focus and then diverge may exhibit nonlinear dynamics that significanly affect propagation dynamics .
these nonlinear effects break the beam s reflection symmetry about the focal plane .
the greater the intensity , the bigger the difference between incoming and outgoing beams .
thg with such an input beam or pulse produces th in the far - field when the phase - mismatch between the fundamental and its third - harmonic is zero or negative ( and small ) .
we have quantitatively demonstrated this for several cases .
when phase - mismatch is positive , where some far - field th power is possible in the essentially linear case , the nonlinear beam dynamics complicate the accumulation of th power .
the non - zero far - field thg for zero or negative phase - mismatch a qualitatively new effect for homogeneous media is affected by , and thus contains information about the medium in the region of the beam focus . for an inhomogeneous medium , and for thg
microscopy , these effects should be understood , either to be utilized or better avoided .
richard tasgal gratefully acknowledges a kreitman foundation fellowship .
this work was supported in part by a grant from the israel science foundation for a center of excellence ( grant no .
8006/03 ) and by kbn as a reasearch grant 2003 - 2006 ( 2p03b04325 ) .
99 y. barad , h. eisenberg , m. horowitz , and y. silberberg , appl .
70 * , 922 ( 1997 ) ; j.m .
schins , t. schrama , j. squier , g.j .
brakenhoff , and m. muller , j. opt .
b * 19 * , 1627 ( 2002 ) .
ward and g.h.c .
new , phys .
rev . * 185 * , 57 ( 1969 ) .
boyd , _ nonlinear optics _ ( academic : ny , 1992 ) .
l. berg , phys . rep .
* 303 * , 259 ( 1998 ) .
miles and s.e .
harris , ieee j. quantum electron .
* qe-9 * , 470 ( 1973 ) .
band , phys .
a * 42 * , 5530 ( 1990 ) .
sammut , a.v .
buryak , and y.s .
kivshar , j. opt .
b * 15 * , 1488 ( 1998 ) .
tasgal and y.b .
band ( unpublished ) .
m. matuszewski , w. wasilewski , m. trippenbach , and y.b .
band , opt .
* 221 * , 337 ( 2003 ) ; m. trippenbach , w. wasilewski , p. kruk , g.w .
bryant , g. fibich and y.b .
band , opt .
comm . * 210 * , 385 ( 2002 ) . j.e .
rothenberg , opt .
* 17 * , 1340 ( 1992 ) .
fleck , j.r .
morris and m.d .
feit , appl .
* 10 * , 129 ( 1976 ) .
agrawal , _ nonlinear fiber optics _
( academic : san diego , 1995 ) .
d. anderson and m. lisak , phys .
a * 27 * , 1393 ( 1983 ) ; p.v .
mamyshev and s.v .
chernikov , opt .
lett . * 15 * , 1076 ( 1990 ) ; g. fibich , phys .
lett . * 76 * , 4356 ( 1996 ) .
m. trippenbach and y.b .
band , phys .
a * 56 * , 4242 ( 1997 ) .
hellwarth , prog . quantum electron . * 5 * , 1 ( 1977 ) .
stolen , j.p .
gordon , w.j .
tomlinson , and h.a .
haus , j. opt .
b * 6 * , 1159 ( 1989 ) . k.j . blow and d.
wood , ieee j. quantum electron . * 25 * , 2665 ( 1989 ) .
d. milam , appl . opt . * 37 * , 546 ( 1998 ) .
u. gubler and c. bosshard , phys .
b * 61 * , 10702 ( 2000 ) .
a. mito , k. hagimoto , and c. takahashi , nonlinear opt .
* 13 * , 3 ( 1995 ) ; c. bosshard , u. gubler , p. kaatz , w. mazerant , and u. meier , phys . rev .
b * 61 * , 10688 ( 2000 ) .
nicacio , e.a .
gouveia , n.m .
borges , and a.s .
gouveia - neto , appl .
* 62 * , 2179 ( 1993 ) .
malitson , j. opt .
55 * , 1205 ( 1965 ) .
. peak intensity of the fundamental ( @xmath68 m ) beam , and power of the third - harmonic ( @xmath69 m ) normalized by the cube of the input power .
the initial conditions have a range of intensities ( 1.01 , 4.05 , 16.2 , 36.4 , 64.7 , 101 , 122 and 145 kw , corresponding to circled points in the appropriate curve in fig . [ fig2 ] ) , but identical beam width , fwhm = @xmath56 m , radial phase factor , @xmath51 ( i.e. , transverse phase @xmath70 $ ] ) , and all are phase - matched .
+ fig . 2 .
far - field third - harmonic beam power , normalized by the cube of the input power , _ vs_. input power .
the medium has th phase - matching .
two curves are shown , one the far - field results from fig . 1 , and
another with smaller radial phase factor , that brings the beam to a focus in about 1.2 mm . + fig . 3 peak intensity of the fundamental beam , and power of the third - harmonic .
the curves represent beams with identical power , width , and radial phase factor , while the phase - mismatch goes from -12.9 to 51.8 mm@xmath58 ( corresponding to circled points in the appropriate curve in fig . [ fig4 ] ) .
the middle plot shows the th increasing with phase - mismatch , and the bottom plot shows th decreasing with phase - mismatch .
far - field conversion efficiency _
phase - mismatch .
two curves are shown , one the far - field results from fig . 3 , and another with smaller radial phase factor , that brings the beam to a focus in about 1.2 mm .
initial intensities are held constant .
peak intensity of the fundamental beam , and power of the third - harmonic versus propagation distance for different fundamental radial curvatures .
the curves show beams with phase - matching and initially identical power and width ; the initial radial phase factors go from zero ( starting at a focus ) up to @xmath71 , by increments of @xmath72 ( the curves here correspond to circled points in the appropriate curve in fig . [ fig6 ] ) .
far - field conversion efficiency _
vs_. radial phase factor .
two curves are shown , one the far - field results from fig .
5 ( phase - matching ) , and another in a medium with positive phase - mismatching .
peak fundamental intensity and third - harmonic power versus position @xmath73 , for various values of @xmath26 .
the third harmonic power scales with @xmath63 when the third - harmonic intensity is small . | beams that experience third - harmonic generation ( thg ) also experience kerr effects . with kerr effects
, beams do not take simple gaussian shapes , but exhibit nonlinear dynamics .
these nonlinear dynamics have an effect on the thg accumulated by focusing and then diverging beams .
we formulate a self - consistent and complete set of nonlinear schrdinger equations for a pair of coupled beams a fundamental and its third - harmonic .
numerical simulations show that the kerr nonlinearities allow some third - harmonic to propagate to the far - field even for zero or negative phase - mismatch .
this is because the nonlinear dynamics break the beams reflection symmetry about the focal plane and therefore increases far - field thg by changing some of the interference from destructive to constructive .
thg conversion efficiencies are computed as functions of several beam parameters . |
[ sec : intro ] the formation and nature of the first generation of so - called population iii ( pop iii ) stars have been studied and speculated about for over thirty years ( schwarzschild & spitzer 1953 ; yoneyama 1972 ; hartquist & cameron 1977 ; palla , salpeter & stahler 1983 ; silk 1983 ; and many others ) , but the extent to which they differed from present day stars in ways other than composition is still debated . recently , three - dimensional cosmological simulations have reached sufficient resolution on small scales to begin to address the star formation problem ( ostriker & gnedin 1996 ; abel et al . 1998 ) .
while considerable uncertainty remains regarding the continued evolution of the dense knots they find in their calculations , these simulations do not exclude the formation of a first generation of quite massive stars @xmath6 ( _ e.g. _ , larson 1999 ; abel , bryan , & norman 2000 ) . using different numerical methods to analyze the fragmentation of primordial clouds , bromm et al .
( 1999 ) reached similar conclusions . without further study , especially including the complexity of radiation transport and molecular opacities , it may be premature to conclude that _ all _ pop iii stars were so massive , but there certainly exists adequate motivation to examine the properties of such stars .
in particular , if such massive pop iii stars were ever born , would they retain their large masses until death ?
that is , would they die as pair - instability supernovae , with possible unique signatures of nucleosynthesis and black hole formation , or would they lose most of their mass and die much the same as present day stars ?
it is well known ( ledoux 1941 ; schwarzschild & hrm 1959 ) that , above a critical mass , main sequence stars are vibrationally unstable due to the destabilizing effect of nuclear reactions in their central regions . for stars of pop
i composition , the critical mass for this instability is @xmath7 ( schwarzschild & hrm 1959 ) .
according to non - linear calculations , such an instability leads to mass loss rather than catastrophic disruption ( appenzeller 1970a ; talbot 1971ab ; papaloizou 1973ab ) .
consequently , it has been thought that stars more massive than about 100 @xmath8would not survive for long before losing a substantial fraction of their mass .
although the stability of very massive stars has well been explored since the work of ledoux ( 1941 ) , somewhat surprisingly , all these analyses have been devoted to solar - like compositions or slightly metal - poor stars ( schwarzschild & hrm 1959 ; stothers & simon 1968 ; aizenman et al .
1975 ; stothers 1992 ; glatzel & kiriakidis 1993 ; kiriakidis et al . 1993 ) , or to helium - core objects such as wolf - rayet stars ( maeder 1985 ; cox & cahn 1988 ) . since the temperature dependence of hydrogen burning and the surface gravitational potential of low metal stars are quite different , it is important to learn whether their stability criteria may also vary .
we have thus performed , to our knowledge , the first stability analysis of zero metallicity stars with @xmath9 and have also studied the stability of stars in this mass range for variable initial metal mass fraction @xmath10 .
the stellar models are presented in sect .
[ sec : model ] .
the results of the linear stability analysis are given in sect .
[ sec : stalin ] , mass loss estimates are presented in sect .
[ sec : mdot ] , and our conclusions follow in sect .
[ sec : concl ] .
[ sec : model ] the evolution of four stars of 120 , 200 , 300 and 500@xmath8was followed from the zero age main sequence ( zams ) to the end of core helium burning using an implicit lagrangian hydrodynamics code adapted to stellar modeling ( langer et al . 1988 , heger 1998 ; heger et al .
details of the code can be found elsewhere ( heger & langer 2000 ) , but to summarize the main physics , the opacity is taken from iglesias and rogers ( 1996 ) and the principal nuclear reaction rates from caughlan & fowler ( 1988 ) with the rate for @xmath11c(@xmath12o multiplied by a constant 1.7 ( see also buchmann 1996 ) .
mass loss is not taken into account , except for what is driven by the pulsations studied here . for zero metallicity stars ,
radiative mass loss is probably negligible ( kudritzki 2000 ) . as was first noted by ezer and cameron ( 1971 ) , massive stars with no metals
do not obtain their energy on the main sequence from the pp - cycle as one might naively expect .
instead the star contracts to @xmath13 , burns a trace of helium to carbon , and then expands slightly to burn hydrogen by the cno cycle .
a cno mass fraction of just a few times @xmath14 is sufficient to catalyze even the enormous energy generation of such stars .
however , owing to this small cno mass fraction , the star retains a central temperature of @xmath15 throughout central hydrogen burning . as ezer & cameron ( 1971 ) also noted , these stars are thus much more compact than their metal - enriched counterparts and their radiation , while on the main sequence , is in the ultraviolet ( table [ tab : zams ] ) .
figure [ fig : z0-hrd ] _ _ shows the theoretical tracks of our @xmath16 stars in a hertzsprung - russell diagram ( hrd ) .
all begin to contract dynamically after core helium burning and the star ceases to evolve in the hrd due to shorter evolutionary time - scale of the core compared to the thermal time - scale of the envelope .
note however that , for @xmath17 , even metal - free stars evolve toward @xmath18 and eventually become red supergiants , as the hydrogen - burning shell becomes more active with increasing stellar mass . in order to analyze the pulsational stability as a function of metallicity
, we repeated our calculations for a wide range of @xmath0 from @xmath19 to solar , @xmath20 , for the 120 @xmath21 , and for selected values of @xmath22 for @xmath23 .
the results are displayed in fig .
[ fig : z - hrd ] _ _ for a @xmath24 star . since most of the stability analyses mentioned in sect .
[ sec : intro ] ( ledoux 1941 ; schwarzschild & hrm 1959 ; appenzeller 1970a ; and others ) are based upon models with solar metallicity , but computed using only electron scattering opacity , we also calculated , for comparison , models with such an opacity ( fig .
[ fig : z - hrd ] _ _ ) .
[ sec : stalin ] a linear non - adiabatic stability analysis of all models was carried out using a radial pulsation code originally developed by umin lee ( 1985 ; see also heger et al .
1997 ; baraffe et al .
1998 ; alibert et al .
1999 for recent applications ) .
the main uncertainty in these calculations is the treatment of convection , which is treated in the `` frozen - in '' approximation , _
i.e. _ , the perturbation , @xmath25 , of the convective flux is neglected in the linearized energy equation .
although this approximation is an unavoidable consequence of our poor knowledge of the interaction between convection and pulsation , it should not be crucial .
as we shall see , the relevant regions for pulsation , which are located in the deep interior of the stars , are fully convective and can be considered as adiabatic . in the outer regions , which may also be important for the pulsation ,
the convective flux is not the dominant energy transport .
the stability of the fundamental mode and the overtones ( up to the fifth overtone ) was analyzed starting from zams models up to the end of core helium burning for @xmath26 stars .
[ sec : z0 ] [ sec : zams ] table 1 gives , for metal - free zams stars , the effective temperature , @xmath27 , stellar radius , @xmath28 , central temperature , @xmath29 , pulsation period , @xmath30 , and the pulsational @xmath5-folding time , @xmath31(cox 1980 ) . adopting a time dependence for the eigenfunctions of the form @xmath32 , where @xmath33 is the eigen - frequency and @xmath34 , the stability coefficient , the @xmath5-folding time for the growth of the instability is given by @xmath35 .
for all masses studied , the zams models are unstable and oscillate in the fundamental mode , whereas the overtones are stable .
the excitation mechanism is related to the temperature and density dependence of the nuclear energy generation rate near the center of the star , the so - called @xmath2-mechanism .
as was shown by ledoux ( 1941 ) , this driving mechanism is efficient because of the large radiation pressure in such massive stars .
the driving and damping zones are indicated by correspondingly positive and negative values of d@xmath36/d@xmath37 , where @xmath36 is the work integral ( cox 1980 ; unno et al .
this is illustrated in fig .
[ fig : dwdm ] _ _ for the 300@xmath8star .
note that the outer damping regions ( see fig .
[ fig : dwdm]_a _ ) give a negligible contribution to the work integral due to the very small amount of mass involved ( see fig .
[ fig : dwdm]_b _ ) . the same behavior of the work integral is also found for the other masses studied .
[ sec : evol ] it is known from previous studies of solar metallicity stars that high mass stars eventually stabilize as they evolve away from the zams ( schwarzschild & hrm 1959 ; stothers and simon 1968 ; aizenman et al . 1975 ; maeder 1985 ) .
stabilization comes about because of the increase of the mean molecular weight in the inner regions which increases the central condensation ( see , _
e.g. _ , maeder 1985 for details ) .
the same properties are also found here for metal - free stars . as they evolve away from the zams , their @xmath5-folding time increases , indicating a decrease of the driving term due to nuclear energy compared to the damping term from heat leakage in the envelope .
this is a direct consequence of the increase of the central density relative to the mean density .
the stars become stable ( negative @xmath34 ) on the main sequence after a decrease of the central hydrogen mass fraction .
this is illustrated in fig .
[ fig : k ] _ _ , which displays the stability coefficient , @xmath34 , as a function of time for different masses , and in fig .
[ fig : a ] _ _ , which shows the time integration of @xmath34 during core hydrogen burning . in fig .
4 , the larger value of @xmath34 at early times and its rapid variation with time are due to the adjustment of the stellar structure when the initial cno is produced ( see 2 ) .
the results are also summarized for the different models in table 2 which gives the number of @xmath5-foldings , @xmath38 , by which the amplitude of an initial perturbation would grow in the framework of linear stability analysis , as derived from integration of the stability coefficient with time ( see also fig .
[ fig : a ] _ _ ) . after reaching pulsational stability during central hydrogen burning ,
all stars studied remained stable for the rest of central hydrogen burning and most of central helium burning as well .
stars of @xmath39 remained stable throughout core helium burning , but stars of initial mass @xmath40 became unstable again towards the end of helium burning .
this resurgence of instability happens for a central helium mass fraction @xmath41 in the 300@xmath8star and @xmath42 for the 500@xmath8star , corresponding to 30 and 50@xmath43before the end of central helium burning , respectively .
the unstable mode is the fundamental mode , characterized by @xmath44 and @xmath45 for the 300@xmath8star and @xmath46 and @xmath47 for 500@xmath8star .
the pulsations in these blue supergiants , with @xmath48 k , exist only in a surface layer that contains very little mass .
the 500@xmath8star , however , expands further and becomes a red supergiant with an extended convective envelope that is characterized by increasing period and decreasing @xmath5-folding time .
the instabilities during core helium burning are characterized by small values of @xmath31compared to the pulsation period .
this indicates the presence of a strong instability that can develop rapidly and reach large amplitudes .
inspection of the work integral shows that the driving zone is located in the hydrogen / helium ionization zone and the excitation mechanism is the standard @xmath3-mechanism .
note also that the helium cores would quickly become unstable to the @xmath2-instability , just like their higher metallicity counterparts , if the hydrogen envelope were removed , _
e.g. _ , to a binary companion .
[ sec : z ] table [ tab : zams ] gives the pulsational properties of zams models as a function of @xmath0 .
all models analyzed for @xmath22 are vibrationally unstable .
the fundamental mode is the pulsation mode and the overtones are stable . only for 120@xmath8with @xmath49 ,
are both the fundamental ( f ) and the first overtone ( 1h ) unstable , with an @xmath5-folding time for 1h much shorter than for the f mode .
table [ tab : zams ] also gives the results for solar metallicity models calculated with pure electron scattering opacity .
for such models , our results are in good agreement with previous results based on pure electron scattering or on simple opacity laws .
schwarzschild & hrm ( 1959 ) find for 121@xmath8 , @xmath50 and @xmath51 , and for 218@xmath8 , @xmath52 and @xmath53 .
aizenman et al .
( 1975 ) obtain for 110@xmath8 , @xmath54 and @xmath55 .
inspection of table [ tab : zams ] shows that for a given mass , the period decreases with metallicity .
this is a consequence of the smaller radius and more compact structure of metal - poor stars when they reach thermal equilibrium on the zams , a well known effect of @xmath0 on evolutionary models ( _ e.g. _ , ezer & cameron 1971 ; baraffe & el eid 1991 ; meynet et al 1994 ) .
the period , @xmath30 , is related to the dynamical time scale of the star , @xmath56 .
another important result of table [ tab : zams ] is the increase of @xmath31as @xmath0 decreases from @xmath20 to 0 .
this indicates that metal - free stars , although unstable , require a much longer time for the growth of the oscillation amplitudes than solar metallicity stars .
we also note a significant difference between @xmath31obtained for @xmath57 models calculated using the most recent opacities , and the electron scattering cases .
this highlights an important contribution of opacity to the instability of very massive stars with solar metallicity .
indeed , the driving zones in the @xmath20 zams models with @xmath58 are located at @xmath59 and are associated with a sharp peak of the opacity in this region due to the contribution of heavy elements ( iglesias & rogers 1996 ) . for 120@xmath8with @xmath60 , our results are in agreement with the analysis of glatzel & kiriakidis ( 1993 ) : the driving for the first overtone is due to the @xmath3-mechanism whereas for the fundamental mode , both the @xmath2- and the @xmath3- mechanisms contribute to the driving .
as @xmath0 decreases , we find a transition between nuclear - driven pulsation and opacity - driven pulsation ( _ i.e. _ , the @xmath2- and @xmath3-mechanisms ) .
the @xmath2-mechanism is dominant for @xmath61 and the @xmath3-mechanism dominates for @xmath62 , for the mass range of interest .
the increase of @xmath31as @xmath0 decreases for @xmath63 is partly due to the different temperature dependence of the nuclear energy generation from the cno cycle .
as shown in table [ tab : zams ] , central hydrogen burning proceeds at higher @xmath64 for decreasing @xmath0 .
locally , the temperature dependence of the specific nuclear energy generation rate , @xmath65 , depends on temperature as @xmath66 , where @xmath67 .
since the temperature sensitivity , @xmath68 , of hydrogen burning by the cno - cycle decreases with increasing temperature ( _ e.g. _ , kippenhahn & weigert , 1992 , figure 18.8 ) the metal - rich stars have a stronger sensitivity than the metal poor stars .
we find @xmath69 for @xmath26 and @xmath68 increases with @xmath0 up to @xmath70 for @xmath57 in the 120@xmath8star .
analytically @xmath68 can be computed from @xmath71 where @xmath72 @xmath73 , @xmath74 , @xmath75 , @xmath76 are mass number and charge number of the two input nuclei , and @xmath77 is the temperature in 9@xmath78(clayton 1968 ) . for @xmath79(p,@xmath80 ) , @xmath81 at @xmath82 and @xmath83 at @xmath84 , in agreement with the values obtained above .
since the driving of pulsations due to the @xmath2-mechanism depends directly on @xmath68 with @xmath85 , it is larger in metal - rich than in metal - free stars , contributing to the decrease of @xmath86 = @xmath87 as @xmath0 increases .
another contribution to @xmath86 comes from the total pulsation energy @xmath88
@xmath89 @xmath90 , since @xmath91 ( see next section eqs .
( [ eq : lpuls ] ) and ( [ eq : ep ] ) ) and @xmath30 increases with @xmath0 , as mentioned above .
as long as the contribution of metals to the opacity is negligible , this explains why @xmath92 decreases as @xmath0 increases .
once the @xmath3-mechanism operates in the envelope , it yields a dramatic increase of @xmath36 and significantly smaller @xmath31 , by up to six orders of magnitude ( see table [ tab : zams ] ) .
rlrrrll & @xmath0 & @xmath27(@xmath93 ) & @xmath28 ( @xmath94 ) & @xmath29(8@xmath78 ) & @xmath30 ( @xmath95 ) & @xmath31(@xmath96 ) + 120 & 0 & 99.2 & 4.6 & 1.38 & 0.064 & 1.44 + & 2@xmath97 & 72.9 & 8.2 & 0.74 & 0.16 & 5.63 + & 2@xmath98 & 67.1 & 9.6 & 0.63 & 0.20 & 4.23 + & 2@xmath99 & 62.6 & 11.0 & 0.55 & 0.25 & 3.73 + & 0.01 & 53.3 & 15.2 & 0.45 & 0.34 ( 0.16 ) & 4.63 ( 2.3@xmath100 ) + & 0.02 & 48.9 & 18.4 & 0.44 & 0.36 ( 0.26 ) & 1.92 ( 1.4@xmath101 ) + & 0.02 ( @xmath102 scatt . ) & 67.1 & 13.2 & 0.45 & 0.34 & 7.82 + 200 & 0 & 104.2 & 6.0 & 1.43 & 0.083 & 3.73 + & 0.02 & 44.9 & 32.1 & 0.46 & 0.76 & 1.3@xmath101 + & 0.02 ( @xmath102 scatt . ) & 70.5 & 17.5 & 0.47 & 0.45 & 5.32 + 300 & 0 & 107.0 & 7.6 & 1.47 & 0.10 & 2.63 + & 2@xmath99 & 68.1 & 18.2 & 0.60 & 0.39 & 1.33 + & 0.02 & 39.8 & 54.1 & 0.47 & 1.79 & 2.1@xmath101 + & 0.02 ( @xmath102 scatt . ) & 72.5 & 22.0 & 0.48 & 0.55 & 4.32 + 500 & 0 & 109.2 & 10.0 & 1.51 & 0.13 & 2.03 + + note. the pulsation period and @xmath5-folding time are given for the fundamental mode .
only for 120@xmath8with @xmath60 the fundamental mode and first overtone ( in parentheses ) are both unstable . for @xmath103 , results based on models calculated with pure electron scattering opacity ( @xmath102 scatt .
) are also given .
note. the table gives the initial mass @xmath37 , the total hydrogen - burning lifetime of the star @xmath104 , the age at which the star becomes stable @xmath105 , the central hydrogen mass fraction at this time @xmath106 ( initial value is 76% ) , and the number of @xmath5-folds an initial pulsation has grown since the zams by then , @xmath107d@xmath108 .
note. the table gives the initial mass , @xmath37 , metallicity , @xmath0 , pulsation energy , @xmath109 , mass loss rate at the zams , and an estimate of the total mass lost during the time the stars are unstable ( @xmath105 , see table [ tab : stab ] ) during central hydrogen burning , _ i.e. _ , @xmath110 .
the results for 120@xmath8 , @xmath111 and pure electron scattering opacity is also given .
[ sec : mdot ] in order to derive a realistic mass loss rate , non - linear calculations would be required to determine the pulsation amplitudes and dissipation due to shock wave formation .
non - linear calculations have been performed for the solar metallicity case by various authors ( appenzeller 1970ab ; ziebarth 1970 ; papaloizou 1973a ; talbot 1971ab ) who , unfortunately , give different results .
appenzeller ( 1970ab ) finds an upper limit @xmath112 for a 130@xmath8star and @xmath113 for a 270@xmath8star , whereas papaloizou ( 1973b ) finds stronger dissipation effects and mass loss rates less than @xmath114@xmath115for similar masses .
since the dissipation effects in appenzeller ( 1970ab ) may be underestimated compared to other studies , yielding larger mass loss rates , we used his results to derive an upper limit for @xmath116for the @xmath16 models .
appenzeller found that when the velocity amplitude at the surface approached the sound speed , mass loss sets in .
an upper limit of the mass loss rate came from assuming that the entire rate of gain of pulsation energy , @xmath117 , was used to accelerate the matter to escape velocity , @xmath118 .
thus one has : @xmath119 @xmath117can be estimated from the total pulsation energy @xmath109 , averaged over one period , and the stability coefficient , @xmath34 ( see cox 1980 ) , with @xmath120 where @xmath121 since the relative pulsation amplitude , @xmath122 , is given by our linear stability analysis , @xmath123and thus @xmath109are estimated by assuming , like appenzeller ( 1970ab ) , that the surface velocity amplitude reaches the sound speed , @xmath124 , _
i.e. _ : @xmath125 for the zero metallicity stars from 120 to 500@xmath8 , @xmath124is typically @xmath126 .
the resulting energy , @xmath109 , and mass loss rates for the @xmath16 models on the zams are given in table [ tab : stru ] .
results are also given for the 120@xmath8star with @xmath111 and pure electron scattering opacity .
our upper limit for @xmath116 in this case is in good agreement with appenzeller s value of 5@xmath98@xmath115for similar masses .
a crude upper limit for the total mass lost @xmath127during hydrogen burning is derived assuming that mass loss proceeds at constant rate as long as the star remains unstable , _
i.e. _ , during the time @xmath105(see table [ tab : stru ] ) . in any case , @xmath127does not exceed 5% of the total mass for zero - metal stars .
note that this estimate neglects the time required for the pulsation amplitude to grow to the value @xmath123characterizing the beginning of mass loss ( eq . ( [ eq : vsurf ] ) ) .
given the non negligible value of @xmath31compared to @xmath105(see table [ tab : zams ] ) , this time may be significant for the 120@xmath8star .
in fact , this star may never encounter pulsationally driven mass loss . moreover , since the stars stabilize as soon as they evolve away from the zams ( see sect . [
sec : evol ] ) with decreasing @xmath34 ( see fig . [
fig : k ] _
_ ) , one can expect decreasing @xmath116along the main sequence , and thus even smaller values for @xmath127 .
we thus find that even for 500@xmath8star , pulsations do not lead to significant mass loss on the main sequence . for post - main sequence evolution , as found in sect .
[ sec : evol ] , only stars with initial masses of @xmath128 become unstable again during helium burning and can undergo another phase of mass loss .
any estimate of mass loss is hampered by the lack of non - linear calculations during these late phases of evolution . moreover , as the star evolves toward low effective temperature and eventually becomes a red supergiant , it develops a deep convective envelope , the opposite of main sequence stars .
this adds additional difficulty to any derivation of @xmath116since convection can have an important contribution to the damping or driving of the pulsation in this case .
[ sec : concl ] our analysis has shown that even though metal - free stars with @xmath129 are radially unstable on the main sequence ( like their metal - enriched counterparts ) , the efficiency of the @xmath2-mechanism is reduced because of the hotter central @xmath64 and more compact structure .
consequently , the @xmath5-folding times characterizing the growth of amplitudes are much longer . after burning some hydrogen ,
the stars stabilize ( table [ tab : stab ] ) . in order to know the effects of such oscillations and accurately estimate the resulting pulsationally driven wind ,
non - linear calculations would be required .
we have not done those and have only estimated the mass loss using simple arguments based upon previous non - linear calculations . adopting the approach of appenzeller ( 1970ab ) that the mass loss can be derived from the energy gain rate of the pulsation when its surface velocity reaches the local velocity of sound
at least gives a quantitative result which is a reasonable upper limit ( sect .
[ sec : mdot ] ) . for the 120@xmath8star
we find negligible mass loss during the central hydrogen burning phase . the integrated mass loss during central hydrogen burning of about 2.5@xmath8for a 200@xmath8star increases to about 25@xmath8for the 500@xmath8star - less than 5% of its initial mass .
we also note that papaloizou ( 1973b ) disagrees with the results of appenzeller ( 1970a ) and finds an important limitation of the pulsation amplitude , which can not drive any mass loss .
if this is the case for the solar metallicity stars , it is even more so for metal - free stars .
therefore we conclude that mass loss during central hydrogen burning remains unimportant for the evolution of all the z=0 stars investigated ( table [ tab : stru ] ) .
finally , it is worth noting that the strong instability found at the end of core helium burning for @xmath130 could have important consequences .
what exactly happens depends on the remaining time before collapse ( only a few 10@xmath43for the 300 and 500@xmath8stars investigated here ) and , especially , on the strength of the mass loss rates achieved in these pulsating giant stars with convective envelopes .
this is an important issue regarding the gamma - ray burst scenario suggested by fryer et al .
( 2000 ) and needs further investigation . from our estimates of the mass loss rates we conclude that primordial very massive stars of @xmath131 ( but over @xmath132 ) reach core collapse with helium cores massive enough to encounter the pair creation instability .
single stars probably do not lose their hydrogen envelope for initial masses of @xmath133 .
the nucleosynthetic yields of these stars can be important for the chemical evolution of the early universe , as they may eject up to @xmath134 of @xmath135 for a @xmath136 star ( heger & woosley 2000 ) .
still heavier stars make black holes ( bond , arnett & carr 1984 ) . extrapolating from our results , we speculate that stars of @xmath137 might experience such strong mass loss that they end their lives as smaller objects . but clearly the stability of these massive `` first stars '' is a subject worthy of additional exploration beyond the first linear stability analysis presented here .
we are grateful to gilles chabrier for many valuable discussion .
ib thanks the astronomy departments of university of california at berkeley and santa - cruz for hospitality during the completion of this work .
this research was supported , in part , by prime contract no .
w-7405-eng-48 between the regents of the university of california and the united states department of energy , the national science foundation ( ast 97 - 31569 , int-9726315 ) , and the alexander von humboldt - stiftung ( flf-1065004 ) . | the stability of metal - free very massive stars ( @xmath0 = 0 ; @xmath1 ) is analyzed and compared with metal - enriched stars .
such zero - metal stars are unstable to nuclear - powered radial pulsations on the main sequence , but the growth time scale for these instabilities is much longer than for their metal - rich counterparts .
since they stabilize quickly after evolving off the zams , the pulsation may not have sufficient time to drive appreciable mass loss in z = 0 stars . for reasonable assumptions regarding the efficiency of converting pulsational energy into mass loss
, we find that , even for the larger masses considered , the star may die without losing a large fraction of its mass .
we find a transition between the @xmath2- and @xmath3-mechanisms for pulsational instability at @xmath4 .
for the most metal - rich stars , the @xmath3-mechanism yields much shorter @xmath5-folding times , indicating the presence of a strong instability .
we thus stress the fundamental difference of the stability and late stages of evolution between very massive stars born in the early universe and those that might be born today . 0.1 in 0.1 in |
we consider the existence of solutions of the following nonhomogeneous problem @xmath6 where the energy functional is defined by @xmath7 the nonlinearity is given by a function @xmath3 of _ double power _ type that is an even function @xmath8 with @xmath9 satisfying the following requirements : 1 .
there exist positive numbers @xmath10 with @xmath11 such that + @xmath12 + @xmath13 2 .
there exists @xmath14 and @xmath15 such that , for all @xmath16 @xmath17 3 . for any @xmath18 we have @xmath19 for example
the required assumptions are satisfied by @xmath20 with @xmath21 small enough , as shown in the appendix .
we assume @xmath22 , for some @xmath23 and @xmath24 moreover , we want @xmath25 and @xmath5 on a set of positive measure . in @xcite the existence of two positive solutions @xmath26 of the equation @xmath27 is proved when @xmath28 satisfies @xmath29 , @xmath30 .
recently , in @xcite , a similar problem for the @xmath31-laplacian is studied .
namely , the author proves , with variational techniques , that the problem @xmath32 in @xmath1 , where @xmath33 , @xmath34 , @xmath35 and @xmath3 a is continuous superlinear function such that @xmath36 and @xmath37 as @xmath38 , admits two positive solutions @xmath26 .
the existence of a positive solution of the problem @xmath39 on @xmath1 , @xmath40 for @xmath41 , was proven in @xcite when @xmath42 and @xmath43 , @xmath44 , @xmath45 and @xmath46 for some @xmath47 . in @xcite
there is a result of multiplicity for this problem . in @xcite
the author shows that the dirichlet problem on a bounded domain @xmath48 in the critical case @xmath49 has two solutions @xmath50 , for @xmath4 satisfying a suitable condition , and if @xmath51 then @xmath52 and @xmath53 .
we are interested in studying the problem with double power nonlinearity . in pioneering work berestycki and lions @xcite showed the existence of a positive solution in the case @xmath54 when @xmath55 , @xmath3 has a supercritical growth near the origin and subcritical at infinity .
more recently in the papers @xcite the double - power growth condition has been used to obtain the existence of positive solutions for different problems of the tipe ( [ pv ] ) .
in particular , in @xcite , the authors proved that in the same hypothesis on @xmath56 the homogeneous problem @xmath57 has a ground state solution ( i.e. least energy nontrivial solution ) .
other results on similar problems with the double power nonlinearity can be found in @xcite . in this paper
we prove the following theorem if @xmath58 , for some @xmath59 , and if @xmath60 is sufficiently small there exist two solutions of problem ( [ pv ] ) in @xmath61 .
the first solution is close to 0 ; if also @xmath62is small enough , the critical value of the second solution is close to the least energy level @xmath63 of the homogeneous problem ( [ ph ] ) .
furthermore , if @xmath44 the two solutions are non negative .
indeed the hypothesis on the sign of @xmath56 is used only to find the second solution , but we prefer a more compact claim for the theorem . anyway , in the proofs we focus out when we use any hypothesis . to get the solutions of ( [ pv ] ) we look for critical points of the functional @xmath64 constrained on the nehari manifold @xmath65
the study of the structure of the nehari manifold will be a fundamental part of this paper .
this paper is organized as follows : in section 2 , we recall some technical results concerning the appropriate function space required by the growth properties of the nonlinearity @xmath3 .
moreover , we study the geometry and the properties of the nehari manifold . in section 3
, we prove a splitting lemma necessary to overcome the lack of compactness .
this lemma is a variant for a well known result of @xcite . in section 4
we prove the existence of two distinct critical points of the functional on the nehari manifold .
we will use the following notations * @xmath66 completion of @xmath67 with respect to the norm @xmath68 * @xmath69 ; notice that , by ( [ condv ] ) , we have that @xmath70 is a norm in @xmath61 equivalent to the usual one ; * @xmath71 ; * @xmath72 ; * @xmath73 ; * @xmath74 ; we call @xmath75 the minimizer of @xmath76 on @xmath77 radially symmetric ; * @xmath78 ; we call @xmath79 the minimizer of @xmath80 on @xmath81 ; * @xmath82 ; * @xmath83 the lebesgue measure of the subset @xmath84 ; * @xmath85 ; * @xmath86 ; * @xmath87 . in order to study the properties of the functional @xmath64 and its nehari manifold , we consider some suitable orlicz space @xmath88 , where @xmath89 , related to the double power growth behavior of the function @xmath3 .
we recall some properties of these spaces to get the smoothness of the functional @xmath64 given @xmath90 , we consider the space @xmath88 made up of the functions @xmath91 such that @xmath92 the space @xmath88 is a banach space equipped with the norm : @xmath93 it is well known ( see , for example @xcite ) that @xmath88 coincides with the dual of @xmath94 . then : @xmath95 and we can introduce the following norm equivalent to the previous one @xmath96 hereafter we recall some results useful for this paper contained in @xcite .
[ lp+lq ] we have 1 . if @xmath97 , the following inequalities hold : @xmath98\leq\\ & \leq&||v||_{l^p+l^q}\leq\\ & \leq&\max[||v||_{l^q({\mathbb r}^n\smallsetminus\gamma_v)},||v||_{l^p(\gamma_v)}]\end{aligned}\ ] ] when @xmath99 ; 2 .
let @xmath100 .
then @xmath101 is bounded in @xmath88 if and only if the sequences @xmath102 and @xmath103 are bounded .
3 . @xmath104 is a bounded map from @xmath88 into @xmath105 [ emb ] by the previous lemma we have @xmath106 when @xmath89 .
then , by sobolev inequality , we get the continuous embedding @xmath107 in order to prove the @xmath108 regularity of the functional @xmath64 , we need the following lemmas proved in @xcite [ stimef ] if @xmath3 satisfies the hypothesis ( [ f1 ] ) and ( [ f2 ] ) , we have that 1 .
if @xmath109 are bounded in @xmath88 , then @xmath110 is bounded in @xmath94 ; 2 .
@xmath111 is a bounded map from @xmath88 into @xmath112 ; 3 .
@xmath111 is a continuous map from @xmath88 into @xmath112 ; 4 .
the map @xmath113 from @xmath114 in @xmath115 is bounded .
the functional @xmath64 is of class @xmath108 and it holds @xmath116=\langle \nabla e_g^v(u),v\rangle&=&\int_{{\mathbb r}^n}\nabla u\nabla v+vuv - f'(u)v - gv;\\ e_g''(u)[v , w]&=&\int_{{\mathbb r}^n } \nabla v\nabla w + vvw -f''(u)vw.\end{aligned}\ ] ] moreover the nehari manifold defined as @xmath117 is of class @xmath118 and its tangent space at the point @xmath119 is @xmath120 at last , we introduce the functions @xmath121 we have that @xmath122 notice that the conditions on @xmath3 assure that also @xmath123 exists . furthermore , if @xmath124 , then @xmath125 , so @xmath126 , and _ vice versa _ , so we want to find the critical points of @xmath127 . to study the manifold @xmath128
it is useful to consider the following manifold : @xmath129 we have that for all @xmath18 there exists an unique @xmath130 such that @xmath131 we have that , using ( [ fmu ] ) and ( [ f1 ] ) , @xmath132 because @xmath133 .
furthermore we have that @xmath134 is strictly concave when @xmath135 , and that @xmath136 , so , for every @xmath18 there exist an unique maximum point @xmath130 for the function @xmath134 .
thus @xmath137 [ infv ] we have that @xmath138 by contradiction , we suppose that there exists a sequence @xmath139 such that @xmath140 converges to 0 .
we set @xmath141 , hence we can write @xmath142 where @xmath143 .
remark [ emb ] , the sequence is bounded in @xmath88 .
since @xmath144 and @xmath145 converges to 0 , we have @xmath146 hence we get @xmath147 and by claim 2 of remark [ lp+lq ] we get the contradiction .
[ l ] let @xmath18 and let @xmath148 the unique positive number such that @xmath131 .
then @xmath149 by contradiction , suppose that there exists a minimizing sequence @xmath150 , with @xmath151 such that @xmath152 .
let @xmath153 .
we have that @xmath154 because @xmath144 .
furthermore , by hypothesis , we have @xmath155 thus , by [ fmu ] , @xmath156 so , because @xmath157 we have that @xmath158 that is a contradiction .
obviously , by lemma [ infv ] we have also @xmath159 and @xmath160 does not depend on @xmath4 . at last
we can give the following characterization of the nehari manifold .
[ gsmall ] let @xmath161 , sufficiently small , and let @xmath18 with @xmath162
. then 1 .
if @xmath163 , then there exists an unique @xmath164 such that @xmath165 and @xmath166 , where @xmath167 is the unique value for which @xmath168 .
if @xmath169 , then there exists an unique @xmath164 such that @xmath165 and @xmath170 .
if @xmath171 , then there exist two positive numbers @xmath172 and @xmath173 such that @xmath174 and @xmath175 , where @xmath148 is the unique value for which @xmath131 .
4 . @xmath164 and @xmath176 depend @xmath118 on @xmath177 and on @xmath178 . furthermore , fixed @xmath119 , we have @xmath179 , when @xmath180 .
\1 . if @xmath181 , with @xmath182 , by [ fmu ] , we have that @xmath183<0,\ ] ]
so @xmath184 is a maximum point for @xmath185 .
furthermore we have that @xmath186 , @xmath187 and @xmath188 . using ( [ fmu ] ) and ( [ f1 ] ) , we have @xmath189 because @xmath133 .
this proves that there is exactly one @xmath164 such that @xmath190 ; it is easy to see that @xmath166 .
\2 . in this case , we can proof , as in ( [ phi - g ] ) that @xmath191 when @xmath192 and that if @xmath182 is a critical point of @xmath185 then ( [ phi2-g ] ) holds . at last , consider that @xmath193 , and so 0 is a local minimum for @xmath185 , and we can conclude .
we have just proved that , for any @xmath18 , we have an unique maximum point @xmath148 of @xmath134 .
so , if we prove that @xmath194 we have that there exist two numbers @xmath172 and @xmath173 such that @xmath195 . set @xmath196 as in lemma [ l ] , and consider that @xmath197 recalling that @xmath162 , if @xmath161 is sufficiently small , that is @xmath198 , we have exactly two positive numbers @xmath172 and @xmath173 such that @xmath195 , and @xmath172 and @xmath173 are respectively the maximum and the minimum point of @xmath185 \4 . for simplicity
we only prove that @xmath199 is a @xmath118 function .
the other case is straightforward .
let us define a function @xmath200 , @xmath201 we have that @xmath202 is a @xmath118 function .
let @xmath203 be such that @xmath204 we know that @xmath205 , thus , by the implicit function theorem there is a @xmath118 function @xmath206 such that @xmath207 .
we have then the claimed result .
the nehari manifold so can be described as : @xmath208 where @xmath209 we have also that @xmath210 on @xmath211 and @xmath212 on @xmath213 .
furthermore , because @xmath81 and @xmath214 are bounded away from 0 , we have also that @xmath215 .
the geometry of @xmath128 is represented in the following picture .
1.5 cm [ u->0 ] there exists @xmath216 such that @xmath217 indeed , by ( [ fmu ] ) we have @xmath218 so @xmath219
we recall that a sequence @xmath220 such that @xmath221 , and @xmath222 is a palais - smale sequence at level @xmath223 for @xmath64 . in the same way we say that @xmath224 such that @xmath221 , and there exists a sequence @xmath225 s.t .
@xmath226 , for all @xmath227 is a palais - smale sequence at level @xmath223 for @xmath64 restricted to @xmath128 .
a functional @xmath3 satisfies the @xmath228 condition if all the palais - smale sequences at level @xmath223 converge .
unfortunately the functional @xmath64 on @xmath128 does not satisfy the _
ps _ condition in all the energy range . in this section by the splitting lemma we get a description of the _ ps _ sequences for the functional @xmath64 .
let @xmath229 and let @xmath221
. then @xmath230 is bounded .
we have that @xmath231 .
furthermore , for ( [ fmu ] ) we have @xmath232.\end{aligned}\ ] ] if @xmath233 we have that @xmath234 so we will have @xmath235 that is a contradiction .
[ splitenergia ] let @xmath236 , and let @xmath221 .
then , up to subsequence @xmath237 in @xmath61 . furthermore
, setting @xmath238 we have 1 .
@xmath239 ; 2 .
@xmath240 . by the previous lemma we have that @xmath241 is bounded .
then @xmath237 and we have that @xmath242 furthermore , we have that @xmath243 indeed , we have the following equation , where @xmath244 @xmath245 using lemma [ lp+lq ] we have that the terms in @xmath246 are arbitrarily small when @xmath247 is sufficiently large . furthermore , since @xmath248 in @xmath249 for all @xmath48 bounded and for all @xmath250 , we get that @xmath251 the proof follows easily .
[ gpsi ] suppose that @xmath252 in @xmath61 .
then we have @xmath253 again we use that @xmath248 in @xmath249 for all @xmath48 bounded and for all @xmath250 .
we have that @xmath254 let @xmath255 a ps sequence at level @xmath223 for the functional @xmath64 restricted to the manifold @xmath128 .
then , up to a subsequence , there exist @xmath256 sequences of points @xmath257 , @xmath258 , with @xmath259 , a solution @xmath260 of the problem @xmath261 , and @xmath256 solutions @xmath262 , @xmath258 , of the problem @xmath263 such that @xmath264 since @xmath150 is a _ ps _ sequence for the functional @xmath64 restricted to the manifold @xmath128 , then @xmath150 is a _
_ sequence for the functional @xmath64 . by the lemma [ splitenergia ] we have that @xmath150 converges to @xmath260 weakly in @xmath61 ( up to subsequence ) , so , given @xmath265 , @xmath266
it is easy to see that @xmath267 arguing as in step 1 of ( * ? ? ?
* lemma 3.3 ) we get also that , for some @xmath268 , @xmath269\varphi= \int_{\text{supp } \varphi}f''(\theta u_n+(1-\theta)u^0)(u_n - u^0)\varphi \rightarrow 0,\ ] ] as @xmath270 , because @xmath271 in @xmath249 , with @xmath272 bounded and @xmath250 .
so we have proved that @xmath260 solves @xmath261 .
now we set @xmath273 then @xmath252 weakly in @xmath61 .
if @xmath274 strongly in @xmath61 , for step 3 of ( * ? ? ?
* lemma 3.3 ) we have that there exists a sequence @xmath275 with @xmath276 such that @xmath277 in @xmath61 , and @xmath278 . because @xmath260 is a weak solution of ( [ pv ] ) and
@xmath150 is a @xmath279 sequence for @xmath64 we have that , for any @xmath265 , @xmath280 so @xmath281 using ( [ f(un ) ] ) we have that @xmath282 is a @xmath279 sequence for the functional @xmath80 .
thus , for any @xmath265 we have @xmath283\varphi(x - y_n)-\int v(x)\psi_n(x)\varphi(x - y_n)+o(1).\end{aligned}\ ] ] using the same argument of lemma [ gpsi ] we can prove that @xmath284 furthermore we have @xmath285\varphi(x - y_n)=\\ & = & \int\limits_{b_r } [ f'(u_0+\psi_n)-f'(u^0)]\varphi(x - y_n)+\\ & & + \int\limits_{b_r^c } [ f'(u_0+\psi_n)-f'(\psi_n)]\varphi(x - y_n)-\\ & & -\int\limits_{b_r^c } f'(u^0)\varphi(x - y_n)+\int\limits_{b_r } f'(\psi_n)\varphi(x - y_n)\leq\\ & \leq&||[f''(u^0+\theta\psi_n)-f''(\theta\psi_n)]\varphi(\cdot - y_n)||_{l^{p'}({\mathbb r}^n ) } & & + ||[f''(\psi_n+\theta u^0)-f''(\theta u^0)]\varphi(\cdot - y_n)||_{l^{p'}\cap l^{q ' } } @xmath268 . because @xmath286 for @xmath287 , and given @xmath247 @xmath288 as @xmath270 , we have that @xmath289 at last , it is easy to see that @xmath290 so we have also proved that @xmath291 solves the problem @xmath292 .
set @xmath293 , we have that @xmath294 , thus @xmath295 by lemma [ splitenergia ] .
so , @xmath296 now , if @xmath297 strongly in @xmath61 , we have the claim , otherwise we can proceed by induction and conclude the proof in a finite number of steps .
we set @xmath298 we show that there exist a solution with critical value @xmath299 and another solution with critical value @xmath300 .
we set also @xmath301 and we recall that there exists a positive radially symmetric function @xmath302 such that @xmath303 finally , we set @xmath304 we know , by @xcite , that for any @xmath305 and @xmath5 on a set of positive measure there exists a function @xmath306 such that @xmath307 and @xmath308 we prove the following results . [ mainneg+ ] there exist a @xmath309 such that @xmath310 .
furthermore , when @xmath311 is small , @xmath312 is unique .
by definition of @xmath213 we have that @xmath313 , and that @xmath314 .
at first we prove that @xmath315 . by contradiction , suppose that there exist a sequence @xmath316 and a sequence @xmath317 with @xmath143 and @xmath318 such that @xmath319 we have also that @xmath320 .
so , if @xmath145 is bounded , we have @xmath321 that is bounded by lemma [ lp+lq ] .
thus we have that , up to subsequence , @xmath322 . finally , arguing as in ( [ eglimitata ] )
we have that @xmath323 that is a contradiction .
now , let @xmath150 a minimizing sequence .
for the ekeland variational principle , we can suppose @xmath150 be a @xmath279 sequence .
for the splitting lemma there exists a @xmath324 and @xmath256 functions @xmath262 , @xmath325 such that @xmath326 we know that @xmath327 for all @xmath328 .
so , if @xmath329 we will have @xmath330 for some @xmath331 and this is a contradiction .
so , we have @xmath332 furthermore , we have @xmath333 , so @xmath334 , and this concludes the proof of the existence . to prove uniqueness , we argue by contradiction .
if @xmath335 are minimizers of @xmath336 on @xmath213 , both @xmath337 and @xmath338 solve ( [ pv ] ) , so we have @xmath339 with @xmath268 .
so @xmath340 , we have that , if @xmath341 in @xmath342 , then both @xmath337 and @xmath338 are small in @xmath88 , so we have that @xmath343 in @xmath112 by lemma [ stimef ] , and , by interpolation , @xmath344 that is a contradiction . [ upos ] suppose that @xmath51
. then there exists an @xmath345 in @xmath213 such that @xmath310 .
take @xmath312 as in theorem [ mainneg+ ] . because @xmath309 we have that @xmath346 .
if @xmath312 changes sign , or @xmath312 negative , we have that @xmath347 so , reminding that @xmath3 is even we have @xmath348 we know that there exists a @xmath349 such that @xmath350 . furthermore we know , by the study of @xmath351 that @xmath349 is a local minimizer of @xmath351 , in fact , @xmath352 for all @xmath353 $ ] .
we have @xmath354 and @xmath355 thus @xmath356 and @xmath357 that concludes the proof .
we want to prove that , under suitable hypothesis on @xmath358 and @xmath56 , there exists another solution of [ pv ] , by minimizing the functional @xmath64 on @xmath211 . in order to prove that a minimizing sequence converges
we will show that , for @xmath4 small , @xmath359 [ limsup ] suppose that @xmath25 and @xmath5 on a set of positive measure .
if @xmath360 sufficiently small , then there exist a @xmath331 such that @xmath361 moreover , @xmath362 by ( * ? ? ? * lemma 4.4(a ) ) and ( * ? ? ?
* theorem 1.1 ) we know that there exists a @xmath363 such that @xmath364 we set @xmath365 , so @xmath366 .
we know that there exists @xmath367 such that @xmath368 by proposition [ gsmall ] .
furthermore , by proposition [ gsmall ] we have that @xmath369 when @xmath370 , and so @xmath371 that concludes the proof .
[ u1 g ] for @xmath370 there exist @xmath372 a solution of ( [ pv ] ) . furthermore ,
if @xmath51 the solution @xmath373 can be chosen positive . by the splitting lemma , to obtain the result it is enough to show that @xmath374 . in the previous lemma
, we have proved that there exists a @xmath331 such that @xmath375 for @xmath360 sufficiently small . by remark [ gsmall ]
we have also that @xmath376 when @xmath377 in @xmath342 .
so there exists @xmath378 a solution of ( [ pv ] ) .
moreover @xmath379 is positive , so @xmath380 . to prove the last claim , consider that @xmath381 .
also , there exists a @xmath184 such that @xmath382 .
then we have @xmath383 so if @xmath373 is a solution , also @xmath382 is a solution of ( [ pv ] ) .
[ mv ] if @xmath384 , then @xmath385 .
we take a sequence of @xmath386 in @xmath94 .
we know that for any @xmath387 there exists @xmath388 such that @xmath389 .
for simplicity we call @xmath390 . also , we set @xmath391 , and @xmath392 .
we have @xmath393\ ] ] and we have that there exist a @xmath331 such that @xmath394 . now ,
suppose , by contradiction , that @xmath395 .
then , @xmath396 and so @xmath397 by ( [ fmu ] ) , we have that @xmath398 so @xmath399 .
now the hypothesis on @xmath3 @xmath400\leq \int \frac{f(t_n v_n)}{t_n}\rightarrow0,\ ] ] so we have that both @xmath401 and @xmath402 vanish when @xmath403 , and so @xmath404 that is a contradiction .
furthermore , by proposition [ gsmall ] , we have @xmath145 bounded away from 0 .
so , we have that there exists two positive constants @xmath405 and @xmath406 such that @xmath407 now , let @xmath408 such that @xmath409
. we can show that @xmath410 when @xmath411 .
the main idea is that @xmath412 because @xmath413 is bounded and @xmath386 in @xmath94 .
the details are omitted for the sake of simplicity .
now we have that @xmath414 we have that @xmath415 is bounded , so , up to subsequences , there exists a @xmath416 such that @xmath417 when @xmath411 , and , because @xmath150 is bounded in @xmath88 , also @xmath418 , and , by ( [ taunun ] ) , @xmath419 .
so , @xmath420 . by lemma [ limsup ] we know also that @xmath421 so we get the claim . by theorems [ mainneg+ ] and [ u1 g ] , we have that there exists a @xmath309 and @xmath372 that solve ( [ pv ] ) .
furthermore , by theorem [ u1 g ] and proposition [ upos ] the solution can be chosen nonnegative .
at least , by remark [ u->0 ] we have that @xmath422 in @xmath61 and by proposition [ mv ] that @xmath385 when @xmath341 .
we want to prove that there exists a function that satisfies all the conditions required in the introduction .
obviously we can choose @xmath432 such that @xmath433 and @xmath434 .
furthermore , we choose @xmath21 sufficiently small such that also @xmath435 and @xmath436 are positive , so the second part of ( [ fmu ] ) is proved .
we can choose @xmath21 sufficiently small , in order to have @xmath441 .
now , set as above @xmath430 , we have @xmath442 } { ( 1+s^{q - p})^{4}}\ ] ] that is positive for all @xmath424 .
so ( [ fmu ] ) is completely proved .
vieri benci , carlo r. grisanti , and anna maria micheletti , _ existence and non existence of the ground state solution for the nonlinear schroedinger equations with @xmath451 _ , topol .
methods nonlinear anal . *
26 * ( 2005 ) , 203219 . | we consider the problem @xmath0 in @xmath1 , under the assumption @xmath2 , and with the non linear term @xmath3 with a _
double power _ behavior .
we prove the existence two solutions when @xmath4 is sufficiently small and @xmath5 .
* keywords : * nonlinear equations , variational methods , orlicz spaces |
the question of how accretion onto a black hole occurs has been of great and ongoing interest .
observations of accreting sources provide a test of astrophysics in the strong gravity regime and will potentially yield a signature of the presence of an event horizon around black holes .
recent advances in x - ray technology have revealed that accretion can occur at extremely low rates across the range of black hole masses from stellar remnants to supermassive black holes at the centre of our own and other galaxies .
however , the nature of that low - level accretion remains unclear ( see e.g. , @xcite ) .
accretion flows at low luminosities do appear to be radiatively inefficient , but we have yet to determine to what extent that arises from advection of hot gas through the event horizon , and what role jets may play in carrying away energy @xcite .
these low - level accreting black holes emit across the electromagnetic spectrum from radio to x - rays , so multiwavelength studies can be used to disentangle different sources from different regions of the inflow , outflow , or jet , and to establish causal connections between behavior in these regions . among the stellar mass black hole population ,
the best studied quiescent system for this purpose has been v404 cyg , which contains a compact object with a mass function of @xmath2m@xmath3 @xcite one of the most secure stellar mass black hole mass determinations we have .
v404 cyg has the longest orbital period of the black hole x - ray transients ( bhxrts ) , at 6.5days , with a k0iv companion star @xcite .
it is also the most luminous stellar mass black hole in quiescence with @xmath4ergs@xmath5 , @xmath6 brighter than the prototypical bhxrt , a062000 @xcite .
these extremes are unlikely to be coincidental and the high @xmath7 may be a consequence of the long orbital period , the evolved donor star , and/or the high mass of the black hole . attempts to relate observations of quiescent black hole binaries to theoretical models of accretion flows have focused upon fitting the broad - band spectral energy distribution , ideally with coverage from radio to x - ray energies .
v404 cyg has been a prime target due to its x - ray brightness @xcite .
these studies mainly focused on pure accretion models and only considered the optical to x - ray regime , as radio data were limited or non - existent .
attempts to apply advective models to the more luminous hard state ( e.g. @xcite ) brought home the fact that these models could not explain the ubiquitous radio emission in this state , that was sometimes seen to extend as far as the near - ir band . a more complete description of the low - state sed appears to require a coupled accretion plus outflow model ( @xcite ; @xcite ) .
more recent studies have revealed significant radio emission in quiescence too , from v404 cyg @xcite , and from a062000 @xcite providing support to the idea that an outflow also plays a significant role in quiescence . to facilitate comparison between developing models and data we present here the most complete sed available for v404 cyg .
it is primarily based on a simultaneous campaign performed in 2003 using _
chandra _ , _ hst _ , and several optical and radio observatories .
we supplement the coverage with non - simultaneous optical and near - ir data from @xcite and near to mid - ir _ spitzer _ data from @xcite and @xcite .
previous results from this campaign were presented by @xcite ; see also @xcite .
ultraviolet photometry used the advanced camera for surveys ( acs ; @xcite ) onboard the _ hubble space telescope ( hst)_. the observations performed are summarised in table [ hstlogtable ] .
v404 cyg is a faint uv source and reddened , so five orbits were used covering the f250w and f330w filters .
llcl date & ut start & duration ( s ) & filter + 2003 july 28 & 19:13:10 & 200 & f330w + 2003 july 28 & 19:19:08 & 400 & f250w + 2003 july 28 & 19:28:22 & 200 & f330w + 2003 july 28 & 19:34:20 & 400 & f250w + 2003 july 28 & 20:28:46 & 200 & f330w + 2003 july 28 & 20:34:44 & 400 & f250w + 2003 july 28 & 21:09:32 & 200 & f330w + 2003 july 28 & 22:00:19 & 400 & f250w + 2003 july 28 & 22:09:33 & 200 & f330w + 2003 july 28 & 22:25:28 & 400 & f250w + 2003 july 28 & 22:34:42 & 200 & f330w + 2003 july 28 & 22:40:40 & 400 & f250w + 2003 july 28 & 23:38:31 & 980 & f250w + 2003 july 29 & 01:18:45 & 980 & f250w + 2003 july 29 & 04:26:38 & 980 & f250w + we used standard acs techniques to reduce the data .
individual images had been pre - processed to the flat - fielded stage using the automatic pipeline , calacs ; we found no need to refine this reduction .
we then combined all of the images offline into a geometrically corrected master image using multidrizzle @xcite with standard settings .
we found excellent registration between individual images , so no need to improve the shifts applied .
we performed photometry using the idl / astrolib aperture photometry routine aper .
as v404 cyg is faint we used a relatively small aperture to perform photometry , of radius 0.125 , with background defined by a 2.5 annulus .
no stars in the field were bright enough to determine per - field aperture corrections , so we use the tabulated values from @xcite .
this aperture collects about 75% of the total light , and almost all of the sharp core of the point spread function .
we measured the position of the target in the f330w bandpass and then fixed it for f250w , after verifying that it was consistently placed in the two .
we note in passing that we do detect both v404 cyg and the nearby contaminating star .
we find an offset of the contaminating star relative to v404 cyg of @xmath8sec ( @xmath9 ) and @xmath10 , an improvement on the relative astrometry of the stars compared to that reported by @xcite .
a significant source of systematic uncertainty in acs photometry is charge transfer inefficiency ( cti ) due to radiation damage to the ccds ( @xcite ; @xcite ) , although fortunately the observations were performed early enough in the lifetime of acs that cti is unlikely to be a severe problem .
the effect is worst at low light levels , so maximised for faint uv sources .
cti is also least well calibrated for these cases , and the correction prescriptions provided by @xcite formally diverge for negligible sky background or source counts .
we face both problems with these data .
sky backgrounds in the near - uv are extremely low , dominated by zodiacal light and earthshine in the f330w bandpass , and geo - coronal emission at shorter wavelengths .
both were essentially unmeasurable in our images , and the background is expected to be dominated by ccd dark current , producing 0.52.0e@xmath11/pixel for our exposures .
we attempted to estimate the cti losses in two ways .
firstly , we followed the prescription in @xcite , except that we used the estimated background ( including the dominant dark current term ) from the acs exposure time calculator .
this prescription predicted losses of @xmath12% .
based on the actual data shown by @xcite however , this prescription appears to overestimate the cti losses for the faintest sources . for the low background case ( 0.51.5e@xmath11/pixel ) at source brightnesses up to a few hundred electrons
, @xcite found losses of around 0.035mag at maximum distance from the readout amplifier .
if we rescale this to the times of our observations and positions of the source we expect losses of about 2.5% .
we use these latter estimates to correct our measured source brightnesses and assign an additional error estimate of 2.5% to account for the uncertainty in applying cti corrections .
the uncertainties in the cti corrections are much smaller than the statistical uncertainties in the measured fluxes , so this is not the dominant source of uncertainty .
we tabulate our photometric results in table [ hstphottable ] .
since we are measuring counts across a broad bandpass rather than monochromatic fluxes , we tabulate the data in two ways ; the corrected number of electrons per second , and the average flux per unit wavelength using photometric zero - points estimated by the acs pipeline .
all have been corrected for cti , and to a nominal infinite aperture .
lcc band & count rate & flux + & ( s@xmath5 ) & ( erg@xmath13s@xmath5@xmath5 ) + + f250w & @xmath14 & @xmath15 + f330w & @xmath16 & @xmath17 + + f250w & 0.073 & + f330w & 0.57 & + + f250w & 0.024 & + f330w & 0.57 & + we obtained simultaneous @xmath18 band observations using the thomson ccd camera at the 0.8 m iac80 from 20:5005:51 ut on the night of 2003 july 28/29 .
we used 300s exposures with 52s readout time between them and bias - corrected and flat - fielded all images in the standard way using iraf .
seeing conditions were not good enough to cleanly separate the contribution of the target and its nearby ( 1.4 ) line - of - sight star @xcite so we applied straightforward aperture photometry using a large aperture of 3.5which adds the flux from both stars as described by @xcite .
we determine an average magnitude @xmath19 in acceptably close agreement with @xcite , indicating little difference in the optical brightness of the system compared to earlier data . to reconstruct the optical sed of the source we use the wht data from the program .
this covered a larger range of wavelengths than the gemini data , and was simultaneous with the _ hst _ observations .
these observations used the isis dual - arm spectrograph on the wht . to maximise efficiency and minimise readout time and noise
, we used the single red - arm mode with the r316r grating and marconi2 ccd .
we set exposure times to 200s , with @xmath20s dead - time between exposures .
to maximise photometric accuracy , we used a 4slit , so our spectral resolution was determined by the seeing ( median @xmath21 ) , and was typically @xmath22(250kms@xmath5 ) .
we performed bias correction and flat fielding using standard iraf techniques .
the slit was oriented in the same was as in our previous observations and covered the same comparison star .
we extracted spectra of both of these stars , and the nearby blended star , with the same techniques previously described ( @xcite ; @xcite ) .
we performed wavelength calibration relative to a single observation of a cune / cuar lamp and corrected for time - dependent variations in the wavelength calibration using telluric absorption features .
we calibrated the on - slit comparison star relative to kopff 27 @xcite , and then calibrated all spectra of v404 cyg relative to this on - slit comparison . to compare the wht data with that from the iac80 we perform synthetic photometry convolving the spectrum with a @xmath18 bandpass .
we found an offset between the two , most likely due to systematic errors in the wht calibration . since the iac80 photometry agrees well with earlier measures we assume that this calibration is correct and rescaled the wht data up by a factor of 1.3 to agree with the iac80 @xmath18 band data .
we used additional optical spectroscopy obtained with the r831 grating and standard eev ccds of the gmos spectrograph on gemini - n to extend our continuum lightcurve .
exposure times were 40s and we binned and windowed the images to reduce the dead - time between exposures to 12s . with a 5 slit and 1.1 median seeing we obtained a spectral resolution of 5.0 ( 230kms@xmath5 ) .
we performed data reduction , spectral extraction , and wavelength and flux calibration in the same way as for the wht data .
wavelength calibration used a cuar lamp , and we performed flux calibration relative to the same on - slit comparison star as used for the wht observations , and applied the same rescaling to ensure consistency with iac80 photometry .
we use data taken from two recent chandra observations of v404 cyg , one on 2000 april 26 for 10,295s ( @xcite , @xcite ) , and another on 2003 july 28/29 for a total of 61,200s @xcite .
the first observation used 1/4 sub - array mode , with time resolution of 1.14s ; while the second used 1/8 sub - array mode , with a resolution of 0.4s . in the original set ,
1587 counts were reported , while the second had 1941 counts .
we reanalyzed the standard pipeline - processed level-2 data from both data sets in ciao v3.2 .
the extraction aperture was set to a radius of 6-pixels and we retained only events between 0.37.0kev to reduce the background .
the background can be particularly significant for the s3 chip , so we used large background regions with a 48-pixel radius .
the background produced approximately 4.6 counts in the source aperture for the 2003 observation , and 0.2 counts in the 2000 observation .
vla data were obtained in the a configuration from 29 july 00:20 ut until 14:15 ut . the data were reduced and each time bin imaged using standard procedures within aips .
absolute flux calibration was obtained relative to 3c 286 .
the westerbork synthesis radio telescope ( wsrt ) observed v404 cyg from 28 july 17:13 ut till 29 july 05:33 ut , corresponding to the period d1f1d2f2 in @xcite and labelled in fig .
[ optvarfig ] .
we used two frequencies : 4.8 and 1.4ghz , with @xmath23 hours on source per frequency .
the first hour was unusable because the source was too low on the horizon .
the full 4.8ghz observation ( with useful coverage 18:4005:13 ut ) gives a flux density of @xmath24mjy . at 1.4ghz
the average flux density was @xmath25mjy .
we supplement our dataset with non - simultaneous points from the literature to fill in gaps in coverage .
the most comprehensive coverage of the optical - ir sed is from @xcite , who present @xmath26 and @xmath27 photometry from a coordinated campaign obtained on 1990 june 10 .
these points do raise some concern that they were taken soon after the outburst and may not be fully quiescent .
there is also a large gap , 13 years , relative to the recent data .
nonetheless , we find excellent agreement in @xmath18 between our simultaneous iac80 photometry and the earlier data .
data in the mid - ir have also been presented by @xcite and reanalysed by @xcite .
we include these 4.5 , 8.0 , and 24@xmath28 m photometry in the sed using gallo s data which differ significantly only in the 24@xmath28 m measurement .
using the wht spectrum we can estimate the contribution of line emission to the photometry and correct for this effect .
the strongest line is h@xmath29 with an equivalent width of 16 .
this translates to a change in @xmath18 magnitude of 0.01mag , and we correct for this before using these data in the sed .
@xcite found a stronger line during 199091 with average equivalent width 38.7 .
this suggests a larger contribution to the earlier photometry of @xmath30% , and we use this value to correct the photometry of @xcite .
we note that in both cases this contamination is effectively negligible in comparison to other uncertainties .
we expect the optical emission in quiescent bhxrts to be a combination of light from the disc and light from the companion star . in the case of v404 cyg ,
this is overwhelmingly dominated by the companion .
traditionally measures of the disc fraction use veiling of the photospheric lines of the companion star and are dependent on having high resolution spectra at the time of the observation
. they can be rather sensitive to assumptions about the spectrum of the companion star ( hynes et al .
@xcite investigated several methods of fitting the veiling at different wavelengths within the optical spectrum . despite large uncertainties in their results ,
they conclude on approximate values of @xmath31 , @xmath32 , and @xmath33 , where the @xmath34 values refer to the fractional contribution of the disc to the total . in the near - ir
, @xcite put an upper limit on the @xmath35 band veiling of @xmath36 .
we can not make a reliable determination of the disc fraction from our low resolution spectroscopy , but can instead use a different method to place a lower limit on the disc fraction .
the optical continuum is significantly variable , and the variability presumably must be attributed to the disc .
if we assume that the optical continuum lightcurve consists of three components : companion star , slowly varying disc light , and flickering disc light , then the fraction of light due to the flickering component provides a robust lower limit on the disc fraction .
we show iac80 photometry together with wht and gemini continuum lightcurves in fig .
[ optvarfig ] .
our observations were centred around photometric phase 0.65 using the ephemeris of @xcite implying that the ellipsoidal modulation should be rising from the deeper minimum at phase 0.5 to the maximum at phase 0.75 .
this rising trend and the approximate amplitude of ellipsoidal modulations expected are consistent with the general upward trend we observe . for a long period system such as v404 cyg , we can approximate the ellipsoidal modulation within a night by a linear trend so we remove the slowly varying light by subtracting a linear fit to the lower envelope of the continuum lightcurve .
we find that the remaining flickering component is about 4percent of the total light in both the wht and gemini segments of the lightcurve and correlates very well with the x - ray variations ( as also found for the h@xmath29 emission by @xcite ) .
this 4% flickering light is actually not far off of spectroscopic estimates based on veiling near h@xmath29 as discussed above .
this makes it plausible that flickering light source does contribute most of the disc light in the optical .
we do not have sufficient information to independently constrain the stellar contribution to the sed at each wavelength , so opt to instead model the stellar spectrum based on estimates of the system parameters .
from @xcite we have a spectral type estimate of g8k2iii v .
@xcite estimates companion star parameters of @xmath37m@xmath3 and @xmath38r@xmath3 implying @xmath39 and a giant classification .
the sed of a giant star should be the best approximation for v404 cyg , although strictly it is a stripped giant @xcite . for a k0iii spectral type
we find @xmath40k using the temperature scale derived by @xcite for g , k , and m giants .
we model the spectrum of the companion by interpolating on the grid of model atmosphere spectra of late - type giants of @xcite after logarithmically re - binning the models to @xmath41 . .
the dotted line corresponds to @xmath42 and the dashed line to @xmath43 .
the solid lines below the @xmath26 and @xmath35 points are the ranges of plausible companion contributions to the sed based on estimates of @xcite and @xcite . ]
@xcite estimated the interstellar extinction to be @xmath44 based primarily on comparing the @xmath45 colour with that expected for a k0iii star after correcting for their estimates of disc contamination .
we tested this by fitting the model k0iii spectrum described above to the full set of bvrjhk magnitudes ( folding the model spectrum with bandpasses to generate synthetic photometry ) constraining the fit to preferably lie within ranges implied by the disc contamination constraints of @xcite and @xcite and shown in fig .
[ optfitfig ] by vertical bars .
our fit confirms the earlier estimate with a formal best fit for @xmath46 , shown in fig .
[ optfitfig ] .
the residuals scatter both above and below the line indicating that intrinsic variability and/or calibration uncertainties are playing a role , so we do not attempt to estimate an error on the extinction but we do show companion spectra reddened by @xmath47 and @xmath43 to indicate the sensitivity to extinction .
extinctions much outside of this range would require attributing a larger fraction of the light to the disc than has previously been estimated . the alternate g8 and k2 possibilities are almost indistinguishable from k0 , so we do not show them .
the k0iii stellar spectrum clearly provides a good description of the uv to near - ir data , consistent with previous estimates that the companion dominates the light in both the optical and infrared .
we expand the view in fig .
[ optsedfig ] to include both near to mid - ir data from _
spitzer _ and near - uv data from _ hst_. as found by @xcite , an ir excess does appear to be seen by _
a weak uv component also seems to be present , although this is not as pronounced as seen in other quiescent bhxrts ( @xcite ; @xcite ; hynes et al . in prep . ) the f250w data ( the highest frequency ) shown in fig .
[ optsedfig ] has already been corrected for an estimate of the red - leak as will be discussed in section [ uvexcesssection ] . ) .
the dashed line is a model k0iii sed reddened with @xmath46 .
the dotted line is a 2000k blackbody component adjusted so that the sum of the two ( shown with a long solid line ) fits the short wavelength _
spitzer _ data . ]
the _ spitzer _ data clearly lie above a plausible extrapolation of the companion star spectrum alone ( fig .
[ optsedfig ] ) . given the insensitivity of the ir spectrum to the temperature of the companion and interstellar extinction
, we can not adjust the companion spectrum in any way to fit these ir data . while these were not simultaneous with the optical data , we do not expect the continuum brightness of the companion to vary from epoch to epoch .
small variations due to ellipsoidal modulations will be present as a function of orbital phase , but these are less than @xmath48 @xcite .
@xcite interpreted the mid - ir excess seen in v404 cyg as the tail of a shakura - sunyaev accretion disc .
this model , however , predicts a uv excess which is inconsistent with our _ hst _ observations , exceeding our f330w measurement by an order of magnitude and the f250w by two orders . during quiescence instead
most of the disc is expected to be in a cool state , 20003000k , with a more isothermal temperature distribution ( see @xcite and references therein ) .
the contamination from this material will then be most pronounced in the mid - ir as it is cooler than the companion star , but no uv excess will be present .
we use a solid line in fig .
[ optsedfig ] to show a model including a 2000k blackbody in addition to the companion star spectrum discussed above .
this can adequately fit both the near - ir to uv sed and the _ spitzer _ near - ir data ( it is less successful at 24@xmath28 m ) .
the blackbody flux is consistent with a disc of radius @xmath49 cm tilted at an inclination of 56@xmath50 @xcite and at a distance of 3.5kpc . for the system parameters discussed above ,
this corresponds to about 60percent of the radius of the roche lobe of the black hole .
this is a very plausible size for a quiescent accretion disc .
we note that while @xcite preferred a shakura - sunyaev disc model for the ir excess in v404 cyg specifically , in other systems they invoked a much cooler circum - binary disc .
such a disc would remain a possibility in the light of our uv observations as it would not contribute to the uv flux . the difficulty in fitting the 24@xmath28 m point supports a more controversial interpretation as proposed by @xcite .
the flux in this band is rather close to an extrapolation of the flat radio spectrum raising the possibility that the ir excess could be due to synchrotron jet emission .
the major difficulty in quantifying this is that the radio is known to be strongly variable ( section [ radiosection ] ) and was not observed simultaneously with the 24@xmath28 m point .
we will return to this topic in section [ jetsection ] .
our data also provide tenuous evidence for a uv excess .
one possible explanation for this is red - leaks in the uv bandpasses @xcite .
although this may make a large contribution in f250w , the effect is not expected for the f330w filter .
the estimates for red - leaks from @xcite are of no use as they only consider unreddened stars , so we perform our own calculations .
in addition to red - leaks , because the spectrum is so steep in the uv it is hard to gauge the expected flux in a broad - band without performing synthetic photometry . to better assess the significance of any uv excess
, we therefore use synphot @xcite to estimate count - rates in each bandpass for our model reddened k0iii spectrum .
we consider two spectra , one with the full spectral coverage , and the other truncated above 4200 to disentangle in - band stellar flux from the red - leak .
we find that the red - leak is negligible in the f330w bandpass , as expected , and that the in - band stellar flux can account for approximately half of the count - rate seen in this band ; this is a little more than a 3@xmath51 excess .
the f250w band appears to have a large contribution from red - leak , about half of the flux in the band .
after subtraction of the companion star , both in - band and red - leak , the residual excess is @xmath52s@xmath5 , so is not a significant detection of disc light in this band .
we therefore appear to have a significant detection of a uv excess in the f330w band only , but even this is at the mercy of uncertainties in the companion spectral classification ( which do become a concern in the uv ) and reddening .
we therefore choose not to pursue the spectrum of the uv excess further .
llllll model & & & & & + & & & & & + + power - law & @xmath53 & @xmath54 & & 1.00 & @xmath55 + bremsstrahlung & @xmath56 & & @xmath57 & 1.01 & @xmath55 + raymond - smith & @xmath58 & & @xmath59 & 1.05 & @xmath60 + + power - law & @xmath61 & @xmath62 & & 0.93 & @xmath63 + bremsstrahlung & @xmath64 & & @xmath65 & 0.92 & @xmath66 + raymond - smith & @xmath67 & & @xmath68 & 1.08 & @xmath69 + we extracted spectra for both _
observations with ciao v3.2 and analysed them with xspec v12.2 .
we used standard ciao techniques to select response files according to the ccd temperature .
we considered many spectral models within xspec , all with interstellar absorption , and report results from fitting with power - law , bremsstrahlung , and raymond - smith models here .
blackbody fits to the data were so poor that we neglect them for discussion . due to the lower count rate in the 2003 observation ,
we were prompted to try using cash statistics in tandem with the churazov approximation for chi - squared statistics ( cash , 1979 ; churazov , 1996 ) .
for each set of statistics to be used , we grouped the spectra into at least 15 counts per spectral bin ; further binning results in a loss of spectral information .
churazov weighting estimates the weight for a given channel by averaging the counts in surrounding channels .
cash statistics is a method designed to estimate the best - fit parameters using unbinned or slightly binned data , which can be particularly useful when the source yields few photons .
we also checked the consistency of cash with unbinned data , and found the results to be equivalent .
all models except the blackbody model give acceptable fits to both data sets with the power - law providing the best fit .
both statistical methods provide results that are equivalent to previously published results by @xcite for the 2000 observation .
the best - fitting model is shown in fig .
[ xspecfig ] . for the 2003 observation
, we find similar results to @xcite , and a plot is also displayed in fig .
[ xspecfig ] .
table [ xraytable ] summarises the results of the spectral fitting . to test for any variation in the spectrum with luminosity
we construct a hardness intensity diagram .
we separated the data into two bandpasses of approximately equal count rates , which were 0.31.75kev ( soft ) and 1.757.0kev ( hard ) for each set . to account for the reduction in sensitivity between the two epochs
, we used an assumed fixed model of a photo - absorbed power - law to calculate the ratio of expected count rates in these bandpasses between the two observations and hence rescale the 2000 observations to match those from 2003 . the resulting hardness
intensity diagram is shown in fig .
[ colorintensityfig ] .
we see no variation in hardness either within or between the two observations , the first of which had a substantially higher count rate .
this indicates no detectable spectral change over a factor of ten in source luminosity .
to constrain the radio spectral shape we first examine each dataset individually .
we obtain @xmath70 ( where @xmath71 ) from the vla data and @xmath72 from the wsrt .
the difference between them is not statistically significant so we also perform a joint power - law fit to both datasets , allowing the relative normalisation of the two to vary as they sample different portions of the lightcurve .
primarily this is constrained by the vla data as the uncertainties in the wsrt fluxes are larger .
we find a spectral index of @xmath73 .
all the fits are consistent with the canonical flat spectrum as commonly seen in hard state black hole binaries .
the vla data exhibit substantial variability .
we show the higher quality 8.5ghz lightcurve together with that from _ chandra _ in fig .
[ optvarfig ] .
the radio is clearly strongly variable and shows both large flares and dips not dissimilar to those seen by _ chandra_. the most dramatic feature rises from almost zero flux to the highest peak in about 30min , as was also the case in the dataset presented by @xcite .
no clear correlation between radio and x - ray is apparent to the eye , however .
to further test this , we calculate the cross - correlation function between the x - ray and radio data and show this in fig .
[ ccffig ] . no compelling correlation ( or anti - correlation ) is present although an unremarkable peak is present at zero lag . and @xmath74 expectations for uncorrelated variations .
] since wsrt is a linear array , it covers a full synthesis in 12 hours and hence is not capable of the high time - resolution achieved by the vla .
nevertheless the source does show significant variability over this period , even if a possible correlation with x - ray / optical is difficult to infer .
when the observation is split in 2 sub - intervals , 18:4024:00 and 00:0005:13 , the resulting flux densities are : @xmath75mjy and @xmath76mjy respectively . the second interval , covering d2f2 , overlaps with the first part of the vla observation , which shows a significant increase at 8.5ghz beginning at 03:00 ut ( with a peak flux density of 0.6mjy ) .
however , the vla flare only lasts for @xmath77 hour , of which only about one - half was covered by wsrt .
finer sub - intervals proved impractical as there was insufficient coverage of the uv plane to isolate flux from v404 cyg .
previous sections of this paper have focused primarily on one or two bands ; we will broaden our scope here to questions best addressed with the whole sed .
we show the complete radio x - ray sed in fig .
[ sedfig ] .
we have also compiled all the data points , as plotted , in table [ datatable ] to allow easier comparison with future models .
we include a point representing just the flickering component in the optical to represent a lower limit on the accretion light .
we adopt the extinction curve of @xcite in the optical and uv , @xcite for the _ spitzer_/irac data and assume @xmath78 following @xcite .
our dataset should supersede that used by @xcite for future modelling of v404 cyg .
compared to earlier work we now have simultaneous x - ray and optical data , shorter wavelength coverage extending into the near - uv , and simultaneous radio observations . in fig .
[ adaffig ] we illustrate the additional constraints now possible , showing model 1 of @xcite overlaid on our sed after scaling downwards to fit our x - ray spectrum .
this is of course not intended to be a detailed model fit .
not only should a new model be computed to fit a lower x - ray luminosity , but there have been refinements in modelling of advective accretion flows in the last decade as well .
our main point is to stress the strong constraint provided by the low near - uv flux .
model 1 was an acceptable fit to the sed data available to @xcite but clearly over - predicts the near - uv substantially compared to our new data .
alternative models with winds presented by @xcite , on the other hand , would be consistent with the uv data .
this constraint on the uv to x - ray flux ratio can be used to constrain current and future models .
the flux of the optical flickering component falls at an intriguing location in the sed ( fig .
[ sedfig ] ) .
it is a little above a straight extrapolation of either the x - ray or radio power - laws .
an extrapolation of the radio is plausible , as a flat - spectrum is expected to continue until a break to optically thin synchrotron at higher frequencies .
we would not expect a straight extrapolation of the x - ray spectrum at all , and all models for the broad - band sed ( e.g. @xcite ) involve curvature between x - ray and optical bands . ) here correlations are more telling evidence than fluxes .
we find an excellent correlation between x - ray and optical continuum ( fig .
[ optvarfig ] ) , but no clear correlation between x - ray and radio ( and by extension optical ) .
this suggests that the optical variability and the x - ray emission are closely connected , either by direct emission from the same region , as expected from early advection dominated accretion flow ( adaf ) models , or indirectly through x - ray irradiation of the disc , as inferred for h@xmath29 emission by @xcite . the x - ray and radio emission , on the other hand , do not obviously appear correlated on the timescales observed ( although a correlation on longer timescales is certainly possible ) suggesting that the mechanisms for variability in the two are not connected .
there is clearly a mid - ir excess prominent in the 24@xmath28 m band . while it is possible to partially explain this with emission from a cool disc , synchrotron emission from the radio jet provides a better description of the excess @xcite .
an extrapolation of the flat spectrum beyond the ir region is problematic for several reasons .
we have argued above that the optical variability appears directly or indirectly associated with the x - ray emitting component rather than with the radio .
in addition , the low uv flux observed is inconsistent with a straight extrapolation of the flat spectrum all the way to the uv , and problematic even if the spectrum is assumed to break in the optical .
this argues that a break in the jet spectrum at longer wavelengths is needed so that the non - stellar optical / uv emission originates from the disc rather than the jet .
the most likely interpretation is then that the break from flat spectrum to optically thin synchrotron occurs in the mid to far - ir , although given the large amplitude of radio variability only simultaneous radio and mid - ir observations could address this question conclusively .
we have compiled the comprehensive dataset describing the spectral energy distribution of the black hole x - ray binary v404 cyg in quiescence .
this includes new and simultaneous radio , optical , uv , and x - ray data , supplemented with additional non - simultaneous data in the optical through infrared .
our specific findings are : 1 .
there is no need for a model to include a substantial contribution from accretion light over the near - uv to optical spectral range .
the sed over this range ( including the vacuum uv ) is well described by a k0iv star reddened by @xmath79 .
the very low uv flux detected only marginally supports an accretion component , and rather tightly constrains some spectral models of quiescent accretion flows .
2 . as suggested by @xcite no plausible companion spectrum and interstellar extinction can fully explain the mid - ir ; an ir excess from a cool disc or possibly a jet @xcite appears to be required .
the x - ray spectrum is consistent with a @xmath0 power - law as found by all other studies to date .
there is no evidence for any spectral variation over the range of a factor of 10 in luminosity that has been observed over two epochs by chandra .
the radio shows a flat spectrum in @xmath1 as is commonly seen in hard state x - ray binaries and associated with a compact jet .
the break frequency between a flat and optically thin spectrum most likely occurs in the sub - mm to far - ir regime , but is not conclusively constrained by existing data .
the radio is substantially variable but puzzlingly there is no clear correlation with x - ray variability .
if the radio variability reflects changes in the jet power , this suggests that the x - ray variations occur independently of changes in the jet , or with a very large lag ( many hours ) between the two .
overall , the likely interpretation of the sed is as follows : 1 .
the k0iii companion star dominates the light from near - ir to near - uv wavelengths . 2 .
the cool outer regions of the accretion disc may produce a modest mid - ir excess seen by _ spitzer _
3 . the inner accretion flow produces highly variable x - ray emission but apparently only weak uv emission .
this irradiates the accretion disc and is responsible for emission line variability @xcite .
the optical continuum variability may also arise from reprocessing but direct emission from the inner accretion flow remains a possibility .
a compact jet is driven from the inner accretion disc producing variable flat - spectrum radio emission likely extending to the far and possibly mid - ir .
the size of the jet is smaller than 4.5au ( about @xmath80 the binary separation ) for a distance of 3.5kpc @xcite .
lrrr & & & + & & & + + wsrt ( 1.4ghz ) & @xmath81 & @xmath82 + wsrt ( 4.8ghz ) & @xmath83 & @xmath84 + vla ( 4.9ghz ) & @xmath85 & @xmath86 + vla ( 8.6ghz ) & @xmath87 & @xmath88 + iac80 ( r ) & @xmath89 & @xmath90 & @xmath91 + hst ( f330w ) & @xmath92 & @xmath93 & @xmath94 + hst ( f250w ) & @xmath95 & @xmath96 & @xmath97 + chandra & @xmath98 & @xmath99 + + spitzer ( 24@xmath28 m ) & @xmath100 & @xmath101 & @xmath102 + spitzer ( 8@xmath28 m ) & @xmath103 & @xmath104 & @xmath105 + spitzer ( 4.5@xmath28 m ) & @xmath106 & @xmath107 & @xmath108 + archival @xmath35 & @xmath109 & @xmath110 & @xmath111 + archival @xmath112 & @xmath113 & @xmath114 & @xmath115 + archival @xmath116 & @xmath117 & @xmath118 & @xmath119 + archival @xmath18 & @xmath120 & @xmath121 & @xmath122 + archival @xmath123 & @xmath124 & @xmath125 & @xmath126 + archival @xmath127 & @xmath128 & @xmath129 & @xmath130 +
we are grateful to valerie mikles for many helpful comments on this manuscript .
chandra observations were supported by nasa grant go3 - 4044x . c. b. acknowledges support from a laspace gsra award .
e. g. acknowledges support from a hubble fellowship .
the wht is operated on la palma by the ing in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias .
gemini observations were obtained under program gn-2003a - q-12 by 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 secyt ( argentina ) koekemoer , a. m. , fruchter , a. s. , hook , r. n. , & hack , w. 2002 , in proc . the 2002 hst calibration workshop : hubble after the installation of the acs and the nicmos cooling system , eds . s. arribas , a. koekemoer & b. whitmore , 337 , baltimore : stsci | we present a multiwavelength study of the black hole x - ray binary v404 cyg in quiescence , focusing upon the spectral energy distribution ( sed ) .
radio , optical , uv , and x - ray coverage is simultaneous .
we supplement the sed with additional non - simultaneous data in the optical through infrared where necessary .
the compiled sed is the most complete available for this , the x - ray and radio brightest quiescent black hole system .
we find no need for a substantial contribution from accretion light from the near - uv to the near - ir , and in particular the weak uv emission constrains published spectral models for v404 cyg .
we confirm that no plausible companion spectrum and interstellar extinction can fully explain the mid - ir , however , and an ir excess from a jet or cool disc appears to be required .
the x - ray spectrum is consistent with a @xmath0 power - law as found by all other studies to date .
there is no evidence for any variation in the hardness over a range of a factor of 10 in luminosity .
the radio flux is consistent with a flat spectrum ( in @xmath1 ) . the break frequency between a flat and optically thin spectrum
most likely occurs in the mid or far - ir , but is not strongly constrained by these data .
we find the radio to be substantially variable but with no clear correlation with x - ray variability .
[ firstpage ] accretion , accretion discs binaries : close stars : individual : v404 cyg |
any theory of low - mass star formation should not only be able to make predictions for one characteristic stage such as the core mass spectrum in molecular clouds or the initial mass function of emerging stars - but also relate those through both pre- and protostellar stages . a number of the very basic questions of star formation still remain unanswered and debated ( see , e.g. , @xcite , @xcite and @xcite for recent reviews and discussions of both observational and theoretical studies ) , which can be addressed with large homogeneous surveys from mid - infrared through ( sub)millimeter wavelengths and well - defined lists of sources such as those presented in this paper . for example , is the star formation process highly dynamical with cores being transient phenomena in a turbulent medium and are protostars only accreting in short periods of time when `` passing by '' these dense environments ( i.e. , bondi - hoyle accretion ) ? or is low - mass star formation a more slowly evolving process as preferred in traditional ambipolar diffusion scenarios and in that case , how important are other effects such as turbulence in regulating the speed by which the process occurs ? how efficient is the star formation process in turning prestellar dust and gas into young stars ?
are there specific relationships between the natal prestellar cores and the emerging protostars , for example : do the properties of prestellar cores reflect in whether a given core actually forms a star and perhaps whether it forms a single or multiple system ? many of these discussions hinge on statistical arguments , for example estimates of the evolutionary rate of protostars through counts of the number of ysos in different stages . previous studies of the deeply embedded stages were hampered by low sensitivity single element ( sub)millimeter receivers and confusion in low resolution , low sensitivity infrared observations such as those from the iras and iso satellites . in the last few years systematic , detailed surveys of larger regions have been made possible using wide area imaging at high resolution and sensitivity using submillimeter telescopes such as the jcmt with scuba and mid - infrared telescopes including the spitzer space telescope and its infrared array camera ( irac ) and multiband imaging photometer ( mips ) . in this paper
we present a census of the deeply embedded low - mass protostars in the perseus molecular cloud utilizing recently obtained large scale spitzer / c2d @xcite and jcmt / scuba maps @xcite and discuss some of the strong constraints for theories of low - mass star formation emerging from these homogeneous datasets .
the earliest , most deeply embedded stages of low - mass protostars or the latest stages of prestellar cores close to the onset of collapse are important for insight into the physical conditions regulating low - mass star formation . approaching the problem from the prestellar stages ,
continuum observations provide strong constraints on the distribution of dust , and through inference the gas , in star forming regions .
in particular , large scale mapping with submillimeter and millimeter wavelength continuum bolometers ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have enabled relatively unbiased tabulations of dust condensations that either just have or are likely to form protostars .
the distribution of dust emission can also be used to infer the column density distribution of the dust in both pre- and protostellar cores and from there potentially derive self - consistent density and temperature distributions through detailed radiative transfer modeling ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
alternatively , observed parameters such as peak flux or size can be used as empirical diagnostics to estimate which cores will undergo collapse ( e.g. , * ? ? ?
* ; * ? ? ?
the dust continuum maps themselves do not answer the question whether a given core already has formed a central protostar : for that deep mid - infrared observations are needed .
furthermore mid - infrared observations are important as an evolutionary indicator once the protostar itself has formed : its infrared excess , spectral slope in the mid - infrared range or combination of colors are typically used to distinguish between objects in different stages ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
still , the infrared colors themselves do not provide an unambiguous mapping to object type and/or evolutionary status .
deeply embedded ysos for example share many of the same sed characteristics as more evolved edge - on disk systems .
also , since the most deeply embedded ysos emit a significant fraction of their luminosity at far - infrared and submillimeter wavelengths ( class 0 objects are for example observationally defined as objects emitting more than 0.5% of their total luminosity at wavelengths
longer than 350 @xmath0 m ; e.g. , @xcite ) establishing their full seds through these wavelengths is important .
all of these issues illustrate the importance of combining submillimeter and mid - infrared observations as is done in this paper .
the spitzer space telescope `` from molecular cores to planet forming disks ( cores to disks ; c2d ) '' legacy team @xcite has surveyed 5 of the nearby star forming clouds using the mid - infrared cameras , irac and mips , with the aim of characterizing the ongoing star formation in each .
the `` coordinated molecular probe line extinction thermal emission ( complete ) '' survey of star forming regions @xcite collect data for the three northern of these regions in a range of molecular lines , extinction and dust continuum . in this paper
we combine the information from the c2d and complete surveys of perseus .
perseus is a good test case for such a comparison with a large number of ysos in different environments from large clusters such as ngc 1333 and ic 348 to objects in relative isolation or in small groups like l1448 , l1455 , barnard 1 and barnard 5 .
this paper builds on the analysis of perseus through spitzer @xcite and jcmt / scuba observations @xcite by examining the relation between scuba cores and spitzer sources and thereby building a sample of deeply embedded objects .
the paper is laid out as follows : [ cluster ] first examines the general association of mid - infrared sources and the scuba cores , the mid - infrared colors of the sources found to be embedded in scuba cores ( [ midir ] ) and the properties of the scuba cores with embedded mid - infrared sources ( [ submm ] ) . [ overallcloud ] examines the properties of the embedded sources compared to the overall cloud environment in terms of extinction ( [ extincttracer ] ) and submillimeter flux ( [ submmtracer ] ) .
the results of [ cluster ] and [ overallcloud ] are combined and an unbiased sample of deeply embedded protostars in perseus is established . [ discuss ] presents the sample of sources ( [ sample ] ) and points to a number of important observational constraints for models of star formation , in particular in terms of dispersal of newly formed protostars ( [ dispersal ] ) , time scales for the evolution through the prestellar stages ( [ timescales ] ) and the star formation efficiency ( [ sfe ] ) .
the results in this paper will also serve as an important path - finder for comparisons to other clouds surveyed at both mid - infrared and submillimeter wavelengths .
the first task in building up an unbiased list of embedded ysos is to define what is meant by an association between a mid - infrared source and a submillimeter core . for this
we first explore the distribution of mid - infrared spitzer sources in comparison to the dust condensations seen in the scuba maps . in this section
we consider the distribution and properties of the nearest mips 24 @xmath0 m source to each submillimeter core and use these to draw general conclusions about the properties of the embedded mid - infrared sources . in [ submmtracer ] we turn the table and consider the properties of all the mips 24 @xmath0 m sources in the catalog .
deeply embedded ysos are expected to have a steeply rising sed from the irac through mips wavelengths due to the cold envelope emitting at longer wavelengths while absorbing and reprocessing emission from the central star+disk system at shorter wavelengths .
we therefore expect any deeply embedded yso detected at the irac wavelengths also to be detected at 24 @xmath0 m whereas the inverse may not be true .
the class 0 protostar iras 16293 - 2422 is for example seen in irs observations at 2335 @xmath0 m @xcite ( and likewise in more sensitive mips observations ) but not at the irac wavelengths . as a first try we therefore examine the relationship between mips 24 @xmath0 m sources from the c2d observations @xcite and the scuba cores from the list of @xcite and h. kirk et al .
( in prep . ) . the one place where this may be problematic is in regions where the 24 @xmath0 m maps are confused or bright emission saturates the images .
we will return to these issues later in this section .
many of the studied sources are also detected at longer wavelengths by mips at 70 and 160 @xmath0 m where the sed of a typical embedded protostar peaks .
on the other hand , these longer wavelength observations are less sensitive , more likely to be saturated and provide a lower angular resolution , and are therefore less useful for identifying individual objects in an unbiased sense ( as is the goal of this paper ) , especially in confused regions . for establishing the full seds of individual objects these data are important of course .
in the following we will refer to the sources detected at 24 @xmath0 m with mips as `` mips sources '' .
one concern is how big the contamination of background sources is and whether these are confused with the yso population .
based on comparison to the extra - galactic swire data , @xcite find that most of the mips sources are background galaxies .
still , the surface density of mips sources in the perseus field is low enough that given a random spatial distribution only 0.2 sources would be found within any 1 arcminute@xmath8 - or likewise that the probability of a chance alignment within 15 of a given scuba core is just a few percent .
[ source_overlay ] shows the distribution of mid - infrared mips sources and submillimeter scuba cores overlaid on scuba maps in a few of prominent regions of perseus .
a number of the scuba cores are associated with mid - infrared sources within rather small distances .
significant clustering is also seen in other manners , for example the clustering of the submillimeter cores or the mid - infrared sources by themselves . in this paper
we only focus on the association of the mid - infrared sources with the submillimeter cores . of the 72 cores from the scuba map , 36 ( 50% )
are found to have a mips counterpart within 15 of the peak ( 31 , or 43% , have a mips source within 10 ) . at the resolution of the spitzer / mips 24 @xmath0
m and scuba observations ( approximately 6@xmath5 and 15@xmath5 , respectively ) it appears that most of the scuba cores are associated with one mips source only ( table [ multiplicity ] ) . [ cols= " < , < , < , <
, < " , ] it is interesting to examine whether there is any evolution of the colors of the embedded ysos compared to the properties of their parental cores . for example do the more massive cores generally have more deeply embedded protostars with redder mid - infrared colors ? as shown in fig .
[ flux_histo ] , cores with stronger submillimeter fluxes all have embedded sources but a significant number of `` fainter '' cores also appear to be forming stars .
[ evol_diagram ] shows the @xmath1-[4.5]$ ] and @xmath2-[24]$ ] colors of the embedded ysos with the fluxes of their parental cores indicated .
there are two interesting effects to note : ( i ) this diagram includes all the mips sources located within 15 of a submillimeter core and for those sources there is a clear trend across the diagram with an increase in both colors ( from zero ; or no infrared excess ) toward higher values , possibly approaching a maximum value of their @xmath2-[24]$ ] colors of about 68 .
( ii ) if we apply our criterion of @xmath1-[4.5 ] > 1 $ ] for embeddedness the @xmath2-[24]$ ] colors do not actually increase systematically with flux , but are rather scattered around a value of about 6
. some of the brighter scuba cores have ysos with very red colors , but note also that some mid - infrared sources associated with fairly bright cores have @xmath1-[4.5 ] \lesssim 1.0 $ ] .
the upturn in colors is roughly as predicted by the models of @xcite but this diagram extends to much bluer [ 8.0]-[24 ] colors than predicted in their models . on the other hand the @xmath1-[4.5]$ ]
colors are much redder than predicted from the models .
@xcite find that only a few deeply embedded ysos should be detectable at the shortest irac wavelengths . that we actually detect a significant fraction of the deeply embedded protostars might again reflect that the envelopes have larger cleared out cavities similar to the case of iras 16293 - 2422a @xcite , in contrast to the assumption of the models of @xcite which extend into 7 stellar radii . at such small inner radii , the mid - infrared emission from the central protostar
would be completely absorbed by the surrounding envelope and the object not detectable at the irac wavelengths .
the fact that most of these sources are detectable at 3.6 and 4.5 @xmath0 m suggest that this is not the case and @xmath1-[4.5]$ ] color therefore is a potentially interesting diagnostic of the properties of the disks and inner envelopes of deeply embedded protostars . as pointed out in the introduction many important constraints on low - mass star formation hinge on statistical arguments such as using the relative number of pre- and protostellar cores to estimate their lifetimes .
previous studies have often been based on compilations of different surveys not necessarily observed under similar conditions or with the same techniques and have often required the combination of data for many clouds to obtain statistically meaningful samples .
the strength of the combination of the spitzer and scuba data presented in this paper is that it represents a relatively uniform survey of just one cloud and still has enough cores and protostars that strong statements can be made without many of the caveats introduced otherwise .
we here discuss the constraints on three important aspects of low - mass star formation introduced by this study .
one of the debated issues in current theories of star formation is the importance of the motions of embedded ysos through the medium on their accretion rates , i.e. , basically whether bondi - hoyle accretion @xcite applies .
the simulations of @xcite and @xcite for example show that the motions of the embedded ysos through their ambient medium will have a significant impact on their accretion histories and thereby the resulting distribution of ysos as a function of mass . on the other hand based on a study of locations of ysos in different evolutionary stages from taurus
, @xcite argues that the velocity dispersion of the ysos once formed is small @xmath9 km s@xmath10 .
the narrowness of the distribution of separations between the mid - infrared sources and submillimeter cores ( fig .
[ clustdia_hist ] ) indicates that the dispersal of newly formed stars in perseus is similar to that in taurus . as shown in fig .
[ clustdia_hist ] the typical separations between the mid - infrared sources and submillimeter cores are less than @xmath11 10 . assuming a typical lifetime of the deeply embedded protostars of @xmath12 years , the velocity dispersion is : @xmath13 which is smaller than the isothermal sound speed , @xmath14 km s@xmath10 for a molecular cloud with a temperature of 1015 k. @xcite examined the motions of dense cores traced by n@xmath15h@xmath16 @xmath17 emission with respect to less dense ambient envelopes traced by @xmath18co and c@xmath19o @xmath17 .
they found only small velocity differences between the `` cores '' and `` envelopes '' , which they took as evidence that dense cores did not have significant velocities with respect to their surroundings .
@xcite examined the models of @xcite using a similar approach as that in @xcite and also found small velocities of the cores relative to their lower density ambient envelopes .
it therefore appears that those simulations are not invalidated by the observations of @xcite what is shown by this analysis , however , is that the newly formed stars still embedded within their parental cores also do not have highly supersonic velocities and therefore do not seem to travel a substantial distance during their main accretion phase .
it is also interesting to note the similarity to taurus : even though taurus is thought to be forming low - mass stars in a more quiescent fashion compared to perseus , the velocity dispersions are very similar and significantly below the 1 km s@xmath10 virial speeds inferred in models of protostars accreting while moving through the molecular cloud ( and halting accretion when `` leaving '' the densest regions ) .
this study finds that approximately half the scuba cores from the study of @xcite have embedded ysos .
this suggests that the dissipation time scale of the cores once a protostar has been formed is similar to the time scale over which the cores evolve through the part of the prestellar stage observable by the scuba studies .
@xcite studied a sample of optically identified cores and found that only 24% had embedded ysos , which would imply a time scale for their optically identified , starless cores three times longer than the duration of the protostellar stage .
a comparison between the numbers from this study and those of @xcite therefore suggests that the scuba maps only detect the denser stages in the evolution of prestellar cores whereas the sample of optically identified cores by @xcite also includes less evolved cores . in a study of more dense nh@xmath20 cores
, @xcite found that the ratio of starless to stellar cores were close to unity or lower in a wide range of environments , likewise implying similar time scales for the evolution through the pre- and protostellar stages .
@xcite observed a large sample of 52 starless nh@xmath20 cores with scuba and found that the cores which were bright at submillimeter wavelengths ( i.e. , having peak 850 @xmath0 m flux densities larger than 170 mjy beam@xmath10 ) constituted approximately half of the cores detected by scuba - and that the cores detected with scuba constituted approximately half of all the studied nh@xmath20 cores .
comparison across such different samples should naturally be treated with care dealing with systematic uncertainties , selection effects , etc .
the big strength of the sample forming the basis of this paper is again that the sources are selected from one large scale systematic and relatively unbiased set of observations with uniform sensitivity . assuming a typical lifetime of the deeply embedded stages studied here of @xmath21 years , years from previous studies - which includes parts of the `` less embedded '' class i stages where the cores associated with the ysos are not picked up by the scuba surveys . ]
the total duration of the starless core stage would be @xmath22 years - with the scuba maps revealing the last @xmath23 years of this evolution where the central density increases above @xmath24 @xmath7 ( j. @xcite , h. @xcite ) . as pointed out by @xcite
these time scales are significantly shorter than those for typical ambipolar diffusion models unless the cores are marginally subcritical @xcite .
they are similar to the time scale for depletion in protostellar cores of @xmath25 years derived by @xcite based on the measurements of chemical profiles of pre- and protostellar cores - again suggesting that the scuba dust condensations reflect the stage of prestellar cores where the density is high and significant depletion can start to occur .
we can naturally also compare these numbers to the time scales for the later stages of protostellar evolution .
as pointed out above the objects identified here represent ysos in their first @xmath26 years of evolution and are an important addition to the sample of @xcite .
that study was missing some of these deeply embedded objects but more complete in terms of evolved class i and class ii objects which evolve over time scales of approximately 10@xmath27 years .
the number of sources in this list compared to those of @xcite appears to be consistent with the ratio between those time scales of about 10:1 - or alternatively if we accept those time scales as `` known '' : that star formation in perseus has occurred at a fairly constant rate continuing to current times .
we can take the arguments a step further and estimate the star formation efficiency of the submillimeter cores and the entire cloud complex . given the amount of mass in each core from the submillimeter observations
the star formation efficiency can be estimated : @xmath28 where @xmath29 is the mass of the new formed stars and @xmath30 the mass of the cores .
the total mass of the 72 scuba cores in the perseus cloud was found to be @xmath31106 @xmath32 @xcite and counting only those ysos within 15@xmath5 of one of these submillimeter cores we find 43 ysos . assuming that the masses of the newly formed stars are 0.30.5 @xmath32 , the star formation efficiency for the entire perseus cloud complex is estimated to be @xmath33 1015% .
this is comparable to the star formation efficiency of nearby embedded clusters @xcite .
approximately half of the total core mass in the cloud resides in the cores in ngc 1333 , but doing the same calculation as above the star formation efficiency for just those cores is found to be similar to the value for the entire cloud .
this again implies that significant star formation is currently occurring in the cores `` outside '' the main ngc 1333 and ic 348 clusters in the dense regions associated with l1448 , l1455 , barnard 1 and barnard 5 amongst others similar to the conclusion of @xcite .
it should be noted that the mass estimates are sensitive to the assumptions about the dust temperature and the dust opacities and to the observational techniques being more or less sensitive to extended structure , for example .
the star formation efficiency quoted above is an estimate of the current core star formation efficiency and does not represent the actual efficiency of assembling the cores from the overall cloud environment : as pointed out by @xcite ( see also @xcite for a similar result for the ophiuchus molecular cloud complex ) only a small fraction ( @xmath34 0.5% ) of the total cloud mass resides in these dense cores with the main fraction ( @xmath35 85% ) of the cloud mass from the extinction maps ( @xmath33 18,500 @xmath32 ; @xcite ) distributed at low @xmath36 .
that number can not be compared to the current mass of deeply embedded protostars from this analysis : a high number of young stellar objects have already formed in the perseus molecular cloud with for example @xcite identifying a `` high quality '' sample of 400 ysos of class ii or earlier .
these spitzer observations are furthermore not picking out a significant number of more evolved , i.e. , class iii ysos with smaller amount of infrared excesses . with these higher number of ysos distributed over all evolutionary stages and the total cloud mass ,
the time averaged star formation efficiency over the entire evolution of the cloud until present time is probably closer to a few percent , which includes the efficiency of assembling cores from the overall cloud dust mass in the first place and subsequently the efficiency of these cores actually forming stars . in this context
it should also be emphasized that the observed mass distribution of scuba core masses is a current snapshot of potentially star forming cores . with their lifetimes of
about @xmath37 years ( sect .
[ timescales ] ) it is not unreasonable to expect that at least 10 times as many cores have previously existed in the cloud and formed what are currently observed as class ii and iii young stellar objects .
this paper has presented a comparison between the dust condensations from scuba maps and mid - infrared sources from spitzer observations .
this study represents a large systematic census of the embedded yso population and thereby allows statements from a large sample of sources .
the main conclusions of this paper are : 1 .
the mid - infrared sources with 24 @xmath0 m detections and red colors are found to be located close to the center of the scuba cores mapped by @xcite - typically within 15@xmath5 ( less than half a core radius ) of their peaks .
the sources are found to have characteristic red colors with @xmath1-[4.5 ] > 1.0 $ ] and @xmath2-[24 ] > 4.5 $ ] but do not show similarly red @xmath3-[8.0]$ ] colors .
the most deeply embedded objects are found in regions with high extinction , @xmath4 ( 90% of the objects with @xmath1-[4.5 ] > 2 $ ] ) , similar to the extinction threshold for the scuba cores .
still , the red colors are not purely an effect of cloud extinction .
all the scuba cores with high `` concentrations '' have embedded ysos , but also a non - negligible number of cores with low concentrations have embedded ysos making this distinction ambiguous .
a few of the strongest sources in ngc 1333 are not picked up by the above criteria described under point 1 .
- due to saturation and confusion .
the cores associated with those sources can be picked out by their high concentrations and submillimeter fluxes .
3 . from the above considerations a relatively unbiased sample of deeply embedded ysos can be constructed . in this fashion
we identify 49 embedded ysos associated with scuba cores from the list of @xcite . out of 72 scuba cores 40
are found to have associated mips sources ( including four cores which associated with saturated mips sources not included in the c2d catalog ) .
only 3 scuba cores are found to have more than one mips sources within 15@xmath5 .
the narrowness of the spatial distribution of red sources within the scuba cores suggests that the velocities of the newly formed protostars relative to the cores are low - close to the isothermal sound speed .
this argues against the suggestion that the motions of protostars relative to the ambient cloud regulates the time scale over which significant accretion can occur .
5 . from the number of scuba cores with and without embedded ysos ( 40 and 32 , respectively )
the time scale of the evolution through the dense prestellar stages , where the cores are recognized in the submillimeter maps and their central densities are larger than @xmath24 @xmath7 , is similar to the time scale for evolution through the embedded protostellar stages .
this result further implies that the observable scuba core stage constitutes a third of the lifetime of less dense prestellar cores , e.g. , those picked up in optical or nh@xmath20 surveys .
the current star formation efficiency estimated from the mass of the embedded protostars and the scuba cores is found to be @xmath33 1015% , similar to the star formation efficiency of nearby embedded clusters .
if we also take into account the efficiency of assembling the cores in the first place , the total star formation efficiency is only a few percent .
+ the survey presented in this paper places important constraints on the formation mechanisms , which should be taken into account in theoretical studies of low - mass star formation . naturally many interesting problems remain which can be pursued using the results of this paper as a starting point : for example a more detailed characterization of the individual star forming cores ( e.g. , studying variations in dynamics from molecular line widths and/or chemical signatures ) and of the embedded ysos ( e.g. , their physical properties / evolutionary stage by completing their seds from near - infrared through ( sub)mm wavelengths ) .
it will also be interesting to test whether the criteria from this paper can be used to build comparable samples of starless and star - forming cores in other nearby clouds .
this will test whether the differing physical conditions in different star - forming regions , such as the external pressure , affects whether a given core forms stars or not .
this might for example reflect in a different value for the concentration above which all submillimeter cores in a given region is found to have embedded protostars .
this paper will in that sense serve as a starting point for such studies and for similar censuses of other clouds surveyed at mid - infrared and ( sub)millimeter wavelengths .
we are grateful to neal evans , melissa enoch and luisa rebull for useful discussions and comments about the paper .
we thank the referee for detailed comments which improved the presentation of the results .
the research of j.k.j . is supported by nasa origins grant nag5 - 13050 .
support for this work , part of the spitzer legacy science program , was also provided by nasa through contract 1224608 issued by the jet propulsion laboratory , california institute of technology , under nasa contract 1407 .
is supported by a natural sciences and engineering research council of canada grant .
is supported by a university of victoria fellowship and a national research council of canada gsssp award .
this paper has made use of the simbad database , operated at cds , strasbourg , france . | we present a census of the population of deeply embedded young stellar objects ( ysos ) in the perseus molecular cloud complex based on a combination of spitzer space telescope mid - infrared data from the `` cores to disks '' ( c2d ) legacy team and jcmt / scuba submillimeter maps from the complete team .
the mid - infrared sources detected at 24 @xmath0 m and having @xmath1-[4.5]>1 $ ] are located close to the center of the scuba cores , typically within 15 of their peaks .
the narrowness of the spatial distribution of mid - infrared sources around the peaks of the scuba cores suggests that no significant dispersal of the newly formed ysos has occurred .
this argues against the suggestion that motions of protostars regulate the time scales over which significant ( bondi - hoyle ) accretion can occur .
the ysos are found to have red @xmath1-[4.5]$ ] and @xmath2-[24]$ ] colors , but not comparable red @xmath3-[8.0]$ ] colors .
the most deeply embedded ysos are found in regions with high extinction , @xmath4 , similar to the extinction threshold observed for the scuba cores .
all the scuba cores with high concentrations have embedded ysos , but not all cores with low concentrations are starless . from the above considerations a relatively unbiased sample of 49 deeply embedded ysos
is constructed .
embedded ysos are found in 40 of the 72 scuba cores with only three cores harboring multiple embedded ysos within 15@xmath5 .
the equal number of scuba cores with and without embedded ysos suggests that the time scale for the evolution through the dense prestellar stages , where the cores are recognized in the submillimeter maps and have central densities of @xmath6 @xmath7 , is similar to the time scale for the embedded protostellar stages .
the current star formation efficiency of cores is estimated to be approximately 1015% .
in contrast , the star formation efficiency averaged over the cloud life time and compared to the total cloud mass is only a few percent , reflecting also the efficiency in assembling cloud material into the dense cores actually forming stars . |
the difference of two trajectories starting from infinitesimally nearby initial conditions , which is called the lyapunov vector , plays an essential role in the description of stability or instability in dynamical systems .
the exponential rate of expansion or contraction of the absolute value of the lyapunov vector is the lyapunov exponent , and its positivity means that the system has a dynamical instability and is called chaotic .
the lyapunov vector is introduced in each independent direction of phase space , so in general we have to consider a set of lyapunov exponents , called the lyapunov spectrum , and the corresponding lyapunov vectors in a high - dimensional chaotic system .
an algorithm to calculate the lyapunov exponents and vectors in many - body systems has been developed by benettin et al .
@xcite , and shimada and nagashima @xcite , and the behavior of lyapunov vectors has been investigated from various points of view , for example , the conjugate pairing rule for lyapunov spectra in some thermodynamic systems @xcite , and the localization behavior of lyapunov vectors @xcite , etc .
recently , wavelike structure of lyapunov vectors , called lyapunov modes , and the corresponding stepwise structure of lyapunov spectra has drawn attention in many - particle chaotic systems @xcite .
this structure appears in the region of lyapunov exponents with small absolute values .
lyapunov exponents have the dimension of inverse time , so the structure in the small lyapunov exponents is supposed to be a reflection of the slow and macroscopic behavior of many - particle systems . since the first observation of lyapunov modes in computer simulations of hard disk systems @xcite
they have been found in systems with soft potentials , independent of dimensionality and system geometry .
it is now believed that the zero modes arise due to the presence of conserved quantities @xcite , and the spatial dependence of the non - zero modes is essentially analogous to higher order fourier components .
it has been shown that the breaking of conserved quantities , by for example a change from periodic to hard wall boundary conditions , leads to the absence of particular modes .
it is also known that there are stationary modes and time - oscillating modes among the non - zero lyapunov modes , and the time - oscillating period of the lyapunov modes is connected to that of an equilibrium momentum auto - correlation function @xcite . in this paper
we consider the effect of breaking energy conservation by adding energy at one end of the system and removing it at the other .
although this leads to a flow of heat energy through the system , this is not intended to be a realistic model of low dimensional heat flow .
quasi - one - dimensional systems have been extensively used to investigate lyapunov modes and the localization or delocalization of lyapunov vectors for many - particle systems at equilibrium @xcite , and these systems can easily be adapted to maintain a heat flow .
using such a system , we investigate nonequilibrium effects of the lyapunov modes and the momentum auto - correlation functions .
we consider a quasi - one - dimensional many - hard - disk system with a mechanism to insert energy at one end of the system and to remove energy from the other end .
it consists of @xmath0 hard disks of radius @xmath1 and mass @xmath2 in a two - dimensional rectangular region of length @xmath3 and width @xmath4 , with @xmath5 so that the disk order remains invariant .
we choose periodic boundary conditions in the transverse @xmath6-direction , and in the longitudinal @xmath7-direction we use a special boundary condition that allows the transfer of energy between the system and walls : @xmath8 where @xmath9 , @xmath10 is boltzmann s constant and @xmath11 is a parameter ( @xmath12 ) . in a collision , @xmath13 is the incoming momentum of the particle and @xmath14 is the outgoing momentum .
the parameter @xmath11 controls the strength of the coupling to the heat reservoir " of temperature @xmath15 ( where @xmath16 is either @xmath17 or @xmath1 ) .
when @xmath18 the incoming momentum @xmath19 is completely replaced by the mean thermal momentum @xmath20 of the reservoir and all information contained in the incoming momentum is lost . at @xmath21 the connection with the reservoir is removed and the collision process is hard wall . here
we consider values of @xmath11 between @xmath22 and @xmath23 so that the reservoir is coupled to the system but the loss of information is not complete .
we choose the temperatures of the reservoirs independently , so the temperature of the left - hand side reservoir is @xmath24 , and the right - hand side reservoir is @xmath25 . a schematic illustration of this system is given in fig .
[ figb1qua1dim ] . for the numerical results that are shown in this paper
, we use the values @xmath26 , @xmath27 , @xmath28 , @xmath29 and @xmath30 . from left to right ) .
when disks collide with walls at either end of the system energy is transferred between the disks and walls depending upon the values of the temperatures @xmath31 and @xmath32 . ]
note that boundary condition ( [ boundcondi2 ] ) with periodic boundary conditions in the @xmath6-direction guarantees that the total momentum in the @xmath6-direction is conserved .
the parameter @xmath11 determines the strength of the coupling between the reservoir and the system , and specifies the fraction of incoming momentum @xmath19 that is retained after the collision . if @xmath33 and @xmath34 then an energy transfer ( heat ) occurs across the system , and a nonequilibrium steady state is established after a long time .
disks with @xmath35 . the reduced lyapunov exponents ( @xmath36 ) with smallest magnitude are shown in the main plot .
the inset is a plot of the positive exponent region of the full lyapunov spectrum .
the circles are the smallest positive exponents , while the crosses are the absolute values of the conjugate negative exponents .
the horizontal axis is the exponent number @xmath37 , numbered from @xmath23 ( the largest ) to @xmath38 ( the smallest ) . ]
figure [ figc1lyapu ] is a graph of the absolute value of the normalized lyapunov spectrum @xmath36 with @xmath39 . here
the largest lyapunov exponent is @xmath40 . from this figure
we see that the positive branch and the negative branch of the lyapunov spectrum are not symmetric , namely @xmath41 particularly for the exponents of smallest magnitude near @xmath42 .
this is different from the equilibrium case ( @xmath43 ) where the dynamics is hamiltonian with the symplectic structure guaranteeing the conjugate pairing rule for the lyapunov spectra @xmath44 @xcite .
the system has 3 zero - lyapunov exponents , due to total momentum conservation and spatial translational invariance in the @xmath45direction , and time - translational invariance . in the nonequilibrium case ( @xmath34 )
the total energy of the system is no longer conserved and this is the reason that @xmath46 , although it is zero at equilibrium .
.classification of the exponents in the lyapunov spectrum for the quasi - one - dimensional system of @xmath47 particles .
the four zero exponents are 199 , 200 , 201 and 202 at equilibrium . [ cols="^,^,^,^,^ " , ] [ default2 ] the period @xmath48 ( in numbers of collisions ) of the auto - correlation function ( acf ) @xmath49 for the longitudinal component of the momentum increases , and the amplitude decreases as a function of @xmath11 ( see table [ default2 ] ) .
the time oscillating period @xmath50 of the lyapunov vectors for the @xmath51-th longitudinal ( and the momentum proportional ) mode in the positive branch of the lyapunov spectrum decreases slightly as a function of @xmath11 , whereas in the negative branch the period increases slightly as a function of @xmath11 .
such changes of @xmath52 are especially large for small @xmath51 s , which correspond to the ones close to the zero - lyapunov exponents .
the relation @xmath53 is not correct away from equilibrium , while it is justified in equilibrium @xcite .
this does not contradict the proof of this relation as that required equilibrium and is based on the functional form of the lyapunov modes , in particular that they are purely sinusoidal .
the changes in the structure of the lyapunov modes for a system maintained away from equilibrium by a boundary imposed heat flow are described in detail .
the breaking of conservation of energy is shown to introduce systematic effects , changing the number of zero exponents , and modifying the form of other modes , with the basic structure of transverse , longitudinal and momentum proportional modes remaining .
this work is largely descriptive and the theoretical origin of the changes in modes is yet to be developed , as it is for the equilibrium case .
however , it is clear that the modes are not only equilibrium properties of many - particle systems , but also fundamental for nonequilibrium steady states . the original purpose in introducing the nonequilibrium boundary conditions [ i.e. eqs .
( [ boundcondi1 ] ) and ( [ boundcondi2 ] ) ] was to keep dynamical properties , like the momentum conservation in the transverse direction and deterministic property of orbit , etc .
, in a nonequilibrium dynamics while breaking energy conservation .
the equilibrium case is recovered in the limit as @xmath54 .
therefore this choice of the boundary conditions was not intended to provide a realistic model for the interaction between a heat reservoir and a particle system .
for example , a consequence of these boundary conditions is that the momentum distribution function for a particle in contact with the reservoir is very different from a gaussian .
indeed , we found that the incoming momentum is close to gaussian but the outgoing momentum is strongly peaked around the mean momentum of the reservoir , with the width of its distribution determined by the degree of mixing with the incoming momentum ( that is , determined by the value of @xmath11 ) .
the authors appreciate a careful reading of the manuscript by e. g. d. cohen , and the financial support of the japan society for the promotion of science .
g. benettin , l. galgani , and j. -m .
strelcyn , phys .
a * 14 * , 2338 ( 1976 ) .
g. benettin , l. galgani , a. giorgilli , and j. -m .
strelcyn , meccanica * 15 * , 9 ( 1980 ) .
g. benettin , l. galgani , a. giorgilli , and j. -m .
strelcyn , meccanica * 15 * , 21 ( 1980 ) .
i. shimada and t. nagashima , prog .
theor . phys . *
61 * , 1605 ( 1979 ) .
u. dressler , phys .
a * 38 * , 2103 ( 1988 ) .
d. j. evans , e. g. d. cohen , and g. p. morriss , phys . rev .
a * 42 * , 5990 ( 1990 ) . c. p. dettmann and g. p. morriss , phys .
e * 53 * , r5545 ( 1996 ) .
t. taniguchi and g. p. morriss , phys .
e * 66 * , 066203 ( 2002 ) .
p. manneville , lecture note in physics , * 230 * , 319 ( 1985 ) .
k. kaneko , physica d * 23 * , 436 ( 1986 ) .
r. livi and s. ruffo , in _ nonlinear dynamics _ , edited by g. turchetti ( world scientific , singapore , 1989 ) , p. 220 .
m. falcioni , u. m. b. marconi , and a. vulpiani , phys .
a * 44 * , 2263 ( 1991 ) . g. giacomelli and a. politi , europhys . lett . * 15 * , 387 ( 1991 ) . lj . milanovi and h. a. posch , j. mol .
liquids , * 96 - 97 * , 221 ( 2002 ) .
t. taniguchi and g. p. morriss , phys .
e * 68 * , 046203 ( 2003 ) .
t. taniguchi and g. p. morriss , phys .
e * 73 * , 036208 ( 2006 ) .
dellago , h. a. posch , and w. g. hoover , phys .
e * 53 * , 1485 ( 1996 ) .
h. a. posch and r. hirschl , in _ hard ball systems and the lorentz gas _ , edited by d. szsz ( springer , berlin , 2000 ) , p. 279 .
eckmann and o. gat , j. stat . phys . *
98 * , 775 ( 2000 ) .
t. taniguchi and g. p. morriss , phys .
e * 65 * , 056202 ( 2002 ) .
s. mcnamara and m. mareschal , phys .
e * 64 * , 051103 ( 2001 ) .
a. s. de wijn and h. van beijeren , phys .
e * 70 * , 016207 ( 2004 ) .
t. taniguchi , c. p. dettmann , and g. p. morriss , j. stat .
phys . * 109 * , 747 ( 2002 ) .
h. l. yang and g. radons , phys .
e * 71 * , 036211 ( 2005 ) .
t. taniguchi and g. p. morriss , phys .
e * 68 * , 026218 ( 2003 ) .
t. taniguchi and g. p. morriss , phys .
e * 71 * , 016218 ( 2005 ) .
eckmann , c. forster , h. a. posch , and e. zabey , j. stat . phys . * 118 * , 813 ( 2005 ) . t. taniguchi and g. p. morriss , phys .
lett . * 94 * , 154101 ( 2005 ) . | we present the first numerical observation of lyapunov modes ( mode structure of lyapunov vectors ) in a system maintained in a nonequilibrium steady state .
the modes show some similarities and some differences when compared with the results for equilibrium systems .
the breaking of energy conservation removes a zero exponent and introduces a new mode .
the transverse modes are only weakly altered but there are systematic changes to the longitudinal and momentum dependent modes . |
some galaxies are known to emit radiation with extremely high luminosities in the @xmath4-ray , x - ray , uv and radio continuum from a very concentrated volume in the nuclear region .
such active cores are the so - called active galactic nuclei ( agn ) and their radiation is believed to be produced by accretion onto a supermassive black hole . the intrinsically complex nature of such systems and the differences in the terminology among different scientific communities ( radio , optical , x - ray astronomers ) has led to an extremely complicated nomenclature for the agn zoo . as the wealth of observations piled up , and with them the number of different agn types , the opposite enterprise of finding unification schemes has progressively gained support . the basic idea behind
the standard unification scheme is that agn are asymmetric and anisotropic systems .
this is natural , as all rotating systems necessarily single out a preferential axis in space and break the full spherical symmetry of non - rotating bodies .
therefore , the orientation of the agn rotation axis with respect to our line of sight becomes another important parameter that can cause apparent observational differences in two sources that are intrinsically identical ( _ unification by orientation _ ) . according to the current
widely accepted paradigm ( urry and padovani 1995 ) , there are two principal causes of anisotropic radiation : obscuration and relativistic beaming .
the former is usually associated with a torus of gas and dust obscuring the optical , uv and ( sometimes ) soft x - ray radiation along some line of sights , the latter with outflows of energetic particles ( jets ) along the symmetry axis : the high velocity plasma in the jet beams radiation relativistically in the forward direction .
the powerful source of radiation and the launching site of the collimated jet lies in the center of the system .
observationally , jet morphologies and spectral properties of both radio and x - ray cores are remarkably similar in the case of black holes of stellar mass ( galactic black holes , hereafter gbh ) and of their supermassive counterparts in the nuclei of galaxies ( hereafter smbh ) .
if jets are launched in the innermost parts of the accretion flows , as commonly assumed , then these similarities suggest that it should be possible to extend the unification scheme further and to understand the physics of both black hole accretion and jet production by studying all those systems as a single class ( _ unification by mass _ ) .
furthermore , the recent discovery that smbh lie at the center of the majority ( possibly all ) galactic nuclei , even in apparently inactive ones ( kormendy and richstone 1995 ; magorrian et al .
1998 ) , naturally leads to the idea that a grand unification is possible by taking into account the differences in fueling rates among different objects ( _ unification by accretion rate _ ) . in the following
, we will show how to proceed quantitatively towards a grand unification of active black holes by studying the multivariate correlation among masses , radio luminosities and hard x - ray luminosities of objects traditionally classified in the more diverse ways .
the underlying theoretical assumption is that both accretion and jet production are fundamental manifestation of black hole activity , and are somehow physically connected ( begelman et al .
1984 ; rawlings and saunders 1991 ; falcke and biermann 1995 ; heinz and sunyaev 2003 ) .
here we briefly describe the main results of the correlation analysis carried on in merloni , heinz and di matteo ( 2003 , hereafter mhd03 ) .
a more detailed description of the sample used , of the selection effects and of the statistical analysis can be found there , together with a comprehensive list of references to the observational data . for future reference
, we define the dimensionless black hole mass @xmath5 and accretion rate @xmath6 , where @xmath7 is the accretion efficiency .
we have selected from the existing literature a sample of black hole - powered systems with measured masses , the nuclei of which have been observed both at 5 ghz ( mostly with arcsecond resolution with the vla ) and in the 2 - 10 kev band .
the main obvious advantage of this choice lies in the fact that obscuration is unimportant , or easily accounted for , in these spectral bands .
we first considered the full sample of @xmath8 nearby inactive , or weakly active galaxies with existing nuclear black hole mass measurements from observations of spatially resolved kinematics . to these
we have added a comparable number of bright agns ( and qsos ) with nuclear black hole mass measured from reverberation mapping of their broad line region . from this sample we selected all objects which have been observed in both the radio and x - ray bands . in order to obtain a more statistically representative sample
, we also searched the existing literature for both nearby low - luminosity galactic nuclei and for relatively bright seyfert nuclei ( either type 1 , type 2 or narrow line seyfert 1 ) and radio galaxies with available radio and x - ray flux measurements .
we assign black hole masses to these systems using the observed correlation between black hole masses and stellar velocity dispersion ( gebhardt et al . 2000 ; ferrarese and merritt 2000 ) .
relativistically beamed sources ( i.e. those whose jet axis points towards our line of sight ) are dominated by the boosted jet emission , and can not be used to test the disc - jet coupling .
we therefore excluded from our sample bl lac objects . among the quasars in our sample ,
only 3c 273 , which has an extremely high radio loudness and a blazar - like spectrum , is likely to suffer from strong doppler boosting of the radio jet . on the other hand , according to the unification scheme , seyfert 2 nuclei should not be preferentially viewed pole on , while for all the other sources ( mainly low - luminosity agn and seyfert 1 ) , for which the nature of the ( relatively faint ) radio emission is not well established , we have assumed that the orientation of their jets with respect to line of sight is randomly distributed . of x - ray to eddington luminosity .
upper limits are marked with arrows , different symbols indicate objects belonging to different spectral classes and different colors objects in different mass bins .
the color - coding of the different mass bins makes the mass segregation more evident.,scaledwidth=90.0% ] the galactic x - ray binaries included in our sample have been selected to have ( a ) simultaneous x - ray and radio observations , or _
rxte _ all - sky - monitor ( asm ) x - ray data in conjunction with radio fluxes available from the literature , and ( b ) publicly available _ rxte_-asm x - ray and green - bank interferometer ( gbi ) radio lightcurves ( from which we estimated the 5 ghz fluxes by interpolating between the 2.25 ghz and the 8.3 ghz channels ) .
the final sample consists of @xmath0100 active galactic nuclei with measured mass , 5 ghz core emission , and 2 - 10 kev luminosity , together with 8 galactic black holes with a total of @xmath0 50 simultaneous observations in the radio and x - ray bands ( see mhd03 for a full list of the sample sources ) . in figure
[ fig : lr_m2 ] we show the radio core luminosity versus the black hole mass for objects of different spectral classes .
we concentrate on the smbh only in order to show how , at a fixed black hole mass , sources which are classically regarded as powerful accretors ( qsos , narrow line seyfert 1 , seyfert 1 ) tend to lie above the so - called low - luminosity agn ( here represented by liners ) , with seyfert 2 galaxies spanning a large area in the vertical direction . figure [ fig : lr_mdot2 ] shows instead the core radio luminosity versus the ratio of the x - ray nuclear luminosity to the eddington luminosity ( probably a good estimator of @xmath9 , see below ) .
we represent objects in different mass bins with different colors .
it is clear that , when the data points are grouped into mass bins , objects in different bins tend to lie on parallel tracks .
the presence of a mass segregation suggests that the radio luminosity of an object likely depends both on its accretion rate and on its mass .
we can proceed to quantify the degree of correlation between our three observables ( radio luminosity at 5 ghz , @xmath1 , x - ray luminosity in the 2 - 10 kev band , @xmath10 and black hole mass ) .
we fit the data with the function @xmath11 , and obtain well constrained values for the correlation coefficients , meaning that , if we define the instantaneous state of activity of a black hole of mass @xmath3 , by the radio and hard x - ray luminosity of its compact core , and represent such an object as a point in the three - dimensional space ( @xmath12 ) , all black holes ( either of stellar mass or supermassive ) will lie preferentially on a plane ( the `` fundamental plane '' of black hole activity , see figure [ fig : fp ] ) , described by the following equation : @xmath13 with a dispersion @xmath14 ( see figure [ fig : fp ] ) .
the value we obtain for the @xmath15 correlation coefficient is consistent , within the errors , with that found in low / hard state gbh ( @xmath16 ) by gallo et al .
this also means that individual gbh sources for which the correlation between radio and x - ray luminosities is well established ( gx 339 - 4 and v404 cyg ) do indeed follow the same global trend defined by black holes of all masses included in our sample . )
in a recent paper , heinz & sunyaev ( 2003 ) have demonstrated that , under the general assumption that the jet formation process is not qualitatively different among smbh of different mass or between smbh and gbh , it is in fact possible to derive a universal scaling between the jet ( radio ) luminosity at a given frequency and both mass and accretion rate . the derived relation is independent of the jet model and has scaling indices that depend only on the ( observable ) spectral slope of the synchrotron emission in the radio band ( @xmath17 ) , and on the accretion model . in particular , it was shown that , if the magnetic field @xmath18 at the base of the jet is in equipartition with the particles pressure and the total jet power scales as @xmath19 ( see also falcke & biermann 1995 ; heinz et al .
, this proceedings ) , and under the standard assumption that the electrons responsible for the jet synchrotron emission follow a power - law distribution in energy , with index @xmath20 , then @xmath21 . for the optically thick , flat spectrum cores we are mostly interested in ( @xmath22 ) , then : @xmath23 to compare this prediction with the observed fundamental plane correlation , one needs to explicate the dependence of the hard x - ray luminosity on the accretion rate , @xmath24 . in mhd03
it was shown how indeed the prediction of standard synchrotron theory for the radio - x - ray - mass correlation of heinz & sunyaev ( 2003 ) are verified if , and only if , @xmath25 , i.e. if the accretion process is radiatively inefficient .
detailed models for the x - ray emission from such flows do indeed show that @xmath26 ( mhd03 ) , and in this case @xmath27 , i.e. , @xmath1 scales with the physical accretion rate only .
this is tantalizingly close to the predicted dependence of eq .
( [ eq : hs03 ] ) , and such a result can be interpreted in the following way : 1 ) the scale invariance hypothesis at the heart of the heinz and sunyaev ( 2003 ) calculations is correct , and black holes can indeed be further unified by considering their masses and accretion rates , and 2 ) the largest area of the fundamental plane is covered by radiatively inefficient sources .
the total power released by the accretion / jet system may be written as @xmath28 , where the first term on the right hand side is the total radiated luminosity and the last one include contributions from the energy advected and/or stored in the convective motions .
our results suggest that the flow must be radiatively inefficient , therefore , for small enough accretion rates we have @xmath29 . on the other hand , @xmath30 .
therefore , the issue of what the relative fraction of the total accretion energy dissipated into the jet is ( or , alternatively , of when a source is `` jet dominated '' ; fender , gallo & jonker 2003 ; falcke , krding & markoff 2004 ) reduces to the determination of the value of the constant @xmath31 .
this requires the specification of a jet model or the direct measure of the total kinetic power carried by the jet ( see , e.g. heinz et al .
2004 ) , together with a dynamical model for the disc - jet coupling .
it is well accepted , both from theory and observations , that accretion can proceed in different modes ( or states ) , with different radiative efficiencies and spectral properties . by fitting the whole dataset at our disposal with a single linear relationship ( [ eq : fp ] ) ,
we have implicitly neglected the possibility of a global accretion mode change .
this is clearly implausible , as the qsos and the bright seyferts in our sample , which occupy the region of high accretion rates , are independently known to have spectral characteristics inconsistent with models of low radiative efficiency .
they should therefore depart from the observed correlations .
for gbhs , it has indeed been shown that the correlation between radio and x - ray luminosity breaks down as the sources switch to their high states ( maccarone 2003 ; gallo , fender and pooley 2003 ) . however , because both such modes of accretion are expected to occur only above accretion rates about a few percent of eddington , and because another advective accretion mode is expected to ensue at around the eddington limit ( due to efficient radiation trapping , see e.g. abramowicz et al .
1995 ) , we would expect the @xmath32 correlation to break down only in a limited range of @xmath9 .
in other words , independently on the actual number of radiatively efficient sources , the area of the fundamental plane covered by them will always be limited , and the overall orientation of the plane will always be dominated by the radiatively inefficient ones .
this also means that any ( statistically significant ) departure from the fundamental plane relation could be used to identify different modes of accretion .
in fact , maccarone , gallo & fender ( 2003 ) have already shown how is possible to use the fundamental plane relation as a baseline against which identify agn in a high / soft state ( hss ) analogous to that of x - ray binaries . ,
divided by @xmath33 as a function of the ratio @xmath34 .
upper limits are marked with arrows , different symbols indicate objects belonging to different spectral classes and different colors objects in different mass bins .
two vertical lines mark the boundary of the region where we expect the critical luminosity for the mode change between radiatively inefficient and efficient accretion . to the left of these lines , gbh
are in the low / hard state and smbh are mostly fr i. to the right , gbh are in the very - high state ( vhs ) and smbh are mostly fr ii . in between , possibly the restricted region of the parameter space where pure discs accretion is allowed ( high / soft state , hss).,scaledwidth=90.0% ] the analogy with gbh can indeed be very useful to classify and understand the properties of bright agn .
let us consider , for example , a recent paper by fender , belloni and gallo ( 2004 ) , where it was shown that the intermittent , powerful radio flares from the so - called microquasars , associated with the rapid variability in the very high state ( vhs ) , seem to follow a similar radio - x - ray correlation as the low / hard state sources , albeit with a much larger scatter .
can we extend this result to the larger family of radio loud agn ? in figure
[ fig : lr_mdotx_fit ] we plot , as a function of the ratio @xmath34 , the radio luminosity divided by @xmath33 of all the sources in our sample . as expected , by rescaling the radio luminosity in such a way all the different tracks corresponding to different mass bins in fig .
[ fig : lr_mdot2 ] collapse into a single one , with some residual scatter .
the region between the two vertical lines corresponds to the theoretically expected values of @xmath34 above which a change of accretion mode , from radiatively inefficient to standard radiatively efficient is expected to occur . to the left of these lines , gbh
are in the low / hard state and smbh are mostly fr i , or low - luminosity , radio loud agn . to the right ,
gbh are in the very - high state ( vhs ) and smbh are mostly fr ii . in between
, possibly lies the restricted region of the parameter space where pure discs accretion is allowed ( hss ) .
interestingly , it appears as if both low and high luminosity sources at the two sides of the hss region obey a similar scaling @xmath35 .
we may thus speculate that the famous ( and still much debated ) radio loud / radio quiet dichotomy of quasars will appear only at the highest values of @xmath9 , and be caused mainly by a switch of accretion mode analogous to the high / very high transition in gbh . at low accretion rates ,
black holes seem to follow the more regular behavior circumscribed by the fundamental plane of eq .
( [ eq : fp ] ) .
such sources not only tend to be radio loud ( ho & peng 2001 ; ho 2002 ) , but also their radio loudness parameter , @xmath36 ( here defined as the ratio of radio to x - ray luminosity ) , obeys the following scaling : @xmath37 .
therefore , the smaller the x - ray luminosity , the more radio loud these sources are . in this regime ,
no dichotomy need be expected , as already suggested by nagar et al .
we have argued that the fundamental plane analysis presented here is a powerful tool to extend the unified scheme of accreting black holes .
such a relation between mass , radio and hard x - ray luminosity is affected very little by obscuration and beaming , provided that sources whose relativistic jets are in our line of sight can be effectively identified and excluded .
thus , the fundamental plane relation does not depend on orientation , and as such is complementary to the standard unification scheme .
moreover , the relation itself is perfectly consistent with the scaling relations predicted by standard synchrotron theory under a scale invariance assumption .
the main scaling parameters are the mass of the black hole and its accretion rate .
finally , we have shown how the observed correlation can be effectively used to classify objects on the basis of their mode of accretion ( and/or accretion / ejection coupling ) rather than just on specific observational characteristics , as in the true spirit of unification models .
heinz , s. , r. a. sunyaev , a. merloni , t. di matteo , 2004 , to appear in the proceedings of `` growing black holes '' , garching , germany , june 21 - 25 , 2004 .
a. merloni , s. nayakshin and r. sunyaev , springer - verlag series of `` eso astrophysics symposia '' . | we examine the disc jet connection in stellar mass and supermassive black holes by investigating the properties of their compact emission in the hard x - ray and radio bands .
we compile a sample of @xmath0100 active galactic nuclei with measured mass , 5 ghz core emission , and 2 - 10 kev luminosity , together with 8 galactic black holes with a total of @xmath0 50 simultaneous observations in the radio and x - ray bands . using this sample , we study the correlations between the radio ( @xmath1 ) and the x - ray ( @xmath2 ) luminosity and the black hole mass ( @xmath3 ) .
we find that the radio luminosity is correlated with _ both _ @xmath3 and @xmath2 , at a highly significant level .
we show how this result can be used to extend the standard unification by orientation scheme to encompass unification by mass and accretion rate . |
starting from the pioneering works of edwards @xcite , statistical studies of winding properties of topologically constrained random processes have been a subject of keen interest in various contexts such as in the physics of polymers @xcite , the fluxlines in superconductors @xcite and many others .
the winding properties of planar brownian paths have also been a subject of interest to mathematicians for a long time . in 1958 , spitzer @xcite studied the distribution of the total angle @xmath1 wound by a planar brownian path around a prescribed point in time @xmath0 .
he showed that in the large time limit the scaled random variable @xmath2 has a cauchy distribution _
i.e. _ with infinite first moment . later
, various generalizations and extensions of this classic result have been put forward @xcite .
further asymptotic laws of planar brownian motion unifying and extending this approach to the case of @xmath3 different points were obtained by pitman and yor @xcite .
recently , winding properties of other planar processes such as time correlated gaussian process @xcite , schramm - loewner evolution@xcite , loop erased random walk @xcite and conformally invariant curves @xcite have also been studied .
for constrained brownian trajectories on a surface , the methods of computing probabilities associated with the winding properties are mathematically similar to the path integral of a quantum particle coupled to magnetic fields @xcite .
this similarity has been exploited in various situations , for example , the algebraic number of full turns a polymer makes around a cylinder on which it lays , can be analyzed by studying the physics of a charged quantum particle interacting with a tube of magnetic flux @xcite .
thermal fluctuation in the trajectory of a vortex defect in the superconducting order parameter @xcite , which are characterized by winding numbers , can be analyzed by studying path integrals of a polymer melt .
these kind of path integral techniques have also been used to study other quantities , like e.g. the area between a brownian path and its subtending chord @xcite .
other functionals of the brownian motion that occur in the context of weak localization have been discussed in @xcite . in this paper
we study winding properties of a brownian particle moving on a ring of unit radius ( see fig .
[ fig1 ] ) analytically . in mesoscopic physics ,
such diffusion on a ring geometry appears important in quantum transport through a connected ring .
the weak localization correction to the classical conductance of the ring can be computed from the knowledge of the net number of complete turns around the ring @xcite .
here we are , in particular , interested in the distributions of total number of turns @xmath4 and the net number of counter - clockwise turns @xmath5 the particle makes around the circle till time @xmath0 in the following two different cases : ( a ) _ free _ : the position @xmath1 of the particle at time @xmath0 is not constrained and , ( b ) _ constrained _ : the position @xmath1 of the particle at time @xmath0 is constrained to be the starting point ( a in fig . [ fig1 ] ) _ i.e. _ @xmath6 where @xmath7 is an integer .
the path configurations in case ( b ) are called brownian bridges on a circle .
if @xmath1 represents the position of the particle on the circle , then its stochastic evolution is given by the following equations @xmath8 where @xmath9 is the diffusion constant and @xmath10 is a mean zero white gaussian noise .
as time increases , the particle turns around the circle both clockwise and counter - clockwise .
let @xmath11 be the number of full counter clockwise turns and @xmath12 be the number of full clockwise turns the particle makes in time @xmath0 , then the total number of full turns and the net number of full counter - clockwise turns are given by @xmath13 and @xmath14 respectively .
clearly , @xmath4 and @xmath5 are random variables and they are called the total winding number and net winding number respectively .
we would like to compute the distributions @xmath15 of @xmath4 , @xmath16 of @xmath5 and also the joint distribution @xmath17 for both the _ free _ and the _ constrained _ cases . in previous studies for example in the context of polymers @xcite , in the context planar brownian trajectories @xcite or in the context of vortex lines in type ii superconductors @xcite , path integral techniques have been successfully used to evaluate winding statistics of net winding numbers , where one maps the problem of finding the probabilities to the problem of finding the propagator of a suitable quantum particle .
in contrast , here it seems adapting the path integral techniques is not easy .
for example , it is hard to implement the path integral techniques for computing the distributions of quantities like the total winding number @xmath4 and the net winding number @xmath5 especially in the _ free _ case .
the main purpose of this paper is to present an alternative method based on _ renewal _ properties of the process that allows us to compute the winding distribution exactly .
next we study the case where the trajectories of the brownian particle are constrained to be exactly at @xmath41 with @xmath42 at time @xmath0 . in this case
we denote the probability distributions of @xmath3 and @xmath18 by @xmath43 and @xmath44 respectively and the joint distribution by @xmath45 where the subscript `` c '' stands for _ constrained _ trajectories .
we find exact expressions of these distributions at any time @xmath0 .
moreover , we show that in the large @xmath0 limit , the scaling forms and the large deviation functions associated to the distributions @xmath43 and @xmath44 are exactly same as in the previous case _ i.e _ the scaling distributions are described by the same functions @xmath46 and @xmath47 whereas the large deviation functions are also described by @xmath39 and @xmath40 .
the paper is organized as follows . in section
[ free ] we study the winding statistics for the _ free _ case _ i.e. _ for unconstrained brownian motion on the ring . in the beginning of this section
we discuss in detail about the connection between making a complete turn around the circle and making first exit from a box of size @xmath48 .
then in the subsection [ mvs - dis ] we compute the time dependence of the moments of @xmath3 and @xmath18 and also their scaling distributions . in the next subsection [ ldfs ]
we find the large deviation forms associated to the distributions @xmath15 and @xmath16 . in subsection [ exact ]
we derive exact explicit expressions of the @xmath15 and @xmath16 for arbitrary @xmath3 and @xmath18 . after completing the analysis for the _ free _ case , in section [ brow - brdg ] we study the moments and the distributions @xmath43 and @xmath44 corresponding to @xmath3 and @xmath18 for brownian bridge on a ring . in the last section [ conclusion ] we present our conclusion .
few details are left in the appendix .
as mentioned in the introduction , the renewal properties of the brownian walker helps us to find the distributions of the number of complete counter - clockwise turns @xmath11 and the number of complete clockwise turns @xmath12 around the circle in time @xmath0 . to see how , let us look at ( see fig .
[ fig1a ] ) a typical trajectory of the brownian particle on the ring , starting at the origin . in fig .
[ fig1a ] we see that the particle makes its first full turn around the circle at time @xmath49 when it makes , starting at the origin , a first exit from the box @xmath50 $ ] ( magenta box ) . in the next time interval
@xmath51 $ ] the particle , starting at @xmath52 , makes a first exit from the box @xmath53 $ ] ( blue box ) and thus completing its second full turn around the circle at time @xmath54 .
similar first exit events happen for each of the successive complete turns at later time intervals except in the last time interval where the particle may not have enough time to perform a complete turn .
so we see that , each complete turn around the circle is associated to a first exit from a box of size @xmath48 given that the particle starts from the center of the box and after each complete turn the location of the center of the box gets shifted to the particle s current position to perform similar first exit for the next complete turn _
i.e. _ to renew the first exit process for the next complete turn . .
this particular trajectory wounds the circle around completely five times in time intervals @xmath55 successively .
each of these complete turn around the circle is associated to the first exit of the particle from a box of size @xmath48 given that the particle started from the center of the box .
after each complete turn this first exit process gets renewed with the center of the box shifted to the particle s current position as shown by the different colored boxes .
the bottom panel shows the increase of @xmath56 and @xmath57 with time corresponding to this particular trajectory . ] the first exit through the upper boundary of the box corresponds to a full counter - clockwise turn around the circle whereas the first exit through the lower boundary corresponds to a full clockwise turn around the circle . using this connection with first exit problem from a box of length @xmath48 and the renewal property of a brownian walker we study the distributions of having total winding number @xmath3 and net winding number @xmath18 in time @xmath0 for both _
free _ and _ constrained _ cases .
if @xmath58 represents the first exit probability distribution of the particle from the box of size @xmath48 given that it had started from the center of the box , then the probability that the particle starting from the origin , makes a complete counter - clockwise turn in time @xmath0 to @xmath59 , is given by @xmath60 .
similarly , by symmetry ( as the particle starts at the center of the box ) , the probability that it makes a clockwise turn in time @xmath0 to @xmath59 , is also given by @xmath60 . to compute @xmath58 , it turns out to be convenient to study @xmath61 which represents the survival probability that the particle , starting from position @xmath62 , stays inside the box @xmath50 $ ] ( see fig .
[ fig2 ] ) till time @xmath0 . knowing @xmath61 , the first exit probability
is then given by @xmath63 the survival probability @xmath61 satisfies the following backward fokker - planck equation @xcite @xmath64 with initial condition @xmath65 and boundary conditions ( bcs ) @xmath66 . to solve the above equation it is convenient to use the laplace transform ( lt ) and inverse laplace transform ( ilt ) which for a general function @xmath67 are defined as follows : @xmath68 = \int_0^{\infty}dt~ e^{-st}g(t)~ , \nonumber \\
\text{ilt~of}~\tilde{g}(s ) : & & g(t ) = \mathcal{l}_t^{-1}[\tilde{g}(s ) ] = \frac{1}{2\pi i } \int_{\mathcal{b } } ds ~\tilde{g}(s)~e^{st } , \label{lt - def}\end{aligned}\ ] ] where @xmath69 stands for the bromwich integral in the complex @xmath70 plane . taking the laplace transform @xmath71 $ ] on both sides of eq . ( [ bfp1 ] )
, we get @xmath72 with bcs @xmath73 , whose solution at @xmath74 is given by @xmath75,~~ \text{which~provides}~~q(t)=\mathcal{l}_t^{-1}[\tilde{q}(s ) ] . \label{qtilde}\ ] ] now taking derivative of @xmath76 with respect to @xmath0 we get @xmath77 from eq .
( [ fpndsvl ] ) whose lt is given by @xmath78=1-s\tilde{q}(s ) = \text{sech } \left(2\pi\sqrt{\frac{s}{d}}\right ) .
\label{ftilde}\ ] ] knowing @xmath76 and @xmath58 , we proceed to compute @xmath15 as follows .
consider the event of @xmath3 complete turns in @xmath79 $ ] and let @xmath80 denote their respective durations ( see fig .
[ fig1a ] ) .
the difference @xmath81 denotes the duration of the last unfinished turn .
the probability of such an event , where both the number of turns @xmath3 as well as their durations @xmath82 are random variables , is denoted by @xmath83 and can be expressed as @xmath84 where @xmath85 is the survival probability in eq .
( [ qtilde ] ) and @xmath86 is the first exit probability from the box .
the first @xmath3 factors involving @xmath87 in eq . ( [ jpdf1 ] ) represent @xmath3 complete turns ( or the first exits from the box of length @xmath48 as explained in fig .
[ fig1a ] ) , the last factor @xmath88 represents the unfinished turn ( _ i.e. _ , the probability to stay inside the box during @xmath89 as shown in fig .
[ fig1a ] ) .
finally the delta function represents the fact that the durations of all these intervals add up to @xmath0 . in writing this joint distribution in eq .
( [ jpdf1 ] ) , we have used the renewal property of the brownian motion , _
i.e. _ , the successive intervals are statistically independent
. similar renewal equations appeared before in the study of the record statistics of discrete time random walkers @xcite . in the context of record statistics
, @xmath76 represents the probability that the walker stays below its starting position up to @xmath0 steps and @xmath90 represents the first - passage probability that the walker crosses its starting position from below for the first time in between steps @xmath91 and @xmath0 .
in contrast here we have continuous time @xmath0 and @xmath76 and @xmath58 represent respectively the survival and the first exit probability from a finite box @xmath92 $ ] .
the probability @xmath15 that the particle makes @xmath3 complete turns either clockwise or anti clockwise in time @xmath0 , can then be obtained from the joint probability density in eq .
( [ jpdf1 ] ) by integrating over all the durations @xmath93 as we will see later , instead of working with these multiple time integrals , it is simpler to work in laplace space . using the convolution structure in eq .
( [ p_n-1 ] ) , we perform laplace transform with respect to @xmath0 and get @xmath94=\int_0^{\infty}e^{-st}p(n , t)~dt , \nonumber \\ & = & \left[\tilde{f}(s ) \right]^n\tilde{q}(s)= \frac{1}{s}~\frac{\cosh\left(2\pi\sqrt{\frac{s}{d}}\right)-1 } { \left[\cosh\left(2\pi\sqrt{\frac{s}{d}}\right)\right]^{n+1 } } , \label{ltp_n}\end{aligned}\ ] ] where we have used eq .
( [ qtilde ] ) and ( [ ftilde ] ) . from the above expression
one can easily check that , @xmath95 as @xmath96 . finally ,
taking ilt of @xmath97 given explicitly above , we get the probability @xmath15 that the particle has total winding number @xmath3 in time @xmath0 as @xmath98^n\tilde{q}(s ) .
\label{iltp_n}\ ] ] once we know @xmath15 , the joint probability @xmath17 that the particle has total winding number @xmath3 and net winding number @xmath18 in time @xmath0 , can simply be obtained as follows : given the particle makes a complete turn around the circle at time @xmath0 to @xmath59 with probability @xmath99 , the probability that it makes a counter - clockwise turn is @xmath100 and similarly , the probability that it makes a clockwise turn is also @xmath101 .
this is because the probabilities associated to the first exits trough the upper and lower boundaries are equal when the particle starts from the center of the box .
hence , given that the particle makes a complete turn at time @xmath0 , the probability that this turn will be a counter - clockwise turn is @xmath102 and the probability that this turn will be a clockwise turn is also @xmath102 .
so the probability that there are @xmath103 counter clockwise turns out of @xmath3 total complete turns is given by @xmath104 where the binomial factor represents the number of ways to choose @xmath103 counter - clockwise turns from @xmath3 total complete turns .
if the particle has total winding number @xmath3 and net winding number @xmath18 , there must be @xmath105 complete counter - clockwise turns and the rest @xmath106 clockwise turns in time @xmath0 .
hence , the joint probability of having total winding number @xmath3 and net winding number @xmath18 in time @xmath0 , is then given by @xmath107 @xmath108 $ ] @xmath109 $ ] _ i.e. _ @xmath110 from this joint probability , the marginal probability @xmath16 of having net winding number @xmath18 in time @xmath0 , is obtained by summing over @xmath3 as follows @xmath111 whose lt can be expressed using eq .
( [ iltp_n ] ) as @xmath112 .
now using the identity @xcite @xmath113 and performing some algebraic simplification we get @xmath114 = \frac{1}{s } \tanh\left(\pi\sqrt{\frac{s}{d}}\right)~\text{exp}\left(-2\pi |k|\sqrt{\frac{s}{d } } \right ) .
\label{ltr - k}\ ] ] one can find explicit expressions of @xmath15 and @xmath16 after performing the inverse laplace transforms of @xmath97 and @xmath115 . before doing that , let us try to understand whether @xmath15 and @xmath16 have scaling distributions by looking at how the mean and variance of @xmath3 and @xmath18 grow with time @xmath0 .
the @xmath103th moment of the total winding number @xmath3 can formally be written as @xmath116 where @xmath97 is given explicitly in eq .
( [ ltp_n ] ) .
one can write a similar expression for the @xmath103th moment of the net winding number @xmath18 , with only difference is that @xmath97 will now be replaced by @xmath115 . looking at the expression of @xmath115 in eq .
( [ ltr - k ] ) closely , we see @xmath117 , which implies that all odd order moments of @xmath18 are zero . using the expressions of @xmath97 and @xmath115 from eqs .
( [ ltp_n ] ) and ( [ ltr - k ] ) , and performing some algebraic manipulations , one can show that @xmath118 ) with the analytically obtained expressions in eqs .
( [ mean - appndx ] ) and ( [ second - appndx ] ) . ]
these bromwich integrals can be computed by computing residues at the poles @xmath119 for @xmath120 of @xmath121 .
exact expressions of @xmath122 , @xmath123 and @xmath124 are given in eqs .
( [ mean - appndx ] ) and ( [ second - appndx ] ) . in fig .
( [ fig3 ] ) we compare these analytical expressions with the same obtained from direct numerical simulations of the langevin equation in eq .
( [ langevin ] ) and see nice agreement . from these analytical expressions
one can easily see the following small and large @xmath0 behaviors @xmath125 such small and large @xmath0 behavior of the mean and variance , can also be obtained , respectively , from the large and small @xmath126 behaviors of the integrands in the eqs .
( [ mean - contour ] ) and ( [ 2ndmom - contour ] ) . following similar procedure one can easily show from eq .
( [ joint - free ] ) and eq .
( [ iltp_n ] ) that the correlation between @xmath3 and @xmath18 for large @xmath0 , grows as @xmath127 . obtained from simulation with the scaling function @xmath46 by plotting @xmath128 vs. @xmath129 along with @xmath46 vs. @xmath130 . in figure ( b ) we compare the distribution @xmath16 obtained from simulation with the scaling function @xmath47 by plotting @xmath131 vs. @xmath132 along with @xmath47 vs. @xmath130 ] the linear growths of @xmath134 and @xmath124 at large @xmath0 , suggest that if we scale the fluctuations of @xmath3 and @xmath18 around their mean by @xmath135 then the distributions @xmath15 and @xmath16 will have the following scaling forms @xmath30 and @xmath31 where the scaling functions @xmath46 and @xmath47 are given in eq .
( [ gauss ] ) . in fig .
( [ fig4 ] ) we verify these scaling forms by comparing them with @xmath15 and @xmath16 obtained from direct simulation of the langevin equation ( [ langevin ] ) . in fig .
( [ fig4]a ) we plot numerically obtained @xmath136 against @xmath129 for @xmath137 and @xmath138 and on comparison with @xmath46 ( blue line ) we see nice agreement . on the other hand , in fig .
( [ fig4]b ) we plot numerically obtained @xmath139 against @xmath140 for @xmath137 and @xmath138 , to compare with the scaling distribution @xmath47 ( blue line ) . here
also we see nice agreement .
the scaling forms of @xmath15 and @xmath16 , specified by the scaling functions @xmath46 and @xmath47 respectively , are valid for large @xmath0 and over the regions , where their fluctuations around their respective means are @xmath33 .
when these fluctuations around their respective means are of @xmath35 , then the probability distributions are described by large deviation tails .
one can find the form of @xmath39 explicitly in the following three limits ( i ) @xmath167 , ( ii ) @xmath168 and ( iii ) @xmath169 as @xmath170 * as @xmath171 , the solution of eq .
( [ beta * ] ) is @xmath172 upto the leading order in @xmath148 . using this solution in eq .
( [ gx- ] ) , we get @xmath173 .
the value @xmath174 at @xmath74 , implies @xmath175 which means , the probability that the particle has not made any complete turn around the circle in time @xmath0 , decays exponentially for large @xmath0 . in the first exit picture , this is exactly the probability that the particle , starting from the origin , stays inside the box @xmath50 $ ] till time @xmath0 and this probability is given by the survival probability @xmath76 introduced in eq .
( [ fpndsvl ] ) . taking an ilt of @xmath176 in eq .
( [ qtilde ] ) , one can show that @xmath76 for large @xmath0 indeed decays as @xmath177 . * for @xmath169 , the solution of eq .
( [ alpha * ] ) is @xmath178 $ ] which implies @xmath179 . putting this value of @xmath150 in eq .
( [ gx+ ] ) we get @xmath180 for large @xmath148 . + for @xmath181 , both eqs .
( [ alpha * ] ) and ( [ beta * ] ) have solutions @xmath182 . hence expanding the left hand sides of both the eqs .
( [ alpha * ] ) and ( [ beta * ] ) for small @xmath153 and for small @xmath183 respectively , we get @xmath184 + for @xmath185 . note the gaussian nature of @xmath39 around @xmath186 and the asymmetry in the shape between @xmath187 and @xmath188 as expressed in eq .
( [ gx - asmtry ] ) .
inset : plot of the large deviation function @xmath40 for @xmath185 . ] + using these values of @xmath153 and @xmath183 in eqs .
( [ gx+ ] ) and ( [ gx- ] ) respectively , one can show that the large deviation function @xmath39 in eq .
( [ ldp_n ] ) for @xmath189 is given by @xmath190 which implies @xmath191 this large deviation form of @xmath15 for @xmath3 around its mean @xmath192 , is consistent with the gaussian scaling distribution @xmath46 given in eq .
( [ gauss ] ) .
we now put our attention on finding the large deviation form of the distribution @xmath16 of the net winding number @xmath18 . as done for @xmath15 , we start with @xmath193 where we have used the explicit form of @xmath115 from eq .
( [ ltr - k ] ) . in the @xmath194 and @xmath195 limit keeping @xmath37 finite , we see that the dominant contribution in the large deviation function comes from @xmath196 where the rest of the terms in the integrand contribute at @xmath197 .
performing the bromwich integral in the above equation we get the following large deviation form of @xmath16 @xmath198 this quadratic form of the large deviation function implies that over full range of @xmath199 $ ] the distribution @xmath16 has a gaussian scaling form under the scaling @xmath140 , which is given by @xmath47 in eq .
( [ gauss ] ) ( see also the discussion in sec.[mvs - dis ] ) .
till now we have discussed about the asymptotic forms of the distributions @xmath15 and @xmath16 describing either typical or atypically large fluctuations .
it turns out that one can find explicit expressions of the distributions @xmath15 and @xmath16 for arbitrary @xmath3 , @xmath18 and @xmath0 .
such explicit expressions for any @xmath3 and @xmath18 may be useful to compare with simulations . in the following
we present the derivation of such expressions . to find an exact expression of @xmath15 for any @xmath3 we need to perform the inverse laplace transform in eq .
( [ iltp_n ] ) which with the help of eq .
( [ ltp_n ] ) can be rewritten explicitly as @xmath200^{n } } .\label{iltp_n-1}\end{aligned}\ ] ] using the following expansion @xmath201^{-n}= \frac{2^n~e^{-n\sqrt{q}}}{(1+e^{-2\sqrt{q}})^n } = \frac{2^n}{\gamma[n]}\sum \limits_{\ell=0}^{\infty } ( -1)^{\ell}\frac{\gamma[n+\ell]}{\gamma[\ell+1]}~e^{-(2\ell+n)\sqrt{q } } , \label{expnsn}\ ] ] and the identity @xcite @xmath202 it follows that @xmath203}\sum \limits_{\ell=0}^{\infty } ( -1)^{\ell}\frac{\gamma[n+\ell]}{\gamma[\ell+1]}~\text{erfc}\left(\frac{2\ell+n}{\sqrt{4\tau}}\right),\ ] ] where @xmath204 $ ] is the gamma function .
hence the probability distribution of having @xmath3 total winding around the circle in time @xmath0 is explicitly given by @xmath205}\sum \limits_{\ell=0}^{\infty } ( -1)^{\ell}\frac{\gamma[n+\ell]}{\gamma[\ell+1]}~\text{erfc}\left(\frac{\pi(2\ell+n)}{\sqrt{dt}}\right)- \frac{2^{n+1}}{\gamma[n+1]}\sum \limits_{\ell=0}^{\infty } ( -1)^{\ell}\frac{\gamma[n+\ell+1]}{\gamma[\ell+1]}~\text{erfc}\left(\frac{\pi(2\ell+n+1)}{\sqrt{dt}}\right ) .
\label{p - n - t - exact}\ ] ] and @xmath16 given in eqs .
( [ p - n - t - exact ] ) and ( [ r - k - t - exact ] ) , respectively , with the same obtained from numerical simulation of the langevin eq .
( [ langevin ] ) for @xmath206 and @xmath185 .
dashed lines are for guidance to the eye . ]
let us now turn our attention to the evaluation of @xmath16 for arbitrary @xmath18 .
once again this can be done by performing the inverse laplace transform @xmath207 where the explicit expression of @xmath115 is given in eq .
( [ ltr - k ] ) .
writing @xmath208 where @xmath209 we see that @xmath16 can be expressed as a convolution @xmath210 of two functions @xmath211 and @xmath212 which are inverse laplace transforms of @xmath213 and @xmath214 respectively , _
i.e. _ @xmath215 $ ] and @xmath216 $ ] .
one can show that , @xmath211 and @xmath212 are explicitly given by @xmath217 injecting these expressions in the convolution and performing the integral over @xmath218 we get @xmath219 , \label{r - k - t - exact}\ ] ] where i m stands for imaginary part .
explicit expression of @xmath220 $ ] is given by @xcite @xmath221 = \frac{2}{\sqrt{\pi}}e^{-x^2}\sum \limits_{l=0}^{\infty } \frac{(-1)^ly^{2l+1}}{(2l+1)!}~h_{2l}(x),\ ] ] where @xmath222 is the @xmath103th order hermite polynomial .
although this expression of @xmath16 in eq .
( [ r - k - t - exact ] ) converges very fast numerically , it is not suitable for extracting the large @xmath0 gaussian behavior described by the scaling function @xmath47 in eq .
( [ gauss ] ) .
however there is another representation @xmath223 where @xmath224.\ ] ] from this expression one can immediately see that in the large @xmath18 and @xmath0 limit keeping @xmath225 fixed , @xmath226 which is consistent with the scaling function @xmath47 in eq .
( [ gauss ] ) and also with the large deviation function @xmath40 in eq .
( [ ldf - r - k - t ] ) . in fig .
[ compare ] we compare the analytical expressions of the distributions @xmath15 and @xmath16 given in eqs .
( [ p - n - t - exact ] ) and ( [ r - k - t - exact ] ) , respectively , with the same obtained from numerical simulation for @xmath206 and @xmath185 and see very good agreement .
another interesting quantity is the distribution of the maximum net winding number @xmath227 in time @xmath0 . denoting this distribution by @xmath228
, one can write @xmath229 with @xmath230 , where @xmath231 $ ] .
one can easily see that , @xmath232 is exactly the probability that the particle , starting at the origin , stays below the level @xmath233 throughout the time interval @xmath79 $ ] .
this means @xmath232 is the probability that the random process @xmath234 , starting from @xmath235 , stays positive till time @xmath0 and it is given by the well known result @xcite @xmath236 using which we get @xmath237
in this section we consider the situation where the brownian particle on the ring , starting from @xmath238 _ i.e. _ from the point a in fig . [ fig1 ] , is constrained to come back to a after time @xmath0 .
this means that the final position of the particle is constrained to be @xmath6 where @xmath7 is an integer .
the probability @xmath239 that the particle reaches @xmath6 at time @xmath0 is given by @xmath240 given that such a constrained brownian trajectory has net winding number @xmath18 in time @xmath0 , implies that the final position of the particle at time @xmath0 is @xmath241 .
hence the probability distribution @xmath44 of having net winding number @xmath18 ( _ i.e _ @xmath18 net counter - clockwise turns ) in time @xmath0 , is proportional to @xmath242 . letting the normalization constant be @xmath243
one can write @xmath244 where @xmath245 from the expression of @xmath244 it is clear that this distribution is symmetric with respect to @xmath18 which implies , all odd order moments are zero .
the lowest non - zero moment is @xmath246 which can be computed and expressed in terms of @xmath243 in eq .
( [ n_k - t ] ) as @xmath247 it is easy to see that , the second moment of @xmath18 has the same large @xmath0 linear growth @xmath248 as in the _ free _ case ( see sec.[mvs - dis ] ) .
we now compute the probability @xmath43 of having @xmath3 complete turns in time @xmath0 for brownian bridges on the circle . as before ( see eq .
( [ jpdf1 ] ) ) let @xmath249 and @xmath89 are time intervals in @xmath0 where @xmath250 represents the time required by the particle to perform the @xmath251th complete turn around the circle and @xmath89 represents the time interval @xmath252 remaining after the @xmath3th complete turn . following sec .
[ free ] and remembering once again the connection between a complete turn and the first exit from the box of length @xmath48 , one can write @xmath253 where @xmath58 is the first exit probability as before and @xmath254 is the probability with which the particle starting from the origin ( point a in fig .
[ fig1 ] ) returns back to the origin in the last time interval @xmath89 without making any complete turn . in the first exit picture
, @xmath254 represents the probability that the particle starting from the center of the box of length @xmath48 ( see fig .
[ fig1a ] ) comes back to the center in time @xmath89 without exiting the box .
taking laplace transform of @xmath255 with respect to @xmath0 we get @xmath256 = \left [ \tilde{f}(s)\right]^n\tilde{r}_0(s ) , \label{lt - mcaln}\ ] ] where @xmath257=\text{sech}\left ( 2\pi\sqrt{\frac{s}{d } } \right ) $ ] from eq .
( [ ftilde ] ) and @xmath258 is the laplace transform of @xmath259 _ i.e. _ @xmath260 $ ] . to compute the probability @xmath259 , one needs to solve the diffusion equation @xmath261 with bcs @xmath262 and the initial condition @xmath263 , where @xmath264 represents the probability that the particle , starting at the origin , reaches @xmath148 at time @xmath0 while staying inside the box @xmath50 $ ] ( see fig .
[ fig2 ] ) .
taking the laplace transform with respect to @xmath0 on both sides of the above diffusion equation we get @xmath265 it is easy to check that the solution of this equation is given by @xmath266 which for @xmath74 provides @xmath267 .
once we know @xmath258 and @xmath268 explicitly , then taking inverse laplace transform of @xmath269 in eq .
( [ lt - mcaln ] ) we get @xmath255 _ i.e. _ @xmath270=\frac{1}{2\pi i}\int_{\mathcal{b}}ds~e^{st } ~\left(\tilde{f}(s)\right)^n\tilde{r}_0(s)=\frac{1}{4\pi}~\frac{1}{2\pi i}\int_{\mathcal{b}}dq~e^{q\frac{dt}{4\pi^2 } } ~\frac{\sinh \sqrt{q}}{\sqrt{q}~[\cosh \sqrt{q}]^{n+1}}. \label{mcaln}\ ] ] at this point , one can easily check that @xmath271 which implies @xmath272 . moreover , using this expression of @xmath255 one can also find exact time dependence of the mean @xmath273 and the variance @xmath274 ( given in appendix [ appndx - f ] ) from which we see that , for large @xmath0 both the mean and variance grow exactly the same linear way as in the _ free _ case ( see sec.[mvs - dis ] ) .
this implies that for large @xmath0 , the scaling distribution function corresponding to the distribution @xmath43 is also described by the same functions @xmath46 as given in eq .
( [ gauss ] ) for the _ free _ case .
furthermore , it can be easily seen that the large deviation function associated to @xmath43 is same as the large deviation function @xmath39 associated to @xmath15 ( see sec.[ldfs ] ) , because what happens in the last time interval @xmath275 after the @xmath3th complete turn do not contribute at the leading order in the @xmath276 and @xmath142 limit while keeping @xmath143 fixed .
to evaluate the inverse laplace transform in eq .
( [ mcaln ] ) , we follow the same steps as done in sec.[exact ] . once again using the expansion in eq .
( [ expnsn ] ) and the identity @xcite @xmath277 one can show that the function @xmath255 is explicitly given by @xmath278}~\sum \limits_{\ell=0}^{\infty}(-1)^{\ell}~\frac{2\ell+n}{\ell+n } ~\frac{\gamma[\ell+n+1]}{\gamma[\ell+1]}~\text{exp}\left(-\frac{(2\ell+n)^2\pi^2}{dt } \right),\ ] ] where @xmath204 $ ] is the gamma function .
hence we have an exact expression of the distribution @xmath279 where @xmath243 is given in eq .
( [ n_k - t ] ) .
let us now look at the distribution @xmath280 of the maximum net winding number @xmath227 in time @xmath0 . once again this distribution can be obtained from @xmath281 with @xmath282 where @xmath283 $ ] .
one can easily see that , the probability @xmath284 is equal to the ratio @xmath285 . here
@xmath286 is the probability of having the particle s final position @xmath1 at integer multiple of @xmath52 while conditioned on the fact that the particle , starting from @xmath238 , stayed below the level @xmath287 throughout .
the function @xmath243 in the denominator is given in eq .
( [ n_k - t ] ) and it represents the probability that the final position @xmath1 of the particle is integer multiple of @xmath52 given that the particle had started from @xmath238 . to evaluate @xmath286
, we consider the shifted random process @xmath288 .
the probability @xmath286 can be obtained from the propagator @xmath289 representing the probability density that the process @xmath290 , starting from @xmath291 , reaches @xmath292 at time @xmath0 while staying positive throughout .
this is actually the propagator of a brownian particle with an absorbing wall at the origin and it is given by @xcite @xmath293 $ ] . hence , the probability @xmath286 is given by @xmath294 .
after performing some algebraic simplification of this infinite sum and using eq .
( [ n_k - t ] ) , we get from eq .
( [ p_max ] ) : @xmath295
path integral techniques have been used to study statistics of net winding number @xmath5 in many situations @xcite where one maps the problem to a suitable quantum problem .
but for other quantities like the total winding number @xmath3 , path integral techniques are hard to adapt . in this paper
we have presented a method based on renewal properties of brownian motion to study winding statistics of a single brownian motion on a ring .
this method is alternative to the standard path integral methods .
more precisely , using the renewal property of brownian motion and the connection between a complete turn around the circle and the first exit from a box of size @xmath48 , we derived analytical expressions of the probability distributions of the total number of turns @xmath3 and the net number of counter - clockwise turns @xmath18 at any time @xmath0 .
such distributions are relevant in quantum transport in ring geometry @xcite . for large @xmath0
, we have shown that these distributions have gaussian scaling forms describing the typical fluctuations of @xmath34 around their respective means .
we have also computed the large deviation functions associated to these distributions , which describe the atypical fluctuations of @xmath35 .
correlation between the total winding number @xmath3 and net winding number @xmath18 is studied from the joint probability distribution of @xmath3 and @xmath18 whose expression have been provided for any @xmath0 .
numerical simulations have been performed to verify our analytical results .
one can extend this problem in different directions .
for example , it would be interesting to see what happens to these distributions when the particle is being subjected to some pure or random potential .
investigating the equilibrium state of a ring coupled to a thermal bath reveals interesting connections with random walk on a sinai potential @xcite . in the context of polymer physics , fluctuations of winding number of a directed polymer in random media
have been studied in @xcite .
another interesting extension would be to consider interacting multiparticle system @xcite .
recently , winding statistics of @xmath296 non - intersecting brownian bridges on unit circle have been studied for a case where the diffusion constant scales with the number of particles @xmath296 @xcite .
it will be interesting to study the winding statistics of non - intersecting walkers in the _ free _ case when the final positions of the walkers are not constrained .
finally , in the ring geometry , quantities other than the winding numbers , like residence time spent inside some given region , local time spent at some specified point etc . would also be interesting to study .
such quantities have been studied for general gaussian stochastic processes in the context of persistence @xcite .
this research was supported by anr grant no .
2011-bs04 - 013 - 01 walkmat and in part by the indo - french centre for the promotion of advanced research under project no.4604 - 3 .
we thank the galileo galilei institute for theoretical physics , firenze for the hospitality and support received .
exact expressions of @xmath122 , @xmath123 and @xmath124 are given as follows @xmath297 \\
\langle n(t)^2 \rangle & = & \left(\frac{d^2t^2}{4\pi^4 } + \frac{dt}{6\pi^2 } -\frac{2}{45}\right ) + \frac{1}{2}\sum \limits_{m=1}^{\infty } e^{-d m^2 t } \left(\frac{d t}{3 \pi ^2}+\frac{1}{6 \pi ^2 m^2}-\frac{d t}{2 \pi ^4 m^2}-\frac{1}{2 \pi ^4 m^4}-\frac{2 d^3 m^2 t^3}{3 \pi ^4}\right ) \label{second - appndx } \allowdisplaybreaks[1]\\ & = & \frac{1}{3}\sum \limits_{\ell=0}^{\infty}~\frac{\gamma[\ell+4]}{\gamma[\ell+1]}~\left[\text{erfc}\left(\frac{(\ell+1)\pi}{\sqrt{dt}}\right ) + \text{erfc}\left(\frac{(\ell+2)\pi}{\sqrt{dt}}\right ) + \text{erfc}\left(\frac{(\ell+3)\pi}{\sqrt{dt}}\right)\right].\end{aligned}\ ] ] second expressions in both the above equations for @xmath122 and @xmath298 are obtained using poisson formula .
edwards s f , proc .
91 , 513 , ( 1967 ) .
edwards s f , j. phys .
a , 1 , 15 , ( 1968 ) .
rudnick j and hu y , j. phys . a : math . gen .
20 , 4421 , ( 1987 ) .
rudnick j and hu y , phys .
60 , 712 , ( 1988 ) .
grosberg a and frisch h , j. phys .
36 , 8955 , ( 2003 ) .
wiegel f w , introduction to path - integral methods in physics and polymer science , world scientific , singapore .
nelson d r , phys .
lett . 60 , 1973 , ( 1988 ) .
drossel b and kardar m , phys .
e 53 , 5861 , ( 1996 ) .
spitzer f , trans . am .
87 , 187 , ( 1958 ) .
pitman j and yor m , ann .
11 , 733 - 79 , ( 1986 ) .
le doussal p , etzioni y and horovitz b , j. stat .
mech , p07012 , ( 2009 ) .
schramm o , isr .
118 , 221 , ( 2000 ) .
hagendorf c and le doussal p , j. stat . phys .
133 , 231 , ( 2008 ) .
duplantier b and binder i a , phys .
lett , 89 , 26 , ( 2002 ) .
brereton m g and butler c , j. phys . a : math
20 , 3955 , ( 1987 ) .
khandekar d c and wiegel f w , j. phys .
a : math . gen .
21 , 56 , ( 1988 ) .
comtet a , desbois j and ouvry s , j. phys . a : math .
23 , 3563 - 3572 , ( 1990 ) .
antoine m , comtet a , desbois j and ouvry s , j. phys .
24 , 2581 - 2586 , ( 1991 ) .
nelson d r and stern a , complex behaviour of glassy systems lecture notes in physics volume 492 , pp 276 - 297 , ( 1997 ) .
nelson d r and seung s , phys
b 39 , 9153 ( 1989 ) , nelson d r and le doussal p , phys . rev .
b 42 , 10113 ( 1990 ) .
duplantier b , j , phys . a : math
22 , 3033 , ( 1989 ) .
comtet a , desbois j and texier c , j. phys .
38 r341 , ( 2005 ) .
texier c and montambaux g , j. phys . a : math .
38 , 3455 - 3471 , ( 2005 ) .
majumdar s n , curr .
89 , 2076 , ( 2005 ) .
redner s , a guide to first - passage processes ( cambridge university press , cambridge , 2001 ) .
bray a j , majumdar s n and schehr g , advances in physics , 62 , 225 ( 2013 ) .
majumdar s n and ziff r m , phys .
lett . 101 , 050601 ( 2008 ) .
majumdar s n , schehr g and wergen g , j. phys . a : math .
45 , 355002 ( 2012 ) .
schehr g and majumdar s n , exact record and order statistics of random walks via first - passage ideas , to appear in the special volume `` first - passage phenomena and their applications '' , eds .
r. metzler , g. oshanin , s. redner .
world scientic ( 2014 ) , preprint arxiv:1305.0639 .
godreche c , majumdar s n and schehr g , j. phys . a : math .
47 , 255001 ( 2014 ) .
schieff l j , the laplace transform : theory and applications , springer , ( 1999 ) .
wolfram functions site : _ http://functions.wolfram.com/gammabetaerf/erf/19/02/ _ hurowitz d , rahav s , and cohen d , phys .
e 88 , 062141 ( 2013 ) .
brunet e , phys .
e , 68 , 041101 , ( 2003 ) .
hobson d g and werner w , bull .
london math .
soc . , 28(6):643650 , ( 1996 ) .
liechty k and wang d , arxiv:1312.7390v2 [ math.pr ] . | we consider a brownian particle moving on a ring .
we study the probability distributions of the total number of turns and the net number of counter - clockwise turns the particle makes till time @xmath0 . using a method based on the renewal properties of brownian walker
, we find exact analytical expressions of these distributions .
this method serves as an alternative to the standard path integral techniques which are not always easily adaptable for certain observables . for large @xmath0
, we show that these distributions have gaussian scaling forms .
we also compute large deviation functions associated to these distributions characterizing atypically large fluctuations .
we provide numerical simulations in support of our analytical results . |
this investigation forms part of a long term project aimed at understanding the nearby spiral structure in the third galactic quadrant of the milky way ( hereinafter tgq ) from an optical point of view by means of open clusters .
the still unresolved question about the grand design structure of the milky way ( see , e.g. choi et al .
2014 ) has been demanding an extraordinary observational effort during the last decades . actually , a crucial starting point took place in the @xmath2 when georgelin & georgelin ( 1976 ) produced the first map of the galaxy utilizing hii regions as spiral tracers .
this first map showed that a four arms model fits the observations very well , a concept widely used yet .
since then , a number of other maps showing the structure of the galaxy have been developed by means of different spiral tracers such as molecular clouds ( efremov 1998 ) or by combining them with hii regions ( hou et al .
dame et al .
( 2001 ) collected co observations from several sources to map the entire milky way . while russeil ( 2003 ) carried out a multiwavelength study of the plane of our galaxy and found evidences that favored a four arms model , nakanishi & sofue ( 2003 ) presented a three dimensional map constructed with hi observations concluding that a three arms model is better to fit the observations .
churchwell et al .
( 2009 ) combined data from a variety of sources in the wavelength range from 3.6 to 24 microns adopting also a four arms model but recognizing some difficulties of interpretation .
this is , since the controversy about the number of spiral arms in the galaxy is still open we suggest the readers to have a look to the summary of spiral arms parameters in the valle ( 2013 , 2014 ) paper series and references therein .
however , in almost all these cases distances depend on a rotational curve and the adopted kinematic model .
thus , strong ambiguities take place and only can be resolved by adopting other tracers of spiral structures with distances independent from the properties of the models . in this respect , trigonometric parallaxes and proper motions of massive star forming regions in combination with information from hii regions , giant molecular cloud and masers is a promising initiative since distance ambiguities can be removed ( reid et al .
2009 , 2014 ; hou & han ( 2014 ) ) .
just to mention , methanol and water masers have been successfully used for some authors to study particular regions of the milky way such as the local , sagittarius and perseus arms ( xu et al .
( 2014 ) , wu et al . 2014 , choi et al . 2014 , among others ) . in this framework ,
young open clusters become excellent spiral tracers since they can be observed at large distances .
the additional advantage of using open clusters is that their distances do not rely on any rotational curve and kinematic model .
contrarily , observations and detection of open clusters are highly dependent of the visual extinction along the line of view in the galactic plane .
several investigations biased to study the large structure of the milky way using open clusters and stellar component were developed by moffat & vogt ( 1973 ) and vogt & moffat ( 1975 ) although with a strong magnitude cut - off .
this precluded from arriving to conclusive definitions about the extension of the local arm ( orion ) , the angular recognition ( and the real entity ) of the perseus arm and the possible trace of the outer arm ( norma - cygnus ) .
some other efforts were carried out by mccarthy & miller ( 1974 ) , moffat & fitzgerald ( 1974 ) , clari ( 1974 ) , reed & fitzgerald ( 1984ab ) , slawson & reed ( 1988 ) , reed & fitzgerald ( 1985 ) , reed ( 1990 ) , all of them concerning the distribution and structure shown by the stellar components and also by open clusters .
most of the optical inspections performed in the tgq were concentrated along the region @xmath3 to @xmath4 in longitude , including the puppis and vela regions .
this could be done , in part , thanks to the presence of a dust - window , the so - called fitzgerald window ( fitzgerald 1968 ) near the puppis association .
a thorough information on related earlier works can be found in reed ( 1989 ) but we also suggest the contribution from kaltcheva & hilditch ( 2000 ) who revealed the presence of very distant blue stars in the longitude interval from @xmath5 .
indeed , the absorption has been the greatest obstacle to observations . during the last two decades
a notorious feature has been confirmed in the tgq from the dust mapping for schlegel et al .
( 1998 ) together with the hydrogen emission maps from shassa ; these two surveys suggest a symmetry brake - up in the sense that both , dust and emission ( neutral gas ) show a tendency to locate below the formal galactic plane at @xmath6 galactic latitude .
this fact is in agreement with previous investigations where it has been shown that the hi layer in the galaxy flares , becomes thicker at larger distances from the galactic center ( henderson et al .
1982 ; kulkarni et al . 1982 ; burton & te lintel hekkert 1986 ) on a side . on the other ,
a tendency for the tgq gas clouds to fall below the galactic plane has been revealed and interpreted as evidence for warp in the molecular disk as well ( may , murphy & thaddeus 1988 ; may et al . 1997 ) .
co observations of the infrared astronomical system ( iras ) sources wouterloot et al .
( 1990 ) concluded that , although the hi layer is substantially thicker than the molecular clouds layer in the inner galaxy , the z - height distribution of the molecular clouds show the same thickness in the outer galaxy .
a few years ago our group ( vzquez et al .
2008 , carraro et al . 2005 , moitinho et al .
2006 , carraro et al .
2015 ) presented and discussed a new picture of the spiral structure in the tgq built by means of a mixture of optical data of open clusters , field blue stars and co sources distributed along 60 degrees in galactic longitude .
this new picture confirms the presence of the warp ( in optical and co ) , reveals the existence of the outer arm and suggests that the local arm extends from the sun to its encounter with the outer arm .
however , none of these studies can give definite evidences about the trace of the perseus arm except for a very few open clusters that , ambiguously , can be related to the local arm too .
our ongoing observational programme at la plata observatory was designed to improve parameters of known clusters and to look for those with no information at all .
this is an essential task that allows us to find far spiral tracers and recognize the grand design spiral structure in the tgq .
so , we report here on results coming from ccd @xmath0 photometry combined with spectroscopic measures for five open clusters , ruprecht 20 , ruprecht 47 , ruprecht 60 , ngc 2660 and ngc 2910 , all of them projected against the vela gum nebula at @xmath7 including vela and puppis associations . to situate the readers the gum nebula ,
one of the biggest hii regions of our galaxy , covers an angular extension of @xmath8 and is centered at @xmath9 , @xmath10 .
sahu ( 1992 ) analyzed the interstellar medium in the zone up to 2 kpc in vela - puppis ; cometary globules and the star formation process in the gum nebula have also been studied by kim et al .
definitely , @xmath11 pup ( o4if ) and @xmath12 vel ( wc8+o7.5 i ) , the ob associations , trumpler 10 and vela ob2 , seem to be the responsible entities of most of the photo - ionization in the nebula itself .
to be remarked is the presence of the iras vela shell ( sahu 1992 ) , an expanding dust shell surrounding vela ob2 with 6 - 8 pc radius enclosing the gum nebula . behind it , the vela molecular ridge ( murphy & may 1991 ) , a superposition of giant co molecular clouds , appears obscuring the farthest part of the galaxy .
the purpose of the present investigation is twofold : on a side we want to re - discuss the parameters of five clusters in the tgq onto the basis of a new set of photometry ( one of them has never before been subject of @xmath0 photometry ) . on the other side , we want to connect the properties of these objects with the structure of puppis and vela associations and look for evidences of the grand design spiral structure .
the photometric observations have been complemented with spectroscopy of several stars located in the cluster regions to discard field interlopers but also to estimate the absorption law .
table 1 lists coordinates of the clusters together with their angular sizes ( from our analysis , in advance ) .
the remaining of the paper is organized as follows . in
2 we present and discuss the data acquisition of photometry and spectroscopy for five open clusters projected against the vela gum nebula . in
3 we illustrate the procedure applied for the derivation of fundamental parameters for the cluster sample .
4 presents a cluster- by- cluster discussion and the trend color excess .
discussion of the derived galactic structure and our conclusions are given in
5 and 6 , respectively .
we show in fig . 1 the surveyed areas with ccd @xmath13 photometry of the five clusters in our sample enclosed by squares @xmath14 on a side . ngc 2660 and ngc 2910 were measured at las campanas observatory , chile , on february 1996 with the 0.6-m telescope of the university of toronto southern observatory ( utso ) equipped with a ccd pm @xmath15 methacrome - ii uv coated camera . in these two clusters we carried out @xmath0 photometry .
the scale of the ccd camera is @xmath16 what generates a field of @xmath17 on a side .
the nights were photometric with seeing values between @xmath18 and @xmath19 .
four frames were necessary in the case of ngc 2910 to get the best possible coverage of the cluster .
.cluster s radii from subsection 3.1 .
[ cols="<,<,<,<,<,^",options="header " , ] [ tab_spectra - cont ]
in the following subsections we describe our method to analyze each cluster .
just to put the reader in the galactic context we show in fig . 7 the location of each of the five clusters in this study projected against a @xmath20 image of the vela gum adapted from the shassa sm-1 survey .
table 5 is a summary of fundamental parameters for all cluster as found in the literature ( last raw in each cluster is the result from this work in advance ) .
c|c|ll|c|c|c + & @xmath21 & & & & & + & @xmath22 & & & & & + & & vogt & moffat ( 1972 ) & & 0.1 & 1.21 & - + & & this work & & - & - & - + & & vogt & moffat ( 1972 ) & & 0.28 & 3.03 & - + & & kharchenko et al .
( 2005 ) & & 0.25 & 3.01 & 0.11 + & & this work & & @xmath23 & @xmath24 & 0.06 - 0.08 + & & bonatto & bica ( 2010 ) & & @xmath25 & @xmath26 & @xmath27 + & & this work & & @xmath28 & @xmath29 & 0.8 - 1.0 + & & hartwick and hesser ( 1973 ) & & @xmath30 & @xmath31 & 1.20 + & & lynga ( 1987 ) & & 0.35 & 2.1 & 1.58 + & & hesser & smith ( 1987 ) & & @xmath32 & 4.35 & - + & & frandsen et al .
( 1989 ) & & 0.36 & 2.9 & 1.2 + & & geisler et al .
( 1992 ) & & @xmath28 & - & 1.7 + & & friel ( 1995 ) & & - & 2.89 & 0.8 + & & sandrelli et al .
( 1999 ) & & 0.37 - 0.42 & 2.63 - 2.88 & @xmath33 + & & this work & & 0.33 & @xmath34 & @xmath35 + & & becker ( 1960 ) & & 0.09 & 1.25 & 0.04 + & & topaktas ( 1981 ) & & 0.05 & 1.32 & - + & & ramsay & pollaco ( 1992 ) & & @xmath36 & @xmath37 & @xmath38 + & & kharchenko et al .
( 2005 ) & & 0.34 & 2.61 & 0.03 + & & this work & & @xmath39 & @xmath40 & 0.06 + [ tab_param ] accuracy in cluster parameters is achieved only when star sequences sweeping a broad range of visual magnitudes are identified and when contamination for field interlopers is minimized . in this last aspect let us mention that cluster sizes are routinely used as criteria for separating cluster dominated from field - dominated regions in photometric diagrams ( vzquez et al . 2008 ) .
although useful for recovering the precise locus occupied by cluster members , it is worth recalling that the cluster s linear size is a fundamental but overlooked quantity to characterize these objects .
in fact , precise sizes are necessary for evaluating the cluster evolutionary status through analysis of its density and other related parameters [ see aarseth ( 1996 ) ; de la fuente ( 1997 ) ; kroupa et al .
( 2001 ) ] such as the slope of the mass spectrum .
usually , cluster sizes are computed by assuming a centrally peaked spherical stellar distribution and then determining the the distance at which its radial density profile merges with the ( flat ) density of the stellar background . in practice , this determination is normally done either by visually setting this limit in a density radial plot or analytically by fitting a king ( 1962 ) profile to an observed radial density profile . in both approaches a spherically symmetric distribution
is assumed what may not be true since open clusters are , by definition , irregular in shape .
moreover , in the case of the king profiles , it is assumed that the system is dynamically relaxed too .
since a good size estimation reduces quite considerably the field star contamination , we restrict the analysis to stars within the cluster limits .
our starting point consists in adopting cluster center coordinates given in available catalogs ( e.g. webda ) and counting the number of stars found in concentric rings @xmath41 in width around them . instead of using our own photometry
we use 2mass data ( the three bands but particularly the deepest @xmath42 band ) that cover areas larger than ours .
this way we ensure the reliability of the cluster sizes and overcome the effects from visual absorption .
the star counts were divided by the area of each ring following well established receipts ( e.g. carraro et al 2007 ) .
8 shows density profiles for the 5 clusters in our sample .
vertical error bars in the density profiles are computed as the square root of the number of stars divided by the ring surfaces .
the radius of each cluster ( see table 1 ) is shown by an arrows in the point where the cluster density merges with the field stellar density level approximately ( solid lines in fig . 8) .
in all the cases we estimated the field density level by eye . finally , comparing figs . 1 and 8
we want to emphasize that since cluster radii are larger than the areas we have surveyed our observations are entirely contained within the cluster boundaries and therefore we do not expect high contamination by field interlopers .
thus , we are confident that our observations are useful to yield realistic cluster parameters .
open clusters are usually not far from the galactic plane and appear therefore projected against crowded stellar fields and immersed in regions of high absorption .
this makes the membership assignment a difficult task , particularly amongst faint stars .
all across this article we address memberships applying the classical method of checking simultaneously the consistency of the location of each star in several photometric diagrams ( two color diagrams , hereinafter tcds , and different color - magnitude diagrams , hereinafter cmds ) .
this method is an efficient tool when membership estimates rely on a careful inspection of the tcd and consistent reddening solutions are applied .
contamination for field interlopers is dramatic in crowded fields and may lead , on a side , to bad membership assignments and , on the other consequently , to wrong cluster parameters .
it happens the turn - off point position may be strongly affected by interlopers even in the bright portion of the cmds .
however , we use the @xmath43 index that is much more sensitive to stellar temperature changes than the @xmath44 index ( meynet et al .
1993 ) so that the @xmath45 vs @xmath43 cmd is quite useful for eliminating most of a - f - type field stars contaminating precisely the vertical part of the cluster sequences , especially of young clusters .
the reddening value @xmath46 in each cluster has been computed by superposition of the schmidt - kaler ( 1982 ) [ sk82 hereinafter ] zams in the tcd of the cluster .
we shift the zams along the reddening line ( given by the standard relation @xmath47 ) until the best fit to the cluster sequence is achieved . on the other hand , the knowledge of the ratio of total to selective absorption , @xmath48 , must be well established to compute a confident cluster distance .
though the analysis in moitinho ( 2001 ) carried out for several open clusters in the tgq suggests that the extinction law is normal , an additional check of the @xmath49value was performed in the @xmath50 plane in ruprecht 60 , ngc 2660 and ngc 2910 . in this diagram , if the slope of the standard reddening line @xmath51 [ dean et al .
( 1978 ) ] follows the stellar distribution this fact means the extinction law is @xmath52 .
apparent cluster distance moduli were derived then by superposing the sk82 zams onto the observed cluster sequences and the cluster distances from the sun ( @xmath53 ) are obtained after removing absorption from apparent moduli . the fitting error in each cluster
was estimated by eye inspection . as for cluster ages they were derived under the assumption clusters
are of solar metal content .
we superposed isochrones produced by girardi et al .
( 2000 ) onto the cluster sequences ( once they have been well established ) putting special emphasis in fitting not only probable members in the upper sequence but also covering the largest possible magnitude range .
however , in four out of the five clusters we were forced to use two ischrones to establish an age range .
various factors such as remaining material from the star formation processes ( affecting young clusters ) , undetected binary stars or unresolved double stars , altogether scatter stars around the mean sequence causing difficulties to achieve a good fit by a single isochrone . in older clusters these difficulties are amplified because of the need to fit simultaneously main sequence and red clump stars as we show in subsections 4.3 . and 4.4 .
spectral types are a powerful and independent tool to estimate distances and color excesses of individual stars and become indispensable to corroborate distances to open clusters . combining mk types and @xmath54 photometry we can easily get the distance of individual stars via the well know spectroscopic parallax method . in the cases of stars in table 4
we proceed to assign intrinsic @xmath55 and @xmath56 colors based in mk types following the relations given in sk82 .
intrinsic colors according to mk types and observed colors allow to get the reddening @xmath46 and derive the @xmath57 to produce free absorption magnitudes @xmath58 mag using the expression given below .
in a couple of late type stars the reddening turned out to be negative .
we think this is due to the low reddening of these stars and uncertainties in the true colors of evolved stars . in these cases
we adopt a reddening value @xmath59 . combining the mk types with absolute magnitudes @xmath60 using sk82 relations we are able to get individual star distances .
the main sources of errors in distances come from photometric uncertainties in color and magnitudes .
this is , the formula : @xmath61 leads to an error in distance @xmath62 of the form : @xmath63\ ] ] where @xmath64 and @xmath65 are typical photometric errors in our sample .
this procedure has been successfully applied in perren et al .
( 2012 ) . in the current case
, we assume no error in the extinction law ( @xmath52 ) and no error in the absolute magnitudes ( @xmath60 ) of the stars .
the photometric errors of individual stars with spectral types are all below 0.04 which means that errors in distances derived with the spectroscopic parallax method are all below 10% .
table 4 , columns 8 , 9 and 10 , contains the estimated @xmath46 excesses , adopted @xmath60 and distances for isolated stars with spectral types .
ruprecht 20 was studied by vogt & moffat ( 1972 ) who obtained @xmath54 photoelectric photometry for 11 stars .
although their work calls into question the existence of a true cluster in this region , they assume that stars number 10 , 3 and 18 ( our i d ) [ 6 , 9 and 11 in vogt & moffat notation ] are potential members of luminosity class v ( their assumed luminosity classes are confirmed by our spectroscopy in table 4 but we do not agree they are potential members of ruprecht 20 ) .
vogt & moffat ( 1972 ) obtained the following cluster parameters @xmath66 , @xmath67 and @xmath68 kpc in case it is a real entity .
this cluster has been subject also of a specific survey aimed at determining radial velocity amongst red giants to secure memberships ( mermilliod et al . 2008 ) .
these authors could find two red giant stars that they assumed likely cluster members . according to our data the cluster radial density profile of ruprecht 20 shown in fig .
8 , upper left panel , does not suggest the presence of an open cluster since no star grouping is evident .
indeed , the density profile resembles the typical profile produced by random stellar fluctuations .
figure 9 includes from left to right the @xmath69 , @xmath70 and the @xmath71 diagrams . in the tcd , middle panel , we superpose the zams shifted by @xmath72 according to vogt & moffat ( 1972 ) and , simultaneously
, we show the zams fitted to the apparent distance modulus @xmath73 ( @xmath74 kpc ) suggested by them too .
it is simple to see that no cluster sequence is accompanied by this zams fitting or , in other words , it is hard to perceive a cluster sequence in these diagrams .
instead , we see a sparse handful of stars , some of them affected by a similar amount of reddening , @xmath72 , but no traces of any cluster
. this is , exactly , what fig .
8 indicates .
we find the respective distances for the 10 stars with mk spectral types ( see table 4 ) .
according to their classifications , stars 10 , 3 and 18 ( our i d ) are located at distances of approximately 2.55 kpc , 1.99 kpc and 0.95 kpc respectively , demonstrating beyond any doubt they do not belong to any stellar aggregate .
the distance of stars 10 and 3 are of the order of the distance of star 6 ( @xmath75 kpc ) .
but this last is of late evolved type , a k7iii , while star 6 is a b8v and star 9 is a b9v . in short ,
the cmds and the tcd of ruprecht 20 confirm that no cluster exists in this location but we are dealing with a sparse group of stars mostly of dwarf and giant - types .
to mention is the fact that above the second knee of the zams there are some stars that are hard to interpret .
they are located along the path of the reddening in the tcd and could be very reddened blue stars . however , they are too faint
. they could be sdb type stars but since their average absolute magnitudes are @xmath76 and our magnitude cut - off is @xmath77 mag we do not expect more than a tenths of sdb stars in our frames .
certainly , spectroscopy could help to solve the nature of these objects .
after the above analysis we understand that evidences provided by our photometry and spectral analysis are irrefutable and we conclude that the open cluster ruprecht 20 is not a physical entity . in our opinion the connection of two red giant stars as suggested by mermilliod et al .
( 2008 ) with this object is unrealistic .
ruprecht 47 also was studied by vogt & moffat ( 1972 ) , who carried out photoelectric photometry on 10 stars . through these observations they placed the cluster at a distance from the sun of 3.03 kpc but , since their observations were limited to only a few stars on the upper main sequence , the authors suggest the true distance value can be quite different .
kharchenko et al .
( 2005 ) also studied this cluster and found 7 probable members at a distance of the order of 3 kpc with a reddening similar to the obtained by vogt & moffat ( 1972 ) .
this case is quite different from the one of ruprecht 20 . in principle , the radial density profile in fig .
8 upper right panel shows a clean star density profile extending up to 4@xmath78 where it merges with the background star density .
this fact suggests we are in the presence of a star overdensity .
figure 10 , the cmds and tcd , show in turn a well defined star sequence proper of a relatively young open cluster because of the presence of several blue stars in the tcd ( middle panel ) .
numbers ( our notation ) in the diagrams point out stars with spectral types .
the superposition of the zams in the tcd yields average color excesses of @xmath79 and @xmath80 , a bit larger than earlier studies in table 5 but still , within 3@xmath81 , in agreement with vogt & moffat ( 1972 ) though quite far from the value found by kharchenko et al .
( 2005 ) . in our interpretation ,
our reddening value is more robust than previous ones since we can cover the cluster sequence till the second knee in the tcd , the scatter of the data is low in this diagram and the mean differences with vogt & moffat ( 1972 ) data are all quite acceptable .
the fitting of the sk82 zams to the main sequence in the @xmath45 vs @xmath44 and @xmath45 vs @xmath43 diagrams yielded an absorption free distance modulus of @xmath82 , corresponding to a distance of @xmath83 kpc .
this result is quite different from earlier values determined by vogt & moffat ( 1972 ) and kharchenko et al .
( 2005 ) for more that 1 kpc .
so far as we can interpret the data the explanation for such a difference must be found in a simple fact already mentioned : our definition of the cluster in the tcd is very good and , in addition , we can track the cluster main sequence down to @xmath84 mag in the cmds .
this means a larger range of apparent magnitudes that allows to achieve a better zams superimposition and therefore more precision in the cluster distance .
the cluster age was derived by matching the girardi et al .
( 2000 ) isochrones ( dashed lines in the cmds in fig . 10 ) .
the best possible matches were achieved for ages between 63 and 80 million years ( @xmath85 ) .
there is , as expected from the difference in distance between this work and earlier ones , a significant difference between our age value and the 117 million years from kharchenko et al .
we think that the age difference found is closely related with our ability to locate precisely the cluster `` turn - off '' point at @xmath86 mag .
in fact , that is why we differ from kharchenko et al .
( 2005 ) who derived the cluster parameters taking into account stars down to @xmath87 mag . certainly , since they had no opportunity to see this point their distance and age were weakly determined .
spectral types in the region of ruprecht 47 include 8 bright stars listed in table 4 and pointed out in the corresponding panel of fig .
three of them are of b - type , one , number 2 in our notation , is an emission b5ve star . because of the small @xmath88-value , vogt & moffat ( 1972 ) suggested it is an emission object that we confirm with our spectroscopy . inspecting distances shown in table 4 we conclude that none of these 8 stars with spectroscopic parallaxes have chances to become cluster members since they are , on the average , twice closer to the sun than the cluster ruprecht 47 situated at ( @xmath894 kpc ) .
this negative result is very interesting since without a spectroscopic analysis most of the bright stars in this cluster would have been assumed cluster members .
the astrophysical impact of such an error is clear ; it produces an over - estimation of the number of massive stars in the cluster head and severely modifies the cluster initial mass function value ( out of the scope of the our investigation ) .
that is why spectroscopy is so valuable .
now , it is obvious the presence of other blue stars in the tcd of ruprecht 47 and one may wonder how much field interlopers modify the cluster parameter .
we do not know the answer but we have been as careful as possible in isolating the cluster region using star counts .
this said , we shall assume as likely cluster members the other blue stars in ruprecht 47 until the contrary is demonstrated .
another interesting result from the of spectroscopic parallaxes has to do with star 10 ( in our notation ) .
this is a very early - type star , b1v , affected by an amount of reddening comparable to the rest of stars .
but quite surprisingly , this object is a remote star located at almost 9.5 kpc from the sun .
if our distance estimate is not wrong , this star becomes a serious candidate to be member of the innermost part of the outer arm .
we discuss in section 5 the interpretation of this star in the framework of the nearby galaxy structure .
ruprecht 60 has been studied by bonatto & bica ( 2010 ) through 2mass photometry using a decontamination algorithm developed by them .
these authors stated the cluster is affected by a reddening over @xmath90 and placed it at a distance of more than 6 kpc .
the age they found for ruprecht 60 is @xmath91 yr .
this is the first time that @xmath0 photometry is undertaken in this overlooked object , probably , due to its weakness .
in fact , images from the _ digital sky survey _ ( dss ) show an almost irrelevant group of faint stars concentrated in a small portion of the sky as seen in fig . 1 ( left middle panel ) . notwithstanding ,
our photometry that reaches stars as faint as @xmath92 mag is quite concluding since the radial density profile shown in the middle left panel in fig .
8 confirms it is a true open cluster .
on the other hand , the extreme weakness of the brightest stars in this object ( they have magnitudes of @xmath93 ) precluded us from taking spectra in a reasonable amount of observing time .
so , our analysis of this cluster rests upon photometric data alone .
figure 11 shows the cmds @xmath69 , upper left panel , and the @xmath94 , upper right panel , of ruprecht 60 .
the middle upper panel represents the corresponding tcd . in the lower left panel
we show the @xmath95 cmd and in the lower right panel the @xmath96 color - color diagram .
the @xmath70 tcd shows very few stars with @xmath97 inside the cluster area and therefore with the highest chances to be physically related to the cluster .
we can fit these stars by shifting the intrinsic line by @xmath98 and @xmath99 .
these values are , in our interpretation of the tcd , the most probable color excesses of ruprecht 60 .
this is , we interpret there is no way the color excess of this cluster is @xmath100 as claimed by bonatto & bica ( 2010 ) .
more interesting is the @xmath69 cmd where a tight cluster main sequence dominates in the range 15 mag @xmath101 19 mag .
a worse isolation of this sequence is achieved in the @xmath102 cmd since the scatter of @xmath103 data is much more pronounced . in all the cmds
, the @xmath69 and @xmath102 particularly , the cluster sequence shows evidences of evolution as stars leave the main sequence at @xmath104 mag . since the color excess of the cluster turned out to be @xmath105 we compute the cluster distance by adjusting the sk82 zams in the cmds @xmath69 and @xmath71 achieving the best fit for an apparent distance modulus of @xmath106 .
this corresponds to a true distance modulus of @xmath107 that places ruprecht 60 at a distance of @xmath108 kpc .
the absorption law , that we use to get the absorption - free parameter above is @xmath52 as evidenced by fig .
11 , lower right panel .
a precise estimation of the age of clusters with some degree of evolution , as this is the case , is certainly complex in terms of the morphology of the turn - off point .
in fact , as summarized in friel ( 1995 ) , the number of binary stars and blue stragglers in the field of old open clusters has important implications in interpreting the color - magnitude diagram and their comparison with theoretical models , since the morphology of the turn - off region is strongly affected by the presence of composite systems .
when binary systems evolve through the turn - off region of single / simple stars , a color enlargement of the cmd diagram occurs near this point and the systems move toward larger luminosity than that of the true turn - off , before evolving into the region of red giants . apart from this ,
many old open clusters show stars in the blue straggler region of the color - magnitude diagram , above and to the blue of the main sequence turnoff .
radial velocity studies indicate that a significant number of stars found in this region are members of clusters and , among them , the spectroscopic binary frequency exceeds 40% [ milone ( 1992 ) , friel ( 1995 ) ] .
the intersection of the evolutionary paths of single stars and binary systems creates thus a more complex color and luminosity distribution precisely in the region near the turn - off , where clarity is necessary in order to adjust the set of isochrones and determine the age based on the position of this point . in ruprecht 60
cmds we do not see a potentially large number of candidates to blue straggler if any . as for the presence of binary systems
there is nothing we can say since our observations were not designed for detecting this type of objects .
therefore , and under the hypothesis that this cluster is solar metal content , the best fit for the set of isochrones by girardi et al .
( 2000 ) yielded , after several tries , an age range between 800 and 1000 million years ( @xmath109 ) as shown in fig .
our data analysis yielded results quite different from bonatto & bica ( 2010 ) .
in fact , they have found , as seen in table 5 , a larger reddening and a larger distance ( almost 50% ) than ours . and , in terms of the cluster age , ours is larger by a factor of more than 2 . even though these authors applied a less personal method ( more sophisticated too ) to estimate cluster parameters , we interpret that these differences ( cluster size aside ) come from the use of @xmath110 2mass photometry .
that is , getting the cluster reddening using @xmath111 photometry may lead to wrong results if one only rest on this photometry alone . as for the cluster age , the isochrone matching shown in their fig .
5 is somewhat weak mainly because the cluster main sequence displays only 2 mag in @xmath112 . on the contrary ,
our optical cmds show a more robust fitting not only because of the larger magnitude range but also because our isochrone matching follows pretty well the bend of the cluster main sequence .
the first detailed study of ngc 2660 was carried out by hartwick & hesser ( 1973 ) by means of @xmath54 , @xmath113 and h@xmath88 photoelectric photometry and @xmath114 photographic photometry . since
then several authors have studied this cluster as reported in table 5 .
it is noticeable that through the years different authors with different techniques converge to a same @xmath46 color excess and the distance within the usual errors .
the exception is the very discrepant value of @xmath115 kpc found by hesser & smith ( 1987 ) well above the mean of @xmath116 2.8 kpc .
the situation is quite different in relation to the age of ngc 2660 that spans a range going from 0.7 to 1.7 gyr from one author to another .
certainly , the critical point resides in the metal content of the cluster , a question that still remains open . since the purpose of our investigation excludes investigation of the metal content in ngc 2660 we rest onto the recent papers of sestito et al .
( 2006 ) and bragaglia et al .
( 2008 ) that concluded this cluster is solar metal content . just for the sake of completeness we want to mention the most recent paper of sandrelli et al .
( 1999 ) based on @xmath0 photometry who found a probable 30% population of binary stars in ngc 2660 .
mlholt et al .
( 2006 ) detected a number of @xmath117 scuti variable stars in this object .
mermilliod et al .
( 2008 ) , in turn , included this cluster in their large survey for red giants in open clusters identifying six members .
the radial density profile of ngc 2660 , right middle panel in fig .
8 , shows a strong star density covering a radius of 6 arcmin ( 2mass data ) that emerges above a quite flat star field .
our survey is therefore covering the very nuclear cluster zone . in fig .
12 we show the cmds @xmath69 , upper left panel , and the @xmath71 , upper right panel .
the middle upper panel shows the @xmath70 tcd .
the lower left panel is the @xmath102 cmd and the lower right panel is the @xmath96 color - color diagram .
the tcd , like in ruprecht 60 , is very useful to give a reasonable approximation to the true color excess affecting ngc 2660 , even if there are no blue - type stars .
the best zams shifting in the tcd is for @xmath118 , a value slightly lesser than previous estimates reported in table 5 but still in good agreement with the historical mean for ngc 2660 .
the cmds show a very well populated cluster main sequence with a well defined turn - off .
the data scatter along the main sequence seen in the @xmath119 and @xmath102 is relatively little what suggests , in our opinion , that contamination for field interlopers is low . in the @xmath69 and @xmath102 cmds of fig .
8 we show the zams of sk82 fitted with an apparent distance modulus @xmath120 and the above mentioned @xmath46 . since the @xmath96 color - color diagram ( lower right panel in fig .
8) confirms that the absorption law @xmath52 is completely valid in this region the absorption free distance modulus is @xmath121 that places ngc 2660 at a distance @xmath122 kpc .
this value agrees with those found in the literature ( see table 5 ) .
a dominant feature in all the cmds of ngc 2660 is the compact red clump at @xmath123 , also noticeable in the tcd at @xmath124 and @xmath125 . adopting the same temperament as in the case of the turn - off point of ruprecht 60 we proceed to estimate the age of the cluster looking for the best set of isochrones embracing not only the turn - off point stars but also the evident red clump .
again , assuming the cluster s metallicity is solar , and by making use of girardi et al .
( 2000 ) set of isochrones , we find that the best fit corresponds to the age interval @xmath126 .
particularly good is the isochrones fitting in the @xmath71 cmd .
notwithstanding , the red clump of this cluster raises some controversy since we could only fit simultaneously the cluster turn - off and the red clump stars in the @xmath102 cmd . in the other two cmds
we note the clump looks blue - shifted and becomes impossible to fit for an isochrone .
this effect has been noticed too in earlier works and that raised the controversy about the metal content of stars in this cluster .
photometry is unable to resolve the controversy .
just for the sake of curiosity we have made several attempts using different metal content isochrones without achieving a satisfactory result . clearly a profound spectroscopic analysis is needed in ngc 2660 .
mk types for stars in this cluster reported in table 4 are illustrative in terms of comparing reddening and distance derived from photometry with the same data from spectroscopy alone .
data in table 4 shows @xmath46 values ranging from 0.18 to 0.72 ( only one star with this extreme value ) while most stars have reddening values near the mean of 0.33 .
looking at individual star distances with mk types in ngc 2660 , we conclude that most of them are in front of the cluster at more than 0.5 kpc from it .
these stars are 2 , 3 , 6 , 9 , 14 , 19 , 44 , 80 , 100 and 110 .
photometric arguments ( the location of these stars in the photometric diagrams in fig .
12 ) favor that a number of them are not members directly or have low chances to be related to the cluster .
two stars , 8 and 61 are farther than the cluster distance ; the star 61 is otherwise bad located in the photometric diagrams .
we draw the attention to star 8 , a b8iii placed at a distance @xmath127 kpc .
we will return to this star in the next section .
star 1 has a chance , from a spectroscopic point of view , to be a likely member ; however , its position in the photometric diagrams excludes it .
stars 99 and 7 are relatively close to the cluster distance . in the case of star 7 , and given its location near the turn - off , it becomes a potential blue straggler star . like in ruprecht 47 a large number of stars with mk types turned out to be non members .
this fact permits us to throw away field interlopers with confidence and get clean diagrams of the cluster .
there is no influence of interlopers on the cluster parameters since these latter depend on the turn - off location and the red clump .
there is no influence of interlopers on the cluster parameters since these latter rely on the location of the turn - off and the red clump .
available information in the literature suggests ngc 2910 is a young open cluster .
it has been previously studied by becker ( 1960 ) , topaktas ( 1981 ) , ramsay & pollaco ( 1992 ) and kharchenko et al .
becker and topaktas made use of the same data set ( the becker ( 1960 ) one ) founding a large number of members amongst the bright stars in the zone . however , the deeper ccd photometric survey carried out by ramsay & pollaco ( 1992 , 1994 ) could only find 12 members .
kharchenko et al .
( 2005 ) revisited this cluster finding only 9 probable members .
evidently there are difficulties in understanding the upper main sequence of ngc 2910 that could be easily solved by undertaking a vigorous spectroscopic program .
although we have few stars with spectroscopy in this cluster we hope we can improve the knowledge of the region as we show below .
despite our photometric survey includes a large portion of the cluster surface it is still insufficient to cover the entire cluster given its radial extension ( see fig . 4 lower panel ) .
photometric diagrams in fig .
13 show an extended cluster main sequence down to @xmath128 mag with very low scatter around the mean location .
contamination by field interlopers is not dramatic and the cluster appears clean against the star background . as for the cluster main sequence it appears affected by a uniform reddening value that we estimate @xmath129 and show in fig .
13 middle upper panel .
this value disagrees with previous reports found in the literature and indicated in table 5 .
our reddening is more than twice the found for some authors but kharchenko et al .
( 2005 ) who found @xmath130 quite unrealistic from the tcd in the middle upper panel in fig . 13 .
we applied the same procedures used in the other clusters in this sample to derive the corresponding distance .
the superposition of the sk82 zams in the @xmath69 and @xmath71 cmds has been done for a distance modulus @xmath131 . after correcting for visual absorption , taking into account that
the extinction law is the standard one ( see fig . 13 , lower right panel ) , the distance of ngc 2910 from the sun turns out to be @xmath40 kpc a coincident value with earlier findings .
again we have a large discrepancy with kharchenko et al .
( 2005 ) who found 2.6 kpc . as for the age of the cluster
we find a single isochrone able to fit the entire main sequence .
this is the @xmath132 for an age of 63 myr .
that is , from our data analysis , the cluster is a little bit older than earlier estimates .
mk types for ngc 2910 in table 4 leave only three stars with chances to be cluster members ; they are stars 1 and 2 of evolved b - types and star 6 , another b - type star .
the other two stars , 4 and 5 in our notation , are not members of the cluster .
they are too of b - types , one of late b - type and the other is a peculiar b - type star . the remaining star with mk - types in ngc 2910 is a background f - type star .
we finally show in fig .
14 the path of @xmath46 against the distance for all the stars and clusters in the present sample .
the upper panel includes stars with mk types in the field of clusters ruprecht 20 and in ruprecht 47 .
the position of ruprecht 47 is indicated . in the lower panel
we do the same for ruprecht 60 ngc 2660 and ngc 2910 .
following well known relations that state that @xmath133 we compare our reddening estimations with the findings of kaltcheva & hilditch ( 2000 ) . for both regions under analysis ,
@xmath134 and @xmath135 the reddening values are entirely in agreement .
the presence of some patches of dust may be responsible for the couple of nearby stars in fig .
14 , in both zones , showing high reddening values .
ruprecht 20 and ruprecht 47 stars show different reddening behavior probably due to the fact that ruprecht 47 is located along the galactic plane ( steadily increasing ) while ruprecht 20 is at more than 5 degrees below the plane ( quite flat ) .
one of the clusters in our sample is ruprecht 20 that we have already shown it is not a physical entity .
this object is projected against the north of the vela gum in a region also known as puppis association .
the other cluster in our survey also in puppis is ruprecht 47 ( in the limit with pyxis constellation ) .
puppis is an intricate zone with a stellar structure hard to elucidate .
humphrey ( 1978 ) found various components of what is designated as puppis ob1 association at 2.5 kpc from the sun , while havlen ( 1976 ) had , a little before , reported a second , more distant association , puppis ob2 at 4.3 kpc .
however , the existence of these two separate associations have been questioned by kaltcheva & hilditch ( 2000 ) , who could not find evidences of them .
actually , these authors suggested the presence in the same galaxy direction of two other star groups ( at 1 kpc the first , and 3.2 kpc the second ) , but significantly lower in latitude , at @xmath136 .
these two young star groups would surround , in projection , the star cluster ngc 2439 , located at 3.5 - 4.5 kpc from the sun ( see fig . 6 in kaltcheva & hilditch ( 2000 ) ) .
ruprecht 20 is situated a little below this last location and we found two stars , numbers 10 and 3 in our notation , of b8v and b9v spectral types in the field of this object at distances of @xmath116 2.5 and @xmath116 2.0 kpc respectively . within distance errors of the order of 15 - 20% as indicated in subsection 3.3 these stars are placed in front of the stellar aggregate 2 proposed by kaltcheva & hilditch ( 2000 ) making this stellar group a long structure extending from 2 to 4 kpc from the sun .
ruprecht 47 is placed well along the galactic plane at a distance @xmath137 kpc .
it becomes therefore clearly coincident with puppis ob2 at 4.3 kpc according to havlen ( 1976 ) assertion and the genetic connection with puppis ob2 is supported by its age ( less than 100 myr ) .
it is to mention that not far from the position of ruprecht 47 there exists the open cluster haffner 20 ( @xmath138 , @xmath139 ) with similar age ( carraro et al . 2015 ) and at a distance of 5.5 kpc in the conjunction with the perseus arm .
it seems , therefore , that puppis ob2 may extend farther , even much more than previously indicated in the literature .
table 4 informs as well the existence of two b - type stars in the field of ruprecht 47 , one b1v and the other b5ve at distances of 2.0 and 2.2 kpc respectively .
if we follow the humphreys ( 1978 ) discussion these two stars belong to puppis ob1 .
connecting our findings with those of kaltcheva & hilditch ( 2000 ) it seems that early type stars in puppis constellation , including puppis ob1 and ob2 and regions 1 and 2 from kaltcheva & hilditch compose a sort of thickening ( maybe a bridge ) of the galactic disk of about 300 pc at 2 - 3 kpc from the sun extending outward up to 6 kpc . in the background of ruprecht 47
there is another star , number 10 in our notation and classified b1v , placed at the huge distance of @xmath140 kpc .
this is a surprising finding since this distance implies we have detected a young stellar component of the innermost side of the outer arm .
this star is very distant no matter the data set adopted . indeed
if we compute its spectroscopic parallax with data from vogt & moffat ( 1972 ) the distance from the sun would be of the order of 10 kpc .
ngc 2660 and ngc 2910 are two clusters placed in the vel ob1 association . according to humphreys ( 1978 ) the nucleus of the association is placed at 1.8 kpc from the sun .
the cluster ngc 2660 is not subject of a discussion of this type in terms of its belonging to vel ob1 since the age , around 1 gyr , excludes it .
however , there is another star , 8 in our notation , classified b8iii and placed at @xmath141 kpc from the sun .
we discuss below the meaning of this star . as for ngc 2910 ,
its distance of 1.3 kpc makes it a sure member of the vel ob1 association and therefore a member of the local arm . in the cluster field
there are also two stars , 4 and 5 , b5vp and b7iii , at @xmath142 kpc from the sun respectively , that are likely members of the vela ob1 association .
figure 15 is the x - y projection of the tgq showing the position of clusters and young stars analyzed in the present study .
the picture in fig .
15 is complementary of similar ones we have recently shown in carraro et al .
( 2015 ) and enlarges the available information by including more clusters and stars in puppis and vela associations .
15 and figs . 1 and 12 in carraro et al .
( 2015 ) show the presence of a large number of blue stars and several young clusters composing a sort of stellar bridge between the perseus arm and the sun position covering from @xmath143 to @xmath144 in galactic longitude .
that is , the young stellar population we have found with spectroscopy forms part of the local arm in the tgq , strictly speaking in vela and puppis associations . ngc 2910 is a young object member of the local arm and close to the border of the third and fourth galactic quadrants .
probably this cluster is one of the farthest member of the local arm in the direction to vela .
the cluster ruprecht 47 is also young and is placed close to the formal galactic plane .
so , we understand that this object is a likely member of the perseus arm ( or a very distant member of puppis ob2 in the local arm ) . as for the blue star , number 8 , in the background of ngc 2660 it may be a member of the local arm though it is still possible for it to become one of the innermost members of the perseus arm .
15 shows the location of another remote blue star , number 10 , seen in the background of ruprecht 47 at 9.5 kpc from the sun .
we have already commented above that this star may become member of the innermost part of the outer arm .
nevertheless , what makes this object a singular star is its location along the galactic plane .
in fact , at @xmath145 the warp is quite pronounced ( vzquez et al . 2008 ) but some young open clusters as haffner 18 and haffner 19 and blue stars in the background of haffner 21 and trumpler 9 can be found ( see carraro et al .
gathering all this information we interpret that the presence of this very remote young star in the galactic plane is not an isolated fact but it is a strong indication that the galactic thin disk is not only warped , but also flared , like the thick disk .
quite recently , xu et al ( 2013 ) have presented evidences of a large number of star formation regions with trigonometric parallaxes that extend the local arm back to the first galactic quadrant .
xu et al . (
2013 ) stated also a series of questions ( not answered yet ) concerning the interpretation of the local arm : 1.- it is a branch of the perseus arm , 2.- it is part of a major arm or 3.- it is an independent spiral arm segment .
unfortunately , we can not speculate about the nature of the local arm because our observations were designed just to establish the existence and the spatial distribution of young objects in the tgq
. more observations ( photometry and spectroscopy ) for a precise estimation of cluster parameters are needed in order to reach a plausible answer to these questions from an optical point of view . meanwhile , a comparison of our fig . 15 with their fig .
10 reveals that our findings are in good agreement with theirs in terms of the distribution of young population throughout the local arm in tgq .
we presented in this article the results of a photometric and spectroscopic survey carried out in five cluster located at @xmath146 and @xmath147 .
this region comprises mostly the vela gum region and includes part of the puppis association .
one of our clusters , ruprecht 60 , has been studied with @xmath54 photometry for the first time and its parameters are now improved .
other of the clusters , ruprecht 20 , is not a cluster since neither from star counts used to construct its radial density profile nor from the photometric diagrams we can conclude that it is a true entity .
it is just a random stellar fluctuation . from the point of view of identification of large stellar structures in this portion of the galaxy
we have found evidences that favor the presence in puppis , at @xmath148 , of a sort of bridge connecting two regions defined by kaltcheva & hilditch ( 2000 ) .
ruprecht 47 is definitely a member of puppis ob2 and in its field we also found two blue stars related with puppis ob1 .
ngc 2910 becomes member of vel ob1 and also a couple of blue stars with no relation to ngc 2910 . as for two peculiar isolated stars we report the presence of a very distant star seen against the field of ruprecht 47 placed at almost 10 kpc from the sun .
ruprecht 47 is along the formal galactic plane ( @xmath149 ) and this remote star in the background may belong to the innermost part of the outer arm .
another interesting star has been found in the field of ngc 2660 at near 5.7 kpc from the sun .
this star could be member of the innermost part of the perseus arm or a far member of the local arm .
as for the grand design structures in the tgq we are in the position to confirm the angular extension of the local arm from @xmath150 to @xmath151 .
our findings are in good agreement with previous studies ( see carraro et al .
2015 ) we have already performed in the sense that our evidences indicate that the galactic plane is not only warped but also flared .
this is , thin disk population shows a pattern similar to the shown by the thick disk .
moreover , there is a good spatial agreement with xu et al .
( 2013 ) findings . finally , the kind of observations we performed do not allow to speculate about the nature of the local arm but our evidences -collected in several papers as indicated in section 1 and concerning
strictly the tgq- suggest that the local arm may even cross the perseus arm and reach the outer arm .
acknowledgements : we warmly acknowledge the casleo staff for the very professional support along the years .
one of us , rav , wishes to thank to dr .
garrison for the allocation of observing time at the university of toronto southern observatory in las campanas observatory , chile .
eg , gs and rav thank the financial support from conicet pips 1359 and 5970 and from la plata observatory .
we especially acknowledge to our anonymous referee for the careful and collaborative report that greatly contributed to improve the value of our manuscript .
this publication makes use of the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
the digitized sky survey was produced at the space telescope science institute under u.s .
government grant nag w-2166 .
the images of these surveys are based on photographic data obtained using the oschin schmidt telescope on palomar mountain and the uk schmidt telescope .
the plates were processed into the present compressed digital form with the permission of these institutions . | we report on @xmath0 photometry and spectroscopy for mk classification purposes carried out in the fields of five open clusters projected against the vela gum in the third galactic quadrant of the galaxy .
they are ruprecht 20 , ruprecht 47 , ruprecht 60 , ngc 2660 and ngc 2910 .
we could improve / confirm the parameters of these objects derived before .
ruprecht 20 is not a real physical entity , in agreement with earlier suggestions .
ruprecht 47 , a young cluster in the galactic plane , at 4.4 kpc from the sun is quite farther than in previous distance estimations and becomes , therefore , a member of the puppis ob2 association .
for the first time ruprecht 60 was surveyed in @xmath0 photometry .
we found it to be placed at 4.2 kpc from the sun of about and 1 gyr old .
ngc 2660 is another old object in our survey for which distance and age are coincident with previous findings .
ngc 2910 turns out to be a young cluster of vela ob1 association at a distance of 1.4 kpc approximately and 60 myr old .
the spectroscopic parallax method has been applied to several stars located in the fields of four out of the five clusters to get their distances and reddenings . with this method we found two blue stars in the field of ngc 2910 at distances that make them likely members of vela ob1 too .
also , projected against the fields of ruprecht 20 and ruprecht 47 we have detected other young stars favoring not only the existence of puppis ob1 and ob2 but conforming a young stellar group at @xmath1 kpc from the sun and extending for more than 6 kpc outward the galaxy . if this is the case , there is a thickening of the thin galactic disk of more than 300 pc at just 2 - 3 kpc from the sun .
ruprecht 60 and ngc 2660 are too old objects that have no physical relation with the associations under discussion .
an astonishing result has been the detection in the background of ruprecht 47 of a young star at the impressive distance of 9.5 kpc from the sun that could be a member of the innermost part of the outer arm .
another far young star in the field of ngc 2660 , at near 6.0 kpc , may become a probable member of the perseus arm or of the inner part of the local arm .
the distribution of young clusters and stars onto the third galactic quadrant agrees with recent findings concerning the extension of the local arm as revealed by parallaxes of regions of star formation .
we show evidences too that added to previous ones found by our group explain the thickening of the thin disk as a combination of flare and warp .
open clusters and associations : general , milky way : structure , galactic spiral arms |
the role of neutrinos in primordial nucleosynthesis has been exploited to infer constraints on the number of leptons and light neutrinos in the standard model of particle physics @xcite .
complimentary to this , the width of the @xmath5 was found to allow only three light weakly - coupled neutrinos @xcite .
this experimental determination of the number of neutrinos , @xmath6 , has led some to describe primordial nucleosynthesis as determined by only one parameter : the baryon - to - photon ratio , @xmath7 . along with the assumptions of homogeneity and isotropy ,
the theory of standard big bang nucleosynthesis ( sbbn ) predicts the abundance of the lightest nuclides within reasonable error through the single parameter @xmath7 .
now , however , indications of neutrino oscillations from the atmospheric muon neutrino deficit , the solar neutrino deficit and the lsnd excess may prove that our understanding of neutrino physics is incomplete @xcite . this uncertainty in neutrino
physics threatens to destroy the aesthetics of a one parameter sbbn with the several parameters of the neutrino mass and mixing matrix .
neutrino mixing in the early universe can induce effects including matter - enhanced neutrino flavor transformation , partial thermalization of extra degrees of freedom , and lepton number generation .
efforts are sometimes made to crowd the multiple effects of neutrino mixing into a single parameter , @xmath8 , the effective number of neutrinos .
the kinetic equations describing the effects of neutrino mixing in the early universe were first discussed by dolgov @xcite .
larger @xmath9 and @xmath10 accelerate population of sterile states .
resonant conversion of active neutrinos to sterile neutrinos may also take place when the mass eigenvalue corresponding to @xmath11 is smaller than that for the @xmath12 , _
i.e. _ , @xmath13 .
both resonant and non - resonant conversion has been used to place constraints on active - sterile neutrino properties @xcite .
several formulations of neutrino dynamics in the early universe evolve the system at a monochromatic average neutrino energy with interactions occuring at an integrated rate .
this approach is appropriate when the time scale of interactions between the neutrinos and the thermal environment is much shorter than the dynamic time scale of neutrino flavor transformations .
however , when tranformations and the hubble expansion are more rapid than thermalization , neutrino energy distributions can be distorted from a fermi - dirac form .
furthermore , such spectral distortions can survive until nucleosynthesis @xcite .
the role of neutrino energy distributions in primordial nucleosynthesis has been discussed since the earliest work on the subject @xcite .
recent work @xcite has emphasized the importance of neutrino energy distributions on the weak nucleon interconversion rates @xmath14 .
kirilova and chizhov @xcite included the distortion of neutrino spectra from nonresonant transformations in their nucleosynthesis calculation . a time - evolving distortion was also included in a calculation of helium production after a neutrino - mixing - generated lepton asymmetry in electron - type neutrinos from @xmath15 or @xmath16 ( active - sterile ) mixing@xcite . in this paper we explore in detail the effects on primordial nucleosynthesis of the nonthermal neutrino spectra produced by the resonant transformations discussed in @xcite . the resonant transformation of electron - type neutrinos to steriles can leave a near absence of neutrinos in the low energy end of the active neutrino energy spectrum . we find that this suppression of low energy neutrinos or anti - neutrinos in the otherwise fermi - dirac spectrum alters the neutron - to - proton weak interconversion reactions .
these alterations stem from initial and final phase space effects . specifically , we find that the absence of low energy electron anti - neutrinos lifts the fermi blocking of final state neutrinos in neutron decay .
in addition , the absence of low energy electron neutrinos capturing on neutrons significantly alters the rate of this interaction . both cases lead to an alteration of the neutron - to - proton ratio at weak freeze out , which is directly related to the produced mass fraction of primordial helium .
the general equations describing the evolution of @xmath17 flavors of neutrinos in a high density environment such as the early universe are derived from the evolution equation for the density operator @xmath18.\ ] ] here @xmath19 can be the density operator for the entire system of @xmath17 neutrinos or any smaller neutrino system that evolves rather independently of the other neutrino mixings .
several authors @xcite explicitly follow the evolution of neutrinos through the density matrix .
the evolution of the density operator for a two state active - sterile system has been popularized in a vector formalism where the density operator is @xmath20 where @xmath21 , @xmath22 and @xmath23 refer to the anti - neutrino states .
for an active - sterile two state system , the hamiltonian of eqn .
( [ hberg ] ) includes forward scattering off of the @xmath24 background , and elastic and inelastic scattering of neutrinos from all leptons ( see , _
e.g. _ , ref .
@xcite ) . for this system , substituting ( [ rhop ] ) into the evolution equation resulting from eqn .
[ hberg ] gives @xmath25 @xmath26 where an average over the momentum ensemble of neutrinos is taken for the vector potential @xmath27 and the interaction rates @xmath28 . here
@xmath29 and @xmath30 are the damping coefficients due to inelastic and elastic neutrino scattering off of @xmath24 and themselves ( see @xcite ) . also , @xmath31 and @xmath32 .
this set of equations ( [ masterp],[masterp2 ] ) can be followed for specific neutrino momenta .
an initial thermal neutrino spectrum can be discretized into @xmath33 energy bins . along with the number density evolution equations of neutrinos that do not undergo transformation
, these equations form @xmath34 equations describing the evolution of the system . here
we assume that the sterile neutrino sea is unpopulated at a temperature @xmath35 ( see ref .
active neutrinos can undergo resonant transformation to sterile neutrinos as the density of @xmath24 decreases with the expansion .
as in the sun , the resonance point is energy dependent .
for example , the resonance , for @xmath36 , occurs at @xmath37 where @xmath38 is the neutrino energy scaled by the ambient temperature .
this resonance adiabatically converts @xmath39 to @xmath40 as long as the timescale of the resonance is much longer than the neutrino oscillation period at resonance , and the resonance position must move slowly through the neutrino energy spectrum ( see ref .
@xcite ) . as the universe expands and the ambient temperature decreases , the resonance moves from lower to higher neutrino energies .
the energy - dependence of the resonance becomes important at temperatures below @xmath41 @xcite .
below this temperature , the neutrinos decouple from the @xmath24 background .
the weak interaction rates of elastic and inelastic scattering of neutrinos above @xmath42 are much faster than the expansion time scale . in this case
, the neutrino energy distributions can reshuffle into a thermal fermi - dirac spectrum before the onset of nucleosynthesis .
however , if the transformation occurs after decoupling , the expansion rate will be too rapid , and therefore the neutrino spectra enter the epoch of nucleosynthesis in a non - thermal state
. this will be the case , usually , when the mass - squared difference between mass eigenvalues corresponding to @xmath2 and @xmath11 is @xmath43 .
direct @xmath44 transformation was explored by foot and volkas @xcite as a possible means to allow for a sterile neutrino within sbbn with minimal modification to the theory , and also to reduce the difference between the @xmath45he abundance predicted by the observed primordial deuterium abundance @xcite and the primordial @xmath45he abundance inferred by olive _
_ @xcite . the change in primordial helium production in this @xmath46 scenario , including alterations of neutrino spectra ,
was described in @xcite .
the resonant transformation of @xmath44 produces a lepton number asymmetry in electron - type neutrinos , @xmath47 .
here we define the net lepton number residing in neutrino species @xmath12 to be @xmath48 , where @xmath49 , @xmath50 , and @xmath51 are the proper number densities of @xmath12 , @xmath52 , and photons , respectively .
the sign of this asymmetry depends sensitively on initial thermal conditions and can be different in causally disconnected regions @xcite . in the following sections , we address how non - thermal spectra affect the production of primordial helium .
since nearly all of the neutrons present during primordial nucleosynthesis go into @xmath45he nuclei , the production of primordial helium in the early universe is dominated by the abundance of neutrons relative to protons .
the weak nucleon interconversion reactions @xmath53 are in steady state equilibrium at temperatures @xmath54 .
the neutron - to - proton ratio @xmath55 freezes out of equilibrium when the weak nucleon interconversion rates fall below the expansion rate ( see fig .
[ ratplot ] ) .
the `` frozen '' @xmath56 ratio slowly evolves as free neutrons decay until the epoch ( @xmath57 mev ) when almost all neutrons are incorporated into alpha particles ( `` nucleosynthesis '' ) . in steady state equilibrium
, @xmath56 can be approximated by the ratio of the nucleon interconversion rates , @xmath58 the evolution of @xmath56 is followed numerically in detail in kawano s @xcite update of wagoner s code @xcite .
the suppression of low energy electron - type neutrinos affects the rates of the six weak nucleon reactions differently .
if sterile neutrinos are partially or completely thermalized , they will increase the expansion rate since they will contribute to the energy density of the primordial plasma .
the expansion rate , @xmath59 , is related to the total energy density , @xmath60 and the cosmological constant , @xmath61 as @xcite @xmath62 in general , an increase to the energy density and expansion rate will increase @xmath56 at freeze - out , and thus increase the primordial @xmath45he yield , @xmath63 . the contribution to the expansion rate is not significant for @xmath44 conversion post - decoupling , since the energy density in electron - type neutrinos is just transferred to steriles without further thermal creation of neutrinos ( since they have decoupled from the @xmath24 plasma before the onset of transformation ) . for the case where @xmath64 , the reactions influenced by the @xmath2 distortion are the forward and reverse reactions of @xmath65 . because of the neutron / proton mass difference , very low energy neutrinos readily participate in @xmath66 . here , the neutrino energy is @xmath67 , where @xmath68 is the nucleon mass difference , and @xmath69 is the electron energy .
the suppression of low energy neutrinos impacts this reaction by significantly limiting the phase space of initial states of the interaction .
the neutrino energy distribution is accounted for in the @xmath70 rate integral in the following manner : @xmath71^{-1 } [ 1+e^{-e_e /
k t}]^{-1}}.\ ] ] the rate of the reaction @xmath72 is given by @xmath73^{-1 } [ 1+e^{-e_\nu / k t_\nu}]^{-1}}.\ ] ] in these equations we have used weinberg s notation @xcite .
the deficit in low energy neutrinos renders the integrand of ( [ rate2 ] ) to be nearly zero for energy ranges where @xmath2 have transformed to steriles ( see fig .
[ intnue ] ) .
the reaction @xmath74 is altered by a lifting of fermi - blocking @xmath75^{-1}$ ] at low energies because of the reduction of low energy neutrino numbers .
however , the rate @xmath76 is changed less significantly than @xmath77 and thus affects @xmath56 less .
this can be seen in fig .
[ deviate ] ( a ) , where the change in rate with respect to the overall @xmath78 rate is shown for rate ( 2 ) @xmath79 and ( 6 ) @xmath80 . for @xmath81 ,
the reactions affected by @xmath4 distortion are @xmath82 and @xmath83 . in the forward and reverse reactions @xmath82 , the escaping or incident @xmath4 must have an energy @xmath84 . even for the case we examined with the greatest spectral distortion ( @xmath85 ) ,
the spectral cutoff never extends above @xmath86 .
therefore , @xmath82 is not altered by this low energy distortion ( see fig .
[ deviate ] ( b ) ) .
the interaction @xmath87 makes a very small contribution to @xmath88 , and modifications of its rate are not important to @xmath56 at freeze - out .
the only significant reaction rate affected by the @xmath4 distortion is neutron decay : @xmath89 ( fig .
[ deviate ] ) .
the reaction is limited in sbbn by fermi - blocking of low - energy @xmath4 products .
this blocking is lifted when low - energy @xmath4 transform to @xmath11 .
the integrand of the neutron decay rate @xmath90^{-1 } [ 1+e^{-e_e / k t_\nu}]^{-1}}\ ] ] can be seen in fig .
[ intnueb ] for the most dramatic case of @xmath91 .
the fermi - blocking term , @xmath92^{-1}$ ] , is unity for low energy neutrinos that have transformed to steriles .
the change in @xmath79 ( for @xmath64 ) is significantly larger than that for @xmath93 ( for @xmath81 ) .
the change in helium mass fraction ( @xmath63 ) produced by bbn for @xmath64 is @xmath94 , while a @xmath81 correspondingly causes smaller a change of @xmath95 @xcite . in summary ,
if electron - type neutrino spectra are altered through matter - enhanced transformation and survive until nucleosynthesis , they can modify @xmath96 interconversion rates through the availability of initial and final neutrino energy states .
this kind of scenario is realized in @xmath97 matter - enhanced transformations with @xmath98 .
non - thermal @xmath99 spectra will arise in any case where flavor transformations occur near or below @xmath42 between a more - populated electron - type neutrino and some other less - populated neutrino flavor , and will produce the same kinds of limitations on reaction phase space as the resonance point goes from lower to higher neutrino energies .
the effects of non - thermal electron - type neutrino distributions will then need to be included . | we examine in detail the changes in the production of primordial helium resulting from nonthermal neutrino momentum distributions produced by resonant transformation of electron - type neutrinos to steriles .
these transformations , @xmath0 ( @xmath1 ) , amplify a positive ( negative ) lepton number asymmetry .
we find that the resulting supression relative to a thermal distribution of low energy @xmath2 reduces @xmath3 conversion to a greater extent than does the enhancement of @xmath3 from an identical suppression of @xmath4 .
thus , equal lepton - number asymmetries of opposite sign have unequal effects on the resulting helium yield in primordial nucleosynthesis .
amssym.tex # 1#23.6pt , _ et al . , _ |
in a recent paper [ 28 ] , we proved existence and multiplicity results for non - linear second - order periodic systems driven by the one - dimensional @xmath0-laplacian and having a non - smooth potential .
our results there extended to the recent works of tang [ 31,32 ] , who examined semilinear ( i.e. @xmath1 ) systems with smooth potential . in this paper
we continue the study of non - linear , non - smooth periodic systems .
we prove new existence theorems under more general growth conditions on the non - smooth potential . in [ 28 ]
all the results assumed a strict sub-@xmath0 growth ( i.e. strictly sublinear potential in the semilinear ( @xmath1 ) case ) . here
the growth conditions are more general .
also we obtain new multiplicity results and we also establish the existence of non - trivial homoclinic solutions . our approach is variational and it is based on the non - smooth critical point theory of chang [ 4 ] .
extensions of this theory were obtained recently by kourogenis and papageorgiou [ 17 ] and [ 18 ] .
problems with non - differentiable potential which is only locally lipschitz in the state variable @xmath2 , are known as ` hemivariational inequalities ' and have applications in mechanics and engineering . for details in this direction
we refer to the book of naniewicz and panagiotopoulos [ 27 ] . in the last decade
there has been an increasing interest for problems involving the one - dimensional @xmath0-laplacian or generalizations of it .
we refer to the works of dang and oppenheimer [ 6 ] , del pino _
et al _ [ 7 ] , fabry and fayyad [ 8 ] , gasinski and papageorgiou [ 9 ] , guo [ 11 ] , halidias and papageorgiou [ 12 ] , kyritsi _ et al _ [ 19 ] , manasevich and mawhin [ 21 ] , mawhin [ 22,23 ] and the references therein .
as we have already mentioned our approach is variational , based on the non - smooth critical point theory . for the convenience of the reader , in this section we recall the basic facts from this theory .
it is based on the clarke subdifferential theory for locally lipschitz functions .
let @xmath3 be a banach space and .
we say that @xmath4 is locally lipschitz , if for every bounded open set @xmath5 , there exists a constant @xmath6 such that @xmath7 for all @xmath8 .
it is a well - known fact from convex analysis that a proper , convex and lower semicontinuous function is locally lipschitz in the interior of its effective domain @xmath9 .
in particular an @xmath10-valued , convex and lower semicontinuous function is locally lipschitz . in analogy with the directional derivative of a convex function , for a locally lipschitz function ,
we define the generalized directional at derivative @xmath11 in the direction @xmath12 , by @xmath13{}x } \frac{\varphi ( x'+ \lambda h)-\varphi ( x ' ) } { \lambda}.\ ] ] it is easy to check that the function @xmath14 is sublinear , continuous and so by the hahn banach theorem it is the support function of a non - empty , convex and @xmath15-compact set @xmath16 the set @xmath17 is known as the generalized ( or clarke ) subdifferential of @xmath4 at @xmath11 . if @xmath18 are both locally lipschitz functions , then for all @xmath19 and all @xmath20 we have @xmath21 and @xmath22 .
moreover , if @xmath4 is also convex , then the subdifferential @xmath23 coincides with the subdifferential in the sense of convex analysis .
recall that the convex subdifferential of @xmath4 is defined by @xmath24 . also if @xmath25 , then @xmath26 for all @xmath11 . given a locally lipschitz function @xmath27 ,
a point @xmath11 is said to be a ` critical point ' of @xmath4 , if @xmath28 .
if @xmath25 , then as we saw above , @xmath26 and so this definition of critical point coincides with the classical ( smooth ) one .
it is easy to see that if @xmath11 is a local extremum of @xmath4 ( i.e. a local minimum or a local maximum ) , then @xmath28 . from the smooth critical point theory , we know that a basic tool is a compactness - type condition , known as the ` palais
smale condition ' ( ps - condition for short ) . in the present non - smooth setting
this condition takes the following form : ` a locally lipschitz function @xmath29 satisfies the non - smooth ps - condition , if every sequence @xmath30 such that @xmath31 is bounded and @xmath32 \rightarrow 0 $ ] as @xmath33 , has a strongly convergent subsequence ' .
a version of the theory based on a weaker condition known as the ` non - smooth @xmath34-condition ' can be found in kourogenis and papageorgiou [ 17 ] .
a @xmath20 is said to be an ` eigenvalue ' of minus the @xmath0-laplacian with periodic boundary conditions , if the problem @xmath35\\[3pt ] x(0)=x(b ) , x'(0)=x'(b ) , \quad 1 < p<\infty \end{array } \right \},\ ] ] has a non - trivial solution @xmath36 , known as corresponding to @xmath37 ` eigenfunction ' .
let @xmath38 denote the set of these eigenvalues .
evidently @xmath39 and if @xmath40 , then for every @xmath41 the periodic problem @xmath42\\[3pt ] x(0)=x(b),\quad x'(0)=x'(b ) \end{array } \right \},\ ] ] has at least one solution .
each element of @xmath38 is non - negative and @xmath43 is the smallest ( first ) eigenvalue . if @xmath44 ( scalar case ) , by direct integration of the equation we obtain all the eigenvalues which are @xmath45 when @xmath1 ( semilinear case ) , then @xmath46 and we recover the well - known eigenvalues of the ` scalar periodic negative laplacian ' which are @xmath47 with @xmath48 . in the case
@xmath49 ( vector case ) , @xmath50 but @xmath38 contains more elements ( see [ 22 ] ) .
in this section we prove an existence theorem for non - smooth periodic systems driven by the ordinary vector @xmath0-laplacian , which will be used in our investigation of homoclinic orbits in 5 .
it concerns the following non - linear and non - smooth periodic system : @xmath51\\[2pt ] x(0)=x(b ) , x'(0)=x'(b),\ ; 1 < p<\infty \end{array } \right \}.\ ] ] our hypotheses on the data of ( 1 ) are the following : @xmath52 : @xmath53 , @xmath54 and for all @xmath55 , @xmath56 .
@xmath57 : @xmath58 is a functional such that @xmath59 , @xmath60 and * for all @xmath61 , @xmath62 is measurable ; * for almost all @xmath63 , the function @xmath64 is locally lipschitz ; * for almost all @xmath55 , all @xmath61 and all @xmath65 , we have @xmath66 @xmath67 with @xmath68 ; * there exists @xmath69 such that for almost all @xmath55 and all @xmath70 with @xmath71 we have @xmath72 * @xmath73 uniformly for almost all @xmath55 ; * there exists @xmath74 such that @xmath75 . if hypotheses @xmath76 , @xmath77 hold , then problem @xmath78 has at least one non - trivial solution @xmath36 with @xmath79 .
let @xmath80 be the locally lipschitz function defined by @xmath81 first we show that @xmath4 satisfies the non - smooth ps - condition . to this end
let @xmath82 be a sequence such that @xmath83 for all @xmath84 and some @xmath85 and @xmath86 .
since @xmath87 is @xmath88-compact , the norm functional in a banach space is weakly lower semicontinuous and @xmath89 is embedded compactly in @xmath90 , from the weierstrass theorem we know that we can find @xmath91 such that @xmath92 we have @xmath93 with @xmath94 being the non - linear operator defined by @xmath95 @xmath96 and @xmath97 a.e .
on @xmath98 ( see [ 5 ] , pp .
47 and 83 ) .
it is easy to check that @xmath99 is monotone , demicontinuous ; thus maximal monotone ( see [ 14 ] , p. 309 ) .
combining hypothesis @xmath57(iii ) with the lebourg mean value theorem ( see [ 20 ] or p. 41 of [ 5 ] ) , we see that for almost all @xmath55 and all @xmath100 , @xmath101 from the choice of the sequence @xmath102 , we have @xmath103 & \;\bigg(\frac{\mu}{p}-1\bigg)\|x'_n\|_p^p+\bigg(\frac{\mu}{p}-1\bigg)\int_0^b g(t)\|x_n(t)\|^p \d t\\[5.7pt ] & \quad \ \,\,-\int_0^b [ ( u_n(t),-x_n(t))_{\mathbb{r}^n}+\mu j(t , x_n(t))]\\[5.7pt ] & \;\;\leq\mu m_1+\varepsilon_n\|x_n\|\\[5.7pt ] \rightarrow & \ ; \bigg(\frac{\mu}{p}-1\bigg)\bigg(\|x'_n\|_p^p+\int_0^b
g(t)\|x_n(t)\|^p \d t \bigg)\\[5.7pt ] & \quad \ \,\,+\int_0^b [ -j^0(t , x_n(t);-x_n(t))-\mu j(t , x_n(t ) ) ] \d t \\ & \;\;\leq\mu m_1+\varepsilon_n\|x_n\|.\end{aligned}\ ] ] using hypotheses @xmath57(iii ) and ( iv ) , we obtain @xmath104 \d t\\ & = \;\int_{\{\|x_n\|<m\ } } [ -j^0(t , x_n(t);-x_n(t))-\mu j(t , x_n(t ) ) ] \d t\ ; \\ & \quad\,+ \;\;\int_{\{\|x_n\|\geq m\ } } [ -j^0(t , x_n(t);-x_n(t))-\mu j(t , x_n(t ) ) ] \d t\\ & \;\geq \;-c_2\quad \t{for some}\;c_2>0\;\;\t{and all } \;n\geq 1.\end{aligned}\ ] ] therefore it follows that @xmath105 & \rightarrow \|x_{n}\|^{p}\leq c_{4}+\varepsilon'_{n}\|x_{n}\| \quad \t{for some } \
\ c_{4}>0 \ \ \t{and } \ \
\varepsilon'_{n}\downarrow 0.\end{aligned}\ ] ] from the last inequality it follows that @xmath106 is bounded and so by passing to a subsequence if necessary , we may assume that @xmath107 in @xmath89 and @xmath108 in @xmath90 .
we have @xmath109 because @xmath99 is maximal monotone , it is a generalized pseudomonotone ( see [ 14 ] , p. 365 ) and so we have @xmath110 .
because @xmath111 in @xmath112 and the latter is uniformly convex , from the kadec
klee property ( see [ 14 ] , p. 28 ) , we have @xmath113 in @xmath112 , hence @xmath108 in @xmath114 . so @xmath4 satisfies the non - smooth ps - condition . because of hypothesis @xmath115(v ) ,
given @xmath116 we can find @xmath117 such that for almost all @xmath55 and all @xmath61 with @xmath118 we have @xmath119 . on the other hand , hypothesis h(j)@xmath120(iii ) and the lebourg mean value theorem , imply that for almost all @xmath55 and all @xmath61 with @xmath121 we have @xmath122 for some @xmath123 .
so finally for almost all @xmath55 and all @xmath124 we can write that @xmath125 for some @xmath126 and with @xmath127 .
therefore for every @xmath128 we have @xmath129 because @xmath130 is embedded continuously in @xmath131 , we have @xmath132 taking @xmath133 we obtain that @xmath134 recall that @xmath135 .
so we can find @xmath136 small so that @xmath137=\xi>0 $ ] . on @xmath138 ,
the function @xmath139 is continuous convex , thus it is locally lipschitz . from ( [ 5 ] ,
we have that @xmath140 is locally lipschitz on @xmath138 for almost all @xmath55 ( hypothesis h(j)@xmath120(ii ) ) and we have @xmath141 using lebourg s mean value theorem , we can find @xmath142 such that @xmath143 & \hskip 8.5pc \geq\frac{r-1}{\lambda^{\mu + 1}}(-\mu j(t,\lambda x)-j^{0 } ( t,\lambda x;- \lambda x))\geq 0 \\[.4pc ] & \hskip 9pc\ , \ \
( \t{see hypothesis } \ \ \hbox{h}(\hbox{j})_{1}(\hbox{iv } ) ) \\[.4pc ] & \rightarrow r^{\mu}j(t , x)\leq j(t , rx ) \quad \t{for almost all } \ \
t\in t,\ ; \t{all } \
\|x\|\geq m \ \
\t{and all } \ \
r\geq 1.\end{aligned}\ ] ] choosing @xmath144 as postulated by hypothesis h(j)@xmath145(vi ) , for @xmath146 large we have @xmath147 \ & \leq \lambda^p \|g\|_{\infty}\|x_*\|^p b-\lambda^{\mu } \int_{0}^{b}j(t , x_*)\d t\\[.4pc ] \rightarrow\ & \
\varphi(\lambda x_*)\rightarrow -\infty \ \
\t{as } \ \ \lambda\rightarrow + \infty \ \ ( \t{recall that } \ \
\mu > p).\end{aligned}\ ] ] thus we can find @xmath148 large so that @xmath149 and @xmath150 .
also note that @xmath151 ( recall that @xmath152 ) .
therefore we can apply the non - smooth mountain pass theorem ( see [ 4 ] or [ 17 ] ) and obtain @xmath128 , @xmath153 such that @xmath154 .
we have @xmath155 and so @xmath156 let @xmath157 .
we have @xmath158 since @xmath159 ( see [ 1 ] , p. 50 ) , we have @xmath160 with @xmath161 denoting the duality brackets for the pair @xmath162 .
since @xmath163 is dense in @xmath164 , it follows that @xmath165 from ( 3 ) it follows that @xmath166 . because the map @xmath167 is a homeomorphism on @xmath168 onto itself and @xmath169
, we infer that @xmath170 , hence @xmath171 .
next if @xmath172 , from green s inequality ( integration by parts ) , we have @xmath173 using ( 2 ) and ( 3 ) , we obtain @xmath174 & \quad \t{for all $ y\in w_{\rm per}^{1,p}(t,\mathbb{r}^{n})$},\\[3pt ] \rightarrow & \|x'(0)\|^{p-2}x'(0)=\|x'(b)\|^{p-2}x'(b),\\[2pt ] \rightarrow & x'(0)=x'(b).\end{aligned}\ ] ] also since @xmath175 , @xmath176 .
therefore @xmath36 is the desired solution of ( 1).qed the following function is a non - smooth potential satisfying hypotheses h(j)@xmath177 ( and does not satisfy the conditions imposed by tang [ 31,32 ] ( for @xmath1 ) and papageorgiou and papageorgiou [ 28 ] ) .
again for simplicity we drop the @xmath178-dependence .
we have @xmath179 since @xmath180 , we have @xmath181 if @xmath182 .
thus hypotheses h(j)@xmath120 hold .
in this section we prove a multiplicity result .
it concerns an eigenvalue version of problem ( 1 ) .
@xmath183\\[2pt ] x(0)=x(b ) , x'(0)=x'(b ) , \lambda\in\mathbb{r}. \end{array } \right \}.\ ] ] we prove a multiplicity result for a whole semiaxis of values of the parameter @xmath184 . our hypotheses on the non - smooth potential are the following : * @xmath185 is a functional such that @xmath59 and * for all @xmath61 , @xmath62 is measurable ; * for almost all @xmath63 , the function @xmath64 is locally lipschitz ; * for almost all @xmath55 , all @xmath61 and all @xmath65 , we have @xmath186 @xmath187 with @xmath188 , * @xmath189 and there exists @xmath190 such that @xmath191 ; * @xmath192 uniformly for almost all @xmath55 . if hypotheses @xmath76 and @xmath193 hold , then there exists @xmath194 such that for all @xmath195 problem @xmath196 has at least two non - trivial solutions @xmath197 such that @xmath198 , @xmath199 . for every @xmath200 we consider the locally lipschitz functional @xmath201 defined by @xmath202 first
we show that @xmath203 satisfies the ps - condition . for this purpose
, we consider a sequence @xmath82 such that @xmath204 for all @xmath84 and some @xmath85 and @xmath86 . as before we can find @xmath205 such that @xmath206 for all @xmath84 .
for every @xmath84 we have @xmath207 where @xmath208 is as in the proof of theorem 1 and @xmath209 , @xmath210 a.e . on @xmath98 . from hypothesis h(j)@xmath177(iii ) and the lebourg mean value theorem , we obtain that for almost all @xmath55 and all @xmath61 , @xmath211 with @xmath212 . for every @xmath84
, we have @xmath213 for some @xmath214 . using young s inequality with @xmath116 , we obtain @xmath215 for some @xmath216 ( recall that @xmath217 ) . therefore we obtain @xmath218 for all @xmath84 and some @xmath219
. choose @xmath116 so that @xmath220 .
from ( 5 ) it follows that @xmath221 is bounded . arguing as in the last part of the proof of theorem 1 , we conclude that @xmath203 satisfies the non - smooth ps - condition .
in fact from ( 5 ) we infer that @xmath203 is coercive . also exploiting the compact embedding of @xmath130 into @xmath131 ( sobolev embedding theorem )
, we can check easily that @xmath203 is sequentially weakly lower semicontinuous .
so from the weierstrass theorem it follows that there exists @xmath222 such that @xmath223 and @xmath224 .
next , let @xmath225 be the integral functional defined by @xmath226 . by virtue of hypothesis h(j)@xmath177(iv )
, we have @xmath227 .
the sobolev space @xmath228 is dense in @xmath229 .
so we can find such that .
therefore there exists @xmath194 large enough such that for all @xmath195 we have @xmath230 .
hence @xmath231 , i.e. @xmath232 . since @xmath224 we verify that @xmath233 , @xmath234 and that it is a non - trivial solution of ( 4 ) . because of hypothesis h(j)@xmath177(v
) we can find @xmath235 and @xmath117 such that for almost all@xmath55 and all @xmath236 , we have @xmath237 . combining this with the growth condition on @xmath238 , we obtain that for almost all @xmath55 and all @xmath61 , @xmath239 for some @xmath123 and with @xmath135 .
so we can write that @xmath240 thus if we choose @xmath241 small enough , we can have that @xmath242=\gamma>0.\ ] ] since @xmath243 , @xmath232 and @xmath244 , we can apply the non - smooth mountain pass theorem and obtain @xmath245 such that @xmath246 , hence @xmath247 , @xmath248 and @xmath249 . as before for @xmath199 we can check that @xmath250 , @xmath251 and it solves ( 4).qed the following non - smooth potential satisfies hypotheses h(j)@xmath177 ( again we drop the @xmath178-dependence ) : @xmath252 \frac{1}{r}\|x\|^{r}+\cos\|x\|+c , & \mbox{if $ \|x\|\geq1 $ } \end{cases},\quad r
< p \quad \hbox{and}\quad c=\frac{1}{p}-\frac{1}{r}-\cos 1.\ ] ] note that @xmath253 \t{conv } \{-x , x-(\sin1)x\ } , & \mbox{if $ \|x\|=1$}\\[3pt ] \|x\|^{r-2}x-\frac{x}{\|x\|}\sin \|x\| , & \mbox{if $ \|x\|>1$}. \end{cases}\ ] ]
in this section we turn our attention to the question of existence of homoclinic solutions ( to 0 ) , for the homoclinic problem in @xmath168 corresponding to ( 1 ) .
namely , we consider the problem : @xmath254 so far the ` homoclinic problem ' for second order systems has been studied only in the context of semilinear equations , primarily with smooth potential .
we refer to the works of grossinho _
et al _ [ 10 ] , korman and lazer [ 16 ] , rabinowitz [ 29],yanheng [ 34 ] and the references therein .
non - smooth semilinear systems were studied only recently by adly and goeleven [ 2 ] and hu [ 13 ] , using different methods .
to our knowledge our result is the first one ( even in the context of smooth systems ) on the existence of homoclinic ( to 0 ) orbits for quasilinear systems .
our approach is based on that of rabinowitz [ 29 ] ( see also [ 10 ] ) .
our hypotheses on the non - smooth potential are the following : h(j)@xmath255 : @xmath256 is a functional such that @xmath257 a.e .
on @xmath10 and * for all @xmath61 , @xmath62 is measurable and @xmath258-periodic ; * for almost all @xmath259 , the function @xmath64 is locally lipschitz ; * for almost all @xmath260 , all @xmath61 and all @xmath65 , we have @xmath261 with @xmath262 ; * there exists @xmath69 such that for almost all @xmath260 and all @xmath70 with @xmath71 , we have @xmath263 * @xmath264 uniformly for almost all @xmath260 ; * there exists @xmath265 such that @xmath266 .
hypothesis h(j)@xmath267(v ) is equivalent to the following one : * @xmath268 uniformly for almost all @xmath260 and all @xmath269 .
first we show that @xmath270 . from the lebourg mean value theorem , we know that for almost all @xmath260 and all @xmath271 , we have @xmath272 \ & = \frac{j(t,\frac{x}{2})}{2^{p}\|\frac{x}{2}\|^{p}}+ \frac{\lambda^{p-1}}{2^{p } } \frac{(u,\frac{\lambda x}{2})_{\mathbb{r}^{n } } } { \|\frac{\lambda x}{2}\|^{p}},\end{aligned}\ ] ] @xmath273 as @xmath274 , we have @xmath275 and so we conclude that @xmath276 uniformly for almost all @xmath260 , i.e. @xmath277 holds . next
we show that @xmath278 . from the previous argument for almost all @xmath260 and all @xmath279 we have @xmath280 and
the convergence is uniform for almost all @xmath260 .
so ( v ) holds .
thus we have proved that @xmath281 . also the hypothesis on the coefficient function
@xmath282 takes the following form : + h(g)@xmath120 : @xmath283 , @xmath282 is 2@xmath284-periodic and for all @xmath285 $ ] , @xmath286 .
if hypotheses @xmath287 and @xmath288 hold , then there exists a non - trivial homoclinic solution @xmath289 for problem @xmath290 .
we consider the following auxiliary periodic problem : @xmath291,\\[4pt ] x(-nb)=x(nb ) , x'(-nb)=x'(nb),\ ; 1 < p<\infty \end{array } \right \}.\ ] ] from theorem 2 , we know that problem ( 7 ) has a non - trivial solution @xmath292 .
let @xmath293 be the locally lipschitz energy functional corresponding to problem ( 7 ) , i.e. @xmath294 hereafter by @xmath295 we shall denote the lebesgue space @xmath296 and by @xmath297 the sobolev space @xmath298 .
consider the integral functional @xmath299 defined by @xmath300 . by virtue of hypothesis h(j)@xmath267(vi ) we have that @xmath301 . because @xmath302 is continuous and
@xmath303 is dense in @xmath304 , we can find @xmath305 such that @xmath306 . then recalling that for almost all @xmath55 , all @xmath182 and all @xmath307
, we have @xmath308 we can easily see that @xmath309 & \quad\ , -\int_{-b}^{b}j(t,\lambda\bar{x}(t))\d t \to - \infty \quad \t{as } \
\lambda\to + \infty.\end{aligned}\ ] ] ( recall @xmath310 . ) so we can find @xmath311 such that for all @xmath312 we have @xmath313 . define @xmath314 as follows : @xmath315 0 , & \mbox{if $ t\in t_{n}\setminus t_{1}$ } \end{cases}.\ ] ] then we have @xmath316 for all @xmath317 ( recall that @xmath257 a.e . on @xmath10 ) . from the proof of theorem 1
we know that the solution @xmath318 of problem ( 7 ) is obtained via the non - smooth mountain pass theorem and so it satisfies ( see [ 17 ] ) @xmath319 } \varphi_{n}(\gamma(t))&=\varphi_{n}(x_{n})\geq \inf [ \varphi_{n}(x)\!\!:\|x\|=\rho_{n}]=\xi_{n}>0\\ & \quad \t{and } \
\ 0\in \partial\varphi_{n}(x_{n}),\end{aligned}\ ] ] where @xmath320,w_{n}^{1,p})\!:\gamma(0)=0,\gamma(1)=\lambda\hat{x}\}$ ] with @xmath312 . by continuous extension by constant
, we see that for @xmath321 , we have @xmath322 therefore we have produced a decreasing sequence @xmath323 of critical values . for every @xmath84 we have @xmath324 since @xmath325 , we can find @xmath326 such that @xmath327 .
so we have @xmath328 we take the duality brackets ( for the pair ( @xmath329 ) ) of ( 9 ) with @xmath330 .
we obtain @xmath331 multiply ( 8) with @xmath332 and then add to ( 10 ) .
we obtain @xmath333 using hypothesis h(j)@xmath267(iv ) , for every @xmath84 we have @xmath334 for some @xmath335 independent of @xmath84 . using this
lower bound in ( 11 ) , we obtain @xmath336 with @xmath337 independent of @xmath84 .
so it follows that @xmath338 with @xmath339 independent of @xmath84 .
moreover , as in ( [ 29 ] , p. 36 ) , we can have that @xmath340 with @xmath341 independent of @xmath84 .
we extend by periodicity @xmath342 and @xmath343 to all of @xmath10 .
from(12 ) and since @xmath344 is embedded compactly in @xmath345 , by passing to a subsequence if necessary , we may assume that @xmath346 in @xmath347 , hence @xmath348 .
also because of hypothesis h(j)@xmath267(iii ) , we have @xmath349 with @xmath350 independent of @xmath84 .
so we may assume that @xmath351 evidently @xmath352 and using proposition vii.3.13 , p. 694
, of [ 14 ] , we have @xmath353 a.e . on @xmath354 for all @xmath84 , hence @xmath353 a.e . on @xmath10 ( recall that the multifunction @xmath355 is upper semicontinuous , see [ 5 ] , p. 29 ) .
for every @xmath356 we have that @xmath357 we can find @xmath358 such that for all @xmath359 we have @xmath360\subseteq t_{n_{0}}$ ] and then using ( 13 ) we have @xmath361 \rightarrow & \int_{-\tau}^{\tau}\|x(t)\|^{p}\d t \leq\xi_{3}^{p}.\end{aligned}\ ] ] because @xmath362 was arbitrary it follows that @xmath363 . next let @xmath364 .
then supp @xmath365 $ ] for some @xmath84 . integrating by parts we have @xmath366 ,
hence we have @xmath367 @xmath96 ( see ( 12 ) and recall that @xmath368 ) .
note that @xmath369 uniformly on compact sets ( i.e. the convergence is in @xmath347 ) and @xmath370 a.e .
on @xmath354 ( see ( 13 ) ) . set @xmath371 0 , & \mbox{otherwise } \end{cases}.\ ] ] then @xmath372 and we have @xmath373 a.e . on @xmath10 . by the dominated convergence theorem we have @xmath374 \rightarrow & \left|\int_{\mathbb{r } } ( x(t),\theta'(t))_{\mathbb{r}^{n}}\d t\right|\leq \xi_{3}\|\theta\|_{l^{q}(\mathbb{r},\mathbb{r}^{n})}.\end{aligned}\ ] ] from proposition ix.3 , p. 153
of [ 3 ] , we obtain that @xmath375 . also since @xmath376 in @xmath377 , we have that @xmath378 , while from the fact that @xmath346 in @xmath347 it follows that @xmath379 also from integration by parts we have @xmath380 because @xmath342 is a solution of ( 7 ) , we see that @xmath381 for all @xmath84 . from this
we obtain that @xmath382 for all @xmath84 . also by passing to a subsequence if necessary
, we may assume that @xmath383 in @xmath384 , hence @xmath385 in @xmath386 ( and in @xmath347 too ) .
if @xmath387 is defined by @xmath388 , we have @xmath389 in @xmath390 and so @xmath391 , hence @xmath392 .
therefore we have @xmath393 \ & \quad = -\int_{\mathbb{r}}((\|x'(t)\|^{p-2}x'(t))',\theta'(t))_{\mathbb{r}^{n}}\d t \ \ ( \t{by integration by parts}).\end{aligned}\ ] ] since for all @xmath84 large we have @xmath394 & \quad + \int_{\mathbb{r}}g(t)\|x_{n}(t)\|^{p-2}(x_{n}(t),\theta(t))_{\mathbb{r}^{n}}\d t = \int_{\mathbb{r}}(u(t),\theta(t))_{\mathbb{r}^{n } } \d t\end{aligned}\ ] ] by passing to the limit as @xmath395 and using the convergences established above , we obtain @xmath396 & \quad + \int_{\mathbb{r}}g(t)\|x(t)\|^{p-2}(x(t),\theta(t))_{\mathbb{r}^{n}}\d t = \int_{\mathbb{r}}(u(t),\theta(t))_{\mathbb{r}^{n } } \d t.\end{aligned}\ ] ] because @xmath364 is arbitrary , it follows that @xmath397 and @xmath398 , @xmath399 a.e . on @xmath10 .
next we show that @xmath400 .
recall that from previous arguments we have @xmath401 .
so from corollary vii.8 , p. 130 of [ 3 ] , we have @xmath402 as @xmath403 .
hence we have @xmath404 .
since @xmath405 a.e . on @xmath10 , from hypothesis h(j)@xmath267(iii )
we have @xmath406 a.e . on @xmath10 . because @xmath375 , we have @xmath407 and so @xmath408
. therefore , @xmath409 and once again from p. 130 of [ 3 ] , we have that @xmath410 as @xmath403 , hence @xmath411 as @xmath403 .
therefore @xmath412 and we have proved that @xmath413 is a homoclinic ( to 0 ) solution . it remains to show that @xmath413 is non - trivial . for every @xmath84
, we have @xmath414 set @xmath415 0 , & \mbox{if $ x_{n}(t)=0 $ } \end{cases}.\ ] ] we have @xmath416 by virtue of hypothesis @xmath417 ( see the remark following h(j)@xmath267 ) , given @xmath116 we can find @xmath418 such that for almost all @xmath260 , all @xmath118 and all @xmath65 we have @xmath419 if @xmath420 , then @xmath421 in @xmath347 and so we can find @xmath358 such that for all @xmath359 and all @xmath422 , we have @xmath423 .
therefore for all @xmath359 and almost all @xmath424 , we have @xmath425 ( see ( 14 ) ) and so @xmath426 for all @xmath359 ( recall that @xmath427 were extended by periodicity to all of @xmath10 ) .
let @xmath428 to obtain @xmath429 , a contradiction .
this proves that @xmath153 .
therefore @xmath430 is the desired non - trivial , homoclinic ( to 0 ) solution of the non - smooth non - linear periodic system.qed
in this last part of the paper we study the scalar ( i.e. @xmath44 ) problem .
we approach the problem using a generalized landesmann
lazer type condition , which is more general than the one used by tang [ 33 ] in the context of smooth semilinear periodic equations .
so our work is a two - fold generalization of the work of tang .
* @xmath432 is a functional such that @xmath433 and + .8pc * for all @xmath434 , @xmath62 is measurable ; * for almost all @xmath63 , the function @xmath64 is locally lipschitz ; * for almost all @xmath55 , all @xmath434 and all @xmath65 , we have @xmath435 @xmath67 with @xmath436 , @xmath437 ; * @xmath438 uniformly for almost all @xmath55 and all @xmath439 ; * there exist functions @xmath440 such that @xmath441 and @xmath442 a.e . on @xmath98 and @xmath443 .
we claim that @xmath447 is bounded .
suppose that this is not the case . by passing to a subsequence if necessary
, we may assume that @xmath451 .
let @xmath452 , @xmath84 .
we may assume that @xmath453 ( recall that @xmath454 is embedded compactly in @xmath455 . ) from the choice of the sequence @xmath447 , we have @xmath456 by virtue of hypothesis h(j)@xmath457(v ) we have @xmath458 .
so from ( 16 ) and the weak lower semicontinuity of the norm in a banach space , we obtain @xmath459 , hence @xmath460 .
if @xmath461 , then we have @xmath462 , hence @xmath463 in @xmath454 .
but for every @xmath84 , @xmath464 and so we have a contradiction . therefore @xmath465 and without any loss of generality we may assume that @xmath466 ( the analysis is the same if instead we assume that @xmath467 ) . recall that @xmath468 with @xmath469 .
we have @xmath470 with @xmath471 , @xmath472 , @xmath84 . then @xmath473 with @xmath474 , @xmath475 , @xmath84 . from the choice of the sequence @xmath476
, we have @xmath477 with @xmath478 , hence @xmath479\leq\varepsilon_{n}.\ ] ] since we have assumed that @xmath466 , we have that for all @xmath55 , @xmath480 as @xmath481 . we claim that this convergence is uniform in @xmath55 .
indeed let @xmath482 be such that @xmath483 ( recall that we have assumed @xmath484 ) .
since @xmath485 in @xmath486 , we can find @xmath487 such that for all @xmath488 and all @xmath489 , hence @xmath490 .
because @xmath491 given @xmath492 we can find @xmath493 such that for all @xmath494 we have @xmath495 .
so for all @xmath496 and all @xmath55 we have @xmath497 . because @xmath492 is arbitrary and @xmath498
, we infer that @xmath499 evidently @xmath500 , while by virtue of hypothesis h(j)@xmath457(iv ) , given @xmath116 we can find @xmath501 such that for all @xmath502 and almost all @xmath503 we have @xmath504 .
so for @xmath502 we have @xmath505 by virtue of hypothesis h(j)@xmath457(v ) , given @xmath116 we can find @xmath513 such that for almost all @xmath55 and all @xmath514 , we have @xmath515 . recall that @xmath480 uniformly in @xmath55 ( i.e. @xmath516 .
so we can find @xmath358 such that for all @xmath517 we have @xmath518 using this and ( 18 ) in ( 19 ) we obtain @xmath521 , hence @xmath522 ( recall that we have assumed that @xmath484 ) .
this contradicts hypothesis h(j)@xmath457(v ) .
therefore @xmath476 is bounded and then arguing as in the proof of theorem 1 we conclude that @xmath4 satisfies the non - smooth ps-condition.qed suppose that the result of the proposition is not true .
then we can find @xmath525 such that @xmath526 and @xmath527 such that @xmath528 for all @xmath84 .
so @xmath529 .
first suppose that @xmath530 .
we may assume that @xmath531 for all @xmath84 .
we have @xmath532 propositions 46 permit the application of the non - smooth saddle point theorem .
so we obtain @xmath552 such that @xmath553 , hence @xmath554 with @xmath555 , @xmath405 a.e . on @xmath98 . as in the proof of theorem 1
we show that @xmath550 , @xmath551 and it solves problem ( 15).qed as we have already mentioned in the beginning of this section , our generalized landesmann
lazer type condition ( see hypothesis h(j)@xmath457(v ) ) generalizes the one used by tang [ 33 ] ( for smooth potentials ) . in the next proposition we are going to show this . for this purpose
we introduce the following functions : @xmath556 and @xmath557 the functions @xmath558 , @xmath559 are essentially the ones used by tang [ 33 ] in the context of smooth , semilinear ( i.e. @xmath1 ) periodic problems . in that case , since @xmath560 , we have @xmath561 , and hence @xmath562 .
this proposition shows that our generalized landesman
lazer type condition ( hypothesis h(j)@xmath457(v ) ) is more general than the one used by tang [ 33 ] .
here is an example of a non - smooth locally lipschitz potential which satisfies h(j)@xmath457(v ) but does not satisfy the condition of tang .
again for simplicity we drop the @xmath178-dependence @xmath591 when dealing with the semilinear ( i.e @xmath1 ) case , we can consider problems at resonance in an eigenvalue of any order .
similar problems ( but with smooth potential ) were studied by mawhin and ward [ 25 ] , p. 67 of mawhin and willem [ 26
] , mawhin and schmitt [ 24 ] ( problems near resonance ) and tang [ 33 ] ( who employed his more restrictive version of the generalized landesman lazer condition ( see proposition 8) . * @xmath432 is a functional such that @xmath433 and + .9pc * for all @xmath434 , @xmath62 is measurable ; * for almost all @xmath63 , the function @xmath64 is locally lipschitz ; * for almost all @xmath55 , all @xmath434 and all @xmath65 , we have @xmath599 @xmath67 with @xmath600 , @xmath437 ; * @xmath438 uniformly for almost all @xmath55 and all @xmath439 ; * there exist functions @xmath601 such that @xmath602 and @xmath603 a.e . on @xmath98 and @xmath604 for all @xmath605 .
let @xmath617 and suppose that the result is not true .
we can find @xmath618 such that @xmath619 .
set @xmath620 , @xmath84 .
we may assume that @xmath621 in @xmath606 and @xmath622 in @xmath623 .
thus in the limit as @xmath481 , we obtain @xmath624 and so @xmath625 .
hence @xmath626 and so @xmath463 in @xmath606 , a contradiction to the fact that @xmath464 for all @xmath84 .
qed again we find @xmath633 such that @xmath206 and @xmath634 , with @xmath635 , @xmath210 a.e . on @xmath98 . here
@xmath636 is defined @xmath637 for all @xmath638 .
of course @xmath99 is a maximal monotone , bounded linear operator .
we claim that @xmath639 is bounded .
if this is not the case , we may assume that @xmath451 .
we set @xmath452 , @xmath84 and we may assume that @xmath640 since @xmath116 was arbitrary , we infer that @xmath647 .
also we have @xmath648 .
so from ( 23 ) and if @xmath649 , we obtain @xmath650=0,\\ \rightarrow & \mathop { \lim}\limits_{n \to \infty } \int_{0}^{b}y_{n}'(y_{n}'-y')\d t \ \ ( \t{recall that $ y_{n}\rightarrow y$ in $ c_{\rm per}(t)$ } ) ,
\\ \rightarrow & \;\|y_{n}'\|_{2}\rightarrow \|y'\|_{2}.\end{aligned}\ ] ] since @xmath653 and @xmath654 ( see hypothesis h(j)@xmath655(v ) ) , if in ( 23 ) @xmath656 and we pass to the limit as @xmath481 , we obtain that @xmath657 , hence @xmath658 with @xmath659 and so @xmath660 with @xmath661 , @xmath662 .
we write @xmath663 with @xmath664 , @xmath665 , @xmath666 , @xmath84 .
using @xmath667 as our test function , we obtain @xmath668 where @xmath669 with @xmath670 , @xmath671 , @xmath672 . because @xmath673 , we have @xmath674 and so for all @xmath84 we have @xmath675 .
since @xmath676 for all @xmath677 , taking @xmath678 and exploiting the orthogonality relations , we have @xmath679 also from lemma 3 of [ 33 ] , we have that @xmath680 for all @xmath84 and some @xmath681 while from lemma 9 , we have @xmath682 for all @xmath84 . using the orthogonality relations among the three subspaces @xmath683 , @xmath684 , and @xmath614
, we obtain @xmath685 thus finally we can write that @xmath686 = & \;\lambda_{m}\|\bar{x}_{n}\|^{2}_{2}-\|\bar{x}_{n}'\|^{2}_{2}+\|\hat{x}_{n}'\|_{2}^{2 } -\lambda_{m}\|\hat{x}_{n}\|_{2}^{2 } \nonumber\\[6pt ] \geq & \ ; \beta_{2}\|-\bar{x}_{n}+\hat{x}_{n}\|^{2}\quad \t{for all } \ \ n\geq 1 \ \ \t{and some } \ \ \beta_{2}>0.\end{aligned}\ ] ] moreover , recalling that for almost all @xmath55 , all @xmath644 and all @xmath439 we have @xmath645 and using also hypothesis h(j)@xmath655(iii ) , we have @xmath687 & \quad\,\t{for some } \
\nonumber\\ \ & \leq\frac{\varepsilon}{\|x_{n}\|}\|w_{n}\|^{2}+\frac{1}{\|x_{n}\|}\beta_{3 } \|w_{n}\|\nonumber\\[3pt ] & \,\quad \t{with } \ \ w_{n}=-\bar{x}_{n}+\hat{x}_{n } , n\geq 1.\end{aligned}\ ] ] recall that @xmath697 with @xmath698 .
so @xmath699 with @xmath700 , @xmath701 , @xmath84 .
we have @xmath702 , @xmath84 .
so @xmath703 \rightarrow & \;\frac{\lambda_{m}}{2}\frac{\|w_{n}\|_{2}^{2}}{\|x_{n}\|}= \frac{\lambda_{m}}{2 } \frac{\xi_{n}^{2}b}{\|x_{n}\|}+\frac{\lambda_{m}}{2}\frac{\|v_{n}\|_{2}^{2}}{\|x_{n}\| } \leq\lambda_{m}\frac{\|w_{n}\|^{2}}{\|x_{n}\|}\rightarrow 0.\end{aligned}\ ] ] also we have @xmath704 & \quad ( \t{by the poincar\'e -- wirtinger inequality}),\\[.2pc ] \rightarrow & \;\frac{1}{2}\frac{\|w_{n}'\|_{2}^{2}}{\|x_{n}\| } \leq\frac{\beta_{7}}{2}\frac{\|w_{n}\|^{2}}{\|x_{n}\|}\rightarrow 0 \quad \t{as } \
\ n\rightarrow\infty.\end{aligned}\ ] ] note that on @xmath706 we have @xmath707 and on @xmath708 we have that @xmath709 .
in addition @xmath710 .
so via fatou s lemma , we have @xmath711 \ & \geq\int_{0}^{b}j_{+}(t)y^{+}(t)\d t -\int_{0}^{b}j_{-}(t)y^{-}(t)\d t.\end{aligned}\ ] ] from ( 29 ) and since @xmath715 , we obtain @xmath716 \ & \quad <
\;r \bigg(\int_{0}^{b}j_{+}(t)\sin(m\omega t+\theta)^{+}\d t -\int_{0}^{b}j_{-}(t)\sin(m\omega t+\theta)^{-}\d t \bigg)\nonumber\\[.2pc ] \ & \quad\;\;\;\;\;(\t{by hypothesis } \ \ \hbox{h}(\hbox{j})_{5}(\hbox{v})).\end{aligned}\ ] ] comparing ( 30 ) and ( 31 ) , we reach a contradiction .
this proves that the sequence @xmath630 is bounded , hence we may assume that @xmath717 in @xmath606 and @xmath346 in @xmath455 . as before ( see the proof of theorem 1 ) we can finish the proof and conclude that @xmath4 satisfies the non - smooth ps-condition.qed let @xmath452 , @xmath84 .
we may assume that @xmath621 in @xmath606 and @xmath622 in @xmath455 .
because @xmath728 is finite dimensional and @xmath729 for all @xmath84 , we have @xmath622 in @xmath606 .
for all @xmath84 we have @xmath730 propositions 1012 , permit the use of the non - smooth saddle point theorem which gives @xmath609 such that @xmath553 . as in the proof of theorem 1
, we can check that @xmath550 , @xmath743 and also @xmath413 solves ( 22 ) .
so we can state the following existence theorem .
theorem 13 generalizes theorems 2 and 3 of tang [ 33 ] .
the generalization is two - fold .
on the one hand we assume a more general landesmann lazer type condition ( see hypothesis h(j)@xmath655(v ) and proposition 8) and on the other hand we have a non - smooth potential function .
moreover , in tang the potential function is independent of the time - variable @xmath55 .
mawhin j , periodic solutions of systems with @xmath0-laplacian - like operators , in : nonlinear analysis and applications to differential equations ( lisbon ) ( 1997 ) ; progress in nonlinear differential equations and applications ( boston : birkhauser ) ( 1998 ) pp . | in this paper we study second order non - linear periodic systems driven by the ordinary vector @xmath0-laplacian with a non - smooth , locally lipschitz potential function .
our approach is variational and it is based on the non - smooth critical point theory .
we prove existence and multiplicity results under general growth conditions on the potential function .
then we establish the existence of non - trivial homoclinic ( to zero ) solutions .
our theorem appears to be the first such result ( even for smooth problems ) for systems monitored by the @xmath0-laplacian . in the last section of the paper
we examine the scalar and semilinear problem .
our approach uses a generalized landesman
lazer type condition which generalizes previous ones used in the literature .
also for the semilinear case the problem is at resonance at any eigenvalue .
= mtsyb at 10pt |
the mechanism of exclusive production of mesons at high energies became recently a very active field of research ( see e.g. ref .
@xcite and references therein ) .
central exclusive production processes represent a very promising and novel way to study qcd in hadron - hadron collisions .
recently , there is a growing interest in understanding exclusive three - body reactions @xmath6 at high energies , where the meson ( resonance ) @xmath7 is produced in the central rapidity region . in particular , these reactions provide a valuable tool to investigate in detail the properties of resonance states .
many of the resonances decay into @xmath8 and/or @xmath9 channels .
the representative examples are : @xmath10 and , as we shall emphasis here , @xmath2 .
various decay channels can be studied .
it is clear that these resonances are seen ( or will be seen ) `` on '' the background of a @xmath8 or @xmath9 continuum .
production ( see e.g. ref .
@xcite ) . ] the two - pion background to exclusive production of @xmath11 meson was already discussed in ref .
the recent works concentrated on the production of @xmath12 mesons ( see e.g. refs .
@xcite and references therein ) where the qcd mechanism is similar to the exclusive production of the higgs boson .
furthermore , the @xmath13 states are expected to annihilate via two - gluon processes into light mesons and may , therefore , allow the study of glueball production dynamics @xcite . recently , the cdf collaboration has measured the cross section for exclusive production of @xmath12 mesons in proton - antiproton collisions at the tevatron @xcite , by selecting events with large rapidity gaps separating the centrally produced state from the dissociation products of the incoming protons . in this experiment @xmath12
mesons are identified via decay to the @xmath14 with @xmath15 channel .
the experimental invariant mass resolution was not sufficient to distinguish between scalar , axial and tensor @xmath12 . while the branching fractions to this channel for axial and tensor mesons are large @xcite ( @xmath16 and @xmath17 , respectively ) the branching fraction for the scalar meson is very small @xmath18 @xcite . on the other hand ,
the cross section for exclusive @xmath2 production obtained within the @xmath3-factorization is much bigger than that for @xmath19 and @xmath20 . as a consequence ,
all @xmath21 mesons give similar contributions @xcite to the @xmath14 decay channel .
clearly , the measurement via decay to the @xmath14 channel can not provide cross section for different @xmath12 .
could other decay channels be used ?
the @xmath2 meson decays into several two - body channels ( e.g. @xmath22 , @xmath23 , @xmath24 ) or four - body hadronic modes ( e.g. @xmath25 , @xmath26 ) .
the branching ratios are shown in table [ tab : hadronic_modes ] . in this paper
we analyze a possibility to measure @xmath2 via its decay to @xmath27 channel .
the advantage of this channel is that the @xmath27 continuum has been studied recently @xcite and is relatively well known .
in addition the axial @xmath19 does not decay to the @xmath22 channel and the branching ratio for the @xmath20 decay into two pions is smaller .
a much smaller cross section for @xmath20 production means , in practice , that only @xmath2 will contribute to the signal . .branching fractions for the @xmath28 two- and four - body hadronic decays , taken from ref .
@xcite . [ cols="^,^,^,^",options="header " , ]
the main experimental task is to measure the distributions in the @xmath2 rapidity and transverse momentum .
can one recover such distributions based on the measured ones in spite of the severe cuts on pion kinematical variables ? in fig .
[ fig : ratio ] we show the two - dimensional ratio of the cross sections for the @xmath2 meson in its rapidity and transverse momentum : @xmath29 the numerator includes limitations on @xmath30 and @xmath31 .
these distributions provide a fairly precise evaluation of the expected acceptances when experimental cuts are imposed .
the experimental data could be corrected by our two - dimensional acceptance function to recover the distributions of interest .
it was realized recently that the measurement of exclusive production of @xmath21 via decay in the @xmath14 channel can not give production cross sections for different species of @xmath21 . in this decay channel
the contributions of @xmath21 mesons with different spins are similar and experimental resolution is not sufficient to distinguish them . in the present paper
we have analyzed a possibility to measure the exclusive production of @xmath2 meson in the proton-(anti)proton collisions at the lhc , tevatron and rhic via @xmath32 decay channel .
since the cross section for exclusive @xmath2 production is much larger than that for @xmath19 and @xmath20 and the branching fraction to the @xmath22 channel for @xmath2 is larger than that for @xmath20 ( @xmath19 does not decay into two pions ) the two - pion channel should provide an useful information about the @xmath2 exclusive production . we have performed detailed studies of several differential distributions and demonstrated how to impose extra cuts in order to improve the signal - to - background ratio .
the two - pion background was calculated in a simple model with parameters adjusted to low energy data .
we have shown that relevant measurements at tevatron and lhc are possible . at rhic
the signal - to - background ratio is much worse but measurements should be possible as well . imposing cuts distorts the original distributions for @xmath2 in rapidity and transverse momentum .
we have demonstrated how to recover the original distributions and presented the correction functions for some typical experimental situations . in the present paper
we have concentrated on @xmath33 channel only .
this is mainly due to the relatively good control of the background .
measurements of other decay channels , e.g. @xmath23 , are possible as well and will be discussed elsewhere .
* acknowledgments * this study was partially supported by the polish grant of mnisw n n202 249235 .
r. barbieri , m. caffo , r. gatto and e. remiddi , nucl .
b192 * ( 1981 ) 61 ; w. kwong , p.b .
mackenzie , r. rosenfeld and j.l .
rosner , phys .
* d37 * ( 1988 ) 3210 ; m.l . mangano and a. patrelli ,
b352 * ( 1995 ) 445 . | we consider a measurement of exclusive production of scalar @xmath0 meson in the proton - proton collisions at lhc and rhic and in the proton - antiproton collisions at the tevatron via @xmath1 decay .
the corresponding amplitude for exclusive double - diffractive @xmath2 meson production was obtained within the @xmath3-factorization approach including virtualities of active gluons and the corresponding cross section is calculated with unintegrated gluon distribution functions ( ugdfs ) known from the literature .
the four - body @xmath4 reaction constitutes an irreducible background to the exclusive @xmath2 meson production .
we calculate several differential distributions for @xmath5 process including absorptive corrections . the influence of kinematical cuts on the signal - to - background ratio is investigated .
corresponding experimental consequences are discussed . |
the shrinking size of spintronics devices and the need for efficient and scalable methods for their manipulation has led to enhanced interest in spin - torque effects of electrical current on the magnetic configuration of nanostructures .
the study in this direction has focused so far on effects related to magnetic nonuniformity ; in particular , the effect of current on ferromagnetic domain walls which yields domain wall motion @xcite and the effect of current on magnetic heterostructures which yields magnetic switching @xcite .
both effects have been observed in different systems and they appear to be useful for novel memory devices .
a more subtle effect of magnetic nonuniformity is expected when a single uniformly magnetized nanostructure is connected via asymmetric contacts to normal metals . in this case ,
asymmetric spin accumulation near the two contacts may induce magnetic instability @xcite . based on several theoretical works ,
electrical current is expected to induce magnetization reversal also in a uniformly magnetized ferromagnet irrespective of the contact geometry @xcite due to a current - induced spin wave instability .
this is a fundamental phenomenon with practical implications on the functionality of spintronic devices ; hence , its experimental study is of particular importance for the field of spintronics .
nevertheless , so far the study of this phenomenon has been very sparse @xcite .
current - induced magnetization reversal in uniformly magnetized films is expected at relatively high current densities .
such currents generate heat and high oersted - fields which may also induce magnetization reversal .
therefore , reliable identification of the phenomenon requires the ability to isolate the effect of the current itself .
here we use the itinerant ferromagnet srruo@xmath0 @xcite , a material characterized by large spin polarization of its conducting electrons @xcite .
this material exhibits large domain wall resistivity @xcite and efficient current - induced domain wall motion @xcite indicating strong effect of the current on the magnetic configuration and vice versa . in this work ,
we observe magnetization reversal induced by current above a temperature dependent threshold and show that this effect is not simply due to sample heating or oersted fields .
our observations are obtained for a wide range of temperatures while applying various fields including a zero field and fields suppressing reversal .
basic features of the results are consistent with the spin wave instability scenario @xcite suggesting that this is the relevant mechanism .
inset : the antisymmetric transverse resistance vs. temperature with reversing fields between 1 koe and 3 koe .
the sharp jumps are attributed to magnetization reversal . ]
our samples are high - quality epitaxial films of srruo@xmath0 which were grown on slightly miscut ( 2@xmath1 ) srtio@xmath0 substrates .
the films are orthorhombic ( @xmath2 , @xmath3 , @xmath4 ) with the @xmath5 axis in the film plane ( perpendicular to the miscut direction ) and the @xmath6 and @xmath7 axes are at 45@xmath1 relative to the film plane .
the curie temperature of the films is @xmath8 k and they exhibit uniaxial magnetocrystalline anisotropy with the easy axis changing in the ( 001 ) plane between the @xmath7 axis at @xmath9@xmath10 to @xmath11 from the film normal at low temperatures @xcite . the ratio between the resistivity at 300 k and the resistivity at the low temperature limit is greater than ten .
the films are patterned for magnetotransport measurements using e - beam lithography and @xmath12 ion milling with a typical current path width of 1.5 @xmath13 m .
the data presented below are for film thickness of 37.5 nm .
the magnetization of the patterns is monitored by measuring the hall effect ( he ) which consists ( as in other magnetic conductors ) of an ordinary hall effect ( ohe ) determined by the perpendicular component of the magnetic field and an anomalous hall effect ( ahe ) related to the perpendicular component of the magnetization . since the easy axis for magnetization in srruo@xmath14
is tilted out of the plane there is a contribution of the ahe in the absence of an external field @xcite .
as we study the contribution of current to magnetization reversal with and without an applied magnetic field , we first determine the temperature dependence of the reversal field , @xmath15 , in the absence of current effects @xcite . in these measurements
we monitor the magnetization with a small current of 30 @xmath13a ( current density @xmath16 ) which is too small to contribute to the magnetization reversal ( see figure [ 1 ] ) .
we fully magnetize the sample at a low temperature and then apply a certain reversing field and warm the sample at a rate of 20 k / min until reversal occurs ( see inset of figure [ 1 ] ) . in this experiment and in the experiments described below the magnetic field
is applied perpendicular to the film plane . to explore current contribution to magnetization reversal , we fully magnetize the sample at an initial temperature , @xmath17 , between 28 and 145 k and apply a current pulse of constant duration . the amplitude of the pulse is gradually increased and the magnetization is monitored after each pulse by measuring the ahe with a probing current of 30 @xmath13a .
we repeat these experiments with various magnetic fields , both assisting and suppressing magnetization reversal . as we show below
, our main observation is that when the amplitude of the pulse exceeds a certain value the pattern is no longer fully magnetized . ) vs. pulse duration .
the pulse amplitude varies between 2.57 and 5.8 ma , and the initial temperature varies between 100 and 30 k , respectively .
@xmath18 is measured during the last 50 @xmath13s of the pulse .
inset : the change in the local temperature vs. the current s power .
the pulse duration is 1 ms .
( c ) the longitudinal resistance during the last 50 @xmath13s of the pulse vs. the pulse amplitude for different initial temperatures .
the total pulse duration is 1 ms . ]
the currents that induce magnetic instability correspond to high current densities ( 1 ma corresponds to current density of @xmath19 ) ; therefore , the effect of heating should be carefully considered . to identify changes in the temperature of the pattern during the pulse , we use the measured temperature dependence of the resistance of the pattern as our calibrated temperature sensor ; see the temperature dependence of the longitudinal resistance , @xmath18 , measured between two neighboring voltage leads ( figure [ 2](a ) ) . figure [ 2](b ) shows changes in @xmath18 during current pulses with various amplitudes .
the duration of the pulse is 2 ms and @xmath18 is measured in 50 @xmath13s intervals .
the temperature of the pattern before the current pulse is applied , @xmath17 , is between 30 k and 100 k and the current amplitude is between 2.57 ma ( for @xmath17=100 k ) to 5.8 ma ( for @xmath17=30 k ) .
we note a fast initial increase in the resistance followed by saturation .
in the current - induced magnetization reversal measurement shown below we use 1 ms pulses with amplitudes which do not exceed 5.84 ma , and we measure @xmath18 during the last 50 @xmath13s of the pulse . in these conditions ,
the measurement reflects well the highest value of @xmath18 during the pulse .
figure [ 2](c ) shows @xmath18 measured during the last 50 @xmath13s of the pulse as a function of the pulse amplitude .
based on the temperature dependence of @xmath18 ( figure [ 2](a ) ) we can determine the effective temperature , @xmath20 , at the end of the current pulse .
as this is practically a steady state , we expect that the power dissipation in the pattern , given by @xmath21 , would be balanced by the heat flow .
the inset of figure [ 2](b ) shows a linear dependence between @xmath22 and @xmath21 suggesting heat flow proportional to @xmath23 with practically temperature - independent thermal conductivity in the relevant temperature regime .
( right ) vs. current amplitude for @xmath17=70 k and a reversing field of @xmath24=100 oe . the dashed lines are guide lines for the magnetization reversal current , @xmath25 , and the magnetization reversal temperature , @xmath26 .
( b ) the magnetization reversal temperature , @xmath27 , at @xmath24=0 and 2000 oe vs. the magnetization reversal current .
the solid lines represent the magnetization reversal temperature in the limit of zero current . ]
figure [ 3](a ) shows a typical measurement of current induced magnetization reversal performed at @xmath17=70 k with a reversing field of @xmath24=100 oe . the amplitude of the current pulse is gradually increased and the figure shows the effective temperature @xmath20 during the pulse and the normalized average magnetization , @xmath28 , as determined after the pulse at the pre - pulse temperature by measuring the ahe with a small probing current and dividing the obtained signal by the signal corresponding to full magnetization ( the sample cools down to its pre - pulse temperature in less than 1 second ) .
based on such measurements we determine the current , @xmath29 , temperature , @xmath27 , and magnetic field , @xmath15 , at which magnetization reversal occurs .
the results are insensitive to current polarity .
figure [ 3](b ) presents @xmath27 ( the magnetization reversal temperature ) as a function of the current pulse amplitude when no magnetic field is applied and when a reversing field of 2000 oe is applied perpendicular to the film plane .
the horizontal lines mark @xmath27 for the two fields in the limit of zero current as determined in temperature sweep experiments ( see figure [ 1 ] ) . to produce the curves we cool the sample to different temperatures and gradually increase the current pulse .
the points represent the current amplitude for which first magnetization reversal occurred and the highest temperature of the sample during the pulse . for @xmath24=2000
oe , @xmath27 is practically independent of @xmath29 for a wide range of current amplitudes , suggesting no current effect beyond the joule heating which we monitor . on the other hand , for @xmath24=0
, @xmath27 significantly decreases with increasing current , indicating non - trivial current effects .
in addition to heating , current may affect the magnetization reversal due to the oersted field it induces @xcite . based on calculations
, we find that the highest field ( at the edges of the current path ) induced by the largest current we use , 5.8 ma , is @xmath30 40 oe normal to the film plane , small compared to @xmath15 at the relevant temperatures .
based on measurements as shown in figure [ 3](a ) , we construct figure [ 4 ] which shows for the vicinity of a given @xmath20 ( within @xmath31 k ) the combinations of currents and fields that produce first reversal or no - reversal .
such plots allow us to explore the nonthermal effect of the current as its heating effect is already taken into consideration .
the dashed line represents @xmath15 in the limit of zero current for the upper bound of @xmath20 .
the solid line marks the threshold current , @xmath32 , above which the current affects the magnetization reversal in a non - trivial way .
we define this current as the current for which @xmath15 is smaller than its zero limit value by 1.5 times the standard deviation of its distribution for lower currents .
the fact that the plot shows multiple values of @xmath15 for a given @xmath33 has several reasons : the spread in @xmath20 , which is related to the finite increments of the current pulse , the stochastic nature of the effect , and weak field dependence of @xmath29 . at @xmath34 k ( figure [ 4](a ) )
we identify a sharp transition from a current independent regime to a field independent regime .
while for currents lower than @xmath32 the field is current independent , for currents higher than @xmath32 the field dependence is very weak and almost the same current induces magnetization reversal for a wide range of fields including fields which oppose reversal . as @xmath20 increases , the transition becomes more gradual .
( @xmath353 k ) between 102 and 120 k. the dashed lines represent @xmath15 in the limit of zero current at the upper bound of the temperature regime , and the solid lines represent the threshold currents , @xmath32 . ] , ( red dots ) and the threshold current , @xmath32 , ( blue squares ) as a function of temperature . ]
figure [ 5 ] shows the temperature dependence of @xmath32 and of @xmath36 defined as the current at which @xmath15 becomes zero ; namely , the current which produces magnetization reversal without external field assistance .
@xmath32 is extracted from figure [ 4 ] ; therefore , @xmath20 has an error bar of @xmath31 k. on the other hand , for @xmath36 the error bar is related to the temperature increment produced by the current increment in our experiments .
we note that ref .
@xcite suggests a threshold current inversely proportional to the ahe coefficient . in srruo@xmath14
the ahe vanishes at @xmath37 k ; however , we see no divergence of the reversing currents in the vicinity of that temperature .
the existence of a threshold current appears to be consistent with the spin wave scenarios @xcite .
following ref .
@xcite , the threshold current density , @xmath38 , in the zero - temperature limit is given by @xmath39 , where @xmath40 , and @xmath41 are the gyromagnetic ratio , electron charge and bohr magneton respectively , @xmath42 and @xmath43 are the domain wall width and the polarization of the current , @xmath44 and @xmath45 are the saturation magnetization , the anisotropy field and the external field , respectively . since the relevant external fields which we apply are negligible relative to the large anisotropy field of srruo@xmath0 ( @xmath46 t ) @xcite , we expect a weak field dependence of @xmath47 , as observed . substituting the parameters of srruo@xmath14 yields at zero temperature current density @xmath48 and a critical wave length of the spin wave excitation of about @xmath304.2 nm , while our measurements at @xmath20 = 102 k yield @xmath49 .
at least some of the discrepancy may be attributed to the fact that the model is developed at zero temperature while our measurements are at @xmath50 .
we note that for co the expected @xmath47 is more than one order of magnitude higher @xcite with a critical wave length of about @xmath3030 nm @xcite .
this comparison highlights the intrinsic advantage of srruo@xmath14 in studying current induced magnetization reversal . in conclusion , we present strong experimental evidence for a non - trivial current contribution to magnetization reversal and we identify features consistent with magnetic instability induced by spin - wave excitations .
we acknowledge useful discussions with a. kent and a. mitra . l.k . acknowledges support by the israel science foundation founded by the israel academy of sciences and humanities .
j.w.r . grew the samples at stanford university in the laboratory of m. r. beasley .
9 l. berger , j. appl .
phys . * 55 * , 1954 ( 1984 ) .
g. tatara , and h. kohno , phys .
lett . * 92 * , 086601 ( 2004 ) .
s. zhang , and z. li , phys .
rev . lett . * 93 * , 127204 ( 2004 ) .
a. yamaguchi , t. ono , s. nasu , k. miyake , k. mibu , and t. shinjo , phys .
lett . * 92 * , 077205 ( 2004 ) .
m. feigenson , j. w. reiner , and l. klein , phys .
lett . * 98 * , 247204 ( 2007 ) .
o. boulle , j. kimling , p. warnicke , m. klui , u. rdiger , g. malinowski , h. j. m. swagten , b. koopmans , c. ulysse and g. faini , phys .
* 101 * , 216601 ( 2008 ) .
l. berger , phys .
b * 54 * , 9353 ( 1996 ) . j. c. slonczewski , j. magn .
magn . mater . *
159 * , l1 ( 1996 ) .
m. tsoi , a. g. m. jansen , j. bass , w .- c .
chiang , m. seck , v. tsoi , and p. wyder , phys .
80 * , 4281 ( 1998 ) .
e. b. myers , d. c. ralph , j. a. katine , r. n. louie , and r. a. buhrman , science * 285 * , 867 ( 1999 )
. j. a. katine , f. j. albert , r. a. buhrman , e. b. myers , and d. c. ralph , phys .
84 * , 3149 ( 2000 ) .
y. ji , c. l. chien , and m. d. stiles , phys .
lett . * 90 * , 106601 ( 2003 ) .
b. zyilmaz , a. d. kent , j. z. sun , m. j. rooks , and r. h. koch , phys .
* 93 * , 176604 ( 2004 ) .
m. l. polianski , and p. w. brouwer , phys . rev . lett . * 92 * , 026602 ( 2004 ) .
m. d. stiles , j. xiao , and a. zangwill , phys .
b * 69 * , 054408 ( 2004 ) .
j. c. slonczewski , j. magn .
magn . mater . *
195 * , l261 ( 1999 ) .
z. li , j. he , and s. zhang , j. appl . phys .
* 97 * , 10c703 ( 2005 ) .
j. shibata , g. tatara , and h. kohno , phys .
lett . * 94 * , 076601 ( 2005 ) .
korenblit , phys .
b * 77 * , 100404(r ) ( 2008 ) .
y. togawa , t. kimura , k. harada , t. matsuda , a. tonomura , t. akashic , and y. otanib , appl .
. lett . * 92 * , 012505 ( 2008 ) .
m. feigenson , j. w. reiner , and l. klein , j. appl .
* 103 * , 07e741 ( 2008 ) . o. riss , a. gerber , i. ya .
korenblit , m. karpovsky , s. hacohen - gourgy , a. tsukernik , j. tuaillon - combes , p. mlinon , and a. perez , phys .
b * 82 * , 144417 ( 2010 ) .
g. koster , l. klein , w. siemons , g. rijnders , j. s. dodge , c. b. eom , d. h. a. blank , and malcolm r. beasley , rev .
phys . * 84 * , 253 ( 2012 ) .
p. raychaudhuri , a. p. mackenzie , j. w. reiner , and m. r. beasley , phys . rev .
b * 67 * , 020411(r ) ( 2003 ) .
l. klein , y. kats , a. f. marshall , j.w .
reiner , t. h. geballe , m. r. beasley , and a. kapitulnik , phys .
rev . lett . * 84 * 6090 ( 2000 ) .
a. f. marshall , l. klein , j. s. dodge , c. h. ahn , j. w. reiner , l. mieville , l. antognazza , a. kapitulnik , t. h. geballe , and m. r. beasley , j. appl .
phys . * 85 * , 4131 ( 1999 ) .
z. fang , n. nagaosa , k. s. takahashi , a. asamitsu , r. mathieu , t. ogasawara , h. yamada , m. kawasaki , y. tokura , and k. terakura , science * 302 * , 92 ( 2003 ) . i. genish , y. kats , l. klein , j. w. reiner , and m. r. beasley , phys .
( c ) * 1 * , 3440 ( 2004 ) .
o. boulle , l. heyne , j. rhensius , m. klui , u. rdiger , l. joly , l. le guyader , f. nolting , l. j. heyderman , g. malinowski , h. j. m. swagten , b. koopmans , c. ulysse , and g. faini , j. appl . phys . * 105 * , 07c106 ( 2009 ) .
m. c. langner , c. l. s. kantner , y. h. chu , l. m. martin , p. yu , j. seidel , r. ramesh , and j. orenstein , phys .
* 102 * , 177601 ( 2009 ) .
y. kats , i. genish , l. klein , j. w. reiner , and m. r. beasley , phys .
b * 71 * , 100403(r ) ( 2005 ) .
z. li , h. he , and s. zhang , j. appl . phys . * 99 * , 08q702 ( 2006 ) . | we inject current pulses into uniformly magnetized patterns of thin films of the itinerant ferromagnet srruo@xmath0 , while monitoring the effective temperature of the patterns during the current injection .
we gradually increase the amplitude of the pulses until magnetization reversal occurs .
we observe magnetization reversal induced by current above a temperature dependent threshold and show that this effect is not simply due to sample heating or oersted fields .
we discuss the applicability of current - induced spin - wave instability scenario . |
superradiance @xcite is a coherent and collective radiation from a multi - atom system .
the radiation intensity is proportional to the square of particle number along with a shorter time variation , which conserves radiation energy .
this collective radiation originates from a common light - matter interaction , through which spontaneously emitted photons can be reabsorbed by the atoms @xcite if they are close to each other , and interatomic phase correlations build up @xcite .
this spontaneous emission decays in an enhanced rate proportional to the number of particles , which is very different from spontaneous emissions of independent and uncorrelated single atoms @xcite .
many theoretical works from different perspectives have investigated the superradiant emission .
the microscopic mechanism of superradiance is shown to be related to induced dipole - dipole interactions @xcite between two atoms , in which spontaneous decay rate and frequency shift @xcite depend on their spatial separations .
the canonical review articles on the frequency shift including cooperative lamb shift ( cls ) and lorentz - lorenz shift from a local field correction in the extended medium of a slab or a sphere can be found in the references of @xcite .
a master equation is formulated @xcite to study the statistical property of superradiance @xcite , and a diagonalization of coupling matrix @xcite reveals the essential information of superradiance decay constants .
one of the characteristics in superradiance is its directionality of emission @xcite , and there is also a semiclassical treatment incorporating propagation effect , which indicates a threshold condition of cooperative emission in an extended medium @xcite . an alternative approach of quantum trajectory method @xcite for superradiance tries to unravel the physics of successive sub- and superradiant photons , and positive - p phase - space method simulations show an enhanced decay in the second order correlation function by including quantum fluctuations in a diamond - type atomic system @xcite .
the preparation of single - photon absorption in the atomic ensemble followed by a collective @xcite and directional emission @xcite raises the interests of single - photon superradiance .
this singly - inverted system reduces dicke s full eigenstates of spin-@xmath0 system into only n state bases , and shows dynamical evolutions @xcite from a specified initial state .
the initially symmetrical excited state for two - level atoms radiates superradiantly while other less symmetrical excited states radiate in a slower rate with smaller probabilities @xcite .
significant collective lamb shift in a spherical geometry is also calculated in such singly - inverted system @xcite . to correctly account for the cooperative effects in a dense medium
, it is crucial to include the counter rotating - wave - approximation ( rwa ) terms @xcite rather than excluding them as incomplete treatments @xcite .
recently some experiments demonstrate superradiance or cls in a variety of atomic systems , including atoms in a planar cavity @xcite , atomic vapor layers @xcite , cold atoms @xcite , and cascade atomic systems @xcite .
the cascade atomic system provides a source for telecommunication bandwidth in its upper transition @xcite .
this correlated photon pair can realize long - distance quantum communication @xcite enabling a low - loss quantum repeater @xcite in the dlcz ( duan - lukin - cirac - zoller ) protocols @xcite . in this paper
we investigate the superradiance of correlated photon pair from a diamond - type atomic ensemble driven by two - color classical fields , and calculate the cls in an optically - thick cylindrical atomic system .
we solve for schrdinger equations for such atomic system , and derive two - photon state function in section 2 .
single atomic excitation adiabatically follows two excitation fields , and subsequently decays through cascade transitions via four - wave mixing ( fwm ) .
counter - rwa terms in the hamiltonian are required to correctly deduce the cls which is a cumulative dipole - dipole interaction energy .
in section 3 we calculate and study the cls of two - photon state in a cylindrical geometry , a conventional setup in cold atom experiments .
we conclude and summarize in section 4 , and the inclusion of virtual photons exchange and adiabatic approximation of excitation process are detailed in appendices a and b respectively .
we consider an ensemble of n four - level atoms excited by two classical fields , and subsequently the signal and idler photons are spontaneously emitted as shown in fig .
[ fig1 ] . without loss of generality , we assume that these identical atoms distribute randomly in a cylindrical geometry with a uniform density as in conventional cold atom experiments @xcite .
we use dipole approximation of light - matter interaction and rotating wave approximation ( rwa ) , and the hamiltonian in interaction picture reads @xmath1 where h.c . denotes hermitian conjugate .
note that the rwa hamiltonian is valid for deriving state functions to calculate transition probabilities while non - rwa or counter - rwa terms are required to account for a complete derivation of cls ( see in appendix a ) .
the collective dipole operators are defined as @xmath2 where single photon detuning @xmath3 @xmath4 @xmath5 @xmath6 @xmath7 , two - photon detuning @xmath8 @xmath4 @xmath5 @xmath9 @xmath10 @xmath6 @xmath11 , and transition frequency @xmath12 @xmath4 @xmath11 @xmath6 @xmath13 . driving rabi frequencies are @xmath14 @xmath15 @xmath16 , @xmath17 @xmath15 @xmath18 , and coupling coefficients are @xmath19 @xmath15 @xmath20 , @xmath21 @xmath15 @xmath22 . the double matrix element of the dipole moment @xmath23 is independent of the hyperfine structure , and @xmath24 @xmath4 @xmath25 where @xmath26 is the quantization volume . polarizations of signal and idler fields are @xmath27 , and the unit directions of dipole operators are @xmath28 respectively . ) and spontaneously emitted photon pair ( @xmath29 ) .
single and two - photon detunings are @xmath30 .
schematic collective couplings between n state bases are denoted by arrows , and slightly shifted energy levels ( not to scale ) are due to light - induced dipole - dipole interaction energy.,width=377,height=207 ] in the limit of large detuned and weak driving fields , @xmath3 @xmath31 @xmath32 , we consider only single excitation , and neglect spontaneous decay during excitations ( @xmath3 @xmath31 @xmath33 , @xmath8 @xmath31 @xmath34 ) .
we may express the state function as @xmath35 where the indices @xmath36 , @xmath37 for light fields , @xmath38 @xmath15 @xmath39 for atomic levels @xmath40 , and @xmath41 is the vacuum photon state .
these bare states make a complete basis of collective excitations for identical particles , which describe a complete cycle of a single atom following the excitation and spontaneously emitted fields .
we apply schrdinger equation @xmath42 @xmath4 @xmath43 , and the coupled equations of motion are @xmath44 in the next subsection , we proceed to solve for two - photon state function from the above equation of motion in the bare state basis .
firstly we solve for signal photon state function by substituting eq .
( [ eqn5 ] ) into eq .
( [ eqn4 ] ) , and the time derivative of @xmath45 becomes @xmath46 we use weisskopf - wigner approach @xcite which assumes a weak - coupling limit and implicitly a markov approximation @xmath13
@xmath31 @xmath47 with an intrinsic relaxation time set by @xmath48 ( see below ) .
in addition the secular approximation suggests an ensemble size ( @xmath49 ) limit of @xmath50 @xmath51 @xmath48 for the approach to be valid @xcite .
the summation of field modes in the above leads to the frequency shift and decay rate which are @xmath52,\nonumber\\ & = \sum_\nu c^\nu_{s}(t)\left[\frac{\gamma_3}{2}f_{\mu\nu}(\xi)-i\omega_{\mu\nu}^-(\xi)\right],\label{eqn_f}\end{aligned}\ ] ] where @xmath53 , and @xmath54 .
@xmath55 involves the lamb shift ( @xmath56 ) and collective radiation shift from dipole - dipole interaction ( @xmath57 ) which we describe in details in appendix a. the other frequency shift of @xmath58 can be derived by including non - rwa terms as shown in appendix a. therefore the differential equation of probability amplitude of @xmath45 becomes @xmath59,\label{eqnc}\end{aligned}\ ] ] where @xmath60 , and @xmath61 is lamb shift .
@xmath62 and @xmath63 represent the collective contributions to spontaneous decay rate and frequency shift , which are induced from the couplings between n state bases as shown schematically in fig .
[ fig1 ] . the differential equation of eq .
( [ eqnc ] ) can be expressed in a matrix form as an eigenvalue problem , and we can solve for @xmath45 by a similarity transformations @xmath64 @xmath4 @xmath65 comprised of eigenvectors @xmath66 , @xmath67 where @xmath68 and @xmath66 are eigenvalues and eigenvectors respectively of n@xmath69n matrix @xmath70 , @xmath71 note that we have absorbed lamb shift into the optical frequency @xmath13 , and the properties @xmath72 @xmath4 @xmath73 , @xmath74 @xmath4 @xmath75 have been used .
the diagonal element demonstrates a single atomic spontaneous decay , and off - diagonal elements represent the couplings between atoms .
the cross couplings are in similar order of magnitude indicating highly dynamical interactions in the system .
we then substitute eq .
( [ signal_c ] ) into eq .
( [ eqn5 ] ) , and derive the two - photon state function in the bare state basis @xmath76 in the adiabatic approximation of the excitation process described in appendix b , we may derive @xmath77 , and the single and two - photon probability amplitudes become @xmath78 where the time evolution of two - photon state incorporates two - time integration of pulse shapes @xmath79 @xmath4 @xmath80 which results from adiabatic driving process ( see appendix b ) .
the cascade two - photon state is expressed of discrete sum of n eigenvalues @xmath81 shown in the time evolution and sum of eigenvectors sandwiched by phase factors induced by four fields interacting with the system . in the above treatment where we use the bare state basis ,
the signal spontaneous emission comes from the adiabatic transfer of detuned two - color driving fields , and the idler spontaneous emission comes from n intermediate excited states .
these solutions of probability amplitudes demonstrate a dynamical coupling between n atoms , and two - photon state is generated from all possible excitations of one of the atomic ensemble . in the next section , we rotate the coupling matrix @xmath70 to the new basis , and we find the superradiant idler photon with its decay constant coming from one of the eigenvalues .
the superradiance ( @xmath81 @xmath82 @xmath47 ) and subradiance ( @xmath81 @xmath83 @xmath47 ) are embedded in these eigenvalues @xmath81 which depend on the density and geometry of atomic ensemble . for a low density where @xmath84 @xmath31 @xmath85 , @xmath72 @xmath86 @xmath87 , @xmath74 @xmath86 @xmath87 , and the coupling matrix @xmath70 becomes an identity matrix times an overall constant @xmath88 .
the transformation matrix @xmath64 is then also an identity matrix .
therefore the two - photon state from n diamond - type atoms behaves as no difference from a single atom . on the contrary in dicke s
limit where the dimensions of ensemble are smaller than radiation wavelength , @xmath72 @xmath86 @xmath89 , and @xmath74 becomes divergent when @xmath84 @xmath90 @xmath87 . without considering this divergent cls ,
one of the eigenvalues becomes @xmath91 @xmath4 @xmath92 proportional to the number of particles , reminiscent of dicke s state @xmath93 @xcite where @xmath94 is the half difference of atomic populations in the excited and ground states . in the next section ,
we construct the basis in terms of the symmetrical and n-1 unsymmetrical states . in this basis
we are able to investigate the superradiance and cls of the two - photon state . here
we investigate the collective radiation decay and energy shift of two - photon state in the basis constructed from the phased symmetrical state @xcite which provides a preference for the system in the limit of large number of atoms .
consider a unitary transformation @xmath95 where a phase factor in an extended ensemble is introduced into basis @xcite , @xmath96\delta_{l\neq n},\end{aligned}\ ] ] where @xmath97 @xmath4 @xmath98 @xmath9 @xmath99 @xmath6 @xmath100 , and the extra phase factor introduced above is from the observation in eq .
( [ two_bare ] ) .
the phased symmetrical state in the new basis is @xmath101 unsymmetrical states are @xmath102 where @xmath103 , and they are orthogonal to each other and normalized to one . in this new rotational basis ,
the signal state function has the identity @xmath104 where these coefficients are related by the unitary transformation matrix that @xmath105 and @xmath106 . a new set of equation of motion in this basis can be derived , and we have for the equation of @xmath107 , @xmath108 where we have used the property of @xmath109 @xmath4 @xmath110 , and @xmath70 is the coupling matrix in bare state basis .
the coupling matrix in this new rotational basis becomes ( let @xmath111 @xmath15 @xmath112e^{-i\bar{\mathbf{k}}\cdot(\mathbf{r}_\mu-\mathbf{r}_\nu)}$ ] ) @xmath113 where the properties @xmath114 @xmath4 @xmath115 , @xmath116 @xmath4 @xmath117 , @xmath118 are used , and @xmath119 means transpose . the diagonal element @xmath120 has the order of @xmath89 , which indicates a spontaneously decayed single atom , and @xmath121 involves the superradiant decay rate and cls of the symmetrical state . similarly we may diagonalize the above matrix with a similarity transformation matrix @xmath64 , and the solutions for probability amplitudes of single and two - photon states become @xmath122 @xmath123 where @xmath124 @xmath15 @xmath98 @xmath9 @xmath99 @xmath6 @xmath100 @xmath6 @xmath125 indicates fwm mismatch .
the state functions show a specific dependence on @xmath126th column of @xmath127 which signifies the difference of the new rotational basis from bare state one .
one way to simplify the solutions is to examine the matrix elements of @xmath128 . in the limit of large number of atoms ,
@xmath129 and @xmath130 with @xmath131 are negligible in the order of @xmath132 compared to @xmath121 .
it can be shown to use a gaussian density distribution in continuous limit @xcite with approximation of @xmath133 where @xmath134 .
the order of magnitude specifies the coupling strength between the phased symmetrical state and the other @xmath135 less symmetrical ones . one can show the cross - coupling elements of @xmath135 non - symmetrical states are not as negligible as @xmath129 or @xmath130 .
it is estimated as less than @xmath136 compared to @xmath121 , but it does not come into play with our two - photon state due to the weak coupling constants of @xmath129 and @xmath130 .
to lowest order of simplification of matrix @xmath128 , we legitimately set @xmath126th row and column of @xmath128 to zero except for the element @xmath121 and we find that @xmath137 @xmath4 @xmath138 and @xmath139 @xmath4 @xmath140 which can be regarded as decoupling from this symmetrical state @xmath141 .
the two - photon state therefore becomes @xmath142 the above is one of the central results in this paper .
the condition of fwm is embedded in the summation of four - wave phase factor @xmath143 which is maximized when @xmath124 @xmath4 @xmath87 .
the cascade emission has the most significant contribution of the phased symmetrical state where @xmath144 represents the timescale of spontaneous decay and cls for the idler transition .
note that the cooperative effect for this correlated photon pair appears only on the idler transition , which is due to the large - detuned fwm driving process that the atomic system is adiabatically prepared to the upper singly - excited state .
in deriving this two - photon state function with induced dipole - dipole interactions in the idler transition , the superradiant decay rate and cls of the idler photon have no difference from the emission of a phased symmetric state in a two - level atomic system .
s dependence on height @xmath145 and radius @xmath146 in a cylindrical geometry .
the enhanced superradiance factor has a stronger dependence on the height , which can be seen in the sectional plot of ( b ) while it saturates as radius increases as shown in ( c ) .
the legends in ( b , c ) denote the values of @xmath147 and @xmath148 respectively .
d1 transition of the idler photon is chosen for rb atoms , and the atomic density is @xmath149 @xmath150.,width=453,height=207 ] the real part of @xmath144 can be expressed in terms of geometrical constant @xcite of a cylindrical ensemble , @xmath151,\nonumber\\ & = -\frac{\gamma_3}{2}\bigg[\frac{1}{n}\sum_{\mu,\nu}f_{\mu\nu}(k_3r_{\mu\nu})e^{-i\bar{\mathbf{k}}\cdot(\mathbf{r}_\mu-\mathbf{r}_\nu)}\bigg],\nonumber\\ & = -\frac{\gamma_3}{2 } n\int d\omega_3\frac{3}{8\pi}\left[1-(\hat{k}_3\cdot\hat{d})^2\right]\gamma(\bar{\mathbf{k}},\mathbf{k}_3)=-\frac{\gamma_3}{2}(n\bar{\mu}+1),\label{eqn_lambda}\end{aligned}\ ] ] where in the second line we use @xmath152 @xmath4 @xmath89 .
@xmath153 is the solid angle of idler photon with wavevector @xmath154 , and @xmath155 @xmath15 @xmath156 @xmath157 which is the ensemble average with randomly distributed atomic positions @xmath158 and @xmath159 . in deriving eq .
( [ eqn_lambda ] ) we first substitute @xmath160 for @xmath161e^{i\mathbf{k}_3\cdot(\mathbf{r}_\mu-\mathbf{r}_\nu)}$ ] @xcite as in eq .
( [ eqn_f ] ) where the property of polarizations @xmath162
@xmath4 @xmath163 is used with @xmath164 in a dyadic form @xcite .
then we go to the continuous limit by replacing @xmath165 with @xmath166 , and relate to @xmath167 of a cylindrical geometry @xcite .
we note that the excitation ( absorbing two - color classical fields with an emitting signal photon ) propagates along the height of the cylinder , which is @xmath97 .
the geometrical constant @xmath167 for a cylindrical ensemble ( of height @xmath145 and radius @xmath146 ) therefore is @xcite @xmath168j^2_1\left[a\sqrt{1-x^2}\right],\end{aligned}\ ] ] where @xmath169 @xmath4 @xmath170 and @xmath171 @xmath4 @xmath172 are dimensionless length scales , and circular polarizations are considered @xcite .
@xmath173 is the bessel function of the first kind . as shown in fig.([fig2 ] ) , we demonstrate how superradiance decay depends on length and radius of a cylindrical ensemble .
the real part retrieves dicke s result when atoms are confined within the radiating wavelength and @xmath174 .
when the average separation of the ensemble is much larger than emission wavelength , dipole - dipole interaction becomes negligible , and we have single atomic spontaneous emission @xmath175
. the frequency shift of this symmetrical state can be significant if atoms are close to each other .
it is a cumulative contribution from dipole - dipole interactions which are oscillatory in space as shown in appendix a.
in this section , we calculate the cooperative lamb shift of phased symmetrical state , which is @xmath176 where p.v . denotes principal value , and we note that geometrical constant @xmath177 has a momentum dependence in the integral .
interestingly the cls @xmath178 relates to the integral of spontaneous decay rate proportional to @xmath179 , which is a hilbert transform @xcite if we put back the lamb shift into our expression .
the lamb shift is an intrinsic level shift induced by vacuum fluctuation while the cls depends on the size and geometry of the atomic ensemble indicating a very different physical mechanism . and
radius @xmath146 , and ( b , c ) cross sections on @xmath145 and @xmath146 respectively with the values of @xmath146 and @xmath145 indicated in the legends .
the atomic density and other parameters are the same as in fig [ fig2 ] .
short radius behavior involves a sharp increase which can be seen in fig .
[ fig4](b).,width=453,height=188 ] to calculate the integral , we may start from the analytic form of @xmath180 , and re - express eq .
( [ f1 ] ) by renormalizing @xmath181 @xmath90 @xmath182 , @xmath183j^2_1\left[ka\sqrt{1-x^2}\right]}{(1-x)^2(1-x^2)}dx.\end{aligned}\ ] ] the function of integral over @xmath184 changes slowly compared to the change of @xmath181 , and we may move it outside of the integral similar to the nonperturbative treatment in the radiation level shift @xcite . taking @xmath181 @xmath4 @xmath89 for the integral of @xmath184 where the integral of @xmath181 is most appreciable , we have @xmath185 where we introduce the infrared ( @xmath186 ) and ultraviolet ( @xmath187 ) energy cutoffs @xcite to the integral , which represent the lowest and largest energy scales in the system .
then we have the cls in the leading order of @xmath188 and @xmath189 , @xmath190 which indicates a redshift @xcite for the level @xmath191 if we let @xmath192 . in fig .
[ fig3 ] , cls shows a geometrical dependence on height and radius in a cylindrical atomic ensemble .
we use @xmath186 @xmath4 @xmath193{\pi r^2l}$ ] as a long - wavelength momentum cutoff for the system , and use atomic radii @xcite as a measure for maximal momentum cutoff which is reasonable for we can not probe atom s internal structure in our quantum optical treatment . for a needle - like cylinder
where @xmath194 , we have a redshift of several khz that can be observable in conventional experimental setup . a blue shift is also shown in fig .
[ fig3 ] for a disk - like geometry ( @xmath195 ) which is due to the counteracting term from system energy cutoff in eq .
( [ shift ] ) . .
the inset in ( a ) shows the crossing of red - blue shifts for different alkali metals , and short radius dependence is shown in ( b ) .
the height is chosen as @xmath145 @xmath4 @xmath196 mm .
the atomic density and idler photon transition are the same as in fig [ fig2].,width=453,height=188 ] in fig .
[ fig4 ] , we show the cls as a dependence of radius , and demonstrate the crossing points of red - blue shifts for other alkali metals in the subplot ( a ) . due to different d1 transition wavelengths of alkali metals , sodium atoms
reach the zero frequency shift at a smaller radius compared to rubidium and cesium ones . by scanning through the geometrical dependence and comparing with different atomic species
, we may determine the system energy cutoff .
the introduction of energy cutoff means lacking knowledge of the system momentum distribution as in the calculation of eq .
( [ cutoff ] ) .
if the experiment shows a redshift as radius varies , it means a unnecessary introduction of the short - wavelength energy cutoff , which suggests that cls is only relevant to the macroscopic ensemble length scales ( @xmath169 and @xmath171 in our case ) . note that when @xmath169 or @xmath171 approaches the condition when @xmath197 , cls is vanishing as can be seen in the general form of eq .
( [ f1 ] ) where @xmath198 . in fig .
[ fig4](b ) , we specifically show a short radius dependence where there is a sharp increase of cls .
the maximum appears at about @xmath146 @xmath86 @xmath199 @xmath200 m , which can be significant if we have an even higher atomic density .
an estimated of frequency shift of mhz is in the reach in the d1 transition of rb atoms for atomic density @xmath201 @xmath202 @xmath150 .
we note that similar red to blue shifts in cls is demonstrated in an ellipsoid geometry @xcite .
we investigate the superradiance and cooperative lamb shift ( cls ) of correlated two - photon emissions in a diamond - type atomic ensemble .
the photon pair is generated from the cascade transitions through four - wave mixing of two classical excitation fields .
the superradiant idler photon is well described by one of the eigenvalues of coupling matrix constructed by the phased symmetrical state .
this symmetrical state is significantly relevant when the number of atoms is large .
we specifically analyze the geometrical dependence of cls on a cylindrical atomic ensemble .
we find a redshift in a needle - like geometry , and suggest it could be blue - shifted in a disk - like one .
we further demonstrate a comparison of the cls for different alkali metals , and propose to determine the system energy cutoff in experiments .
we can take advantage of manipulating system s geometry and density to control the cls , which enables robust frequency qubits in quantum network . in the same spirits of using dual - species matter and light - frequency qubits from a @xmath203rb-@xmath204rb isotopic mixture @xcite
, we may prepare atomic ensembles with two different geometries or densities that generate entangled photon - pair states @xmath205 assuming the same excitation processes , where we denote ( 1,2 ) for two different frequencies and ( s , i ) for signal and idler bosonic fields @xmath206 respectively .
we may construct multimode frequency qubits without acquiring multiple atomic species , and control the qubits via optical means of atom traps which can be dynamical and efficient . the multimode frequency qubits may offer a speedup of entanglement generation as in temporal modes of rare - earth - metal ions @xcite , spatial quantum registers @xcite , and multiplexing quantum repeaters @xcite .
this work is supported by most ( ministry of science and technology ) grant in taiwan , and we thank t. a. b. kennedy for the initiation and guidance of the work .
in this section , we derive the self - interaction energy shift for single and many two - level atoms interacting with a single photon .
we demonstrate that a correct formulation of the interaction energy , including lamb shift and cls , is derived when non - rwa terms are included in the hamiltonian .
these non - rwa terms indicate the energy nonconserving process .
it involves virtual excitations which are unphysical but crucial in determining the frequency shift . in interaction picture for single - atom light - matter interactions ,
we express the hamiltonian as @xmath207 where @xmath208 and @xmath209 are ground and excited states respectively , and high frequency parts @xmath210 are non - rwa terms .
if the atom is in the excited state , it may decay to the ground state with an emitted single photon , whereas the atom in the ground state may be excited also by emitting a photon from non - rwa terms .
the state function can be expressed as @xmath211 solving for the probability amplitudes of atomic energy levels , we have @xmath212 the summation of spontaneous emission modes in the above for the ground ( sign @xmath9 ) and the excited ( sign @xmath6 ) states becomes @xmath213,\nonumber\end{aligned}\ ] ] where the part of delta function simply represents the spontaneous decay rate , @xmath214 @xmath15 @xmath215 , with the dipole moment @xmath216 for the transition . p.v .
denotes principal value .
moreover the lamb shift can be identified as the difference of two level shifts , @xmath217 \equiv\omega^-_{\alpha\alpha}-\omega^+_{\alpha\alpha}=\omega.\nonumber\end{aligned}\ ] ] now for many atoms interacting with a single photon including non - rwa terms , the hamiltonian becomes @xmath218 where @xmath200 denotes the atomic indices .
the state function can be written as @xmath219 where the last unphysical state ( two atomic excitations with one photon present ) is introduced to couple back to the singly - excited state @xcite .
the coupled equations of motion for the probability amplitudes are @xmath220,\label{c1}\\ \dot{c}_{1,\mathbf{k}}^\mu&=-i g_\mathbf{k } e^{i(\omega+\omega_1)t}e^{-i\mathbf{k}\cdot\mathbf{r}_\mu}c_0,\label{c1k}\\ \dot{c}_{0,\mathbf{k}}&=-i \sum_\mu g_\mathbf{k } e^{i(\omega-\omega_1)t}e^{-i\mathbf{k}\cdot\mathbf{r}_\mu}c_1^\mu,\label{c0k}\\ \dot{c}_{2,\mathbf{k}}^{\mu\nu}&= -i g_\mathbf{k } e^{i(\omega+\omega_1)t}e^{-i\mathbf{k}\cdot\mathbf{r}_\nu}c_1^\mu,\label{c2k}\end{aligned}\ ] ] where again the ground state energy level has a lamb shift contribution similar to single atomic one except that an extra factor of @xmath126 , the number of atoms , appears when substituting eq .
( [ c1k ] ) into eq .
( [ c0 ] ) , which shows its intrinsic , not collective , property by vacuum fluctuations . carefully calculating the contribution from nonconserving energy states in the excited state by substituting eq .
( [ c2k ] ) into eq .
( [ c1 ] ) , we have @xmath221\nonumber\\ & = -\sum_\mathbf{k}|g_\mathbf{k}|^2e^{-i(\omega+\omega_1)t}\int^tdt'e^{-i(\omega+\omega_1)t'}\left[(n-1)c_1^\mu(t')+\sum_{\nu\neq\mu } c_1^\nu(t')e^{i\mathbf{k}\cdot(\mathbf{r}_\nu-\mathbf{r}_\mu)}\right]\label{c1_2},\end{aligned}\ ] ] where we can combine the first term in the above with the lamb shift contribution from the ground state ( a factor of @xmath126 ) , and we deduce the positive frequency part of lamb shift @xmath222 . by substituting eq .
( [ c0k ] ) into eq .
( [ c1 ] ) for the energy conserving term and picking out the atomic index @xmath200 , we deduce the other part of lamb shift @xmath223 .
again the lamb shift is derived as @xmath224 which is the same as single atom case .
the second term in eq . ( [ c1_2 ] ) is the dipole - dipole interaction energy from energy nonconserving terms . along with the energy conserving terms by substituting eq .
( [ c0k ] ) into eq .
( [ c1 ] ) and picking out the atomic indices other than @xmath200 , we deduce the collective lamb shift and spontaneous decay rate as @xmath225,\end{aligned}\ ] ] where the property of frequency shift ,
@xmath226 @xmath4 @xmath227 , is used , and @xmath228 @xmath4 @xmath229 , @xmath230 @xmath4 @xmath231 with the transition wave vector @xmath232 . here
we show that the treatment of schrdinger s equation is equivalent to heisenberg s picture that @xmath233 and @xmath234 are defined as @xcite @xmath235\frac{\sin\xi}{\xi}+\left[1 - 3(\hat{d}\cdot\hat{r}_{\alpha\beta})^2\right ] \left(\frac{\cos\xi}{\xi^2}-\frac{\sin\xi}{\xi^3}\right)\bigg\},\\ g_{\alpha,\beta}(\xi)&\equiv\omega_{\alpha\beta}/\gamma_3\equiv-(\omega^-_{\alpha\beta}+\omega^+_{\alpha\beta})/\gamma_3,\nonumber\\ & \equiv\frac{3}{4}\bigg\{-\big[1-(\hat{d}\cdot\hat{r}_{\alpha\beta})^2\big]\frac{\cos\xi}{\xi}+\big[1 - 3(\hat{d}\cdot\hat{r}_{\alpha\beta})^2\big ] \left(\frac{\sin\xi}{\xi^2}+\frac{\cos\xi}{\xi^3}\right)\bigg\},~{\rm for}~\alpha\neq\beta,\end{aligned}\ ] ] where @xmath23 is the unit direction of electric dipole .
the dependence of @xmath236 in the above earns the name of dipole - dipole interaction which is induced by the common light - matter interaction and radiation reaction .
in this section , we derive the probability amplitudes of the excitation process in the adiabatic approximation .
firstly , we substitute the signal photon state of eq .
( [ signal_c ] ) into eq .
( [ eqn3 ] ) , and the summation of signal field modes in eq .
( [ eqn3 ] ) becomes @xmath237\hat{u}_{\mu l}\hat{u}^{-1}_{lm}b_m(t),\nonumber\\ & = \frac{\gamma_2}{2}b_\mu(t)+{\rm lamb~ shift~ term},\end{aligned}\ ] ] where @xmath238 @xmath4 @xmath239 .
the summation of the signal field modes is determined by the fast oscillating exponential factor of optical frequency , which is valid for the eigenvalues @xmath81 @xmath51 @xmath12 @xmath9 @xmath8 . in the last step of derivation
, we may absorb lamb shift into the signal transition frequency , and the above result indicates that the upper excited state radiates in a single atomic decay rate . in the limit of large detunings , @xmath240 ) and ( [ eqn3 ] ) , we use integration by parts to express probability amplitudes in the first order of @xmath241 . note
that we allow time - varying excitation fields , and let @xmath242@xmath15 @xmath243 , @xmath244@xmath15@xmath245 , we have @xmath246,\nonumber\\ & = -\frac{\omega_a(t)\mathcal{e}(t)}{2\delta_1}-\frac{\omega_b^*(t)\tilde{b}_\mu(t)}{2\delta_1}+\mathcal{o}(\frac{1}{\delta_1 ^ 2}),\end{aligned}\ ] ] @xmath247,\nonumber\\ & = -\frac{\omega_b(t)\tilde{a}_\mu(t)}{2(\delta_2+i\gamma_2/2)}+\mathcal{o}(\frac{1}{\delta_2 ^ 2}),\end{aligned}\ ] ] where the initial conditions @xmath248@xmath4@xmath249 are applied .
the adiabatic approximation requires the driving pulses to be smoothly turned on .
therefore in the first order of adiabatic approximation , we derive @xmath250 the above results show that the probability amplitudes develop by following the driving fields , and the ground state is approximately unity in the limit of large detunings .
the ac stark shift in the ground state can be ignored if @xmath251@xmath51@xmath3 which is also required for the validity of assumption of single excitation in our scheme .
99 r. h. dicke , coherence in spontaneous radiation processes , " phys .
rev * 93 * , 99 - 110 ( 1954 ) . c. r. stroud , jr .
, j. h. eberly , w. l. lama , and l. mandel , superradiant effects in systems of two - level atoms , " phys .
a * 5 * , 1094 - 1104 ( 1972 ) .
m. gross and s. haroche , superradiance : an essay on the theory of collective spontaneous emission , " phys . rep . *
93 * , 301 - 396 ( 1982 ) .
l i menshikov , superradiance and related phenomena , " phys .
uspekhi * 42 * , 107 - 147 ( 1999 ) .
l. mandel and e. wolf , _ optical coherence and quantum optics _
( cambridge university press , 1995 ) .
m. j.stephen , first - order dispersion forces , " j. chem .
phys . * 40 * , 669 - 673 ( 1964 ) .
r. h. lehmberg , radiation from an n - atom system .
i. general formalism , " phys .
rev a * 2 * , 883 - 888 ( 1970 ) .
f. t. arecchi and d. m. kim ,
line shifts in cooperative spontaneous emission , " opt .
* 2 * , 324 - 328 ( 1970 ) .
h. morawitz , superradiant level shift and its possible detection in a transient optical experiment , " phys . rev a * 7 * , 1148 - 1159 ( 1973 ) .
r. friedberg , s. r. hartmann , j. t. manassah , frequency shifts in emission and absorption by resonant systems of two - level atoms , " phys .
* 7 * , 101 - 179 ( 1973 ) .
j. t. manassah , statistical quantum electrodynamics of resonant atoms , " phys
* 101 * 359 - 427 ( 1983 ) .
g. s. agarwal , master - equation approach to spontaneous emission , " phys .
rev a * 2 * , 2038 - 2046 ( 1970 ) .
r. bonifacio , p. schwendimann and f. haake , quantum statistical theory of superradiance .
i , " phys . rev a * 4 * , 302 - 313 ( 1971 ) .
v. ernst and p. stehle , emission of radiation from a system of many excited atoms , " phys .
rev * 176 * , 1456 - 1479 ( 1968 ) .
e. ressayre and a. tallet , quantum theory for superradiance , " phys .
rev a * 15 * , 2410 - 2423 ( 1977 ) .
n. e. rehler and j. h. eberly , superradiance , " phys .
rev a * 3 * , 1735 - 1751 ( 1971 ) .
r. bonifacio and l.a .
lugiato , cooperative radiation processes in two - level systems : superfluorescence , " phys .
rev a * 11 * , 1507 - 1521 ( 1975 ) .
j. c. macgillivray and m. s. feld , theory of superradiance in an extended , optically thick medium , " phys .
rev a * 14 * , 1169 - 1189 ( 1976 ) .
h. j. carmichael and k. kim ,
a quantum trajectory unraveling of the superradiance master equation , " opt .
commun . * 179 * , 417 - 427 ( 2000 ) .
j. p. clemens , l. horvath , b. c. sanders and h. j. carmichael , collective spontaneous emission from a line of atoms , " phys .
rev a * 68 * , 023809 ( 2003 ) .
h. h. jen , positive - p phase - space - method simulation of superradiant emission from a cascade atomic ensemble , " phys .
a * 85 * , 013835 ( 2012 ) .
j. h. eberly ,
emission of one photon in an electric dipole transition of one among n atoms , " j. phys .
b : at . mol .
opt . phys . * 39 * , s599-s604 ( 2006 ) .
m. o. scully , e. s. fry , c. h. raymond ooi , and k. wdkiewicz , directed spontaneous emission from an extended ensemble of n atoms : timing is everything , " phys .
lett . * 96 * , 010501 ( 2006 ) .
i. e. mazets and g. kurizki , multiatom cooperative emission following single - photon absorption : dicke - state dynamics , " j. phys .
b : at . mol .
. phys . * 40 * , f105-f112 ( 2007 ) .
a. a. svidzinsky , j .-
chang , and m. o. scully , dynamical evolution of correlated spontaneous emission of a single photon from a uniformly excited cloud of n atoms , " phys .
lett . * 100 * , 160504 ( 2008 ) .
a. svidzinsky and j .- t .
chang , cooperative spontaneous emission as a many - body eigenvalue problem , " phys .
rev a * 77 * , 043833 ( 2008 ) .
r. friedberg and j. t. manassah , effects of including the counterrotating term and virtual photons on the eigenfunctions and eigenvalues of a scalar photon collective emission theory , " phys .
a * 372 * , 2514 - 2521 ( 2008 ) .
r. friedberg and j. t. manassah , time - dependent directionality of cooperative emission after short pulse excitation , " opt . commun . * 281 * , 4391 - 4397 ( 2008 ) .
m. o. scully , collective lamb shift in single photon dicke superradiance , " phys .
lett . * 102 * , 143601 ( 2009 ) .
r. rhlsberger , k. schlage , b. sahoo , s. couet , r. rffer , collective lamb shift in single - photon superradiance , " science * 328 * , 1248 - 1251 ( 2010 ) .
j. keaveney , a. sargsyan , u. krohn , i. g. hughes , d. sarkisyan , and c. s. adams , cooperative lamb shift in an atomic vapor layer of nanometer thickness , " phys .
* 108 * , 173601 ( 2012 ) .
j. pellegrino , r. bourgain , s. jennewein , y. r. p. sortais , a. browaeys , s. d. jenkins , and j. ruostekoski , observation of suppression of light scattering induced by dipole - dipole interactions in a cold - atom ensemble , " phys .
lett . * 113 * , 133602 ( 2014 ) .
t. chanelire , d. n. matsukevich , s. d. jenkins , t. a. b. kennedy , m. s. chapman , and a. kuzmich , quantum telecommunication based on atomic cascade transitions , " phys .
lett . * 96 * , 093604 ( 2006 ) . b. srivathsan , g. k. gulati , b. chng , g. maslennikov , d. matsukevich , and c. kurtsiefer , narrow band source of transform - limited photon pairs via four - wave mixing in a cold atomic ensemble , " phys . rev
. lett . * 111 * , 123602 ( 2013 ) .
a. g. radnaev , y. o. dudin , r. zhao , h. h. jen , s. d. jenkins , a. kuzmich , and t. a. b. kennedy , a quantum memory with telecom - wavelength conversion , " nat . phys .
* 6 * , 894 - 899 ( 2010 ) .
h. h. jen , spectral analysis for cascade - emission - based quantum communication in atomic ensembles , " j. phys .
b : at . mol .
* 45 * , 165504 ( 2012 ) .
briegel , w. dr , j. i. cirac and p. zoller , quantum repeaters : the role of imperfect local operations in quantum communication , " phys .
lett . * 81 * , 5932 - 5935 ( 1998 ) .
duan , m. d. lukin , j. i. cirac and p. zoller , long - distance quantum communication with atomic ensembles and linear optics , " nature * 414 * , 413 - 418 ( 2001 ) .
m. o. scully and m. s. zubairy , _ quantum optics _
( cambridge university press , 1997 ) . c. cohen - tannoudji , j. dupont - roc , and g. grynberg , _
atom - photon interactions : basic process and applications _
( john wiley & sons . inc . ,
m. donaire , electromagnetic vacuum of complex media .
ii . lamb shift and total vacuum energy , " phys .
a * 85 * , 052518 ( 2012 ) .
j. c. slater , atomic radii in crystals , " j. chem .
phys . * 41 * , 3199 - 3204 ( 1964 ) .
r. friedberg and j. t. manassah , cooperative lamb shift in an ellipsoid , " phys .
a * 81 * , 063822 ( 2010 ) .
lan , s. d. jenkins , t. chanelire , d. n. matsukevich , c. j. campbell , r. zhao , t. a. b. kennedy , and a. kuzmich , dual - species matter qubit entangled with light , " phys .
lett . * 98 * , 123602 ( 2007 ) . c. simon , h. de riedmatten , m. afzelius , n. sangouard , h. zbinden , and n. gisin , quantum repeaters with photon pair sources and multimode memories , " phys .
98 * , 190503 ( 2007 ) .
m. afzelius , c. simon , h. de riedmatten , and n. gisin , multimode quantum memory based on atomic frequency combs , " phys .
a * 79 * , 052329 ( 2009 ) .
k. tordrup , a. negretti , and k. mlmer , holographic quantum computing , " phys .
* 101 * , 040501 ( 2008 ) .
a. grodecka - grad , e. zeuthen , and a. s. srensen , high - capacity spatial multimode quantum memories based on atomic ensembles , " phys .
lett . * 109 * , 133601 ( 2012 ) .
o. a. collins , s. d. jenkins , a. kuzmich , and t. a. b. kennedy , multiplexed memory - insensitive quantum repeaters , " phys
* 98 * , 060502 ( 2007 ) . | we investigate superradiant cascade emissions from an atomic ensemble driven by two - color classical fields .
the correlated pair of photons ( signal and idler ) is generated by adiabatically driving the system with large - detuned light fields via four - wave mixing .
the signal photon from the upper transition of the diamond - type atomic levels is followed by the idler one which can be superradiant due to light - induced dipole - dipole interactions .
we then calculate the cooperative lamb shift ( cls ) of the idler photon , which is a cumulative effect of interaction energy .
we study its dependence on a cylindrical geometry , a conventional setup in cold atom experiments , and estimate the maximum cls which can be significant and observable .
manipulating the cls of cascade emissions enables frequency qubits that provide alternative robust elements in quantum network .
superradiance ; cold atoms ; cascade atomic transitions ; two - photon spontaneous emissions ; quantum telecommunication |
what counts as a successful theory in computer science ? consider obvious exemplars such as * process calculi * type systems * model - checking it is not the case that there is a single agreed model , notation , formalism , tool or language in any of the above areas .
in fact there are a profusion of all of these , although some have been particularly influential .
( insert your favourite examples here ) [ [ the - next-700-cdots - syndrome ] ] the ` next 700 @xmath0 ' syndrome + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + is this profusion a ` scandal ' of our subject ?
i used to think so and i was nt alone ( e.g. robin milner s quest to find the ` @xmath1-calculus of concurrency ' ) .
now i am not so sure . [ [ its - the - paradigms ] ] it s the _ paradigms ! _
+ + + + + + + + + + + + + + + + + + + + + the paradigms and tool - kits , both technical and conceptual , provided by these theories have been deeply absorbed by the research communities and have increasingly influenced applications .
* examples : * * labelled transition systems and bisimulation * naming and scope restriction and extrusion * the automata - theoretic paradigm for model - checking * the type systems paradigm , with compositional typing rules for terms - in - context , and key structural properties such as subject reduction .
[ [ by - their - fruits - shall - ye - know - them . ] ] by their fruits shall ye know them .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + these tool - kits are _ the real fruits of these theories_. they may be compared to the traditional tool - kits of physics and engineering : differential equations , laplace and fourier transforms , numerical linear algebra , etc .
they can be applied to a wide range of situations , going well beyond those originally envisaged , e.g. security , computational biology , quantum computing , etc .
so , is everything in the garden rosy ? [ [ dreams - of - final - theories ] ] dreams of final theories + + + + + + + + + + + + + + + + + + + + + + + + but can we do better than this ?
after all , in physics there _ are _ great theories which transcend mere tool - kits .
we largely lack such theories , in computer science as a whole , and in concurrency and process calculus in particular .
is this unavoidable , as part of the nature of our subject , or will such theories emerge ? some may find such questions uninteresting , or even meaningless ; they can safely stop reading here .
1980 marked the start of a new era in concurrency theory , but not its beginning .
a meaningful theory of concurrency , incorporating some profound insights , had been originated by petri in the 1960 s , and net theory , as well as other approaches to concurrency , continues to be actively developed .
there is no doubt that the advent of algebraic process calculi marked a decisive advance in concurrency theory , in particular in the use of compositional algebraic methods for the description of complex systems .
it is often the case , though , that when an advance is made , something valuable is also lost , or at least , temporarily forgotten .
let us start with the problem of _ canonicity _ the ` next 700 process algebras ' syndrome . in a sense ,
the very success of the paradigmatic tool - kit , as described in the previous section , is also the source of the problem .
it is too easy to cook up yet another variant process calculus or algebra ; there are too few constraints .
this _ plasticity of definitions _ has become so familiar in our field that we may not be aware of it as an issue .
the mathematician andr weil apparently compared finding the right definitions in algebraic number theory which was like carving adamantine rock to making definitions in the theory of uniform spaces ( which he founded ) , which was like sculpting with snow . in concurrency theory , we are very much at the snow - sculpture end of the spectrum .
we lack the kind of external reality , whether it comes from fundamental mathematical objects like the integers , or manifolds , or differential equations , or from physical reality as determined by experiment , which is hard and obdurate , and resistant to our definitions .
is this a necessary feature of our existence , or have we just not yet found the real bedrock ?
an important quality of petri s conception of concurrency is that it _ does _ seek to determine fundamental concepts : causality , concurrency , process , etc . in a syntax - independent fashion .
another important point , which may originally have seemed merely eccentric , but now looks rather ahead of its time , is the extent to which petri s thinking was explicitly influenced by physics ( see e.g. @xcite . as one example , note that k - density comes from one of carnap s axiomatizations of relativity ) . to a large extent , and by design , net theory can be seen as a kind of _ discrete physics _ : * lines * are time - like causal flows , * cuts * are space - like regions , * process unfoldings * of a * marked net * are like the solution trajectories of a differential equation .
this acquires new significance today , when the consequences of the idea that ` information is physical ' are being explored in the rapidly developing field of quantum informatics .
moreover , the need to recognize the spatial structure of distributed systems has become apparent , and is made explicit in formalisms such as the ambient calculus , and milner s bigraphs . * the genius , the success , and the limitation of process calculi is their _
linguistic character_. this provides an ingenious way of studying processes , information flow , etc . without quite knowing , independently of the particular linguistic setting , what any of these notions are .
one could try to say that they are implicitly defined by the calculus .
but then the fact that there are so many calculi , potential and actual , does not leave us on very firm ground .
+ we lack syntax - independent , _ intrinsic _ definitions of the fundamental notions of concurrency theory .
net theory and some related approaches ( e.g. event structures ) still offer the best extant accounts of these issues .
but we are still far from home .
+ thus for example consider the issue of _
there are some fragmentary results , but there is no single compelling notion of ` expressive completeness ' for a process calculus , or of a ` church s thesis for concurrency ' . * we must now also acknowledge that we do not have sole ownership of the notions of information , process , etc .
physics and biology are also interested and they are at our gates ! this presents us with a challenge , and perhaps also an opportunity for some new thinking on these issues .
a major recent development in process calculi has been their application to biological modelling .
this represents perhaps the first substantial example of a trend which , in my view , will form a major part of the future development of our subject : the spreading outwards of ideas developed in computer science , of the tool - kits we discussed in section 1 , to other scientific disciplines . provided there is a real engagement between the cs bio - concurrency community and the biologists , this development has great promise .
however , while biological modelling will surely make new demands on process calculi , and hence lead to new developments ( the next 700 biological process calculi ? ) , i do nt believe it is likely to lead to foundational advances for the issues we are discussing .
biology s foundational and conceptual structures are , if anything , much more plastic than those of computer science for which , of course , it compensates by the exuberant richness and the sheer concrete reality of the existence proofs which it studies
. there is , perhaps , more prospect for guidance in finding fundamental notions of process , information flow , etc . from the rapidly developing interface between computer science and physics , which has grown up around quantum informatics .
we have already discussed how petri s development of net theory was influenced by ideas from physics , and indeed provides some of the ingredients of a discrete physics .
( one feature conspicuously _ lacking _ there is an account of the non - local information flows arising from entangled states , which play a key role in quantum informatics .
locality is so plausible to us and yet , at a fundamental physical level , apparently so wrong ! ) .
meanwhile , there are now some matching developments on the physics side , and a greatly increased interest in discrete models .
as one example , the causal sets approach to discrete spacetime of sorkin et al .
@xcite is very close in spirit to event structures .
my own recent work with bob coecke on a categorical axiomatics for quantum mechanics @xcite , adequate for modelling and reasoning about quantum information and computation , is strikingly close in the formal structures used to my earlier work on interaction categories @xcite which represented an attempt to find a more intrinsic , syntax - free formulation of concurrency theory ; and on geometry of interaction @xcite , which can be seen as capturing a notion of interactive behaviour , in a mathematically rather robust form , which can be used to model the dynamics of logical proof theory and functional computation .
this work admits a striking ( and very useful ) diagrammatic presentation , which suggests a link to geometry and indeed there are solid connections with some of the central ideas relating geometry and physics which have been so prominent in the mathematics of the past 20 years .
we note also that , in a rather different style , the geometry of concurrency has been developed by eric goubault @xcite and others .
so , geometry may yet have an important role to play in concurrency theory .
if anything like these speculations comes to pass , i think process calculus will be raised to a new level .
it will , perhaps , become truly _ the _ calculus of a fundamental science of information dynamics . | process algebra has been successful in many ways ; but we do nt yet see the lineaments of a fundamental theory . some fleeting glimpses are sought from petri nets , physics and geometry .
concurrency , process algebra , petri nets , geometry , quantum information and computation . |
transiting planets provide a wealth of information about exoplanetary systems .
short - term variations in the mid - eclipse times of the transits may reveal the presence of moons , trojans or other planets ( ( * ? ? ?
* holman & murray 2005 ) , ( * ? ? ?
* agol 2005 ) , ( * ? ? ?
* ford & holman 2007 ) ) , whereas long - term variations could result from orbital precession ( ( * ? ? ?
* miralda escud 2002 ) ) .
this provides further constraints on theories of planetary system formation and evolution , as well on theories of planetary atmospheres and their composition .
we observed three transits of tres-1 on ut 2007
september 12 , 15 , and 18 , corresponding to epochs @xmath0 , 153 , and 154 of the ephemeris given by ( * ? ? ? * winn ( 2007 ) ) : @xmath1 and one transit of wasp-2 on ut 2007 september 13 , corresponding to epoch @xmath2 of the ephemeris given by ( * ? ? ? * charbonneau ( 2007 ) ) : @xmath3 we used ag2 , a frame - transfer ccd mounted at the folded cassegrain focus of the william herschel telescope of the isaac newton group , la palma , spain .
ag2 is an ing - designed autoguider head with e2v ccd having a field of view ( fov ) @xmath4 and a pixel scale of @xmath5 .
we used a kitt peak r filter in order to minimize the effect of color - dependent atmospheric extinction on the differential photometry and the effect of limb - darkening on the transit light curve .
we observed under nearly perfect conditions on the nights of ut
september 12 , 13 , and 18 . on the night of ut
september 15 , we observed under scattered clouds in the second half of the night . on all nights , autoguiding ensured the positions of all stars on the ccd varied by no more than 3 pixels over the course of each night .
we strongly defocused the telescope in order to minimize the effect of the pixel - to - pixel sensitivity variations , and also to enhance the duty cycle .
the full - width at half - maximum ( fwhm ) of stellar images was 7 pixels for wasp-2 and ranged from 8 to 13 pixels for tres-1 .
for tres-1 for the three nights , we acquired 5 , 10 and 10 second exposures , while for wasp-2 we used 7 second exposures .
we used standard iraf procedures for the overscan correction , bias subtraction and performing the differential photometry .
we did not use flat - fields to minimise the impact of position - angle dependent scattered light entering the aperture .
different aperture sizes were tried in order to find out the one that produced the minimum noise in the out - of - transit data . for tres-1 for the three nights ,
the optimum aperture radii were 31 , 31 , and 34 pixels , while for wasp-2 it was 30 pixels .
the sky background was subtracted , using an estimate of its brightness determined within an annulus centered on each star .
because of the strong defocusing and other field stars , we set an inner radius of the annulus equal to the aperture radius and used the width of 10 pixels .
the comparison stars used were 2mass : j19041058 + 3638409 , j19040934 + 3639195 , and j19040792 + 3640116 for tres-1 , and 2mass : j20304846 + 0627355 , and j20305168 + 0628008 for wasp-2 .
for each night , differential photometry was obtained by taking the ratio of the signal of the variable to the mean of the comparisons , both normalized by a smooth function of time . to estimate appropriate error bars for our data
, we used a procedure similar to that by ( * ? ? ?
* narita ( 2007 ) ) .
we first fitted the light curves with the analytic formulae of ( * ? ? ?
* mandel & agol ( 2002 ) ) to find the differences between the data and the best - fitting model .
we rescaled the error bars to satisfy @xmath6 , where @xmath7 is the number of the measurements in each light curve .
for the three nights of tres-1 we found that the true errors are higher by factors of 2.2 , 1.6 and 1.9 respectively than the errors including only the poisson noise , and for the one night of wasp-2 by a factor of 2.2 .
differential photometry with these rescaled error bars and the best - fitting model ( solid line ) are shown in fig .
[ wt_lc ] . in order to estimate the amount of the time - correlated red noise
, we solved the equations given e.g. by ( * ? ? ? *
narita ( 2007 ) ) : @xmath8 @xmath9 where @xmath10 is the standard deviation of each residual and @xmath11 is the standard deviation of the average of the successive @xmath12 points .
we selected @xmath12 to correspond to about 20 - 50 minutes depending on the length of the interval with the correlated errors .
@xmath13 is the white noise , which is uncorrelated and averages down as @xmath14 , @xmath15 is the red noise , which is correlated and remains constant for specified @xmath12 .
we adjusted the error bars by multiplying @xmath16^{1/2}$ ] and used these rescaled uncertainties for the subsequent fitting procedure .
.resulting barycentric transit times of tres-1 and wasp-2 systems with their uncertainties given by the 68% confidence limits . [ cols="<,^,^,^",options="header " , ]
we used the markov chain monte carlo simulations with the metropolis - hastings algorithm ( ( * ? ? ?
* ford 2005 ) ) to estimate the statistical uncertainties of the resulting parameters .
we assumed a quadratic law of the limb darkening .
for tres-1 , we fixed the system parameters according to ( * ? ? ?
* winn ( 2007 ) ) and solved only for the transit times . for wasp-2 , we fixed the parameters derived by ( * ? ? ?
* charbonneau ( 2007 ) ) and fitted the limb - darkening coefficients and the transit time . for each night
, we created 10 independent chains , with the length typically 100,000 points in each chain .
we checked the convergence of generated chains using the ( * ? ? ? * gelman & rubin ( 1992 ) ) r statistic .
the resulting barycentric transit times can be found in table [ times ] .
their uncertainties are given by the 68% confidence limits .
the resulting residuals are plotted in fig .
[ tres_oc ] for tres-1 , and in fig .
[ wasp_oc ] for wasp-2 , using the ephemerides ( [ efe - tres ] ) , and ( [ efe - wasp ] ) , respectively . in summary ,
our data provide no conclusive evidence of other bodies in either system but the systematic trend of residuals for tres-1 , found by ( * ? ? ?
* winn ( 2007 ) ) , is worthy of further study . for tres-1 , our 3 new transit times plus 1 by ( * ?
* narita ( 2007 ) ) extend the interval covered by data for more than 150 cycles .
we therefore used a weighted linear fit to the transit times as a function of epoch ( weights inversely proportional to the squares of rms errors of individual transit times ) to derive a more accurate ephemeris for the future transit predictions : @xmath17 + the research was supported by the grants 205/08/h005 and 205/06/0304 of the czech science foundation and also from the research program msm0021620860 of the ministry of education of the czech republic . | searching for transit timing variations in the known transiting exoplanet systems can reveal the presence of other bodies in the system . here
we report such searches for two transiting exoplanet systems , tres-1 and wasp-2 .
their new transits were observed with the 4.2 m william herschel telescope located on la palma , spain . in a continuing programme ,
three consecutive transits were observed for tres-1 , and one for wasp-2 during september 2007 .
we used the markov chain monte carlo simulations to derive transit times and their uncertainties .
the resulting transit times are consistent with the most recent ephemerides and no conclusive proof of additional bodies in either system was found . |
in this work we study the long term behavior of continuous time markov processes whose states assign either @xmath0 or @xmath1 ( usually called a spin value ) to each vertex @xmath5 in a graph @xmath2 .
the graphs @xmath2 we consider are related to homogeneous trees of degree @xmath6 and include infinite stacks of homogeneous trees .
these graphs will be specified in section [ sec : themodel ] , where we will also discuss some earlier papers where such stacks of trees have been studied .
the geometric and percolation theoretic methods we introduce to carry out our analysis ( see especially section [ sec : main ] and appendix [ app : geometric ] ) are potentially of independent interest .
we denote by @xmath7 the value of the spin at vertex @xmath8 at time @xmath9 .
starting from a random initial configuration @xmath10 drawn from the independent bernoulli product measure @xmath11 the system then evolves in continuous time according to an agreement inducing dynamics : at rate 1 , each vertex changes its value if it disagrees with more than half of its neighbors , and tosses a fair coin in the event of a tie .
our arguments and results easily extend to many other types of dynamics , as discussed in remark [ remark1.3 ] below ; these include processes in discrete time , as in @xcite , and different rules for tie - breaking .
our process corresponds to the zero - temperature limit of glauber dynamics for a stochastic ising model with ferromagnetic nearest neighbor interactions and no external magnetic field ( see , e.g. , @xcite or @xcite ) .
this process has been studied extensively in the physical and mathematical literature primarily on graphs such as the hyper - lattice @xmath12 and the homogeneous tree of degree @xmath6 , @xmath13 .
a physical motivation is the behavior of a magnetic system following the extreme case of a deep quench , i.e. , when a system has reached equilibrium at infinite temperature and is instantaneously reduced to zero temperature . for references on this and related problems
see , e.g. , @xcite or @xcite .
the main focus in the study of this model is the formation and evolution of boundaries delimiting same spin cluster domains : these domains shrink or grow or split or coalesce as their boundaries evolve .
an interesting question is whether the system has a limiting configuration , or equivalently does every vertex eventually stop flipping ? whether @xmath14 exists for almost every initial configuration , realization of the dynamics and for all @xmath15 in the underlying graph depends on @xmath4 and on the structure of the underlying graph @xmath2 .
nanda , newman and stein @xcite investigated this question when @xmath16 and @xmath17 and found that in this case the limit does not exist , i.e. , every vertex flips forever .
their work extended an old result of arratia @xcite , who showed the same on @xmath18 for @xmath19 or @xmath20 .
one important consequence of the methods of @xcite is that @xmath21 does exist for almost every initial configuration , realization of the dynamics and every @xmath15 if the graph is such that every vertex has an odd number of neighbors , such as for example @xmath22 for @xmath6 odd .
another question of interest is whether sufficient bias in the initial configuration leads the system to reach consensus in the limit .
i.e. , does there exist @xmath23 , such that for @xmath24 , @xmath25 we will refer to as * fixation to consensus * ( of @xmath0 ) .
kanoria and montanari @xcite studied fixation to consensus on homogeneous trees of degree @xmath26 for a process with synchronous time dynamics .
their process has the same update rules as ours , except that all vertices update simultaneously and at integer times @xmath27 . for each @xmath6 , kanoria and montanari
defined the consensus threshold @xmath28 to be the smallest bias in @xmath29 such that the dynamics converges to the all @xmath0 configuration , and proved upper and lower bounds for @xmath30 as a function of @xmath6 .
fixation to consensus was also investigated on @xmath12 for the asynchronous dynamics model .
it was conjectured by liggett @xcite that fixation to consensus holds there for all @xmath31 .
fontes , schonmann and sidoravicius @xcite proved consensus for all @xmath32 with @xmath33 strictly less than but very close to 1 and morris @xcite proved that @xmath34 as @xmath35
. in @xcite howard investigated the asynchronous dynamics in detail on @xmath36 and showed how fixation takes place . on this tree graph , vertices fixate in spin chains ( defined as doubly infinite paths of vertices of the same spin sign ) .
though no spin chains are present at time 0 when @xmath37 , howard showed that for any @xmath38 , there are ( almost surely ) infinitely many distinct @xmath0 and @xmath1 spin chains at time @xmath39 .
he also showed the existence of a phase transition in @xmath4 : there exists a critical @xmath40 such that if @xmath41 , @xmath0 spin chains do not form almost surely , whereas if @xmath42 they almost surely form in finite time .
our work is motivated by that of howard , but for more general tree - related graphs .
let @xmath43 denote a doubly infinite stack of homogeneous trees of degree @xmath44 , i.e. , the graph with vertex set @xmath45 and edge set specified below .
the main focus of this paper is proving fixation to consensus on @xmath43 for the process started with an independent identically distributed initial configuration of parameter @xmath4 .
such infinite stacks of trees have been studied before in the context of bernoulli percolation @xcite and ising models @xcite .
more general nonamenable graphs have also been studied see , e.g. , @xcite .
one motivation for these studies is that as the parameter of the model varies , the behavior is sometimes like that on a simple homogeneous tree and sometimes like that on a simple amenable graph like @xmath46 .
we express @xmath47 as @xmath48 where @xmath49 , and think of this as a decomposition of the infinite stack @xmath43 into layers @xmath50 .
let the edge set of @xmath43 , @xmath51 , be such that any two vertices @xmath52 are connected by an edge @xmath53 if and only if : 1 .
@xmath54 for some @xmath55 , and the corresponding @xmath56 and @xmath57 are adjacent vertices in @xmath22 ; or 2 .
@xmath58 for some @xmath55 and @xmath59 ; or 3 .
@xmath58 for some @xmath55 and @xmath60 . for a more detailed description of the markov process than the one given in section [ sec : introduction ] , we associate to each vertex @xmath61 a rate 1 poisson process whose arrival times we think of as a sequence of clock rings at @xmath5 .
we will denote the arrival times of these poisson processes by @xmath62 and take the poisson processes associated to different vertices to be mutually independent .
we associate to the @xmath63 s independent bernoulli@xmath64 random variables with values @xmath0 or @xmath1 , which will represent the fair coin tosses to be used in the event of a tie .
let @xmath65 be the probability measure for the realization of the dynamics ( clock rings and tie - breaking coin tosses ) , and denote by @xmath66 the joint probability measure on the space @xmath67 of initial configurations @xmath68 and realizations of the dynamics ; an element of @xmath67 will be denoted @xmath69 .
the main result of this paper is the following theorem , which shows fixation to consensus for nontrivial @xmath4 ; its proof is given in section [ sec : main ] .
unlike kanoria and montanari @xcite , here we do not attempt to obtain good upper bounds on @xmath33 though we expect @xmath33 to approach @xmath70 with increasing degree @xmath6 .
we restrict ourselves to proving fixation to @xmath0 for @xmath4 close enough to 1 with the standard majority update rule : when its clock rings , each vertex updates to agree with the majority of its neighbors or tosses a fair coin in the event of a tie .
see remark [ remark1.3 ] below for other update rules to which our arguments and results apply .
[ theorem1 ] given @xmath44 , there exists @xmath71 such that for @xmath72 the process on @xmath43 fixates to consensus .
the same fixation to consensus result holds for the following graphs , as stated in theorem [ theorem2 ] below , whose proof is also given in section [ sec : main ] : * homogeneous trees @xmath22 of degree @xmath44 . * finite width stacks of homogeneous trees of degree @xmath44 with free or periodic boundary conditions .
these are graphs , which we will denote by @xmath73 or @xmath74 , with vertex set @xmath75 and edge set @xmath76 or @xmath77 . @xmath76 and @xmath77 are defined similarly to the edge set @xmath51 of @xmath43 : two vertices @xmath78 are connected by an edge @xmath79 if and only if either condition i above holds ; or 1 .
@xmath80 for @xmath81 and either condition ii or iii holds ; or 2 .
@xmath82 and @xmath83 ; or 3 .
@xmath84 and @xmath85 .
+ any two vertices @xmath86 are connected by an edge @xmath87 if and only if either condition i holds ; or 1 .
@xmath80 for @xmath81 and either condition ii or iii holds ; or 2 .
@xmath82 and @xmath83 or @xmath88 ; or 3 .
@xmath84 and @xmath85 or @xmath89 .
* semi - infinite stacks of homogeneous trees of degree @xmath44 with free boundary conditions .
these are graphs , which we will denoted by @xmath90 , with vertex set @xmath91 and edge set @xmath92 .
two vertices @xmath93 are connected by an edge @xmath94 if and only either condition i holds ; or 1 .
@xmath80 for @xmath95 and either condition ii or iii holds ; or 2 .
@xmath82 and @xmath83 .
[ theorem2 ] fix @xmath44 and @xmath96 and let @xmath2 be one of the following graphs : @xmath22 , @xmath73 , @xmath74 or @xmath90 .
there exists @xmath71 such that for @xmath72 the process on @xmath2 fixates to consensus .
[ remark1.3 ] our results have natural extensions to other dynamics .
let @xmath97 be the maximum number of neighbors of a vertex in the graph @xmath2 where @xmath2 is any of the graphs of theorem [ theorem2 ] ; for some @xmath98 , we can change ( arbitrarily ) the update rules for those vertices whose number of @xmath0 neighbors is strictly less than @xmath99 , and the conclusions of theorem [ theorem1 ] or [ theorem2 ] remain valid with the same proof . for large @xmath97
, @xmath99 can be taken much larger than @xmath100 .
for example , on the infinite stack @xmath43 of @xmath6-trees , @xmath101 and for @xmath102 , one can take @xmath103 ( as is readily seen from the proof of theorem [ theorem1 ] ) .
a special case of this type of extension of our results is to modify the update rule in the event of a tie : e.g. , instead of flipping a fair coin , flip a biased coin with any bias @xmath104 $ ] or do nothing .
we can also change from two - valued spins to any fixed number @xmath105 of spin values , say @xmath106 .
the initial configuration is given by the measure @xmath107 where @xmath108 and @xmath109 and the updating is done via a majority rule , e.g. , by a rule that respects majority agreement of neighbors on , say , color 1 .
we can then think of color @xmath20 as the @xmath0 spin from before , and the other @xmath110 colors together representing the @xmath1 spin . if @xmath111 is close enough to 1 , we again obtain fixation to @xmath0 consensus .
all our results also apply to the synchronous dynamics of @xcite .
in order to prove theorem [ theorem1 ] we will show that if we take @xmath4 close enough to 1 , then already at time 0 there are stable structures of @xmath0 vertices , which are fixed for all time .
we will choose these structures to be subsets ( denoted @xmath112 ) of the layers @xmath50 in the decomposition of @xmath43 such that they are stable with respect to the dynamics .
we will define a set @xmath113 as the union of @xmath112 for all @xmath114 , and show that for @xmath4 close enough to 1 , the complement of @xmath113 is a union of almost surely finite components .
[ t+ls ] for @xmath114 fixed , let @xmath115 be the union of all subgraphs @xmath116 of @xmath50 that are isomorphic to @xmath117 with @xmath118 .
we point out that @xmath119 is stable for @xmath102 , since every @xmath120 has @xmath121 out of @xmath122 neighbors of spin @xmath123 and @xmath124 for @xmath125 . not only is this set stable with respect to the dynamics on @xmath43 as in theorem [ theorem1 ] , but it s also stable with respect to the dynamics on @xmath50 and the other graphs of theorem [ theorem2 ] .
let @xmath113 represent the union of @xmath126 across all levels @xmath50 , i.e. , @xmath127 where for shorthand notation , @xmath128 . if @xmath129 , @xmath113 as defined above is not stable with respect to the dynamics . in these cases
the argument will be changed somewhat as discussed in section [ sec : main ] .
the goal of this subsection is to state and prove a geometric probability estimate , proposition [ propindept ] , which concerns asymmetric site percolation on @xmath13 distributed according to the product measure @xmath130 with : @xmath131 this equals the distribution of @xmath132 restricted to the layers @xmath50 , and therefore applies to these graphs as well .
the statement and proof of proposition [ propindept ] require a series of definitions .
the first of these defines graphical subsets of @xmath22 , whereas the second concerns probabilistic events for subgraphs of @xmath22 that have a specific orientation .
later , in the proof of theorem [ theorem1 ] which is given in section [ sec : main ] , proposition [ propindept ] will be applied to certain subsets of @xmath50 .
[ treegraphs ] _ * certain rooted subtrees of @xmath22 * _ let @xmath133 in @xmath22 be two adjacent vertices , and denote by @xmath134 $ ] the connected component of @xmath5 in @xmath135 see figure [ fig : mainpart1 ] .
let @xmath136 be three adjacent vertices in @xmath22 , such that @xmath5 and @xmath137 are neighbors of @xmath138 .
denote by @xmath139 $ ] the connected component of @xmath138 in @xmath140 see figure [ fig : mainpart2 ] .
$ ] and @xmath141 $ ] are tree graphs whose roots have coordination number @xmath142 $ ] is a tree graph whose root has coordination number @xmath143 [ fig : mainpart2 ] [ k-1-arytrees ] _ * random @xmath144-ary trees of spin @xmath0 with a certain + orientation * _ let @xmath145 be a deterministic subtree of @xmath22 with at least two vertices , and @xmath146 be a leaf of @xmath145 ; i.e. , @xmath146 has a neighbor @xmath147 in @xmath145 and @xmath144 neighbors in @xmath148 .
@xmath149 $ ] is the event that there is a subgraph @xmath116 of @xmath150 $ ] isomorphic to @xmath151 and containing @xmath146 , such that @xmath152 see figure [ fig : mainpart3 ] . + let @xmath145 be a deterministic subtree of @xmath22 with at least five vertices , and @xmath146 be a * 2-point * of @xmath145 ( i.e. , a vertex of @xmath145 with exactly two neighbors in @xmath145 ) that is also * good * ( i.e. , both its neighbors in @xmath145 are also 2-points of @xmath145 ) .
let @xmath153 be the two neighbors of @xmath146 in @xmath145 and let @xmath154 be @xmath155 s other neighbor in @xmath145 .
@xmath156 $ ] is the event that there is a subgraph @xmath116 of @xmath157 \cup a_{v , w ' } [ w]$ ] isomorphic to @xmath151 and containing @xmath146 and @xmath155 , such that @xmath152 ; here @xmath158 \cup a_{v , w ' } [ w]$ ] is the graph with vertex set @xmath159 } \cup \mathbb{v}_{a_{v , w ' } [ w]}$ ] and edge set @xmath160 } \cup \mathbb{e}^\infty_{a_{v , w ' } [ w ] } \cup e_{v w}$ ] see figure [ fig : mainpart4 ] .
$ ] asserts the existence of a random @xmath161-ary tree of spin @xmath0 that contains a leaf , @xmath146 , of @xmath145 ] $ ] asserts the existence of a random @xmath161-ary tree of spin @xmath0 that contains a good two 2-point , @xmath146 , of @xmath145 , and one of its neighbors , @xmath155 ] for distinct leaves @xmath146 of @xmath145 , the events @xmath164\}_{v \in t}$ ] are defined on disjoint subsets of @xmath13 , and are therefore independent ; they are also identically distributed .
the same is true for @xmath165\}_{v , w \in t}$ ] for disjoint pairs @xmath166 .
the following is essentially the same as definition [ t+ls ] , with the only difference being that here we define the graph @xmath167 on @xmath22 , whereas before we defined the same random graph on @xmath50 .
[ t+l ] let @xmath167 be the union of all subgraphs @xmath116 of @xmath13 that are isomorphic to @xmath168 with @xmath169 .
the next proposition estimates the probability that none of the vertices of a given set @xmath170 belong to any random @xmath161-ary tree of spin @xmath0 ( see definition [ k-1-arytrees ] ) .
this proposition is a main ingredient in the proof of theorem [ theorem1 ] .
[ propindept ] for any @xmath171 such that for @xmath172 and any deterministic finite nonempty subset @xmath170 of @xmath22 , @xmath173 let @xmath145 be the minimal spanning tree containing all the vertices of @xmath170 .
we will call the vertices of @xmath170 the * special * vertices of @xmath145 . note that all the leaves of @xmath145 are special vertices .
we first suppose @xmath174 ; the simpler case @xmath175 will be handled at the end of the proof . by the distinctness and disjointness results of lemma [ disjointlemma ] from the appendix ,
there exist constants @xmath176 depending only on @xmath6 , such that for each such tree @xmath145 , one or both of the following is valid : there are at least @xmath177 leaves @xmath146 in @xmath145 , with the events @xmath178 mutually independent , and/or there are at least @xmath179 edges having endpoints @xmath180 in @xmath145 with @xmath146 a good special 2-point , and the events @xmath181 mutually independent .
let us first suppose that a ) holds .
we claim that , for @xmath146 any leaf of @xmath145 , @xmath182)]^{\epsilon_1 | \lambda | } .\ ] ] the claim follows from a string of inclusions .
first , @xmath183 but if @xmath146 is a leaf of @xmath145 , then @xmath184^c,\ ] ] so that @xmath185^c.\ ] ] labeling @xmath177 of the leaves in a ) as @xmath186 , we restrict the above intersection to the leaves @xmath186 of @xmath145 , so that @xmath187^c \subseteq \bigcap_{j=1}^{\epsilon_1 | \lambda | } { \ensuremath{\operatorname{tree}}}^+ [ v_j]^c.\ ] ] since the events @xmath188 $ ] are mutually independent , @xmath189^c),\ ] ] implying the claim .
alternatively , suppose that b ) holds .
now we claim that @xmath190)]^{\frac{1}{2 } \epsilon_2 | \lambda |},\ ] ] where @xmath146 is a good special 2-point of @xmath145 and @xmath155 is one of @xmath146 s neighbors .
this claim also follows from a string of inclusions .
first , @xmath191 if @xmath166 are adjacent and @xmath146 is a good special 2-point of @xmath145 , then @xmath192^c.\ ] ] as with the proof of the previous claim , we label @xmath193 of the pairs of vertices given in b ) as @xmath194
. then @xmath195^c.\ ] ] the second claim follows from the mutual independence of the events @xmath196 $ ] .
the two claims imply ( [ eqlambda ] ) for @xmath197 by taking @xmath198)^{\epsilon_1 } , ( 1 - \mu_{\theta } ( { \ensuremath{\operatorname{tree}}}^+ [ v , w])^{\frac{1}{2 } \epsilon_2 } \right\},\ ] ] and using lemma [ app : lemmaa3 ] of appendix [ app : galtonwatson ] .
if @xmath175 , suppose the only vertex in @xmath170 is 0 , a distinguished vertex .
then @xmath199 \big ) \\ & \leq & 1-\mu_\theta ( { \ensuremath{\operatorname{tree}}}^+[a_1]),\end{aligned}\ ] ] where @xmath200 are the neighbors of 0 and @xmath201 $ ] is defined as in definition [ k-1-arytrees ] with @xmath145 the tree containing only vertices 0 and @xmath202 .
then ( [ eqlambda ] ) follows in this case by taking @xmath203)$ ] and using equation ( [ taugoesto1 ] ) and lemma [ app : lemmaa1 ] .
this completes the proof .
we study the connected components of @xmath204 as a subgraph of @xmath47 , and show that if @xmath4 is close enough to 1 these connected components are finite almost surely .
we will show that each of these finite connected components of @xmath1 vertices shrinks and is eliminated in finite time leading to fixation of all vertices to + 1 . for any @xmath205
, @xmath206 is the connected component of @xmath5 in @xmath207 : @xmath206 is the set of vertices @xmath208 s.t . @xmath209 , i.e. , there exists a path @xmath210 in @xmath43 with every @xmath211 .
[ finitecomponents ] given @xmath6 , there exists @xmath71 such that for @xmath72 , @xmath204 is a union of almost surely finite connected components .
it suffices to show that @xmath212 is finite almost surely , where 0 is a distinguished vertex in @xmath47 . since @xmath213 < \infty \right .
\right.$ ] implies @xmath214 a.s . , it suffices to show @xmath215
< \infty \right . \right.$ ] .
let @xmath216 represent any site self - avoiding path in @xmath43 of length @xmath217 starting at 0 , then by standard arguments @xmath218 \right .
\right . \leq \sum_{n = 0}^{\infty } \sum_{\gamma_n , | \gamma_n | = n } \mathbb{p}(\gamma_n \in d_0),\ ] ] where by @xmath219 we mean that all the vertices of @xmath216 belong to @xmath212 . to show the sum is finite we need to bound @xmath220
. suppose the vertex set of @xmath216 is @xmath221 , where for each @xmath222 , @xmath223 is a nonempty subset of @xmath224 for some @xmath225 with the @xmath226 distinct .
we now apply proposition [ propindept ] to @xmath223 in each of the layers @xmath224 , which are isomorphic to @xmath22 .
this shows that for any @xmath227 such that for @xmath172 , @xmath228 since the @xmath223 are subsets of distinct levels @xmath224 of @xmath47 , the events @xmath229 are mutually independent .
therefore for @xmath172 , @xmath230 equation ( [ finitesum ] ) and the above bound on @xmath231 imply @xmath232 \right .
\right . &
\leq & \sum_{n = 0}^{\infty } \lambda^n \sum_{\gamma_n , | \gamma_n | = n } 1 \\ & = & \sum_{n = 0}^{\infty } \rho ( n ) \lambda^n,\end{aligned}\ ] ] where @xmath233 is the number of self - avoiding paths of length @xmath234 starting at 0 .
it is easy to see that @xmath235 thus @xmath236 \right . \right . &
\leq & ( k + 2 ) \sum_{n = 0}^{\infty } ( k + 1)^{n - 1 } \lambda^n .\end{aligned}\ ] ] the proof is finished by choosing @xmath237 for @xmath238 .
taking @xmath33 as in proposition [ finitecomponents ] , @xmath239 is a union of almost surely finite connected components : @xmath240 where the @xmath241 s are nonempty , disjoint and almost surely finite with @xmath242 for some @xmath243 .
fix any @xmath114 ; it suffices to show that @xmath241 is eliminated by the dynamics in finite time . by this
we mean that there exists @xmath244 such that for any @xmath245 , and so the droplet @xmath241 fixates to @xmath123 .
we proceed to show this . for any set @xmath246 ,
let @xmath247 @xmath248 so @xmath249 is stable with respect to the dynamics and for any @xmath250 , @xmath251 .
since @xmath241 is finite it contains a longest path , @xmath252 .
since @xmath253 can not be extended to a longer path , @xmath137 must have @xmath254 neighbors in @xmath255 . when @xmath137 s clock first rings , @xmath137 flips to @xmath123 and fixates for all later times .
this argument can be extended to show @xmath241 is eliminated ( i.e. , the @xmath1 vertices are all flipped to @xmath0 ) by the dynamics in finite time as follows .
consider the set of vertices in @xmath241 which have not yet flipped to @xmath123 by some time @xmath256 , and take @xmath256 to infinity .
suppose this limiting set is nonempty .
since this set is finite , it contains a longest path @xmath257 .
but now @xmath254 of @xmath258 s neighbors have spin @xmath123 as @xmath259 , implying that @xmath258 had no clock rings in @xmath260 for some finite @xmath145 .
this event has zero probability of occurring , which contradicts the supposition of a nonempty limit set .
the proof of theorem [ theorem1 ] for @xmath261 and 4 is slightly different than for @xmath102 , since for @xmath261 ( respectively , @xmath262 ) the @xmath112 s of definition [ t+ls ] are not stable with respect to the dynamics : each vertex @xmath263 has 2 ( resp . , 3 ) neighbors of spin @xmath0 , which is always less than a strict majority .
the proof for @xmath261 or 4 requires a different decomposition of the space @xmath43 and definition of stable subsets . with this purpose in mind
, we express @xmath43 as @xmath264 where @xmath265 ( see equation ( [ sinfinity ] ) for a comparison ) .
we call a vertex @xmath266 or its partner in @xmath267 , @xmath268 , * doubly open * if both @xmath269 and @xmath270 ; this occurs with probability @xmath271 .
we proceed by defining a set of fixed vertices in @xmath43 in the spirit of section 2.1 .
for @xmath114 fixed , let @xmath272 be the union of all subgraphs @xmath116 of @xmath267 that are isomorphic to @xmath273 such that @xmath274 , @xmath5 is doubly open .
it is easy to see that @xmath275 is stable for @xmath261 or 4 with respect to the dynamics on @xmath43 .
let @xmath276 denote the union of @xmath275 across all levels @xmath267 , i.e. , @xmath277 where @xmath278 .
we map one independent percolation model , @xmath279 on @xmath43 with parameter @xmath4 , to another one , @xmath280 on @xmath43 with parameter @xmath271 , by defining @xmath281 ( resp . ,
@xmath1 ) if @xmath282 is doubly open ( resp . , is not doubly open ) .
propositions [ propindept ] and [ finitecomponents ] applied to @xmath283 imply that proposition [ finitecomponents ] with @xmath113 replaced by @xmath276 is valid for the @xmath284 percolation model .
the rest of the proof proceeds as in the case for @xmath125 .
the proof proceeds analogously to that of theorem [ theorem1 ] , except that the conclusion of proposition [ finitecomponents ] , that @xmath285 almost surely has no infinite components ( for @xmath4 close to @xmath20 ) , is replaced by an analogous result for @xmath286 with an appropriately defined @xmath287 .
we next specify a choice of @xmath287 for each of our graphs @xmath2 and leave further details ( which are straightforward given the proof of proposition [ finitecomponents ] ) to the reader . for @xmath288 with _ any _ @xmath44
, we simply label @xmath289 ( see definition [ t+l ] ) . for @xmath290 ,
@xmath287 depends on @xmath6 like it did for @xmath291 - i.e. , for @xmath292 , we take @xmath293 and for @xmath261 or @xmath294 we take @xmath295 for @xmath296 or @xmath297 with @xmath102 , we take @xmath298 for @xmath296 or @xmath297 with @xmath261 or @xmath294 , the choice of @xmath287 depends on whether @xmath299 is even or odd since in the odd case the layers can not be evenly paired .
if @xmath299 is even , then we take @xmath300 for @xmath299 odd ( and @xmath301 ) , we pair off the first @xmath302 layers and then use the final @xmath303 layers to define @xmath304 in which the use of doubly open sites for @xmath305 is replaced by * triply open * sites ; then we take @xmath306
the goal of this section is to show that the quantity @xmath307)^{\epsilon_1 } , ( 1 - \mu_{\theta } ( { \ensuremath{\operatorname{tree}}}^+ [ v , w])^{\frac{1}{2 } \epsilon_2 } \right\},\ ] ] which appears at the end of the proof of proposition [ propindept ] , converges to @xmath308 as @xmath309 . here
@xmath146 is a leaf of a subtree @xmath145 of @xmath22 , @xmath166 is a pair of adjacent vertices of @xmath145 such that @xmath146 is a good 2-point ( as in definition [ k-1-arytrees ] ) , and @xmath310 are fixed constants .
for this purpose we consider independent site percolation on @xmath22 and let @xmath311 denote the @xmath6 neighbors of 0 , a distinguished vertex in @xmath22 .
we associate to each @xmath312 a tree @xmath313 $ ] ( see definition [ treegraphs ] ) , for @xmath314 see figure [ fig : app1 ] .
-ary tree with labeled vertices and branches ]
let @xmath145 be a subtree of @xmath22 such that @xmath145 contains @xmath315 and @xmath308 is a leaf of @xmath145 see figure [ fig : app2 ]
. is a subtree of @xmath22 and 0 is a leaf of @xmath145 ] let @xmath316 be one of the neighbors of @xmath317 ( other than 0 ) and @xmath318 be a subtree of @xmath22 containing @xmath319 and @xmath315 ( but not @xmath320 ) such that @xmath317 is a good 2-point of @xmath318 see figure [ fig : app3 ] . is a subtree of @xmath22 that contains @xmath319 and @xmath315 such that @xmath317 is a good 2-point of @xmath318 ] we consider the events @xmath322 $ ] with respect to @xmath145 and @xmath323 $ ] with respect to @xmath318 ( see definition [ k-1-arytrees ] ) and estimate @xmath324)$ ] , @xmath325)$ ] by analyzing a related galton - watson process . for any vertex @xmath326 ,
let @xmath327 denote the @xmath0 spin cluster of @xmath146 , that is , @xmath327 is the set of vertices @xmath328 in @xmath22 such that the path from @xmath146 to @xmath328 ( including @xmath146 and @xmath328 ) includes only vertices @xmath155 , with @xmath329 .
let @xmath330 \cap c ( a_1 ) : d(a_1 , v ) = n|,\ ] ] where @xmath331 represents the graph distance(i.e .
, the minimum number of edges ) between @xmath315 and @xmath146 .
@xmath332 if and only if @xmath333 , and in general @xmath334 is the number of vertices in @xmath335 $ ] at distance @xmath336 from @xmath315 that are in @xmath315 s @xmath0 spin cluster .
@xmath334 is a galton - watson branching process with offspring distribution @xmath337 .
let @xmath338 $ ] denote a tree with root @xmath5 , such that @xmath5 has coordination number @xmath339 and all the other vertices have coordination number @xmath340 .
the following definition is close to that of @xmath341 $ ] ( see definition [ k-1-arytrees ] and figure [ fig : mainpart3 ] ) , except that here the @xmath161-ary tree in question is rooted .
_ * random rooted @xmath144-ary trees of spin @xmath0 * _ consider two vertices @xmath342 such that @xmath147 is a neighbor of @xmath146 .
let @xmath343 $ ] denote the event that there exists a subgraph @xmath116 of @xmath22 isomorphic to @xmath344 $ ] which contains @xmath146 and is contained in @xmath345 $ ] , such that for all @xmath346 .
+ consider three vertices @xmath136 such that @xmath5 and @xmath137 are neighbors of @xmath138 .
let @xmath347 $ ] be the event that there exists a subgraph @xmath116 of @xmath22 isomorphic to @xmath338 $ ] which contains @xmath138 and is contained in @xmath139 $ ] ( see figure [ fig : mainpart2 ] ) , such that for all @xmath346 .
define @xmath348 as @xmath349).\ ] ] @xmath350 $ ] and @xmath351 $ ] are independent for @xmath352 by construction .
the event @xmath322 $ ] is equivalent to the spin at @xmath308 being @xmath0 and the vertices @xmath353 being the roots of @xmath161-ary trees of spin @xmath0 , so that @xmath354 ) = \theta \ , \tau(\theta)^{k -1 } .\ ] ] [ app : lemmaa1 ] @xmath355 as @xmath309 .
the proof is a consequence of proposition 5.30 from @xcite ( about occurrence of @xmath356-ary subtrees in galton - watson processes ) .
define @xmath357 as @xmath358).\ ] ] the event @xmath323 $ ] is equivalent to @xmath359 \cap { \ensuremath{\operatorname{part}}}_{b , 0}^+ [ a_2 ] \}$ ] , so that , by the independence of the events @xmath360 $ ] and @xmath361 $ ] , @xmath362 ) = \tilde{\tau}(\theta)^2.\ ] ] [ app : lemmaa2 ] @xmath363 as @xmath309 .
this result follows as in the proof of lemma [ app : lemmaa1 ] .
equations ( [ taugoesto1 ] ) and ( [ tautildegoesto1 ] ) imply that @xmath324)$ ] and @xmath364)$ ] converge to @xmath20 as @xmath309 , which immediately implies : [ app : lemmaa3 ] @xmath365 as @xmath309 .
let @xmath145 be a finite tree with @xmath234 vertices and maximal coordination number @xmath366 .
@xmath367 of @xmath145 s vertices are labeled special , such that all of @xmath145 s leaves are special vertices .
we remark that in section [ sec : main ] , we start with @xmath368 special vertices in @xmath22 and then @xmath145 is the minimal subtree of @xmath22 that contains all the special vertices .
[ grimmet ] let @xmath369 be the number of leaves in @xmath145 , @xmath370 the number of @xmath371-points ( vertices with exactly two edges in @xmath145 ) , @xmath372 , @xmath373 the number of @xmath6-points ( vertices with exactly @xmath6 edges in @xmath145 ) ; @xmath374 .
then @xmath375 for @xmath376 .
the proof can be found , for example , as part of theorem 8.1 in @xcite .
recall that a _ good _ 2-point in @xmath145 is a 2-point both of whose neighbors are 2-points .
2-point is a 2-point that is not a good 2-point
see figure [ fig : app4 ] .
[ geometriclemma ] there exist constants @xmath377 , depending only on @xmath6 , such that either : @xmath378 , and/or there are at least @xmath379 special good 2-points . by lemma [ grimmet ]
@xmath380 and since @xmath381 , @xmath382 thus either @xmath383 or @xmath384 . in the first case , since @xmath385 @xmath386 and letting @xmath387 gives a ) . in the second case @xmath384 , and if a ) is not valid with @xmath387 , then @xmath388 . to prove b ) we need to count the various types of special vertices in @xmath145 .
the set of special vertices is comprised of : * special good 2-points ; let _ good _ denote the set of such vertices , * special bad 2-points ; let _ bad _ denote the set of such vertices , * special leaves , special 3-points , @xmath372 , special @xmath6-points ; let _ other _ denote the set of such vertices . since @xmath389 , we need to upper bound @xmath390 and @xmath391 .
by lemma [ grimmet ] , @xmath392 now @xmath393 and it is easy to see that the latter is upper bounded by @xmath394 .
thus by lemma [ grimmet ] , @xmath395 since @xmath396 .
thus @xmath397 we let @xmath398 and so @xmath399 .
consider site percolation on @xmath22 distributed according to the product measure @xmath400 with @xmath401 let @xmath145 be a finite subtree of @xmath22 with @xmath402 of its vertices labeled special , such that all the leaves are special .
as in lemma [ geometriclemma ] , in the following lemma @xmath111 and @xmath403 are strictly positive , finite and depend only on @xmath6 . for the events
@xmath149 $ ] and @xmath156 $ ] , see definition [ k-1-arytrees ] .
[ disjointlemma ] * disjoint events * for each such tree @xmath145 , one or both of the following is valid : there are at least @xmath404 leaves @xmath146 in @xmath145 , with the events @xmath178 mutually independent , and/or there are at least @xmath179 edges having endpoints @xmath180 in @xmath145 with @xmath146 a good special 2-point , and the events @xmath181 mutually independent .
lemma [ disjointlemma].a follows from lemma [ geometriclemma].a , since for each of the @xmath404 leaves of @xmath145 we can define an event @xmath149 $ ] , and these events depend on the spins of disjoint sets of vertices and are therefore mutually independent . otherwise , by lemma [ geometriclemma].b there are at least @xmath405 good special 2-points in @xmath145 .
these are arranged into @xmath406 nonempty maximal chains of adjacent vertices along @xmath145 .
we order the chains and let @xmath407 denote the number of vertices in the i^th^ chain , for @xmath408 ; @xmath409 and @xmath410 .
we also order the @xmath411 good special 2-points , @xmath412 , so that they are consecutively ordered in each chain .
suppose @xmath413 for some @xmath114 , and the good special 2-point in this chain is @xmath414 .
let @xmath415 be one of @xmath414 s neighbors in @xmath145 and consider the event @xmath416 $ ] . if @xmath417 , the i^th^ chain contains two adjacent special points @xmath418 and we consider the event @xmath419 $ ] .
generally for the i^th^ chain , we pair adjacent good special 2-points ( other than the last if @xmath407 is odd ) so as to consider @xmath420 events @xmath421 $ ] , where the last event is @xmath422 $ ] if @xmath407 is odd ; these events involve disjoint sets of vertices and are therefore independent .
thus in total we can construct @xmath423 mutually independent events .
* acknowledgments * : the research reported in this paper was supported in part by nsf grants 0ise-0730136 and dms-1007524 .
thanks the institute of mathematical sciences at nyu
shanghai for support .
the authors thank an anonymous referee for carefully reading the paper and making several useful suggestions .
1 nanda , s. , newman , c. m. , stein , d. l. : dynamics of ising spins systems at zero temperature . in : _ on dobrushin s way ( from probability theory to statistical mechanics ) _
r. milnos , s. shlosman , y. suhov , eds . , am .
( 2 ) * 198 * , pp .
183 - 194 ( 2000 ) | we study a continuous time markov process whose state space consists of an assignment of @xmath0 or @xmath1 to each vertex of a graph @xmath2 .
the graphs that we treat are related to homogeneous trees of degree @xmath3 , such as finite or infinite stacks of such trees .
the initial spin configuration is chosen from a bernoulli product measure with density @xmath4 of @xmath0 spins .
the system evolves according to an agreement inducing dynamics : each vertex , at rate 1 , changes its spin value to agree with the majority of its neighbors .
we study the long time behavior of this system and prove that , if @xmath4 is close enough to 1 , the system reaches fixation to consensus .
the geometric percolation - type arguments introduced here may be of independent interest . |
the spectroscopic studies to determine form factors of the @xmath15 semi - leptonic decays , @xmath16 ( @xmath17 ) , are of importance both in studying low energy properties of the strong interaction in terms of effective theories @xcite , and also in studying fundamental interactions . in our previous work @xcite , we reported a result testing the exotic couplings in @xmath2 ( @xmath3 ) decay , showing the non existence of scalar and tensor interactions , contradicting the current world average adopted by particle data group @xcite . in the present work , the @xmath0 ( @xmath1 ) events , which were collected simultaneously , were analyzed to determine the ratio of the @xmath1 and @xmath3 decay widths @xmath6 .
this quantity is one of the most important observables to evaluate the @xmath13 form factors .
assuming that only the v@xmath18a interaction contributes to the @xmath19 decay , the decay amplitude can be described by two dimensionless form factors , @xmath20 and @xmath21 , which are functions of the momentum transfered to the leptons @xmath22=@xmath23 where @xmath24 and @xmath25 are the four momenta of the @xmath26 and @xmath27 , respectively .
they are given as , @xmath28 , \nonumber \\ f_{0}(q^{2})&=&f_{0}(0)[1+\lambda_{0}(q / m_{\pi})^{2 } ] .
\nonumber\end{aligned}\ ] ] assuming @xmath11-@xmath12 universality , the form factors between @xmath29 and @xmath30 decays are identical and @xmath6 can be written as @xcite , @xmath31 this equation can not determine the @xmath32 and @xmath10 parameters uniquely but simply fixes a relationship between them .
however , if the @xmath32 value derived from @xmath30 data analyses is assumed , the @xmath10 parameter can be obtained from eq .
( [ kihon1 ] ) . in this analysis ,
our @xmath30 result for the @xmath32 parameter , @xmath33 , was employed @xcite .
it should be noted that the @xmath10 parameter can be also determined by studying the dalitz plot distribution of @xmath29 decay or the muon polarization in @xmath29 decay @xcite .
the determination of the @xmath10 parameter from the @xmath4 ratio is based on the assumption of @xmath11-@xmath12 universality .
this is not the case for the other two methods .
thus , @xmath11-@xmath12 universality in @xmath19 decay can be tested by comparing the @xmath10 parameter obtained from a branching ratio measurement @xmath34 with that obtained from a dalitz plot measurement @xmath35 and/or @xmath36 polarization measurement @xmath37 . in this case , however , a precise acceptance function has to be carefully determined over the whole dalitz space for the @xmath29 dalitz plot analysis . in this letter
, we present a new precise measurement of the @xmath4 branching ratio and the @xmath10 parameter .
the experiment used a stopped @xmath15 beam in conjunction with a 12-sector iron - core superconducting toroidal spectrometer @xcite .
@xmath4 was obtained by measuring the ratio of the number of accepted @xmath38 and @xmath30 events corrected for detector acceptance .
the experiment was performed at the kek 12 gev proton synchrotron .
the experimental apparatus was constructed for a t - violation search in @xmath38 decay @xcite . a schematic cross sectional side view of the detector is shown in fig .
[ fg : side ] .
we detected both of the @xmath39 decay gammas which enabled us to reconstruct the complete kinematics of a @xmath19 event in contrast to the previous experiment @xcite ; this increased the reliability of the event selection . because of the rotational symmetry of the 12 identical gaps in the spectrometer and the large directional acceptance of the @xmath39 detector @xcite , distortions due to detector acceptance were drastically reduced .
moreover , the similarity of the @xmath29 and @xmath40 kinematics reduced the systematic error due to the imperfect reproducibility of the experimental conditions in the simulation of the ratio of the calculated acceptance for the @xmath29 and @xmath40 decays .
the measurement was carried out for two values of the central magnetic field strength , b=0.65 and 0.90 t , yielding a consistency check with regard to the spectrometer acceptance and energy loss estimation in the target .
@xmath29 ( @xmath40 ) events were identified by analyzing the @xmath36 ( @xmath41 ) momentum with the spectrometer and detecting the two photons in the csi(tl ) calorimeter .
charged particles from the target were tracked and momentum - analyzed using multi - wire proportional chambers ( mwpcs ) at the entrance ( c2 ) and exit ( c3 and c4 ) of the magnet gap , as well as by the active target and an array of ring counters @xcite surrounding the target .
particle identification between the @xmath42s and @xmath41s was carried out by time - of - flight ( tof ) between tof1 and tof2 scintillation counters .
tof1 surrounds the active target and tof2 is located at the exit of the spectrometer .
the @xmath39 detector , an assembly of 768 csi(tl ) crystals , covers 75% of the total solid angle .
since photons produce electromagnetic showers , their energy was shared among several crystals .
the photon energy and hit position were obtained by summing the energy deposits and energy - weighted centroid , respectively .
timing information from each module was used to identify a photon cluster and to suppress accidental backgrounds due to beam particles .
the two - photon invariant mass ( @xmath43 ) and the @xmath39 energy and direction were obtained from the photon momentum vectors .
@xmath38 and @xmath30 decays at rest were selected by the following procedure , which is similar to our @xmath30 study .
the @xmath15 decay time , defined as the charged lepton signal at the tof1 counter , was required to be more than 5 ns later than the @xmath15 arrival time measured by the @xmath44erenkov counter to remove in - flight @xmath15 decays .
events with @xmath45 decays in - flight and scattering of charged particles from the magnet pole faces were eliminated by a track consistency cut in the ring counters .
events with two clusters in the csi(tl ) calorimeter were selected as @xmath39 decays and events with other cluster numbers were rejected .
the acceptance cut on the invariant mass was 50@xmath46140 mev/_c_@xmath47 .
requirements for the charged particle momentum corrected for the energy loss in the target ( @xmath48 175 mev/@xmath49 ) plus opening angle between a charged particle and the @xmath39 ( @xmath50 154@xmath51 ) removed the @xmath52 events .
the mass squared ( @xmath53 ) of the charged particles , obtained from the tof and momentum , are shown in fig .
[ fg : massgmc ] integrated over the entire momentum region .
the @xmath36 and @xmath41 selection regions are also indicated in the figure .
the timing resolution of @xmath54270 ps provides a good separation between the @xmath29 and @xmath40 events .
it should be emphasized that only these requirements were imposed for the @xmath38 and @xmath40 event extraction .
the numbers of good events after those cuts mentioned above are 12882(10704 ) and 23122(16850 ) at 0.65(0.90 ) t for @xmath29 and @xmath30 decays , respectively , where small fractions due to the background events are still included . fig .
[ fg:65kl3 ] and [ fg:90kl3 ] show the @xmath55 and @xmath30 spectra for the setting of b=0.65 and 0.90 t , respectively .
in order to obtain the detector acceptance and estimate the background fraction , the monte carlo simulation was carried out for both the charged particle measurement by the spectrometer and the @xmath39 measurement by the csi(tl ) detector .
the initial dalitz distributions were generated with the values of @xmath56 and the current world average @xmath57 , while the dalitz distribution of @xmath30 decay is insensitive to the @xmath58 parameter .
the radiative corrections were taken into account by following the ginsberg procedure @xcite .
the simulation data were analyzed in the same manner as the experimental sample .
the dotted lines in fig .
[ fg:65kl3 ] and fig .
[ fg:90kl3 ] show the simulation spectra for the setting of b=0.65 and 0.90 t , respectively , using the assumed @xmath32 and @xmath10 parameters .
these spectra are normalized so that the total number of events is the same as the experimental one . here , it is to be noted that much higher statistical accuracy for the monte carlo simulation is necessary to determine the form factors from the @xmath38 dalitz plot analysis . by using the simulation data , the detector acceptance ( @xmath59 )
is calculated as , @xmath60 where @xmath61 and @xmath62 are the number of events accepted by our event selection requirements and the number of generated @xmath15 events , respectively .
@xmath63 denotes the detector acceptance at the assumed form factor for the production of the simulation samples .
they were @xmath64 , @xmath65 for the setting of b=0.65 t and @xmath66 , @xmath67 for the setting of b=0.90 t. the detector acceptance depends only slightly on the form factor ( @xmath68 , the shape of the dalitz distribution ) , therefore a correction factor @xmath69 is introduced .
@xmath70 was calculated by taking into account the event selection requirements such as @xmath71 , @xmath72 , and @xmath73 cuts . as shown in fig .
[ fg : lacc](a , b ) , the @xmath70 distribution for @xmath38 decay was obtained as a contour plot in the ( @xmath74,@xmath58 ) space , while @xmath70 for @xmath30 decay was obtained as a function of the @xmath32 parameter .
@xmath53 spectra were calculated by assuming a tof resolution of @xmath75=270 ps .
they are shown as dotted lines in fig .
[ fg : massgmc ] .
since the charged particles were identified by @xmath53 measurement , the choice of the tof response function was important for the determination of the background contamination .
we took into account @xmath52 , @xmath30 , @xmath76 decays for the @xmath29 background and @xmath52 , @xmath29 , @xmath77 decays for the @xmath30 background .
the background fraction depends on the magnetic setting of the spectrometer , as summarized in table [ tb : bgke3 ] .
the most dominant background is due to @xmath52 decay in - flight .
the accidental background fraction due to beam particles was estimated to be 0.3% for both @xmath38 and @xmath30 samples .
.background fractions included in the @xmath38 and @xmath30 samples .
[ cols="^ , < , > , > " , ]
the ratio of the @xmath0 ( @xmath1 ) and @xmath2 ( @xmath3 ) decay widths , @xmath4 , has been measured for stopped positive kaons . assuming @xmath11-@xmath12 universality in @xmath13 decay , the coefficient of the @xmath8 dependent term of the @xmath9 form factor was determined from the measured @xmath4 ratio . in contrast to the previous experiments , a large detector acceptance and its symmetrical structure enabled us to reduce the statistical errors while suppressing the systematic errors .
our results are + @xmath78 the @xmath10 parameter obtained from the present work is consistent with that from recent @xmath14 dalitz plot analyses @xcite , which supports the validity of @xmath11-@xmath12 universality in @xmath13 decay .
if the assumption of @xmath11-@xmath12 universality is removed , @xmath6 can be written as @xcite , @xmath79 ^ 2 \nonumber \\ & \times&(0.6457 + 2.2342\lambda_+^{\mu}-2.3873\lambda_+^{e}+1.5646\lambda_0 ) , \label{eqnot } \end{aligned}\ ] ] where @xmath80 is weak coupling constant for the lepton current . substituting + @xmath81 ( present result with no @xmath11-@xmath12 universality assumption ) , @xmath82 ( the current world average ) , @xmath83 ( our @xmath30 result ) , and @xmath84 ( error weighted average of ref .
@xcite and @xcite ) into eq .
( [ eqnot ] ) , @xmath85 is derived to be @xmath86 which is consistent with unity within the experimental error . from the viewpoint of the theoretical framework ,
the @xmath13 form factors are of importance in studying low energy properties of the strong interaction in terms of effective theories .
the predicted values of the @xmath10 parameter are @xmath87 @xcite and @xmath88 @xcite .
the present result is consistent with the former and inconsistent with the latter .
also , the @xmath891/2 rule leads to identical @xmath10 parameters between @xmath13 and @xmath90 decays @xcite .
the @xmath10 parameter from the @xmath90 analyses has been determined to be @xmath91 @xcite , which is consistent with the present result .
this work has been supported in japan by a grant - in - aid from the ministry of education , science , sports and culture , and by jsps ; in russia by the ministry of science and technology , and by the russian foundation for basic research ; in canada by nserc and ipp , and by the triumf infrastructure support provided under its nrc contribution ; in korea by bsri - moe and kosef ; in the u.s.a by nsf and doe ; and in taiwan by nsc .
the authors gratefully acknowledge the excellent support received from the kek staff .
00 j. gasser and h. leutwyler , nucl . phys .
* b250 * , 517 ( 1985 ) .
kazakov , v.n .
pervushin , and m.k .
volkov , phys .
b64 * , 201 ( 1976 ) .
s. shimizu _ et al .
_ , phys . lett . *
b495 * , 33 ( 2000 ) .
particle data group , eur .
j. * c15 * , review of particle physics ( 2000 ) .
fearing , e. fischbach , and j. smith , phys .
d2 * , 542 ( 1970 ) ; phys .
* 24 * , 189 ( 1970 ) .
j. imazato _
et al . _ , in _ proceedings of the 11th international conference on magnet technology _
( elsevier applied science , london , 1990 ) , p.366 .
et al . _ ,
. lett . * 83 * , 4253 ( 1999 ) .
j. heintze _
_ , phys . lett . *
b70 * , 482 ( 1977 ) .
d.v . dementyev _
et al . _ ,
nucl . instr .
method * a440 * , 151 ( 2000 ) .
a.p . ivashkin _
et al . _ ,
nucl . instr .
method * a394 * , 321 ( 1997 ) .
ginsberg , phys . rev . * 42 * , 1035 ( 1966 ) . l.b .
et al . _ ,
phys . rev . *
155 * , 1505 ( 1967 ) .
r. whitman _
et al . _ ,
phys . rev .
* d21 * , 652 ( 1980 ) .
v.m . artemov _
_ , physics of atomic and nuclei . * 60 * 2023 ( 1997 ) | the ratio of the @xmath0 ( @xmath1 ) and @xmath2 ( @xmath3 ) decay widths , @xmath4 , has been measured with stopped positive kaons . @xmath1 and @xmath3 samples containing 2.4@xmath5 and 4.0@xmath5 events , respectively ,
were analyzed .
the @xmath6 ratio was obtained to be [email protected](stat.)@xmath70.008(syst . ) calculating the detector acceptance by a monte carlo simulation .
the coefficient of the @xmath8 dependent term of the @xmath9 form factor was also determined to be @[email protected](stat.)@xmath70.004(syst . ) with the assumption of @xmath11-@xmath12 universality in @xmath13 decay .
the agreement of our result with the @xmath10 value obtained from @xmath14 dalitz plot analyses supports the validity of the @xmath11-@xmath12 universality . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , and |
ams-02 2014 measurements of cosmic rays ( cr ) leptons @xcite , which have confirmed the rise of the positron fraction for kinetic energy above @xmath3 , up to @xmath4 , have stimulated different interpretations of this excess of positrons as primary evidence of dark matter ( dm ) annihilation .
the interpretation of dm as a new source of positrons , to explain the departure from the pure secondary positron fraction , can be tested with other complementary and independent indirect measurements , as the cosmic microwave background ( cmb ) anisotropies .
in fact , if dm particles self - annihilate at a sufficient rate , the expected signal would be directly sensitive to their thermally averaged cross section and it could have drawn an imprint on cmb temperature and polarization anisotropies .
the aim of the present letter is to discuss the link between the dm annihilation cross sections at freeze - out , recombination and in the milky way galactic halo , which is essential to compare different indirect constraints on dm itself .
the three different physical quantities are defined by the environment in three different epochs .
the relation between them is not trivial , not a simply decreasing function and so it must be carefully analyzed to put coherent constraints on dm properties . once defined a consistent framework , one can compare information from cr physics with cmb observations .
+ the need of a high thermally averaged annihilation cross section , not purely thermal , _
i.e. _ @xmath5 , comes from the above mentioned observation of a huge excess in the cr positron fraction , in contrast with the expected behavior of secondaries produced in the interstellar medium ( ism ) . in order to describe this experimental evidence
, a very high annihilation cross section has to be invoked , compared with typical expectation for a s - wave annihilating thermal relic matching the observed dm abundance @xcite .
+ the main way to interpret the positron excess is the sommerfeld enhancement @xcite , a non - perturbative quantum effect which modifies the annihilation cross section in the regime of small relative velocity of the annihilating particles and in presence of an effectively long - range force between them . indeed , this well - known quantum mechanical effect can occur in dm annihilations in the galactic halo , if the two annihilating particles exchange an interaction mediated by a force carrier . from a feynman diagram point of view , the weakly interacting massive particles ( wimp ) interact with the new boson through a multi - box diagram , depicted in fig . [ fig : feynman ] , annihilate into @xmath6 , and the decay of @xmath7 produces more light leptons than expected .
the structure of the paper is the following : in section [ sec : two ] the basic setup of the sommerfeld enhancement and the relation between the galactic halo cross section and the one at recombination are briefly reviewed ; in section [ sec : three ] we report a collection of cr best fits for tev - ish dm candidates capable to reproduce the observed positron excess and we discuss the main sources of uncertainties ( from cr physics and the dm sector ) , in order to compare predictions with cmb constraints and obtain some general remarks in section [ sec : conclusion ] .
the sommerfeld enhancement is fundamental to interpret indirect dm searches .
the thermally averaged dm annihilation cross section ( at any time ) can be generally decomponed into powers of the velocity @xmath8 @xcite : @xmath9 where @xmath8 is the relative velocity of the annihilating particles , that is @xmath10 , @xmath11 is the constant s - wave term , @xmath12 is the p - wave term and @xmath13 the sommerfeld term where @xmath14 stands for a proper asymptotic ( cut - off ) value for the low velocity regime .
the quadratic and quartic terms in eq .
are commonly neglected at a freeze - out description , whereas the so called sommerfeld term is suppressed in the relativistic velocities regime @xcite .
+ the enhancement of the s - wave for these velocities is defined by the ratio of the masses of the dm candidate and the sommerfeld boson @xmath7 which provides the boosted annihilation , and by the effective coupling @xmath15 . defining the dimensionless parameter @xmath16 ,
the condition for the enhancement is @xmath17 , that is @xmath18 and @xmath19 .
( @xmath8 in natural unit on the abscissa ) .
see @xcite for insights.,height=226 ] in fig . [
fig : crosssection ] the different contributions to the total dm annihilation cross section are shown according to eq . .
it can be noted that the sommerfeld contribution is a function of the redshift @xmath20 but , after a particular value of @xmath8 ( @xmath21 ) , which is very close to the one expected for our dark halo , the quantum effect reaches a plateau that is approximately 100 times the thermal annihilation cross section , _
i.e. _ @xmath22 . here
one can notice that there should be necessarily a slow velocity limit , given by the finite range of the attractive force mediated by @xmath7 : once the de broglie wavelength of the particle @xmath23 exceeds the range of the interaction @xmath24 , the quantum effect saturates reaching the value @xmath25 , that is a constant and does not depend on @xmath8 @xcite
. moreover , for specific values of @xmath26 and @xmath27 , where @xmath28 is the ratio of @xmath8 to the speed of light @xmath29 , resonant threshold states can be produced ( see fig . [
fig : sommerfeld ] ) which are capable of further boosting @xmath30 @xcite , inducing @xmath31 .
plane @xcite.,height=226 ] after cosmological recombination , dm had a velocity proportional to the inverse of the scale factor of the expanding universe because of its free stream , justifying very small velocity values ; on the other hand , the virialized velocities in a today galactic halo might be at least three order of magnitude greater .
+ the status of the three cross sections in the previously mentioned regimes can be basically summarized as follows : 1 .
_ freeze - out_. during this phase the cross section is usually assumed to be nearly constant , without the general contribution of other even powers of @xmath8 .
this is an absolute minimum from which the annihilation cross section evolves during the cooling of the universe . as shown in fig .
[ fig : crosssection ] , for too relativistic
freeze - out velocities , the p - wave term would enhance the annihilation cross section of one order of magnitude , generating an incompatibility w.r.t . the expected relic density at redshift @xmath32 .
for this regime the relative velocity is : |_z 1000 = _ 0.3 c and the following relation for the cross section holds : ^ _
~310 ^ -26 cm^3 s^-1 .
recombination_. here the annihilation cross section has a greater value than the freeze - out one , _
i.e. _ @xmath33
. this does not imply the relation @xmath34 , because the quantum sommerfeld enhancement saturates after a certain velocity @xmath35 , which is @xmath36 .
so , for @xmath37 the quantum sommerfeld boost @xmath38 does not increase the annihilation cross section further .
this is the annihilation cross section measured by cmb experiments . in this regime
the relative velocity of the dm particles is very low : |_600<z<1000 = _ ~10 ^ -(86)c and the annihilation cross section is clearly dominated by the sommerfeld enhancement : ^ _ = s__. 3 .
_ galactic halo_. as for recombination epoch , the dominant term is the sommerfeld one . in our galaxy
the velocity of the dm particles is supposed to be of the order of ( @xmath39 , which implies @xmath40 . in the galaxy
the velocity regime is : |_z=0 = _ ~510 ^ -4c with an annihilating cross section : ^ _ = s__. if we take into account that the current value of @xmath8 in our galaxy does not completely saturate the sommerfeld effect , a conservative numerical factor @xmath41 can be assumed for the ratio @xmath42 between the enhancement at recombination and the one today ( see fig .
[ fig : crosssection ] ) .
the relation between the galactic annihilation cross section measured by ams-02 and the one measured by planck at the recombination can be written as : [ eqn : crosssection ] ^_(12)^_.
the cmb experiments can constrain the dm annihilation cross section from the quantity introduced in refs .
@xcite : p_f _ as a function of the factor @xmath43 that encodes the efficiency of the energy absorption at the recombination .
therfore , another parameter has to be taken into account for the comparison between cr and cmb experiments . in principle @xmath44 , but it has been demonstrated that this can be taken as a constant at @xmath45 @xcite , around the recombination epoch ; it lies in the theoretical range between @xmath46 and @xmath47 , with recombination values generally chosen between @xmath48 and @xmath49 as a function of the annihilation channel ( see refs .
so , for small @xmath43 values the annihilation constraints are relaxed . in refs .
@xcite have been obtained the following bounds for the saturated annihilation cross section and the sommerfeld factor from wmap 5 yr data : @xmath50 for @xmath51 dm candidate and @xmath52 eqs . -
lead to @xmath53 and @xmath54 . + now
, with the latest planck 2015 data @xcite this constraint can be improved of about one order of magnitude ( see in fig .
[ fig : planckannihilation ] the comparison between wmap9 and planck constraints ) , leading , for the previous case , to @xmath55 and @xmath56 . + , @xmath57 , @xmath58 ) , ( @xmath59 , @xmath60 , @xmath61 ) and ( @xmath62 , @xmath63 , @xmath64 ) for @xmath65 , @xmath66 and @xmath67 respectively.,height=226 ] in fig . [
fig : planckannihilation ] the constraints from planck @xcite , obtained with the full temperature data and the inclusion of low- and high-@xmath68 polarization data , are shown . here
, the allowed region of parameters space for charged cr measurements was taken from cholis and hooper @xcite , under the assumption that the cr positron excess was due to pure dm annihilation .
+ from fig .
[ fig : planckannihilation ] it can be noted that , for a tev - ish particle , _
e.g. _ with a mass of @xmath69 , and a _
pessimistic case _ with an ideal absorption efficiency @xmath70 , the annihilation cross section at recombination from planck is @xmath71 , which implies @xmath72 at most . instead , in an _
optimistic case _ with @xmath73 and @xmath74 ,
the constraint for a tev - ish dm becomes @xmath75 .
+ several studies suggest values less than @xmath76 for @xmath43 at recombination , that is a rather optimistic scenario for indirect search constraints @xcite .
consequently , constraints of the order of @xmath77 for a heavy dm candidate can be still in agreement with the boosted cross sections which are necessary to reproduce cr positrons excess @xcite ; as a first step , only resonant sommerfeld boosts , that induce an overall enhancement @xmath78 w.r.t @xmath79 , can be for certain excluded .
it follows that it is interesting and necessary to discuss and update the small rectangle in the right upper corner of fig .
[ fig : planckannihilation ] , related to ams-02 2013 @xcite , fermi @xcite and pamela @xcite data , in light of new cr results and speculations , to avoid falling into misleading conclusions . + in fact , the analysis in @xcite was performed with only 2013 ams-02 data along with pamela , fermi and ams-01 ones .
it must be also noted that fermi all electrons channel @xmath80 is in tension with ams-02 positron fraction , as stressed in @xcite , and it has not been reprocessed and confirmed after the recent pass 8 calibration performed in @xcite .
data ( and consequently of positron fraction and @xmath81 data ) at the ams - days at cern on april 2015 , see https://cds.cern.ch/record/2010841 at @xmath82 . ]
+ furthermore , putting all these cr data together could produce some inconsistencies and issues : since the data used for fits were collected by different instruments , the fits errors are hard to estimate . the most interesting constraints in literature for tev - ish dm candidates , _
i.e. _ @xmath83 , using ams-02 positron data are the following : 1 . from the previously discussed ref .
@xcite it must be noted that the reduced @xmath84 for tev - ish dm fits is much better for fermi data rather than for ams-02 2013 ones .
the good fits , with @xmath85 , for ams-02 data in @xcite , for a dm which annihilates into a pair of intermediate states @xmath7 are @xmath86 with @xmath87 , @xmath57 with @xmath88 , @xmath60 with @xmath59 and @xmath63 with @xmath62 for @xmath89 , @xmath90 and @xmath91 combinations .
the overall result is a dm with a mass up to @xmath69 and an annihilation cross section in the range of few @xmath77 ; 2 . in ref .
@xcite some fits to ams-02 2014 positron fraction data with tev - ish dm are obtained : @xmath92 with @xmath93 for @xmath94 channel , @xmath95 with @xmath96 , with little deviations as a function of the annihilation channels combinations .
some fits are achieved using also fermi - lat data , which have a great uncertainty w.r.t .
ams-02 2014 data .
3 . in ref .
@xcite the privileged regions , using ams-02 2013 data , are about @xmath97 with @xmath98 .
they also used fermi and hess data ; 4 . in @xcite , using ams-02 2013 data , along with pamela and fermi ones , they obtain @xmath97 with @xmath99 ; 5 . in ref .
@xcite , using ams-02 2013 , the results for a sommerfeld boost @xmath100 from a scalar , pseudoscalar or vector particle are approximately the same : @xmath97 with @xmath101 and @xmath102 for the @xmath66 channel , @xmath103 with @xmath104 and @xmath105 for the @xmath106 channel , @xmath107 with @xmath108 and @xmath109 for the @xmath110 channel ; 6 . in ref .
@xcite , from ams-02 2013 positron fraction data the upper limit of @xmath111 is achieved , as a function of the annihilation channels , associated to a @xmath112 .
they also use ams-02 positrons from icrc 2013 , which were not finalized nor official data ; 7 . in ref .
@xcite , the best - fits obtained from ams-02 2014 data are about @xmath113 with @xmath114 , for @xmath115 and @xmath116 channels ; 8 . in ref .
@xcite , the ams-02 2014 data lead to a best - fit via sommerfeld boson that is of the order of @xmath117 with @xmath118 , for @xmath66 and @xmath116 channels .
these studies share common properties and weaknesses .
first of all the allowed and privileged leptonic annihilation channels for dm with mass @xmath119 are generally the muonic and tauonic ones .
also annihilation into quarks @xmath120 @xcite and @xmath121 @xcite are suitable for heavy dm .
these best - fits are usually computed with dm masses @xmath122 , however mass values in the @xmath123 range , the most interesting one for a heavy wimp scenario @xcite , are poorly tested in literature ; on the other hand , some studies are performed for very heavy dm with @xmath124 , which is disfavored by recent astrophysical observations @xcite .
in addition , the greater the mass , the greater the uncertainty of the annihilation scheme and the greater the degrees of freedom to tune the annihilation chain and fit the data .
+ it must be stressed that all recent observations point toward a tev - ish paradigm and suggest to look above @xmath51 for wimp dm masses @xcite ; at the same time too high dm masses @xmath125 could introduce some conflicts and could be not too suitable to fit the ams-02 2014 positron fraction and its supposed flattening , due to the achievement of a maximum of the positron production .
it is more advisable to avoid the @xmath126 annihilation scenarios associated to very high masses and too high annihilation cross sections @xcite .
+ the previous fits have one order of magnitude of span in the annihilation cross section ; the ensemble of the fits prescriptions could be approximately described as a rectangle in the @xmath127 plane : @xmath128 .
this must be translated in the @xmath129 plane : if we take @xmath130 , the most general permitted window becomes @xmath131 .
+ but such a rectangle is very loose and it is based on standard assumptions which can be easily extended .
in fact the constraints from ams-02 positron data greatly relax introducing a dark disk ( dd ) in addition to a dark halo : this ensures dynamical enhancements of the dm annihilation cross section within the high density dd , @xmath132 , without the need of @xmath133 , @xcite . for indirect detection ,
the dd scenario could easily accommodate a large boost factor from local density enhancement in the range @xmath134 , depending on the disk height .
+ in addition , the constraints become less stringent if one consider both dm and pulsars as sources of primary positrons @xcite .
pulsars and dd hypothesis can only relax the dm bounds , enlarging the allowed region in the @xmath129 plane .
+ for what concerns the uncertainties which afflict cr propagation physics , we still do not have a complete and well - posed understanding of the cr lepton problematic : non - standard propagation models may be introduced to account for part or all of the positrons measured in space @xcite ) .
hence the constraints obtained from the positron sector ( positron spectrum and positron fraction ) strongly depend on the underlying cr propagation model and certainly on the ams-02 measurements errors .
thanks to ams-02 unprecedented precision it will be soon possible to fix an almost univocal scheme of propagation of cosmic rays in our galaxy , allowing to correctly compute the effective background for dm indirect searches : in fact , tiny variations of the most significant parameters , such as the diffusive halo thickness , the diffusion coefficient and exponent and the electrons spectral indices , may lead to misleading interpretations of ams-02 data , pointing to an incorrect scenario . in fig .
[ fig : pppc4 ] a qualitative illustration of propagation and measurement uncertainties , performed with pppc4dmid @xcite , show how unstable these fits are : a best - fit point in the @xmath127 plain carries up to 50% uncertainty in the choice of the annihilation cross section and dm mass values . and a set with about half the annihilation cross section ( blue curve ) and the ratio between a reference positron fit with @xmath87 and one with @xmath135 ( purple curve ) .
as long the two curves lie below the dark yellow line , which represent the combination of the two main uncertainties sources , from cr propagation ( the min - max sets span in @xcite ) and from ams-02 errors , the best - fit point can be adjusted in the @xmath127 plain.,height=226 ] from fig .
[ fig : pppc4 ] one argues that dm fits to ams-02 positrons and positron fraction with standard propagation models , up to @xmath136 , suffer large uncertainties , of the order of @xmath137 , if taken individually .
this is the degree of uncertainty from the most precise space experiment which measures cr fluxes : when a fit on pamela , fermi ( or ams-01 , heat , hess ) data is performed , an uncertainty at least three times the ams-02 ones should be addressed : it could imply almost one order of magnitude in the annihilation cross section and more than @xmath51 for tev - ish dm masses .
+ finally , when exploring the dm @xmath127 space , other underlying parameters are fixed , such as the dark matter halo shape and the next - to - leading order ( nlo ) corrections to primary positrons from dm annihilation .
the choice of the dm radial profile is not too significant , whereas full calculations of the nlo and nnlo electroweak ( ew ) corrections may modify the primary dm fluxes up to one order of magnitude @xcite , producing more light final states than expected in lo calculations and allowing lower values of the annihilation cross section : they are relevant for spectra predictions especially when @xmath138 is much larger than the ew scale and ew bremsstrahlung is permitted .
we consider the collection of fits discussed in section [ sec : three ] : applying the constraint @xmath139 , according to the up - to - date calculations derived in @xcite , where @xmath76 is preferred only for the electron channels for non tev - ish dm candidates , an overall cr allowed region is obtained ( fig .
[ fig : planckannihilationrev ] ) , which is something like @xmath140 .
the qualitative fraction of this region permitted by cmb observations is about 9% , 4% and 2% of the total area obtained combining cosmic rays fits , assuming @xmath141 respectively .
+ our analysis is compatible with the one presented in @xcite ; in @xcite the region allowed by planck is @xmath142% .
plane from ams-02/fermi / pamela ( dot - dashed rectangle ) , along with planck 2015 bound ( blue line ) . only a cosmic variance limited ( cvl ) experiment ( black line ) could completely falsify the cr indirect search constraints for tev - ish candidates .
the points corresponding to the best - fits are the ones presented in section [ sec : three ] for @xmath58 .
the rectangles for @xmath74 ( yellow ) and @xmath143 ( purple ) are also drawn.,width=340 ] + planck 2015 cmb data appear not to completely exclude annihilating dm as primary source for ams-02 positron excess , regardless the choice of the recombination efficiency value . from fig .
[ fig : planckannihilationrev ] it emerges that future improvement on cmb measurements @xcite could hardly falsify the dark matter interpretation of cosmic - rays antiparticles , for what concerns tev - ish dark matter candidates .
in addition , fundamental uncertainties from cr propagation physics and from dd hypothesis , pulsars contributions and alternative cr propagation models , not reported in fig . [
fig : planckannihilationrev ] , could almost arbitrarily enlarges the window in the @xmath129 plane , up to @xmath144 particles and down to nearly thermal cross sections .
incoming cr high statistics measurements of leptons and nuclei from ams-02 will allow us to deepen the examination of this important issue , narrowing cr uncertainties and dm properties .
besides that , the information from the antiproton channel is mandatory to put more consistent and coherent constraints on the annihilation capability of the dm candidate , because cr antiprotons have less important backgrounds @xcite , so granting more precise estimations .
a cross check based on a hadron - lepton channels comparison will improve our understanding of dm annihilation and probably enlarge the analysis up to the @xmath144 scale .
we wish to thank fabio finelli and daniela paoletti for useful comments and suggestions on the draft .
mb acknowledge support by the `` asi / inaf agreement 2014 - 024-r.0 for the planck lfi activity of phase e2 '' .
l. accardo _ et al . _ [ ams collaboration ] , phys .
lett . * 113 * , 121101 ( 2014 ) . | in view of the current interest in combining different observations to constraint annihilating wimp dark matter , we examine the relation between the sommerfeld effect at the recombination epoch and in the galactic halo . by considering an up - to - date collection of interpolations of cosmic rays lepton data ( ams-02 2014 , fermi and pamela ) , as dark matter annihilation signals ,
we show that current cosmic rays measurements and recent planck 2015 constraints from cmb anisotropies almost overlap for dark matter masses of the order of few @xmath0 , although great theoretical uncertainties afflict cosmic rays and dark matter descriptions . combining cosmic rays
fits we obtain proper minimal regions allowed by cmb observations , especially for @xmath1 and @xmath2 annihilation channels , once assumed viable values of the efficiency factor for energy absorption at recombination : the results are consistent with those obtained by the planck collaboration but allow a slightly larger overlap between cosmic rays constraints from the lepton sector and cmb .
incoming ams-02 measurements of cosmic rays antiprotons will help to clarify the conundrum .
dark matter , ams-02 , planck |
granular matter is a generic name given to a system composed of macroscopic , athermal particles that have mutual repulsive , dissipative interactions @xcite .
it is an intensely studied field in the physics community given the several distinct behaviors shown by such systems as a consequence of different external conditions imposed on them .
one of such conditions is that which imposes a flow of particles , named granular flow @xcite . within the several granular flow examples ,
the flow around immersed obstacles has received some attention lately @xcite .
one of the objectives of such investigations is to measure the force in the obstacle due to interactions with the flowing grains , the so called granular drag @xcite , analogously to the viscous flow force on an obstacle . on the other hand
, one knows that a viscous fluid flow around an obstacle produces an additional force called lift , which is perpendicular to the flow .
given the analogy between a viscous flow and the granular flow , a lift force should exist in an obstacle immersed in a granular flow under suitable conditions .
however , most investigations focus only on the drag , while the lift studies are restricted only to a few experiments and simulations @xcite . soller and koehler @xcite showed that the lift on a rotating vane inserted in a granular packing scaled with geometric parameters of the system , such as an effective aspect ratio and immersion depth .
the word effective means that the referred quantity , say the immersion depth , is considered taking into account the finite grain size . ding _
et al_. @xcite , dragging an intruder at constant velocity through a granular packing , showed that a lift force is induced on the intruder and it depends on its geometry .
common to both investigations is the fact that the lift arises as a consequence of the hydrostatic nature of the stress in granular packings . given that these experiments were carried on such hydrostatic stress systems , variations of this condition might reveal different aspects of this force .
for instance , is there a lift force in intruders immersed in flows where the stresses are not hydrostatic ?
if so , in what conditions this force arises ?
this paper is aimed at studying the lift force on an obstacle due to a dilute granular flow through numerical simulations .
the approach will be identical to the one used in probing the drag on a cylinder due to a dilute granular flow @xcite .
the argument drawn here to obtain an analytical expression for the lift shows that there is no net lift in such dilute conditions on a circular obstacle .
hence , the obstacle chosen here is an elliptical one .
the dependence of this force on the flow parameters will be obtained and studied numerically .
section [ sec1 ] is reserved for developing the theoretical argument and presenting its predictions . in section [ sim ] , the simulation is described along with the numerical results for the lift .
section [ analy ] holds analyses regarding the argument given in the previous section in order to understand the discrepancies between the theoretical and the numerical results . finally , section [ conclusions ] has the conclusions .
in this section , the theoretical argument leading to an expression for the lift force on the ellipse is developed .
also , some predictions are shown in order to be compared to the numerical results in the next section . in @xcite ,
the line of thought that led to an expression for the drag force on the circular obstacle was based on the dissipative collisions among grains and obstacle .
the same line of thought can be drawn here in order to obtain an expression for the lift force on the ellipse due to collisions with the incoming stream .
first of all , all disks are assumed to have only the horizontal velocity component , @xmath0 .
therefore , for the ellipse given in fig .
[ local_ellipse_frame ] , which is located at @xmath1,@xmath2 , a collision will occur only if the impact parameter ( the vertical distance between a disk s center and the horizontal line through the ellipse s center ) is in the range @xmath3,@xmath4 $ ] .
this is so because @xmath5 and @xmath6 are , respectively , the lowest and the highest points in the ellipse , in the same way as @xmath7 and @xmath8 are the leftmost and the rightmost ones .
any collisions that happen in the segment @xmath9 will exert a downward lift , while collisions that occur in the @xmath10 segment will produce an upward lift ( from now on , the @xmath9 segment , and all other segments in the text , will be referred to only as @xmath9 ) .
moreover , since the lengths of both segments are , in general , unequal , the longest of them , in this case @xmath9 , will suffer more collisions than the other , which implies a net , negative , lift force .
it is this mechanism that prevents any net lift to take place on a circular obstacle under these conditions ( in general , any body that is symmetrical with respect to the flow direction , and is fully immersed in the flow , does not suffer a net lift ) .
therefore , the determination of the coordinates of these four points in the ellipse , @xmath7 , @xmath8 , @xmath5 and @xmath6 , is very important . in order to do this
, one should notice that the tangent line to the ellipse is horizontal at @xmath6 and @xmath5 and vertical at @xmath7 and @xmath8 .
therefore , starting from the tilted ellipse s equation in the global frame : @xmath11 where @xmath12 and @xmath13 , one can write for @xmath6 and @xmath5 : @xmath14 and @xmath15 where the plus(minus ) sign is for point @xmath6(@xmath5 ) . a similar calculation yields , for points @xmath7 and @xmath8 : @xmath16 and @xmath17 where in both eqs .
the plus(minus ) sign is for point @xmath8(@xmath7 ) .
the force due to a collision between a disk and the ellipse can be obtained by calculating the change in linear momentum of the disk .
the total force is obtained by integrating the product of this individual momentum change and a suitable collision rate over the obstacle section facing the flow , i.e. , integrating over @xmath18 .
suppose a disk hits the obstacle in @xmath9 .
figure [ coll_geo ] presents schematically the geometry of the collision ( assumed frictionless ) .
the grain velocity is @xmath19 , and the impact angle is @xmath20 , which is the one between the direction @xmath21 and the normal at the contact point .
the post - collisional velocity is @xmath22 . in order to obtain the final components as a function of @xmath0 , one has two equations that relate the normal and tangential velocities before and after the collision : @xmath23 and @xmath24 , where @xmath25 , @xmath26 are the normal and tangential unit vectors at the collision point and @xmath27 is the normal restitution coefficient , assumed velocity independent . from these considerations , the components of the disk velocity after the collision are @xmath28 and @xmath29 .
therefore , the vertical momentum change for a general collision is given by : @xmath30 the number of disks that strike the ellipse in a time @xmath31 is given by the number of particles within the area of the parallelogram with sides @xmath32 and @xmath33 , this last quantity being the arc length around the collision point .
given the geometry of the collision ( figure [ coll_geo ] ) , the area of this parallelogram is @xmath34 .
hence : @xmath35 where @xmath36 is the area fraction of the incoming flow .
the collision frequency is simply this number divided by @xmath31 .
the arc length @xmath33 is given by the product of the collision radius , @xmath37 , by the arc element , @xmath38 .
this product is readily evaluated for a circle . however , in the ellipse case , this is not so simple , since this radius varies with @xmath39 . besides , the grain size introduces an additional complication for calculating @xmath37 , as seen in fig .
[ coll_tri ] , which shows the triangle formed by the disk center , the ellipse center and the collision point .
hence , the ellipse is assumed much larger than the disks and the disk size only enters the expression for @xmath37 as a minor correction . with this assumption ,
the arc length is given by @xmath40 .
in parametric form , @xmath41 and @xmath42 , this length reads : @xmath43 finally , the effective collision radius is given by : @xmath44 another difficulty to obtain the expression for the lift force is that the momentum change ( i.e. , the force ) depends explicitly on @xmath20 , while the collision frequency depends on @xmath39 .
therefore , any hope of integrating the lift force over all collisions should pass through obtaining the relationship between the angles @xmath20 and @xmath39 .
this is done in the following paragraph .
since @xmath45 gives the tangent of the angle between the normal line to a point in the ellipse and the @xmath46 axis , from the ellipse equation ( [ ellipse_eq_glob_frame ] ) , one has : @xmath47 where @xmath48 , @xmath49 and @xmath50 are the coefficients of the terms @xmath51 , @xmath52 and @xmath53 in eq .
( [ ellipse_eq_glob_frame ] ) . since @xmath54 and from eqs .
( [ x_dc ] ) , ( [ y_dc ] ) , ( [ x_ab ] ) and ( [ y_ab ] ) , the relation between @xmath20 and @xmath39 can be cast in terms of the coordinates of the points @xmath6 and @xmath7 , since : @xmath55 and @xmath56 hence : @xmath57 the lift force can now be evaluated as : @xmath58 where @xmath59 . by using eqs .
( [ py_change ] ) , ( [ coll_freq ] ) and ( [ radius_param ] ) , the total lift on the ellipse due to collisions is given by : @xmath60 where @xmath61 is the flow mass density and the arc length is to be calculated bearing in mind ( [ theta_alpha_rel ] ) .
the lift force scales with the obstacle size @xmath62 , as expected , since it depends linearly on the arc length .
also , it is seen that , for more inelastic grains the lift is smaller , since more inelasticity implies less intense change of momentum , see eq .
( [ py_change ] ) .
the dependence of the lift on the tilt angle is hidden in the integral over @xmath18 , because the lengths of @xmath10 and @xmath9 , which contribute forces with distinct signs , are unequal for a general @xmath63 .
apart from the dependence of @xmath64 in these parameters , it also depends on the ellipse eccentricity , i.e. , on the ratio of the ellipse s axes @xmath65 . for @xmath66 ,
the obstacle is a circle , and the lift vanishes by symmetry . in the other extreme value , @xmath67 , which corresponds to a flat plate
, the lift vanishes only for @xmath68 ( horizontal plate ) and @xmath69 ( vertical plate ) , which are the only symmetrical orientations with respect to an horizontal flow . in fig .
[ lift_theo_k ] , it is shown the predictions of eq .
( [ lift_ad ] ) for distinct values of @xmath70 ( the parameter values in plotting these curves were chosen to agree with those used in the simulations ) .
one can see that the maximum ( absolute ) value of @xmath64 increases when @xmath70 decreases .
this happens because when @xmath71 , the points @xmath6 and @xmath5 merge with points @xmath8 and @xmath7 , as can be inferred from eqs .
( [ x_dc ] ) , ( [ y_dc ] ) , ( [ x_ab ] ) , and ( [ y_ab ] ) . in this case ,
@xmath72 and @xmath73 .
since the lift is the difference of the contributions from @xmath9 and @xmath10 , all collisions will give positive contributions to the net lift value , and it should reach its maximum for a particular @xmath63 . the objective of the simulations is to study the lift force on the obstacle as a function of four parameters that appear on eq .
( [ lift_ad ] ) , namely , the obstacle size , @xmath62 , the restitution coefficient , @xmath27 , the ellipse eccentricity , @xmath70 , and the tilt angle , @xmath63 .
in this section , the simulation is detailed , along with the parameter values used in the computations , and the numerical results for the lift are shown .
the system is composed of @xmath74 soft disks , with unity diameter @xmath75 and unity mass @xmath76 , located in a workspace of lengths @xmath77 and @xmath78 in the horizontal and vertical directions , respectively .
they interact through normal and tangential forces , according to the model used in @xcite .
the normal force between two disks is : @xmath79 where the first term is the conservative part , given by a simple harmonic spring force @xmath80 where @xmath81 is the spring constant , @xmath82 is the @xmath83-th disk diameter , @xmath84 is the distance between the two disks and @xmath85 is a unit vector along the normal between the disks centers . the second term is a dissipative , velocity dependent force , given by @xmath86 where @xmath87 is the normal damping coefficient and @xmath88 is the relative velocity between the contacting disks .
the tangential force at the contact point is given by the following expression : @xmath89 where @xmath90 is the sliding friction constant and @xmath91 , the static friction coefficient , while @xmath92 is the unit vector along the direction of the relative velocity at the contact point .
this vector is calculated as : @xmath93 where @xmath94 is the @xmath83-th disk angular velocity and @xmath95 .
the total contact force @xmath96 vanishes if the disks are not in contact , i.e. , if @xmath97 .
the values of the elastic parameters used were : @xmath98 , @xmath99 and @xmath100 , @xmath101 and @xmath102 .
the restitution coefficient @xcite for these parameters were @xmath103 , for @xmath99 , and @xmath104 , for @xmath100 .
the ellipse s major and minor half - axes are @xmath62 and @xmath105 , respectively .
the parameter @xmath62 has the values @xmath106 , @xmath107 , @xmath108 , @xmath109 and @xmath110 , with @xmath111
. for the studies of the dependence on @xmath70 , size @xmath112 was used , with @xmath113 , @xmath114 and @xmath115 .
all these cases were studied for both @xmath27 values .
for the two smaller obstacle sizes , results were obtained with a @xmath116 and @xmath117 packing , while for the others , a @xmath118 and @xmath119 packing was used . in both cases ,
the packing fraction was about @xmath120 .
the ellipse is located in @xmath121 and @xmath122 , and its major half - axis is tilted related to the horizontal axis ( flow direction ) by @xmath63
. the values of the tilt angle were in the range @xmath123,@xmath124 $ ] , divided in @xmath125 increments , which gives a total of @xmath126 distinct @xmath63 values . at the beginning ,
all disks are randomly generated without overlap among them and the ellipse . at this stage ,
the obstacle is modeled as a circle with diameter @xmath127 since this facilitates the overlap check .
all disks have initial velocity @xmath19 , where @xmath21 is the unit vector in the horizontal ( flow ) direction and @xmath128 .
the system has periodic boundaries perpendicular to the flow direction , while , along the flow , the conditions are the same as in @xcite : whenever a disk leaves the system through the right boundary , it is placed at the left one in a random vertical position . in fact , the code searches for a position where the incoming grain overlaps with no other disk , which is fairly easy , giving the low density that is used here .
its velocity is set as the incoming flow velocity @xmath129 plus a small random vertical component chosen uniformly in the interval @xmath130 $ ] , where @xmath131 ( results obtained with @xmath132 differ little from those shown here ) .
the interactions between the disks and the ellipse are the same as those given above for two disks .
the main problem is to resolve particle - ellipse contacts . in order to do this ,
all disks are checked for overlap with a circle of diameter @xmath127 .
if one disk overlaps with the circle , it is mapped in the ellipse s local frame of reference ( elfr ) .
then , the contact point coordinates , in the elfr , are calculated using the algorithm proposed in @xcite .
the direction of the normal force is along the line joining the disk center and the contact point , while the tangential force is perpendicular to this direction .
lengths , forces and time are given in units of @xmath75 , @xmath133 , and @xmath134 , respectively .
equations of motion are integrated with a leapfrog scheme @xcite , with a time step of @xmath135 .
each simulation is performed during @xmath136 molecular dynamics ( md ) cycles for thermalization and @xmath137 md cycles for measurements .
all results are averaged over @xmath108 and @xmath107 independent runs for the @xmath116 and @xmath118 packings , respectively .
the main interest is to measure the force exerted on the obstacle by the stream of grains due to the collisions among them .
therefore , the components of this force , drag and lift , are measured at all cycles after thermalization . at regular intervals , the forces are recorded as averages over this period .
this accumulation interval is @xmath138 md cycles long ( results were also obtained with @xmath139 md cycles long accumulation periods and do not differ appreciably from those reported here ) .
the flow velocity and particle number density fields , @xmath140 and @xmath141 , were measured as follows : the workspace is divided in square bins of side @xmath75 . at each cycle , all particles are mapped in an appropriate bin and its velocity components are added to the respective field element . similarly ,
angular profiles related to the contact angle between the grains and the ellipse , the collision angle @xmath39 , were measured .
they are the collision number and the velocity components .
each time there is a disk - ellipse contact , the angle formed by the line joining the collision point and the ellipse center with the flow direction is calculated .
then , the above mentioned quantities are added to the appropriate profile bins . the field and the profiles
are presented as averages over cycles and runs . in fig .
[ lift_k ] , the numerical results for the lift as a function of the tilt angle and the eccentricity are shown .
comparing figs .
[ lift_theo_k ] and [ lift_k ] , it is clear that eq .
( [ lift_ad ] ) captures the qualitative features of the lift force , even though the correspondence becomes weak for @xmath142 and @xmath103 .
also , the results agree with the prediction that the lift should be higher for less inelastic flows . finally , the theoretical result overestimates the numerical ones by , roughly , a factor of @xmath143 , for these cases . in fig .
[ lift_phi ] , the numerical results for the lift force as a function of the tilt angle and obstacle size are shown .
as expected , the lift increases with the obstacle size . however , the lift does not grow linearly with @xmath62 .
plotting these data against the obstacle size shows that @xmath144 .
since the results are not conclusive , it suffices to acknowledge that more simulations are needed in order to obtain a more reliable scaling with the obstacle size . also , this figure shows that the lift force for @xmath104 packings grows faster with @xmath62 than for those with @xmath103 . finally , the numerical results , as seen in fig .
[ lift_k ] , also are smaller than their theoretical counterparts ( for @xmath145 , the factor is about @xmath146 ) .
this shows that the difference between model and theory depends only a little on the tilt angle , while the obstacle size , inelasticity and eccentricity are the factors that accept the most on this difference .
the reasons behind the failure to reproduce the numerical results will be analyzed in the next section , where the assumption that led to eq .
( [ lift_ad ] ) will be reviewed in detail .
there are three basic hypotheses on which the argument for the lift force in section [ sec11 ] was built .
the first one is that particle - obstacle interactions were frictionless , despite the force model , eq .
( [ tang_force ] ) , is not .
second , the arc length expression , @xmath40 , does not take into account the fact that the grains have a finite diameter @xmath75 . finally , the collision frequency expression ( [ coll_freq ] )
was obtained under the hypothesis that only one particle at a time hits the obstacle , i.e. , a disk - ellipse collision does not affect the next collision , a condition met only in very dilute or in ideal gas flows ( which is not the case here ) .
therefore , its applicability here is clearly questionable , as inferred from the system configuration shown in fig .
[ config_4_6 ] .
there is a dense region that begins above point @xmath7 and forms a separation boundary in front of the ellipse .
this is the typical granular shock wave @xcite that forms around bodies immersed in fast granular flows . clearly , the dilute flow assumption is not valid .
figure [ vel_fields ] shows the clear signature of this structure in the velocity field .
also , this picture shows that the two branches of the shock wave meet behind the obstacle .
this fact could , at first sight , invalidates the lift results because this clearly does not allow the use of periodic boundaries perpendicular to the stream .
this is not the case , however , as seen in the results for the lift obtained from simulations that used rigid walls in the @xmath147 direction .
these results agree with those shown in figs .
[ lift_k ] and [ lift_phi ] within numerical error .
nevertheless , periodic boundaries in a particular direction should only be used when the system is independent in that direction .
c + + before discussing the influence of the shock wave in the results , a few brief comments will be made regarding the first two assumptions in the theoretical argument . to take into account the grain size in computing
the collision radius is to consider a larger ( effective ) obstacle facing the flow , which would increase the theoretical prediction for the lift .
since this quantity is already larger than the numerical results , it was safe to ignore it in the calculations .
the influence of the friction force on the lift can be inferred from the velocity field results , fig .
[ vel_fields ] .
as seen in this figure , particles follow tangential trajectories along the ellipse @xcite , which means that they slide along the obstacle .
since friction is a tangential force , one concludes that a grain sliding along @xmath9 exerts an upward lift . by symmetry ,
a disk exerts a downward lift while sliding along @xmath10 .
therefore , the net effect of friction is to decrease the lift exerted by collisions in each segment . the combination of both effects might decrease or increase the net lift .
the numerical results for the lift rising from friction show that it is opposite to the one resulting from the normal force .
in other words , friction decreases the net lift .
the effect is small , though , because the sliding friction is also small , @xmath101 . in most simulations of granular matter ,
regardless of the tangential force model , the sliding friction is only mildly lower than the normal interaction parameter . here
, a much lower value was used , which could greatly affect the results .
simulations performed with a distinct set of parameters , @xmath148 and @xmath149 , values which are more common to general simulations of granular matter , show that the results do not change significantly , and are still qualitatively the same as those in figs .
[ lift_k ] and [ lift_phi ] .
in fact , the lift data are reduced only by @xmath150 to @xmath151 .
this happens due to the fact that the velocities involved in the simulations are high enough for the tangential force to reach the coulomb static friction criterion , which caps its maximum value and limits its influence on the lift .
this conclusion is supported by the fact that the fraction of all disk - obstacle contacts , for @xmath148 , that reach the failure condition is about @xmath152 , with only a very small @xmath63 dependence . as stated earlier
, the dilute flow assumption does not hold .
a rigorous calculation of the force in the obstacle , in which all dense packing effects are taken into account , is not a simple matter . in @xcite
an attempt was made in this direction .
the formation of the shock wave produces a dense region that shields the obstacle from the incoming particles .
since the interactions are dissipative , particles should reach the obstacle with reduced velocities .
this clearly reduces the force exerted by the disks .
also , given the large density within the shock wave , the collision rate might be affected . finally , as seen from the velocity field data , fig .
[ vel_fields ] , particles should hit the obstacle , on average , with a non vanishing vertical velocity component , i.e. , particles suffer oblique inelastic collisions , which could affect the force on the obstacle .
both effects will be discussed in more detail in the following two subsections .
this discussion will by made only qualitatively , since more simulations are needed to determine the amount of influence each of them has on the results .
one of the consequences of the existence of the shock wave is that the obstacle is shielded from the flow , in a way that the grains hit it with lower horizontal velocity than the upflow value .
this fact alone indicates that the numerical net lift should be lower than its theoretical prediction . aside from reducing the incoming flow velocity , the shock wave forms a dense region around the obstacle through which grains should go through in order to reach it .
this could affect the collision rate and , in turn , the net lift .
both effects must play a role in order to explain the faster growth of the lift force with the obstacle size for stronger inelasticity flow compared to that of lower inelasticity one .
the results for the collision profiles show that , for @xmath104 , the shock wave is more localized around point @xmath7 ( where , for the @xmath113 obstacles the peak of the profiles is located ) and that the overall collision number is greatly increased compared to the number when @xmath103 .
this is a consequence of the aggregation that grains suffer due to the inelastic interactions and is common to all cases studied .
this fact implies that the force on the obstacle should increase due to this increase in the collision rate and could compensate , in part , for the decrease in the force due to the reduced incoming velocity .
the collision profiles are changed , in a very different way , for obstacles with distinct eccentricities , @xmath70 .
[ coll_ang_prof_2 ] has results that illustrate this effect .
the most striking feature of these curves is that the profiles for @xmath142 develop two peaks , instead of the more familiar one around point @xmath7 , which is still present , but it moves away from this point as the ellipse is oriented vertically .
the other one develops around point @xmath6 and is seen at tilt angles as high as @xmath153 .
such features are also seen in the @xmath154 results , although with smaller peaks .
these results can be explained by the fact that in some cases , for stream velocities below a critical value , there is the appearance of a gap , filled with a hot granular gas , between the obstacle and the shock wave @xcite . in the present case ,
the appearance of the gap is a function of @xmath70 alone , since the stream velocity is the same for all simulations .
figure [ rho_f ] has a result for the density field that indicates the presence of the gap .
notice the denser arc formed right in front of the ellipse , the region limited by this arc and the obstacle is the gap . as a last note , the collision rate increases faster with @xmath63 for @xmath142 than for @xmath113 .
[ configs ] has two configuration snapshots that illustrate the shock wave for two obstacles with distinct eccentricities .
c + + one can infer from these pictures the reason behind the asymmetrical peaks in the collision profiles when @xmath155 and @xmath142 .
there is a very small concentration of particles in front of @xmath10 , which certainly allows for more collisions to occur .
these considerations still left unanswered the question about the faster growth of @xmath156 with @xmath62 for stronger inelasticity compared to those at lower inelasticity . since the net lift is the difference of the forces exerted in @xmath10 and @xmath9 , one should seek a relative measure of the speeds in each of them in order to account for the size of the contributions each has in the net value .
this is done by calculating the ratio between the average horizontal disk velocities in @xmath9 and @xmath10 . if this ratio is greater than @xmath157 , the collisions on @xmath9 exert , on average , a stronger force than those on @xmath10 .
these averages are evaluated directly from the simulations .
the results , for all tilt angles and obstacle sizes , are given in fig .
[ vx_ratio ] .
first it can be seen that except in a few cases , markedly at small obstacles , all ratios are larger than @xmath157 , and those for @xmath104 cases are larger than those for @xmath158 .
it is possible to identify a trend in these graphs , despite the noise : the ratio grows with obstacle size .
this is consistent with the results in fig .
[ lift_phi ] , where the net lift grows faster with @xmath62 for the @xmath104 flows compared to those with @xmath103 .
[ vx_ratio_k ] has the data for the same quantity measured for obstacles with distinct eccentricities .
it is seen that the ratio increases as @xmath70 decreases ( it reaches a factor of @xmath143 for @xmath142 ) .
these data , together with those for the collision profiles , show that the large lift observed for low @xmath70 obstacles is a consequence not only of a larger number of collisions at @xmath9 , but also due to collisions with larger speeds than those at @xmath10 .
the last piece of this analysis is the role of the vertical velocity component in the lift value . before proceeding ,
the collision argument of section [ sec1 ] is reviewed by allowing the incoming disks to have a vertical velocity , @xmath159 , and calculating the new momentum change in each collision .
this will elucidate the effect of the vertical velocity on the lift value . the velocity of a disk which collides with the obstacle is @xmath160 . the final disk velocity for a inelastic , frictionless collision is obtained as : @xmath161 and @xmath162 the vertical momentum change due to this collision is given by : @xmath163 using the result for @xmath164 , ( [ new_vy ] ) , and performing some algebra , the new momentum change is : @xmath165,\ ] ] where @xmath166 . by looking at eq .
( [ new_p_change ] ) one can see that a collision that happens in the range @xmath167 , which corresponds to @xmath10 , will produce a lower momentum change if @xmath168 .
similarly , for collisions that take place in the range @xmath169 , which is @xmath9 , a lower momentum change will occur if @xmath170 . from the velocity field data , fig .
[ vel_fields ] , it is clear that the average vertical velocities in @xmath9 and @xmath10 are positive and negative , respectively .
the oblique impacts decrease the positive lift exerted by the flow in @xmath10 , but also decrease the negative lift in @xmath9 .
therefore , it is not obvious if both effects combined increase or decrease the net lift . in order for this combination of effects to increase the net lift ,
the vertical velocity in @xmath9 should be lower than the corresponding velocity in @xmath10 . in figs .
[ vy_ratio ] and [ vy_ratio_k ] , the absolute value of the ratio of the average vertical velocities in both segments is shown for distinct obstacle sizes eccentricities , respectively .
it is seen that the ratio , for @xmath104 , depends little on the obstacle size and restitution coefficient , and that those for @xmath103 grow with @xmath62 .
it varies strongly with tile angle and eccentricity . despite a few cases , the ratio is larger than @xmath157 which implies that the net lift decreases due to oblique impacts , as provided by the existence of the shock wave
a theoretical argument and numerical results on the net lift force exerted by a granular stream of equal disks on an ellipse were presented .
the argument used to obtain a theoretical expression for the lift relied on inelastic , frictionless collisions of a very dilute granular flow , in the horizontal direction , and the ellipse .
the expression obtained , eq . ( [ lift_ad ] ) , captures nicely the qualitative features of the numerical results .
it does not , however , reproduce the results quantitatively , neither does it reproduce the difference in the lift force for @xmath103 and @xmath104 . the numerical data , shown in figs .
[ lift_k ] and [ lift_phi ] , are lower than the theoretical prediction by a factor that depends mostly on the obstacle size ( for @xmath145 , eq .
( [ lift_ad ] ) is about @xmath146 times larger than the measured results ) .
additional analyses of the flow properties showed that , from the assumptions drawn in the text , the dilute flow one is the most problematic .
the existence of the granular shock wave @xcite invalidates this hypothesis .
this structure affects directly the velocity components of the disks at contact with the obstacle as well as the collision rate . the horizontal velocity at contact decreases due to dissipative collisions that occur before the disks reach the shock wave .
also , some of the horizontal momentum is deviated by the shock wave around the obstacle , introducing a vertical component whose net effect is to decrease the lift .
perspectives to this work include a deeper analysis of dense packing effects on the lift .
this would answer questions about the scaling of the lift with obstacle size , flow speed and density .
such analysis would also allows for a better understanding of the mechanism behind the appearance of the net lift , since an obvious consequence of the presence of the shock wave is that any collisions that occur on its edge should transmit some linear momentum through the dense region until pushing the disks closets to the obstacle , and that is when the actual force is made .
force transmission models , such as those in @xcite can be good starting points for this analysis .
moreover , the effect of eccentricity on the shock wave is something worth investigating since , as seen here , for a constant flow speed , there should be some eccentricity value that allows for the appearance of the gap between the shock wave and the obstacle .
finally , the hydrodynamical approach to granular flow could also be tested in such a situation .
as argued in the text , ignoring the disk size does not affect the results for large obstacles .
hence , if the obstacle is much larger than the disks , the predictions of such theory should be realized in simulations , since scale separation could be achieved , at least approximately .
i thank a. p. f. atman and j. c. costa for a critical reading of this manuscript .
this work is financially supported by cnpq and fapespa .
100 h. m. jaeger , s. r. nagel , r. p. behringer , rev .
phys . * 68 * , 1259 ( 1996 ) .
k. wieghardt , annu .
fluid mech .
* 7 * , 89 ( 1975 ) . c. s. campbell , annu .
fluid mech .
* 22 * , 57 ( 1990 ) .
i. goldhirsch , annu .
fluid mech . * 35 * , 267 ( 2003 ) .
gdr midi , eur .
j. e * 14 * , 41 ( 2004 ) .
y. amarouchene , j. f. boudet , h. kellay , phys .
lett . * 86 * , 4286 ( 2001 ) .
e. c. rericha , c. bizon , m. d. shattuck , h. l. swinney , phys .
lett . * 88 * , 014302 ( 2002 ) .
j. f. boudet , y. amarouchene , h. kellay , phys .
. lett . * 101 * , 254503 ( 2008 ) .
r. albert , m. a. pfeifer , a .-
barabasi , p. schiffer , phys .
* 82 * , 205 ( 1999 ) .
d. chehata , r. zenit , c. r. wassgren , phys .
fluids * 15 * , 1622 ( 2003 ) .
v. buchholtz , t. poeschel , granular matter * 1 * , 33 ( 1998 ) . c. r. wassgren , j. a. cordova , r. zenit , a. karion , phys .
fluids * 15 * , 3318 ( 2003 ) .
pica ciamarra m. _ et al_. , phys .
lett . * 92 * , 194301 ( 2004 ) .
r. soller , s. a. koehler , phys .
e * 74 * , 021305 ( 2006 ) .
y. ding , n. gravish , d. i. goldman , phys .
* 106 * , 028001 ( 2011 ) .
d. c. rapaport , _ the art of molecular dynamics simulation _
, 2nd ed . , cambridge university press ( cambridge , 2007 ) .
l. e. silbert _
et al_. , phys .
e * 64 * , 051302 ( 2001 ) .
a. dziugys , b. peters , int .
. anal . meth .
geomech . * 25 * , 1487 ( 2001 ) . w. h. press , b. p. flannery , s. a. teukolsky , w. t. vetterling , _ numerical recipes in c _
, 2nd edition , cambridge university press ( cambridge , 1986 ) .
y. h. deng , j. j. wylie , q. zhang , phys .
e * 82 * , 011307 ( 2010 ) .
et al_. , science * 269 * , 513 ( 1995 ) .
s. ostojic , d. panja , europhys . lett . * 71 * , 70 ( 2005 ) . | this paper investigates the lift force exerted on an elliptical obstacle immersed in a granular flow through analytical calculations and computer simulations .
the results are shown as a function of the obstacle size , orientation with respect to the flow direction ( tilt angle ) , the restitution coefficient and ellipse eccentricity .
the theoretical argument , based on the force exerted on the obstacle due to inelastic , frictionless collisions of a very dilute flow , captures the qualitative features of the lift , but fails to reproduce the data quantitatively .
the reason behind this disagreement is that the dilute flow assumption on which this argument is built breaks down as a granular shock wave forms in front of the obstacle .
more specifically , the shock wave change the grains impact velocity at the obstacle , decreasing the overall net lift obtained from a very dilute flow . |
with the mosaic ccd imager on the 0.92 m telescope at kpno , we are able to cover an 1@xmath1 field in a single observation , and reach limiting fluxes corresponding to star formation rates ( sfrs ) of about 0.1 m@xmath2/yr for extended galaxies and @xmath30.01 m@xmath2/yr for compact emission - line galaxies
. the broad goals of the survey are to construct a complete inventory of star forming galaxies in a well - defined volume limited sample , to characterize the demographics of the local star forming population , and to quantify the completeness of other methods for measuring the local sfr density .
we are also carrying out a parallel survey of a local field sample , so we also will be able to study the influences of cluster environment on the sfrs .
we have observed four clusters so far during an observing run in february 1999 .
these clusters have been observed previously by moss , whittle , & pesce ( 1998 : mwp98 ) using an objective prism technique . in figure 1 , we show a flux distribution of h@xmath0-emitting galaxies in a1367 . also shown by shaded histograms are the distributions of the galaxies detected by mwp98 . although the analysis is still in the preliminary stage , the comparison between our data and those from mwp98 s prism survey suggests that @xmath4% of h@xmath0 flux is missing from the latter .
this is due to a combination of insensitivity of the prism survey to low - luminosity emission - line galaxies and to brighter galaxies with weak extended h@xmath0 emission .
in addition , we note that we have detected a large number h@xmath0-emitting galaxies without any velocity data .
since membership is uncertain for these objects , we have not included them in any of the analysis .
however , it is evident that the imaging survey has revealed the incompleteness in the detection of h@xmath0 fluxes in the previous prism surveys .
while visually searching for emission - line galaxies in the central field of abell 1367 , we found a particularly interesting field around n3860 , cgcg 97 - 114 and cgcg 97 - 125 , scattered with emission - line dwarf knots .
each `` knot '' is only a few arcsec in diameter , however , some are clustered in small groups that likely are par of dwarf galaxies . with the present data
, we can not answer as to what triggered such widespread star formation in this particular region of the cluster .
we have obtained spectra for some h@xmath0-emitting objects and they all coincide with those of cgcg97 - 114 and 97 - 125 ( @xmath5 km / s ) , confirming that they are associated with the larger galaxies in this group .
( ngc 3860 is not in this group ; its velocity is @xmath6km / s ) .
the star formation rates for these objects , calculated using kennicutt ( 1983 ) , vary from @xmath7 up to @xmath8m@xmath9/year . | we present a progress report on an ongoing h@xmath0 imaging survey of nearby clusters of galaxies .
four clusters have been surveyed to date : a1367 , a1656 , a347 and a569 .
a preliminary comparison of h@xmath0 luminosity functions obtained from our imaging survey with those from the prism survey reveals a significant level of incompleteness in the latter .
this in turn is due to a combination of insensitivity to low - luminosity emission - line galaxies and to brighter galaxies with weak extended h@xmath0 emission .
the survey has also revealed a unique population of clustered dwarf emission - line objects which may be the results of recent tidal encounters between larger gas - rich galaxies .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
the adsorption of polymers on flat and homogeneous attractive surfaces has been the subject of many investigations , see e.g. @xcite , but naturally occurring surfaces usually are rough and/or energetically inhomogeneous , the heterogeneity leading to an enhancement of adsorption under quite general conditions @xcite : already simple physical arguments contain the statement that upon increasing the surface irregularity , the number of polymer - surface interactions is strongly enhanced relative to the idealized planar surface ( see figure [ fig : flory1 ] ) .
this is a consequence of a larger probability of polymer - surface intersection with increasing roughness .
the study presented here is partly motivated by the theoretical investigation of reinforcement mechanisms in carbon black filled elastomers , where the polymer adsorption is substantial as there is a strong binding of the polymers to the surface ( leading to a layer of localized polymers , the so called `` bound rubber '' @xcite ) .
these surfaces are rough and even fractal on many decades of their size , down to the molecular size range , as well as highly energetically ( i.e. chemically ) heterogeneous : the distribution of interaction strengths can be characterized by high energy sites surrounding a relatively low energy background . therefore in principle both kinds of disorder should be incorporated in a theory which is supposed to explain the bound rubber phenomenon . in the literature
so far spatial and energetical heterogeneities were always treated separately . in this paper
we want to present a treatment which covers both sorts of surface heterogeneity , thereby making possible a comparison concerning the strength of adsorption enhancement .
thus the general aim of the present paper is to produce a theory for the adsorption and localization of polymers on such surfaces .
this is a nontrivial problem . for flat surfaces the description of ideal polymers by schrdinger type equations
is regarded as the method of choice , but for rough surfaces this method can not be used in general , because the equations can be solved only for some highly specific boundary conditions , e.g. , regular fractal surfaces @xcite . here
we want to investigate the polymer behavior near surfaces with any type of surface disorder , as well as surfaces with a wide range of heterogeneities in the attractive potential .
therefore a path integral formalism seems to be appropriate , since the surface disorder can be dealt with explicitly up to an advanced stage of the calculation and disorder averages are less difficult . however , some technical problems appear , forcing us to restrict the investigation to an idealized model system : a single ideal polymer chain in interaction with an attractive , penetrable , infinite surface . the treatment of real chains would require field theoretical approaches @xcite which are beyond the scope of the present paper .
the penetrability of the surface ( both sides of the surface are accessible for the polymer ) , which is of interest in the context of polymer chains at membranes or interfaces , here has to be assumed in order to avoid complex boundary conditions . in the problem of polymer adsorption on flat surfaces ,
non - penetrability is relatively easy to handle : the polymers are considered in half space @xcite . for random surfaces
an half space treatment is no longer possible .
therefore we will loose information on the localization behavior , but nevertheless expect general insight into the problem .
the results of the calculations are expressions for the size of the chain , depending on the mean interaction strength @xmath0 , the chain length @xmath1 and the disorder parameters .
the paper is organized as follows .
the size of the chain @xmath2 perpendicular to the surface is obtained * in section [ scaling ] for strong spatial disorder , using a scaling argument la de gennes @xcite this calculation intends to give a better insight into the main mechanism of adsorption enhancement ; * in section [ vario ] for weak spatial disorder and energetically heterogeneous surfaces , using a variational calculation . in the latter case ,
first the hamiltonian is formulated which explicitly contains the surface geometry and a distribution of interaction strengths .
then the free energy is approximated using a feynman variational procedure .
it turns out that for our problem an extension of a trial hamiltonian introduced by garel and orland @xcite is much more suitable than the replica trial hamiltonian frequently used in connection with polymers in disordered environments @xcite . in comparison with garel and orland ,
our main modification is the separation of the monomer coupling parameter into components of different space direction , thus enabling the polymer size parallel and perpendicular to the surface normal to be distinguished .
after minimization of the free energy , the effect of surface heterogeneity can be summarized in an effective interaction strength which in most cases is larger than the mean interaction strength .
the variational scheme is first tested on a flat and homogeneous surface , where it is shown that the result agrees with the scaling estimate and the results given earlier @xcite for ideal chains .
then the effective interaction strength is calculated for the special cases of a periodic and random distribution of interaction strength on a flat surface and for a periodic and random surface profile at constant interaction strength , respectively .
this also allows us to consider the localization transition .
the results are discussed and summarized in section [ conc ] .
first we briefly review a simple scaling treatment of an ideal chain adsorbed on a flat surface , as introduced by de gennes @xcite .
let @xmath3 and @xmath4 be the mean size of an ideal polymer ( with @xmath1 monomers and effective monomer length @xmath5 ) perpendicular and parallel to the surface , respectively .
the monomer density is assumed to be constant in a region of size @xmath6 .
then the number @xmath7 of monomers bounded to the surface is estimated as @xmath8 consequently the free energy can be written as @xmath9 the first term being the confinement energy , the second one due to contact interactions with the surface ( @xmath10 is the effective monomer attraction , @xmath11 the inverse temperature ) .
minimization of the free energy @xmath12 gives an expression for the polymer thickness @xmath2 perpendicular to the surface , @xmath13 thus the thickness of the polymer reduces with growing attractive interaction strength , as expected .
the independence of the chain length @xmath1 indicates that the polymer is in the so called localized regime @xcite .
a fractal surface may be characterized by its fractal dimension @xmath14 , @xmath15 , a value @xmath16 corresponding to a flat surface .
the limit @xmath17 produces an extremely rough , space - filling surface , brownian surfaces @xcite are characterized by @xmath18 .
now the number of bounded monomers is written as @xmath19 running through the same procedure as above yields @xmath20 so that the result ( [ flat ] ) for the case of a flat surface @xmath16 is recovered . from ( [ flat ] )
we have @xmath21 because of @xmath22 for weak adsorption , where no complete collapse on the surface takes place .
in fact for most materials values about @xmath23 are found @xcite .
therefore the polymer adsorption on rough surfaces ( @xmath24 ) generally is enhanced compared to the case of a flat surface , i.e. @xmath25 .
although this is a crude argument , it gives an insight into the main aspects of adsorption enhancement : the crucial point is the competition between the gain in potential energy obtained by binding to the surface and the loss in chain entropy associated with the confinement of chains in comparison to free chains .
therefore , the dominating mechanism in our consideration above is the increasing number of binding sites at a rough surface without paying an entropy penalty , which means that a chain has to lose less configurational entropy to adsorb on a rough surface .
this is in agreement with results of much more expensive previous calculations by douglas et al .
@xcite and hone et al .
@xcite .
a similar argument holds for the case of energetical heterogeneity @xcite : with a distribution of the interaction strength on the surface , the chain can select the strong binding points without changing its configuration too much , thus seeing a larger effective interaction strength .
for a systematic study of @xmath2 in the case of spatial and energetical heterogeneity , the free energy is calculated via a variational procedure , where the disorder is treated as a quenched ( i.e. frozen ) randomness .
the replica method is avoided by introduction of an additional variational parameter , see next section .
we consider an ideal chain at an infinite , penetrable , well defined surface with low profile .
furthermore we assume an attractive contact ( i.e. extremely short range ) interaction between chain and surface that can be mimicked by a delta - potential .
now this system is represented by its edwards - hamiltonian , which reads @xmath26 @xmath27 being the chain segment vector position .
the potential contains the polymer - surface coupling , @xmath28\ ; b\,w({{\underbar{x}}})\ , \delta({\textbf{r}}(s ) - { \textbf{h}}({{\underbar{x}}}))\ ] ] with @xmath29 , where @xmath30 is an internal surface vector .
the surface disorder is described by @xmath31 for energetical disorder , i.e. an interaction strength varying on the surface , and by @xmath32 for spatial disorder , i.e. a rough surface profile .
the factor @xmath33=(1+|\nabla h({{\underbar{x}}})|^2)^{1/2}$ ] takes account of the local deflection of the surface in cartesian coordinates .
in order to approximate the free energy we make use of a feynman variational procedure , @xmath34 where @xmath35 denotes the average with respect to a trial hamiltonian @xmath36 .
this gives an upper bound to the free energy @xmath37 with the abbreviation @xmath38 @xmath39 has to be minimized to give the best estimate for the true free energy @xmath40 .
the appropriate choice of the trial hamiltonian is most significant for the utility of the variational procedure .
here we take an extension of a form suggested by garel and orland @xcite , @xmath41 its features are : ( a ) it is quadratic in @xmath27 , so that an exact calculation of @xmath42 is possible ; ( b ) the coupling of chain segments is mediated by the variational parameters @xmath43 , one for each direction of space : the indices 1 and 2 are identified with the coordinates @xmath44 and @xmath45 of the surface parameterization , index 3 corresponds to the @xmath46 coordinate parallel to the average surface normal ; ( c ) there is an additional variational parameter @xmath47 , equivalent to a translation of the centre of mass of the chain . it should be mentioned that this type of variational principle was originally designed to avoid replica theory in random systems @xcite .
this is another reason why the garel - orland method is chosen here .
if the polymer is assumed to stick permanently at some place along the disordered surface , the problem falls into classes dealing with `` quenched disorder '' and difficulties with replica symmetry breaking arise . in the following
we will show that the garel - orland method is indeed useful to treat the problem of polymer adsorption on disordered surfaces as it yields physically sensible results .
assuming cyclic boundary conditions @xmath48 , the variational free energy ( [ betaf ] ) can now be calculated to give @xmath49 with @xmath50 .
here the interaction energy @xmath51 is the only term depending on the interaction potential , @xmath52 \ , w({{\underbar{x}}})\ , \exp\left\ { - \sum_{i=1}^3 { ( b_i - h_i({{\underbar{x}}}))^2 \over 2g_i}\right\},\end{aligned}\ ] ] the @xmath53 being the components of @xmath54 , i.e. @xmath55 .
the parameters @xmath56 are defined by @xmath57 @xmath56 can be identified with the mean square radius of the polymer parallel ( @xmath58 ) or perpendicular ( @xmath59 , and @xmath60 ) to the surface normal . following the lines of garel and orland , the minimization of @xmath39 with respect to @xmath61 and @xmath47 leads to @xmath62 and @xmath63 because of @xmath64 . as already discussed by garel and orland @xcite ,
one does in general expect the variational equations to have several solutions .
this especially applies to ( [ minb ] ) , since we consider an infinite surface , e.g. leading to an infinite number of solutions in the case of a periodic surface heterogeneity .
all these solutions have equal free energy .
introducing the abbreviation @xmath65 the optimized parameter @xmath56 is calculated from ( [ ming ] ) as @xmath66 -\displaystyle{nb^2\over4}\ \displaystyle{\cot(\alpha_j)-\alpha_j^{-1}\over \alpha_j } & w_j^{\text{eff } } < 0 .
\end{array}\right.\ ] ] the effective interaction strength @xmath67 contains all relevant surface and polymer properties and is given by @xmath68 in two special cases results can be obtained very easily : * if there is no interacting surface at all , i.e. @xmath69 , then we immediately have @xmath70 and therefore @xmath71 , the chain conformation is purely gaussian in all directions , as expected ; * for an ideal surface , which means @xmath72 and @xmath73 , the effective interactions strengths are calculated as @xmath74 and @xmath75 .
so the definition ( [ wjdef ] ) of @xmath67 guarantees correct results for this case . the explicit form of @xmath67 for various special sorts of surface heterogeneity is calculated in the next section .
the discussion of ( [ gallgemein ] ) is complicated by the fact that the effective interaction strength itself is a function of the polymer extensions in different directions .
but in general ( [ gallgemein ] ) can for @xmath76 be expanded in the limits of small and large @xmath77 .
this yields the mean polymer extension into the different directions of space , @xmath78 being parallel to the surface normal , @xmath79 and @xmath80 perpendicular to it , if @xmath81 is assumed , @xmath82 \displaystyle{b\over \beta w_j^{\mathrm{eff } } } \left\ { 1 - \displaystyle{\pi \over n ( \beta w_j^{\text{eff}})^2}\right\ } & \beta w_j^{\text{eff } } \gg n^{-1/2 } \end{array}\right.\ ] ] where , as above , @xmath83 denotes the radius of gyration of the corresponding gaussian chain
. thus in the limit of small effective interaction strength the chain has gaussian conformation ( see figure [ fig : adso1 ] ) , whereas for high @xmath84 the chain is localized at the surface , leading to a mean polymer size which in lowest order shows the same characteristics as the result of the scaling argument , equation ( [ flat ] ) . from the conditions for the limiting cases ,
a localization criterion @xmath85 can be found , which means @xmath86 for long chains .
therefore very long chains always are adsorbed , i.e. localized , at an attractive surface .
this of course is a consequence of our assumption of a penetrable surface , since in the opposite case of impenetrable surfaces adsorption takes place only from a finite interaction strength @xcite ,
i.e. @xmath87 . for negative effective interaction strength @xmath88 ,
only the case @xmath89 is important for us , as we are mainly interested in the adsorption behavior .
expansion of ( [ gallgemein ] ) in this case yields @xmath90 the full general form of the effective interaction strength is @xmath91\ ; w({{\underbar{x } } } ) \left ( 1 - { ( b_j - h_j({{\underbar{x}}}))^{2}\over g_j } \right ) \ , \exp\left\ { - \sum_{i=1}^3 { ( b_i - h_i({{\underbar{x}}}))^2 \over 2g_i}\right\}\end{aligned}\ ] ] where the translational parameter * b * has to be chosen such that @xmath92 . in the following the surface
is assumed to be symmetrical with respect to the coordinates @xmath93 and @xmath94 .
then we immediately have @xmath95 and @xmath96 as a solution of the minimization equation ( [ minb ] ) . for simplicity
, we additionally assume the surface heterogeneity to depend only on one space direction @xmath93 , which means @xmath97 and @xmath98 .
hence @xmath99 and the polymer extension into the direction of @xmath94 equals that of a gaussian chain . in this case
the expressions for @xmath100 and @xmath101 reduce to @xmath102\ ; w(x ) \left ( 1 - { x^2\over g_1 } \right ) \exp\left\ { - { x^2
\over 2g_1 } - { ( b_3-h(x))^2 \over 2g_3}\right\},\\ w_3^{\text{eff } } \label{w3eff } & = & { 1\over ( 2\pi g_1)^{1/2 } } \int\!{\text d } x\ , k[h(x)]\ ; w(x)\,\exp\left\ { - { x^2 \over 2g_1 } - { ( b_3-h(x))^2 \over 2g_3}\right\}.\end{aligned}\ ] ] now a straightforward calculation for various types of surface heterogeneity is possible .
( 1 ) : : for a flat surface with energetical heterogeneity , @xmath103 , the minimization condition ( [ minb ] ) results in @xmath104 , so that the centre of mass of the chain is located on the surface . inserting @xmath105
leads to @xmath106 as can be seen from the notation @xmath107 , the effective interaction strength parallel to the surface is independent of the mean interaction strength @xmath0 . * for a periodic interaction strength @xmath108 with amplitude @xmath109 and wave number @xmath110 , we have @xmath111 thus @xmath101 takes on its maximum @xmath112 , if the wavelength of the heterogeneity exceeds the polymer size parallel to the surface , @xmath113 , because in this case the polymer chain , which is located at a maximum of @xmath114 , does not notice the existence of the minima of the interaction strength . in the inverse case , @xmath115 ,
the fluctuations of @xmath114 can not be resolved any more , @xmath101 is minimal and equals the mean interaction strength .
@xmath100 is always small ( leading to a polymer size @xmath116 parallel to the surface ) , except for the case of a period of the fluctuation fitting the size of the chain , @xmath117 , and large @xmath109 . *
a randomly distributed interaction strength is best handled by identifying the amplitude in fourier space @xmath118 with the square root of the spectral density @xmath119 , so that @xmath120 for a gaussian distribution with variance @xmath121 , correlation width @xmath122 and constant @xmath123
. then the effective interaction strengths read @xmath124 the magnitude of the heterogeneity is determined by both @xmath122 and @xmath125 : the smaller the correlation width and larger the variance , the stronger are the fluctuations , which leads to an increase of the effective interaction strength .
the limiting cases perpendicular to the surface are @xmath126 \ w_0 + c\ , \delta_w\ , \bar{r}_1^{-1 } & \xi \ll \bar{r}_1\ .
\\[3pt ] \end{array}\right.\ ] ] ( 2 ) : : in the case of a heterogeneous surface profile ( whereas the interaction strength @xmath127 is assumed constant ) the disorder has to be weak in order to make the @xmath128 integration feasible . therefore we only investigate the case @xmath129 and @xmath130 , where @xmath131 , and restrict the calculation to first order in the fluctuation of @xmath132 . with @xmath133 , the minimization ( [ minb ] ) yields @xmath134 .
this means that the centre of mass of the chain to some extent follows the surface profile . for an example
see figure [ fig : bzmin1 ] , where @xmath47 is sketched for a periodic surface profile .
+ + now the deflection factor can be approximated by @xmath135=(1+|\nabla h(x)|^2)^{1/2 } \approx 1 + { 1\over2 } \int\limits_0^\infty\!{\text d}q \int\limits_0^\infty\!{\text d}q^\prime\ , \tilde{h}(q)\,\tilde{h}(q^\prime)\ , qq^\prime\,\sin(qx)\,\sin(q^\prime x).\ ] ] if additionally the part of the exponent in ( [ w1eff ] ) and ( [ w3eff ] ) which depends on @xmath132 is expanded , we obtain in lowest order of @xmath136 @xmath137\\ w_3^{\text{eff } } & \approx & w_0 \left\ { 1 + { 1\over2g_3 } \int\limits_0^\infty\!{\text d } q \int\limits_0^\infty\!{\text d } q^\prime\ , \tilde{h}(q)\,\tilde{h}(q^\prime)\ , \exp\left\ { - { g_1
( q^2 + { q^\prime}^2 ) \over 2}\right\ } \times \right . \nonumber\\ & & \hspace*{3cm}\times\ , \bigl [ 3 + g_3qq^\prime \sinh(g_1qq^\prime ) - \cosh(g_1qq^\prime ) \bigr ] \biggr\ } \end{aligned}\ ] ] * a periodic surface geometry @xmath138 leads to @xmath139 for a flat surface , the polymer extension @xmath78 parallel to the mean surface normal ( i.e. in @xmath46 direction ) is identical to the size perpendicular to the surface .
this is different for a rough surface profile , which now turns out to be important for the interpretation of ( [ w1pergeo ] ) and ( [ w3pergeo ] ) . if for example @xmath58 is small compared with the squared wavelength @xmath140 , then the polymer sticks to the surface , following the deflections .
therefore the result for @xmath141 exceeds the polymer size perpendicular to the surface by the amount of the deflection , the effective interaction strength @xmath101 accordingly is smaller than @xmath0 .
adsorption enhancement therefore only can be obtained in the opposite case @xmath142 , where the effective interaction strength parallel to the surface normal takes on its maximum value @xmath143 .
+ here we have used @xmath144 , which results from the fact that @xmath100 always is small according to the condition @xmath145 . as can be seen from ( [ w1pergeo ] ) , in the case @xmath146 discussed above @xmath100
is negative , so that the mean polymer extension in @xmath128 direction exceeds the size of a corresponding gaussian chain , see ( [ rjminus ] ) . *
similar to the case of a randomly distributed interaction strength , the amplitude in fourier space @xmath118 of a randomly distributed surface profile is identified with the square root of the spectral density @xmath119 .
this means @xmath147 for a gaussian distribution with mean @xmath148 , variance @xmath149 and correlation width @xmath122 . in order to satisfy the requirement of weak disorder
, we have to assume @xmath150 and @xmath151 .
then the result for the effective interaction strength in @xmath46 direction is @xmath152 \right\}. \end{aligned}\ ] ] for a very large correlation width , which in the limit @xmath153 corresponds to a flat surface , we again have the effect of a reduction of the effective interaction strength compared with the flat surface , @xmath154 .
therefore the result which is relevant for adsorption enhancement here is obtained in the case @xmath155 , where the effective interaction strength has its maximum value @xmath156 an analytic expression for @xmath100 is not available , but the main features of the result can be estimated to strongly resemble those of @xmath100 for a periodic surface profile discussed above .
the variational calculation presented here is valid for weak spatial disorder only ( therefore it does not reproduce the scaling behavior for fractal surfaces ) .
nevertheless the mechanism of adsorption enhancement is well reproduced , we find agreement of the results in all special cases which were already investigated in the literature .
a special feature of the variational method employed here is the possibility of quantifying the localization transition , i.e. , the transition from a slightly deformed gaussian coil to a localized conformation , where the polymer size perpendicular to the surface no longer depends on the chain length . according to the result ( [ loccrit ] )
the localization can be obtained by increase either of the effective interaction strength or of the chain length .
this helps to compare the strength of adsorption enhancement for the two sorts of disorder considered here : as can be seen from the maximum values of ( [ w3perint ] ) and ( [ w3pergeo ] ) or from a comparison of ( [ w3ranint ] ) and ( [ w3rangeo ] ) , the localization transition is only slightly affected by a rough surface profile , whereas energetical heterogeneity can induce the transition even at vanishing mean interaction strength .
therefore we conclude that the disorder - induced enhancement of polymer adsorption is much more significant for a heterogeneous interaction strength than for spatial roughness .
our findings concerning the localization behavior are affected qualitatively by the assumption of surface penetrability : for infinitely long chains at a flat and homogeneous impenetrable surface , the localization transition in contrast to ( [ loccrit ] ) only occurs for some nonzero value of the attractive potential @xcite .
nevertheless we expect our main statement on the significance of adsorption enhancement to hold also for impenetrable surfaces , since the comparison of transparent and opaque surfaces in simple soluble cases by hone et al .
@xcite shows that they should not be affected differently by weak surface heterogeneities .
the polymer size parallel to the surface does not directly depend on the mean interaction strength , but only through the extension perpendicular to the surface .
thus , because it is less affected by heterogeneity , the former always exceeds the latter , except for one special case : for a flat , neutral surface with a periodic interaction strength , the polymer size parallel to the surface is smaller than perpendicular to it , if the period fits the polymer size such that it is concentrated to a maximum of @xmath114 and even restricted by the neighboring repulsive regions .
we would like to thank g. heinrich for stimulating discussions . financial support by the deutsche kautschuk gesellschaft ( dkg ) is gratefully acknowledged .
99 -5pt e. eisenriegler , _ polymers near surfaces _ , world scientific , singapur ( 1993 ) g.j .
fleer et al . , _ polymers at interfaces _ ,
chapman & hall , london ( 1993 ) g. heinrich , t.a .
vilgis , _ rubber chem .
technol . _
* 68 * ( 1995 ) , 26 t.a .
vilgis , g. heinrich , _ macromolecules _ * 27 * ( 1994 ) , 7846 c.c . van der linden , b. van lent , f.a.m
. leermakers , g.j .
fleer , _ macromolecules _ * 27 * ( 1994 ) , 1915 d. andelman , j.f . joanny , _ j.phys.ii ( france ) _ * 3 * ( 1993 ) , 121 a. baumgrtner , m. muthukumar , _ j. chem
. phys . _
* 94 * ( 1991 ) , 4062 m. blunt , w. barford , r. ball , _ macromolecules _ * 22 * ( 1989 ) , 1458 r. ball , m. blunt , w. barford , _ j. phys .
a : math . gen . _ * 22 * ( 1989 ) , 2587 j.f .
douglas , _ macromolecules _ * 22 * ( 1989 ) , 3707 d. hone , h. ji , p.a .
pincus , _ macromolecules _ * 20 * ( 1987 ) , 2543 j.b .
donnet , r.c .
bansal , m.j .
wang ( eds . ) , _ carbon black .
science and technology _ , marcel dekker , new york ( 1993 ) b. meissner , _ j. appl .
* 50 * ( 1993 ) , 285 h.w .
diehl , a. nsser , _ z. phys .
b _ * 79 * ( 1990 ) , 69 & 79 t. garel , h. orland , _ phys .
b _ * 55 * ( 1997 ) , 226 s.f .
edwards , m. muthukumar , _ j. chem .
* 89 * ( 1988 ) , 2435 s.f .
edwards , y. chen , _ j. phys . a : math .
* 21 * ( 1988 ) , 2963 m.e .
cates , r.c .
j. phys .
france _ * 49 * ( 1988 ) , 2009 p.g .
de gennes , _ scaling concepts in polymer physics _ , cornell university press , ithaca ( 1979 ) j. feder , _ fractals _ , plenum press , new york ( 1988 ) j. klein , in _ liquids at interfaces , les houches 1988 summer school xlviii _ , j. charvolin , j.f .
joanny , j.zinn-justin ( eds . ) , north holland , amsterdam , ( 1990 ) a. baumgrtner , m. muthukumar , _ adv .
* xciv * ( 1996 ) , 625 | the adsorption of a single ideal polymer chain on energetically heterogeneous and rough surfaces is investigated using a variational procedure introduced by garel and orland ( _ phys .
rev .
b _ * 55 * ( 1997 ) , 226 ) . the mean polymer size is calculated perpendicular and parallel to the surface and is compared to the gaussian conformation and to the results for polymers at flat and energetically homogeneous surfaces .
the disorder - induced enhancement of adsorption is confirmed and is shown to be much more significant for a heterogeneous interaction strength than for spatial roughness .
this difference also applies to the localization transition , where the polymer size becomes independent of the chain length .
the localization criterion can be quantified , depending on an effective interaction strength and the length of the polymer chain . |
in the spectra of @xmath0ca , the existence of low - lying deformed states has long been known since 1960s@xcite .
the coexistence of the spherical and various deformed states in the doubly magic nuclei @xmath0ca has attracted a great interest . in the positive parity states below 7 mev , besides the spherical ground state , the existence of two deformed rotational bands built on the @xmath7 state at 3.35 mev and the @xmath2 state at 5.21 mev is known experimentally .
the structure of these bands has been often discussed in relation to many - particle many - hole states of nuclei in this mass region for a long time .
grace and green @xcite firstly suggested that the former band is dominantly a @xmath8-@xmath9 state , while the latter band with a large deformation is understood as an @xmath3-@xmath4 state .
the member states of the first @xmath10=@xmath11 rotational band built on the @xmath7 state are known to be strongly populated by @xmath12-transfer reactions and to have the large @xmath12 spectroscopic factors @xcite .
the @xmath12+@xmath13 cluster structure of this band has been studied theoretically based on semi - microscopic cluster model calculations@xcite .
the states in the second @xmath10=@xmath11 band have been observed in the experimental work searching for the @xmath3-@xmath4 states with @xmath14s(@xmath5c,@xmath12)@xmath0ca reactions , which were performed by middleton _
et al . _ in 1972 @xcite .
due to the strong population in the multi - nucleon transfer data and the strong @xmath1 transitions @xcite , the @xmath11(5.21 mev ) , @xmath15(5.63 mev ) and @xmath16 ( 6.54 mev ) have been thought to belong to the superdeformed band with a dominant @xmath3-@xmath4 configuration .
recently , by using the gammasphear array detectors , the level structure of the deformed bands in @xmath0ca has been explored and many excited states up to very high spins have been discovered @xcite .
consequently , the existence of the superdeformed band of @xmath0ca built on the @xmath17(5.21 mev ) has been experimentally well established .
in contrast to the experimental situation , theoretical interpretation of the superdeformed band is not well established and is still complicated . with
the symmetry restricted skyrme - hartree - fock(hf ) calculations , zheng _ et al .
_ studied the energy systematics of @xmath18particle-@xmath18hole(@xmath19-@xmath20 ) states in @xmath0ca@xcite .
they suggested an @xmath3-@xmath4 deformed states as a local minimum of the energy as a function of @xmath18 .
they also predicted highly deformed @xmath21-@xmath22 states in a high excitation energy region .
however , the recent mean - field calculations suggest that the @xmath3-@xmath4 configuration becomes unstable , especially when the model space is extended .
based on the three - dimensional(3d ) hf calculations without any assumption of the spatial symmetry , inakura _ et al .
_ studied the energy surface of @xmath23 as the function of the quadrupole deformation .
they have shown that the local minimum of the superdeformed @xmath3-@xmath4 configuration on the energy curve becomes quite shallow or even disappear due to the @xmath24 and octupole deformations depending on the effective interaction , though the @xmath21-@xmath22 state and an oblate state remain to be local minima in any of used effective interactions .
they have pointed out that the extreme softness of the superdeformed @xmath3-@xmath4 configuration against the octupole deformation ( @xmath25 and @xmath26 ) .
the instability of the @xmath3-@xmath4 configuration and the stability of the @xmath21-@xmath22 and oblate solutions in @xmath0ca had been also reported by leander and larsson@xcite based on the macroscopic - microscopic model calculations .
further calculations beyond mean - field theory were performed by bender _
et al . _ with gcm calculations in hf+bcs
approach @xcite . in their results ,
the superdeformed band is not dominated by a certain @xmath19-@xmath20 configuration but it contains various configurations such as @xmath8-@xmath9 , @xmath27-@xmath28 , and @xmath3-@xmath4 states due to the strong mixture between these configurations .
moreover , shell model calculations were performed for description of the superdeformed bands@xcite .
although the level structure and quadrupole moments of the superdeformed band are described by the spherical shell model with the fixed @xmath3-@xmath4 configurations , however , when the model space is extended to include other particle - hole configurations , the quadrupole moments are much underestimated due to the configuration mixing .
these facts may imply that a stable solution for the intrinsic state of the superdeformation is hardly obtained in the modern calculations with mean - field approaches .
it seems to be somehow inconsistent with the experimental energy spectra of the superdeformation in which no large deviation from the rigid rotor model is seen @xcite .
therefore , it is natural to expect that there might be some mechanism beyond mean - field approaches to stabilize the superdeformation .
the cluster aspect is one of the essential features for nuclear deformation even in @xmath29-@xmath30 shell nuclei together with the mean - field aspect .
indeed , it has been already shown that two kinds of nature , i.e. cluster and mean - field coexist in the deformed states in @xmath29-shell nuclei @xcite .
we expect that the clustering can play an important role also in the formation of the superdeformed state of @xmath23 . in case of @xmath0ca
, we consider that the softness against octupole deformations in the superdeformed state suggested by inakura _ et al .
_ can be associated with asymmetric di - cluster structure , though , to our knowledge , there has been no studies to investigate its cluster aspect .
we also note that some @xmath29-shell nuclei have the remarkable cluster aspect , molecular resonances , whose properties have been investigated by cluster models .
recently , there are theoretical efforts to link the molecular resonances with low - lying deformed states with full microscopic calculations .
it is suggested that the molecular resonances and the low - lying deformed states in @xmath29-shell are regarded to be a series of molecular bands .
for example , in @xmath14s system , it has been suggested that @xmath31o-@xmath31o molecular resonances are built on the superdeformed band of @xmath14s @xcite owing to the excitation of the inter - cluster motion .
therefore , we consider that molecular resonances can be good probes to understand the cluster aspect of the low - lying deformed states . in @xmath0ca system , there are experimental implications of @xmath6si+@xmath5c molecular resonances in the low - energy fusion cross section@xcite and also elastic scattering data@xcite , where distinct structures have been found in the energy region @xmath32=20 - 30 mev above the @xmath6si+@xmath5c threshold .
it is interesting to look into possible appearance of molecular resonances and its relation to the superdeformation in @xmath0ca . in this paper
, we study deformed states in @xmath0ca with antisymmetrized molecular dynamics(amd ) .
the method of amd has been proved to be useful in description of shape coexistence as well as cluster structure in @xmath29-shell nuclei @xcite .
one of advantages of this method is that both cluster aspect and mean - field aspect can be described within the amd framework . in order to study the coexistence of deformations in @xmath0ca and to investigate the corresponding rotational bands , we apply a constraint amd method .
parity projection , which is essentially important to describe the parity asymmetric shape of nuclei , is performed before energy variation , while total - angular - momentum projection is done after the energy variation .
we also perform superposition of the wave functions with different configurations obtained by the constraint amd to obtain better wave functions and to describe possible configuration mixing .
level scheme and @xmath1 transition strengths are calculated from these superposed wave functions . in the present work ,
we adopt the effective nuclear interaction which contains finite - range 2-body and 3-body forces proposed in ref.@xcite .
this interaction is suitable for describing the nuclear structure properties such as binding energies and radii over a wide mass number region from @xmath33he to @xmath0ca .
moreover , it is also useful to describe nucleus - nucleus potential .
these features can not be described by zero - range forces such as the skyrme forces , and are superior points of the effective forces with finite - range three - body terms .
we believe that they are essential to represent cluster aspect and to investigate properties of many - particle - many - hole states with large deformations .
this paper is organized as follows . in section [ sec : formulation ] , the formulation of amd is briefly explained . in sec.[sec : results ] , we show the calculated results of the energies , quadrupole transitions and moments , and also the intrinsic structure .
we give discussions of molecular aspect and single - particle orbits in sec.[sec : discuss ] .
finally , a summary is given in sec.[sec : summary ] .
we here briefly explain the formulation of the present calculations . for the details of the amd formulation for nuclear structure studies ,
the readers are referred to refs .
@xcite .
we start from the intrinsic wave function for a @xmath34-nucleon system which is expressed by a slater determinant of single - particle wave packets , @xmath35 where @xmath36 is the antisymmetrizer , and @xmath37 is the @xmath38th single - particle wave packet consisting of the spatial part @xmath39 , the spin part @xmath40 and the isospin part @xmath41 . the spatial part of the @xmath38th single - particle wave packet is written by a gaussian wave packet located at @xmath42 .
@xmath43 here , the center @xmath42 of gaussian is a complex number and treated as a independent free parameter for each nucleon .
the width parameter @xmath44 takes a common value for all nucleons , and is chosen to be an optimum value for each nucleus .
the orientation of the intrinsic spin is expressed by a variational complex parameter @xmath45 , and the isospin function is fixed to be up(proton ) or down(neutron ) .
thus , an amd wave function is specified by a set of variation parameters , @xmath46 , which are optimized by the energy variation .
as the variational wave function , we adopt the parity projected amd wave function , @xmath47 thus the parity projection is performed before the variation .
this is essentially important to describe the parity asymmetric intrinsic structure such as the asymmetric cluster structure as discussed in refs.@xcite .
the hamiltonian consists of the kinetic energy , nuclear and coulomb forces . in the present work
, we use the effective nuclear force consisting of two - body and three - body forces . then the hamiltonian is written as , @xmath48 where @xmath49 is a kinetic operator , @xmath50 and @xmath51 are two - body and three - body forces , respectively .
@xmath52 is the kinetic term for the center - of - mass motion . in the amd wave function
, the energy of center - of - mass motion is subtracted exactly .
we apply a constraint amd method , where we find the minimum energy solution in the amd model space under certain constraints .
since the particle - hole structure of the low - lying state is one of the main interests in the present work , we adopt the constraint of a total number of harmonic oscillator(h.o . )
quanta , @xmath53,\ ] ] where the width parameter @xmath44 is taken to be the same value as that of the single - particle wave packets in eq.([eq : varphi ] ) .
this is the principal quantum number of the spherical harmonic oscillator with the oscillation number @xmath54(@xmath55 is the nucleon mass ) . in order to obtain the minimum energy state with the constraint @xmath56 , where @xmath57 is a given number , we perform energy variation for the expectation value @xmath58 of the hamiltonian , @xmath59 with respect to the variational parameters @xmath60 by the method of frictional cooling @xcite under the constraint .
we note the amd wave function obtained with the energy variation for positive or negative parity state under the constraint @xmath56 as @xmath61 or @xmath62 , respectively . in the obtained wave function @xmath63
, there exist local minimum solutions in the energy curve as a function of @xmath57 .
as shown later , those are usually the local minima also with respect to the deformation parameter @xmath64 and considered to be approximate intrinsic wave functions for the corresponding deformed bands . in the calculation of level scheme and physical observables ,
we choose several intrinsic wave functions @xmath65 and superpose the spin - parity projected wave functions , @xmath66 , in order to satisfy the orthogonality among the calculated states and to obtain better wave functions . here , @xmath67 indicates @xmath68 and stands for the label for the obtained intrinsic wave function(@xmath69 ) .
the wave function of the @xmath18th @xmath70 state is written as , @xmath71 where coefficients @xmath72 are determined by digonalizing the hamiltonian and norm matrices : @xmath73 physical quantity for an operator @xmath74 for the state is calculated as , @xmath75 if we use an enough number of base wave functions , this procedure of the superposition is equivalent to the generator coordinate method with respect to the generator coordinate @xmath57 .
however , the number of the basis in the superposition is 10 at most because of huge computational time in the numerical calculations of the total angular momentum projection in the present work .
it means that @xmath76 stands for the summation of discrete basis , and hence , the present calculations do not correspond to the complete gcm calculations .
we use an effective nuclear force which consists of central force , spin - orbit and coulomb forces . as the central force
, we use f3b force@xcite , which contains a finite - range three - body term . the parameter set ( 2 ) of the f3b force
is adopted . as for the spin - orbit force
, we use the same interaction as that adopted in ref.@xcite , the two - range gaussian spin - orbit term in g3rs force@xcite . in the present calculations ,
the strength parameter of the spin - orbit force is chosen to be ( i)@xmath77 mev or ( ii)@xmath78 mev .
coulomb force is approximated by the sum of seven gaussians . in fig .
[ fig : be ] , we show the binding energies of @xmath79 nuclei from @xmath33he to @xmath0ca obtained by using the present interaction parameters .
the calculations are the simple amd calculations without constraints .
the binding energies calculated by variation after parity projection are shown in fig .
[ fig : be](a ) , and the energies with total - angular - momentum projection after the variation with parity projection are plotted in fig .
[ fig : be](b ) .
we also show the results of ref.@xcite for the interaction(ii ) , where amd calculations with constraint and superposition were performed .
the energies for such deformed nuclei as @xmath80be , @xmath5c , @xmath81ne and @xmath82 mg are gained by the total - angular - momentum projection . as shown in fig .
[ fig : be](b ) , the binding energies of medium @xmath29 shell nuclei are overestimated by the interaction(i ) due to the stronger spin - orbit force than that of the case(ii ) .
= 0.35 we apply the constraint amd method to @xmath0ca .
the width parameter @xmath44 for the single - particle wave packets is chosen to be as @xmath83 @xmath84 which optimizes the energy of the ground state .
we impose the constraint of the total number of harmonic oscillator quanta as @xmath56 on the parity projected amd wave function , and obtain energy curves as a function of @xmath57 . with a given constraint value @xmath57
, we obtain a few states with different shapes as local minima .
a prolate shape and an oblate shape are obtained by randomly choosing initial states in the variational procedure . in order to find possible cluster states
, we also start the variation from initial states with various di - nuclear structure and decrease the @xmath57 .
then we find that a @xmath12+@xmath85ar - like clustering appears as a local minimum solution .
the energy curves as a function of the oscillator quanta(@xmath57 ) are shown in figs.[fig : beqc - pp ] and [ fig : beqc - np ] . in principle , the minimum value of @xmath57 for the positive - parity state is 60 , which is given by the @xmath29-shell - closed configuration , and it is 61 for negative - parity states .
the results shown in the figure are calculated with the total - angular - momentum projected wave functions @xmath86 , where @xmath87 is chosen to obtain the lowest energy .
the energies are also plotted as a function of quadrupole deformation @xmath88 . here
the deformation parameter @xmath88 is defined by using the sharp edge liquid drop relation between @xmath89 and @xmath88 as follows ; @xmath90 where the expectation values are calculated for the intrinsic states @xmath91 , and @xmath92 is chosen to be an approximately symmetric axis . here
we use the fixed radius parameter , @xmath93 , which is related to the nuclear mass @xmath94 according to the formula @xmath95 fm .
we first mention the results of positive - parity states with the interaction(i ) ( figs .
[ fig : beqc - pp](a ) and ( b ) ) .
we find three states with different shapes as the local minima on the energy surface .
the lowest state in the @xmath96 curve is an almost spherical state at @xmath97 .
there are two local minimum solutions in the energy curves , which almost degenerate to each other , at 7@xmath988 mev higher energy than the lowest solution .
one is an oblate state at @xmath99 and the other is a large prolate deformation at @xmath100 with @xmath101 . as shown later , the former contains a dominant @xmath8-@xmath9 configuration , and the latter is dominated by a ` @xmath3-@xmath4 ' configuration and corresponds to the superdeformed state .
the appearance of the oblate state is consistent with the macroscopic - microscopic calculation@xcite and also the 3d - hf calculations@xcite . on the other hand ,
the calculated energy curve for the prolate states differs from the macroscopic - microscopic calculation by leander and larsson@xcite where there is no local minimum for the @xmath3-@xmath4 configuration but is a highly deformed @xmath21-@xmath22 state .
the @xmath12+@xmath85ar - like cluster state appears in @xmath102 region , but it is much higher than the superdeformed state in the case of interaction(i ) . with a smaller @xmath57 value of the constraint
, the @xmath12-cluster is absorbed into the @xmath85ar core and the @xmath12+@xmath85ar - like state changes into the normal prolate state during the energy variation . in the @xmath15 states , the energy curve for the prolate states is more gentle than the @xmath11 curve at the small deformation region , @xmath103 .
next we look into the results obtained with the interaction(ii ) ( figs .
[ fig : beqc - pp](c ) and ( d ) ) , which has the weaker spin - orbit force than the interaction(i ) .
although there exist the oblate state and the superdeformed state as local minima as same as in the results(i ) , the excitation energies of these states are much higher than those in the case of the interaction(i ) .
the excitation energy of the superdeformed state is about 20 mev , which largely overestimates the experimental energy , 5.21 mev , of the bandhead of the superdeformed band . on the other hand ,
the energy of @xmath12-cluster state does not so much change from ( i ) to ( ii ) because the effect of the spin - orbit force is smaller than other excited states . as a result
, @xmath12-cluster state almost degenerates with the superdeformed state and does not vanishes even in the smaller @xmath57 region(as @xmath104 ) in the results(ii ) than the case(i ) .
comparing figs .
[ fig : beqc - pp](c ) and ( d ) , we find that the total number of h.o .
quanta has a good correspondence with the quadrupole deformation @xmath88 in the @xmath105 region for the prolate deformation . on the other hand , in the small @xmath57 region ,
all the solutions for @xmath106 have almost spherical shapes with @xmath107 .
it means that different configurations with the same @xmath88 value are obtained as optimum solutions for the given @xmath57 .
we illustrate the density distribution of the intrinsic state @xmath63 in fig.[fig : ca - dense ] and fig.[fig : ca - dense18 ] .
a remarkable point is that parity asymmetric structure appears in the prolate states .
especially , one of striking features is the prominent cluster structure like @xmath5c+@xmath6si in the large prolate deformation . comparing the results of ( i ) and ( ii ) , we remark that the intrinsic structure for a given @xmath57 is almost the same as each other .
it means that the structure of coexisting shapes is not sensitive to the choice of either interaction ( i ) or ( ii ) except for the @xmath12-cluster state .
in contrast , the excitation energies strongly depend on the strength of the spin - orbit force as shown in the comparison between the results with a stronger spin - orbit force(i ) and with a weaker one(ii ) .
particularly , the oblate state and the superdeformed state appear at the low excitation energies in the results with the interaction(i ) , but they exist at much higher excitation energies with the interaction(ii ) .
considering the @xmath8-@xmath9 configuration in the oblate shape and the @xmath5c-@xmath6si clustering in the superdeformed state , it is easily understood that the excitation energies of these states decrease with the stronger spin - orbit force(i ) than with the weaker one(ii ) .
although the systematics of the binding energy in the @xmath29-shell region seems to be reproduced better by the force(ii ) than the force(i ) , however , the excitation energy of the superdeformed state is reasonably reproduced with the force(i ) while they are largely overestimated by the force(ii ) .
the force(i ) also reproduces the excitation energy of the lowest negative - parity state(@xmath108 ) rather well than the force(ii ) as shown later .
therefore , we examine mainly the results(i ) , because our interest is in excited states of @xmath0ca , especially in the properties of the superdeformed state .
we should note again that the structure of the superdeformed state is stable in both cases ( i ) and ( ii ) , and most of the intra - band properties of the superdeformed band does not so much depend on the choice of the interaction(i ) or ( ii ) except for the relative energy to the ground state and other rotational bands .
now we turn to negative - parity states . in fig.[fig : beqc - np ] , we show the energies of @xmath109(@xmath110 ) and @xmath111(@xmath112 ) , obtained with the interaction(i ) , as functions of ( a ) @xmath57 and ( b ) @xmath88 .
it is found that the constraint on @xmath113 acts as the constraint on deformation @xmath88 in negative - parity states as well as in the positive - parity ones . in the small deformation region ,
the lowest state is the @xmath108 which corresponds to a @xmath114-@xmath115 state .
the difference between the minimum energy of the @xmath111 state and that of the @xmath11 state is 4.7 mev , which well corresponds to the experimental excitation energy 3.74 mev of the lowest @xmath111 state .
we here comment that the interaction ( ii ) gives a much larger energy difference as 6.9 mev than the interaction ( i ) and fails to reproduce the experimental excitation energy of the @xmath111 state . as @xmath57 becomes large , the @xmath109 state becomes lower than the @xmath108 state and a @xmath116 rotational band with a prolate intrinsic state is built on this state .
we illustrate the density distributions of the intrinsic states for the negative parity states , @xmath117 in fig.[fig : ca - dense](g)-(i ) . for each @xmath57
, it is found that intrinsic structure of @xmath117 for the negative - parity state is quite similar to that of the prolate solution @xmath118 for the positive - parity state . in particular , the negative - parity state with @xmath119 has the parity - asymmetric shape because of the clustering which is regarded as the @xmath5c+@xmath6si - like structure as well as the case of positive parity state . as a result , the largely deformed state forms the @xmath116 rotational band consisting of @xmath120 due to the @xmath25 deformation , which may have a link to the positive - parity bands as a parity doublet .
= 0.5 = 0.5 in the obtained wave function @xmath63 , there exist local minimum solutions in the energy curve as a function of @xmath57 .
they are also local minima with respect to the deformation @xmath88 and considered to be approximate intrinsic wave functions for the corresponding deformed bands . in order to satisfy the orthogonality among the levels and to obtain better wave functions
, we calculate the level scheme by superposing the spin - parity projected wave functions , @xmath66 . in the calculations with the case(i )
interaction , we use @xmath121@xmath122 , @xmath123 , @xmath124 , @xmath125 , @xmath126 , @xmath127 , @xmath128 , @xmath129 , @xmath130 with prolate shapes and @xmath131 with the oblate shape as the basis for the positive - parity states . for the negative - parity states , we adopt @xmath121@xmath132 , @xmath133 , @xmath134 , @xmath135 , @xmath136 , @xmath137 , @xmath138 , @xmath123 , @xmath124 , @xmath139 .
the obtained level scheme is shown in fig.[fig : level ] . by analysing such properties as the dominant components and @xmath1 transition strengths of the obtained levels , we classify the levels into groups , which are labeled in fig.[fig : level](b ) and ( c ) . in the positive - parity levels ,
we find @xmath140 rotational bands corresponding to the shape coexistence:(b)the oblate band and ( a3)the superdeformed band .
the states in the groups ( a1 ) and ( a2 ) are built from the mixing of the spherical intrinsic state @xmath141 and the normal prolate state @xmath142 , while the side band ( a2(k=2 ) ) of the normal deformation appears from the dominant @xmath143 component due to the triaxial nature of the intrinsic state . in fig.[fig : band ] , we plot the calculated excitation energies as a function of a spin @xmath144 .
we also show the experimental data of the bands labeled `` band 2 '' and `` band 1 '' in ref.@xcite , which are assigned to be the normal - deformed band and the superdeformed band , respectively .
since the band ( a2 ) contains the configuration mixing , the calculated level structure is not consistent with the rigid rotor levels and does not well agree to the experimental data of the normal - deformed band ( labeled `` band 2 '' in ref.@xcite ) .
moreover , it is higher than the superdeformed band in the present results(i ) , however , we tentatively assign this band as the normal - deformed band because of rather strong @xmath1 transitions in ( a2 ) band . on the other hand ,
the calculated energy spectra of the superdeformed band behave as that of the rigid rotor and agrees to the experimental data for `` band 1 '' , though the excitation energy of the band - head state is slightly higher than the experimental data .
the dominant components of the superdeformed band in the low - spin levels is @xmath145 , which have a @xmath5c+@xmath6si - like clustering .
another interesting character of the superdeformed band is its particle - hole property . as discussed later in detail
, @xmath145 has about 90% overlap with the band - head of the superdeformed band , and its intrinsic state @xmath146 is regarded as a `` @xmath3-@xmath4 '' state .
this particle - hole feature of the superdeformed band is supported by the experimental fact that the @xmath1475.21 mev ) , @xmath1485.63 mev ) , and @xmath1496.54 mev ) in the superdeformed band are strongly populated in 8-nucleon - transfer reactions , @xmath14s(@xmath5c,@xmath12)@xmath0ca@xcite .
however , it is not consistent with the gcm calculations of skyrme hf+bcs@xcite , where the superdeformed band can not be described by a certain @xmath19-@xmath20 configuration but contains a mixture of various configurations such as @xmath8-@xmath9 , @xmath27-@xmath28 , and @xmath3-@xmath4 states . in order to check the stability of the superdeformed band for the choice of the intrinsic wave functions to be superposed
, we use another set of the basis as @xmath150@xmath122 , @xmath151 , @xmath123 , @xmath152 , @xmath153 , @xmath124 , @xmath125 , @xmath154 and find that the level scheme and the @xmath1 transition strengths in the superdeformed band are stable against the choice of the basis , and the properties of the states in bands ( a1 ) and ( a2 ) are qualitatively unchanged . in the presetn calculations , the number of the basis in the superposition is 10 at most and is much smaller than the gcm calculations in ref.@xcite .
it also should be noted that , we did not obtain the basis with the quadrupole deformation @xmath155 as shown in fig.[fig : beqc - pp ] . for the detailed investigation of the band mixing ,
it is expected to be important to perform gcm calculations with a large number of the basis concerning the two - dimensional constraints on the @xmath57 and @xmath88 .
the present results of the weak mixings between different @xmath156 configurations seem to be consistent with the experimental facts , however , we comment that the pairing correlations , which were taken into account in ref.@xcite , are ignored in the present calculations though we think they are partially included through the spin - parity projections and superposition of the basis .
since the pairing correlations increase configuration mixings in general , it would be valuable to estimate the pairing effects on the stability of the superdeformed band .
in addition to the prolate bands , the oblate band appears at almost the same excitation energy as that of the superdeformed band .
the moment of inertia is smaller than the superdeformed band .
the oblate deformation is predicted also in the macroscopic - microscopic calculations@xcite and in the mean - field calculations@xcite .
however , since an oblate band has not been experimentally assigned yet , the existence of an oblate shape in @xmath0ca is an open problem .
we also find highly excited bands ( a4 ) and ( a5 ) which arise from the developed @xmath5c+@xmath6si clustering .
the details of these bands are discussed in the next section . in the negative - parity states(fig.[fig : level]c ) , the lowest @xmath111 and @xmath157 states appear from almost the spherical states .
the theoretical excitation energies are in good agreement with the experimental data of the @xmath158(3.74 mev ) and the @xmath159(4.49 mev ) .
we find two @xmath116 rotational bands ( c2 ) and ( c3 ) which arise from the parity - asymmetric shape in the prolate states .
since the energies of the @xmath111 and @xmath157 states in these ( c2 ) and ( c3 ) bands are raised by the mixing with the lowest @xmath111 and @xmath157 states , the level structure is somehow out of that for the typical @xmath116 rotational band .
the @xmath160 state(@xmath161 ) in the band ( c2 ) is dominated by @xmath162 with the overlap @xmath163 90% , while the @xmath164 state(@xmath165 ) in the band ( c3 ) has the dominant @xmath166 with 70% overlap .
we should note that the @xmath160 and @xmath164 states have large overlap with also the @xmath167 states projected from the intrinsic states for the positive - parity , @xmath142 and @xmath168 , respectively .
it is reasonable because @xmath117 is almost the same state as the @xmath118 when the same constraint @xmath169 is imposed . in order to find the possible parity doublet of the superdeformed band
, we pay special attention to the @xmath168 component in the largely deformed negative - parity band ( c3 ) and in the superdeformed band ( a3 ) with positive - parity .
the overlap @xmath170 in the bandhead of the @xmath116 band(c3 ) is about 70% , while the bandhead @xmath11 of the superdeformed band(a3 ) has 80% overlap with @xmath171 .
then it leads to an interpretation that the largely deformed negative - parity band ( c3 ) and the positive - parity superdeformed band ( a3 ) are approximately the parity doublets @xmath116 and @xmath140 arising from the intrinsic state @xmath168 , which has the parity asymmetric shape with the @xmath25 deformation due to the @xmath5c+@xmath6si - like clustering as shown in fig.[fig : ca - dense ] .
next we discuss the energy levels calculated with the interaction(ii ) .
we use @xmath121@xmath122 , @xmath123 , @xmath124 , @xmath125 , @xmath126 with prolate deformations , @xmath131 with the oblate shape , and @xmath172 , @xmath173 , @xmath174 with @xmath12-cluster - like structure as the basis in superposing @xmath175 .
the calculated excitation energies are shown in fig.[fig : band-2 ] .
the rotational bands of oblate deformation , prolate one and superdeformation coexist as well as in the results(i ) .
the excitation energies of these bands with the interaction(ii ) are much higher than those of the results(i ) , and also are inconsistent with those of the experimentally known low - lying states . for example , the oblate band ( b ) starts from 17 mev excitation energy , while the bandhead energy of the superdeformed band ( a3 ) is 26 mev .
one of the characteristics of the results(ii ) is that a @xmath12-cluster - like band ( d ) appears at the relatively lower energy than the superdeformed band .
it has been known that the low - spin states in the normal - deformed band built on the @xmath7(3.35 mev ) state are strongly populated in @xmath12-transfer reactions , and they have relatively larger @xmath12-spectroscopic factors than other low - lying states @xcite .
therefore , it is expected that the components of the @xmath12-cluster state should be contained in the normal - deformed band . however , in the present calculations we failed to describe the @xmath12-cluster components in the low - lying states , and even in the results(ii ) , the components of the @xmath12-cluster state do not mix with the normal - deformed band ( a2 ) except for the @xmath15 state .
it is because the energy difference between the bands ( a2 ) and ( d ) are large in the present calculations . in order to solve this problem
, it may be necessary to introduce a suitable nuclear interaction and extended model wave functions .
= 0.4 = 0.4 = 0.4 we calculate the @xmath1 transition strengths and the electric quadrupole moments for the excited states .
we also extract quantities related to the intrinsic deformation from the physical observables such as @xmath176 and electric quadrupole moments . by using the collective model approximation @xcite
, we define the intrinsic quadrupole moment and the deformation by the @xmath1 transitions as , @xmath177 where @xmath178 is a clebsch - gordan coefficient .
@xmath179 stands for `` transition '' .
we also extract the intrinsic quantities from the observable electric quadrupole moments @xmath180 as follows , @xmath181 here @xmath182 stands for `` spectroscopic '' . in the table [ tab :
be2 ] , [ tab : be2np ] and [ tab : be2 - 18 ] , we show the calculated @xmath176 and @xmath180 moments , and related quantities .
the @xmath1 transitions are small among the states in the group ( a1 ) , which are dominated by spherical states and considered not to be a rotational band of a deformed state .
on the other hand , the prolate band ( a2 ) and the oblate one(b ) have moderate transition strengths , while the @xmath1 transitions are remarkably strong in the superdeformed band ( a3 ) . in the results with the interaction(i ) , the deformation parameter @xmath183 extracted from the @xmath176 is consistent with @xmath184 evaluated with the @xmath180 moment except for negative - parity bands and high spin states in the prolate bands ( a1),(a2 ) , and ( a3 ) . in fig.[fig : beta ] , we show the behavior of the transition deformation @xmath183 with the increase of the total - angular momentum @xmath185 . in the oblate band ( b ) , the intrinsic deformation @xmath64 is almost constant with the increase of @xmath185 , because the oblate states do not mix with the other shapes .
on the other hand , @xmath183 in the bands ( a1 ) , ( a2 ) and ( a3 ) changes as @xmath185 increases .
the deformation of the low - spin states in the superdeformed band is the largest as @xmath186 . with
the increase of @xmath185 , @xmath183 of the superdeformed band ( a3 ) decreases , while that of the normal - deformed band ( a2 ) enlarges , and the @xmath183-lines for these two bands cross each other around @xmath187 .
we remark that the state mixing occurs in the region @xmath188 , and the components contained in ( a2 ) and ( a3 ) change around @xmath187 . as a result ,
the transition from the @xmath189 to @xmath190 is significant strong(see table [ tab : be2 ] ) .
therefore , another assignment may be possible as @xmath189 , @xmath191 and @xmath192 belong to the superdeformed band .
in fact , @xmath183 extracted from the @xmath193 among the @xmath194 , @xmath195 , @xmath196 , @xmath190 , @xmath189 , @xmath191 and @xmath192 are so large as shown by filled triangles in fig.[fig : beta ] , that these states can be assigned to compose a rotational band with the strong @xmath1 transitions . in the experimental data by ideguchi et al.@xcite , they observed the deviation from the constant moment of inertia in the superdeformed band and also the inter - band gamma transitions between normal deformation and superdeformation .
these facts may give an indication of the possible band mixing or some structure change around @xmath197 in the superdeformation .
further fine measurements of @xmath1 transition strengths will be helpful to see the details of the band mixing and to establish the assignment for the high spin states of the superdeformed band . in the results with the interaction(ii ) ( table [ tab : be2 - 18 ] ) , we also find strong @xmath1 transitions in the superdeformed band ( a3 ) .
the transitions of the prolate band ( a2 ) are weaker in the results(ii ) than in the case(i ) . as mentioned before , the @xmath140(d ) and @xmath198(d ) bands with the @xmath12-cluster structure
appear below the superdeformed band in the results(ii ) .
these bands have strong intra - band @xmath1 transitions , which give the larger transition deformations @xmath199 than the prolate band ( a2 ) .
= 0.4 let us make a comparison of intra - band transitions between the present results and the experimental data , and also compare them with other theoretical calculations .
the theoretical and experimental values of @xmath200 for the normal - deformed band and the superdeformed band are listed in table [ tab : q2 t ] .
the present results for the superdeformed band are in reasonable agreements with the experimental data .
they are also consistent with the results by gcm+hfbcs calculations . in the shell model calculations by poves _
, large @xmath200 values were obtained with the truncated calculations within @xmath3-@xmath4 configurations by using effective charges @xmath201 and @xmath202 for protons and neutrons , however , the @xmath200 get eroded when other particle - hole configurations are mixed in the calculations .
the @xmath200 moments of the band ( a2 ) are @xmath203 @xmath204 and comparable to the experimental values @xmath205
@xmath204 for the normal deformation(band 2 ) in ref.@xcite , but are smaller than the experimental data @xmath206 and @xmath207 @xmath204 in ref.@xcite
. our results are consistent with the gcm+hfbcs calculations except for the @xmath16 state , which has a small @xmath200 in the gcm+hfbcs result . as shown in table [
tab : be2 - 18 ] , we obtain the @xmath12-cluster band ( d ) with strong @xmath1 transitions in case of the interaction(ii ) .
the transition moment of the band ( d ) is @xmath208 @xmath204 and well agrees to the experimental data in ref.@xcite .
therefore , there might be an alternative assignment that the band ( d ) corresponds to the normal - deformed band built on the @xmath7(3.35 mev ) , though the excitation energy of the band ( d ) is much higher than the experimental data in the present calculations .
the @xmath209 band were experimentally established based on the measurement of @xmath176 @xcite .
the observed intra - band transitions are @xmath210 ( @xmath211 fm@xmath33 ) and @xmath213 ( @xmath211 fm@xmath33 ) @xcite .
the properties of this band were studied by sakuda and ohkubo with @xmath12+@xmath85ar(@xmath15)-cluster model @xcite , which well repoduces these intra - band transitions . in the present calculations , we find the @xmath198 band ( a2 ) with the prolate deformation in the results(i ) , and the @xmath198 band ( d ) with the @xmath12-cluster structure in the case of ( ii ) . in the band ( a2 ) , the calculated @xmath1 transition strengths are @xmath214=95 ( @xmath211 fm@xmath33 ) and @xmath215=38 ( @xmath211 fm@xmath33 ) , which are smaller than experimental @xmath176 values . on the other hands , the @xmath1 transition strengths in the band ( d )
are large enough as @xmath216 ( @xmath211 fm@xmath33 ) and @xmath217 ( @xmath211 fm@xmath33 ) , which well correspond to the data within the experimental error bars .
it means that the deformation of the @xmath209 band can be described by the @xmath12-cluster band ( d ) , which is obtained with the interaction ( ii ) .
next we discuss the moments of the negative - parity states . for the negative - parity states in the group ( c1 ) , the magnetic dipole moments of the @xmath111 and @xmath157 states are @xmath218 and @xmath219 , which well agree to the experimental values for the lowest @xmath158(3.74 mev ) and the lowest @xmath159(4.49 mev ) ; @xmath220 and @xmath221 . in the prolate bands ( c2 ) and
( c3 ) , the @xmath1 transitions are fragmented among the bands because the @xmath111 and @xmath157 states of these bands contain configuration mixing due to the existence of the lowest @xmath111 and @xmath157 states in the group ( c1 ) ( see table [ tab : be2np ] )
. then , we estimate the intrinsic deformation by looking into @xmath184 extracted from the @xmath180 moments instead of @xmath183 .
according to the calculated @xmath184 of the band - head state(@xmath167 ) , the deformation of the band ( c3 ) is evaluated as @xmath222 .
this is smaller than that of the superdeformed band ( a3 ) , which we raughly assign to the parity doublet of ( c3 ) . .
[ tab : be2 ] electric quadrupole moments @xmath180 ( @xmath223 @xmath204 ) and @xmath176 ( @xmath211fm@xmath33 ) for the positive - parity states calculated with the interaction(i ) . the extracted intrinsic quadrupole moments ( @xmath204 ) and deformation parameters are also listed . the definitions of @xmath224 and @xmath225 are explained in the text .
the labels of the bands correspond those in fig.[fig : level ] . [
cols="^,^,^,^,^,^,^,^,^,^ " , ]
appearence of molecular resonances is one of the important cluster aspects in @xmath29-shell nuclei . in the large prolate deformations of @xmath0ca
, we find the trend of developed @xmath5c+@xmath6si clustering which leads to the superdeformed band as mentioned in the previous section .
therefore , it is natural to expect possible existence of @xmath5c+@xmath6si molecular bands and their link with the superdeformed band in @xmath0ca
. it has been known that molecular resonances appear in such systems as @xmath5c+@xmath5c , @xmath5c+@xmath31o or @xmath31o+@xmath31o @xcite . in the microscopic calculations ,
there have been theoretical efforts to connect the molecular resonances with the low - lying deformed bands @xcite .
we should stress that the recent amd studies @xcite are the first theoretical works which can microscopically describe both the coexisting deformed states in low - energy region and the molecular resonances in the high - energy region without relying on assumptions of constituent clusters . in ref.@xcite ,
the relation between the theoretically predicted superdeformed band in @xmath14s and the @xmath31o+@xmath31o molecular bands has been studied .
it has been suggested that the superdeformed band with a considerable amount of the @xmath31o+@xmath31o component is regarded as the lowest nodal band in a series of @xmath31o+@xmath31o molecular bands , while the excited @xmath31o+@xmath31o bands arise due to the inter - cluster excitation and correspond to the observed molecular resonances . in the present results of the case(i ) , we find the excited rotational bands ( a4 ) and ( a5 ) as shown in fig.[fig : band ] .
the former and the latter exist at about 15 mev and 25 mev higher excitation energies than that of the superdeformed band , respectively .
these bands appear due to the superposition of the large prolate deformations with the @xmath5c+@xmath6si cluster structure . in fig.[fig : amp ] , we show the overlap of the states(@xmath226 ) in the prolate bands ( a1 ) , ( a2 ) , ( a3 ) , ( a4 ) and ( a5 ) with a single amd state : @xmath227 .
the low - spin(@xmath228 ) states in the superdeformed band is dominated by @xmath229 with about 90% overlap .
we should stress again that the @xmath5c+@xmath6si - like clustering actually appears in the @xmath146 ( see fig.[fig : ca - dense ] ) . on the other hand ,
the states in the excited band ( a4 ) contain the amd states with larger @xmath57 , and the components in the band ( a5 ) shift toward the further large @xmath57 region .
since the di - nuclear clustering develops further and two clusters go away from each other with the increase of @xmath57 , these excited bands ( a4 ) and ( a5 ) should be described by the excitation of inter - cluster motion .
in fact , we calculate the oscillator quanta of the subsystem consisting of the nucleons within each cluster(@xmath5c or @xmath6si ) , and find that each of the cluster is almost written by @xmath230 configurations . in other words ,
the increase of @xmath57 corresponds to just enlarging the inter - cluster distance , and hence , the rotational bands ( a4 ) and ( a5 ) arise due to the excitation of relative motion between @xmath5c and @xmath6si clusters .
moreover , it is suggested that the bands ( a4 ) and ( a5 ) may have tails in the inter - cluster motion because the components of these bands are spread into the broader @xmath57 ranges than that of the superdeformed band .
these results suggest an interpretation that the superdeformed band ( a3 ) and the higher bands ( a4 ) and ( a5 ) are regarded as a series of @xmath5c+@xmath6si molecular bands .
namely , the bands ( a4 ) and ( a5 ) appear as the higher nodal states built above the the superdeformed band .
these results have a good analogy to the feature of the superdeformed band and the @xmath31o+@xmath31o bands in @xmath14s discussed in ref.@xcite . in the experimental side
, the @xmath5c+@xmath6si resonances have been observed in the elastic scattering data in the backward - angle region @xcite .
the resonances at the energy @xmath231 and 30.2 mev , which correspond to the excited states of @xmath0ca at @xmath232 and 44 mev , are assigned to be @xmath233 states from the angular distributions .
there are other experimental implications of @xmath5c+@xmath6si molecular states in the low - energy fusion cross section@xcite , where distinct structures have been found in the energy region @xmath32=@xmath234 mev above the @xmath5c+@xmath6si threshold .
these are the candidates of the molecular resonances which might correspond to the @xmath5c+@xmath6si molecular bands ( a4 ) and ( a5 ) obtained in the present results . in order to make further investigations of the molecular bands and give more quantitative discussions
, it should be important to superpose a large number of the amd states .
it also may be effective to take into account the excitation inside the clusters as coupled channel calculations for the description of the detailed band structure .
= 0.35 in an amd wave function , the single - particle wave packets , @xmath37 , in eq.([eq : varphi ] ) are not orthogonal to each other . in order to study the mean - field character , it is useful to transform them to the hf - like single - particle orbits @xmath235 , which are orthonormal to each other and form the total wave function equivalent to the original amd wave function .
we extracted the single - particle orbits @xmath235 from @xmath63 , as explained in refs@xcite , and analyze them .
as mentioned before , the low - spin states of the superdeformed band is dominated by the amd state , @xmath236 .
we find that the intrinsic state , @xmath146 , contains eight @xmath237-like single - particle levels , which are the highest 4 levels among proton orbits and the highest 4 levels among the neutron orbits .
each of these levels has dominant parity - odd component as the ratio @xmath238 .
the density distributions of these levels are shown in fig.[fig : ca - single](a ) and ( b ) .
hereafter , we use the label @xmath239(=1,@xmath240,20 ) of @xmath235 for the @xmath239th highest single - particle levels in the proton orbits .
it is found that the properties of the eight valence levels in @xmath146 are associated with the longitudinal @xmath237 orbits which have @xmath241 and @xmath242 angular dependent terms .
therefore , we regard the superdeformed state as a `` @xmath3-@xmath4 '' state , though it is not equivalent to pure particle - hole states in the mean - field picture .
the origin of the parity asymmetric shape with @xmath5c+@xmath6si cluster in @xmath146 will be discussed later .
the prolate band ( a2 ) contains the significant component of the amd state projected from the intrinsic state , @xmath142 .
the valence orbits in this state are the eight levels which contain 50% parity - odd components and 50% parity - even ones as shown in figures [ fig : ca - single](d ) and ( e ) .
it means that the parity projected state @xmath243 can be approximately described by the mixing of @xmath244-@xmath245 , @xmath8-@xmath9 and @xmath27-@xmath28 configurations .
we comment that @xmath142 is not a local minimum in the energy curve as a function of @xmath57 nor @xmath64 .
the oblate solution @xmath246 does not mix with the prolate states , and independently forms the oblate band ( b ) .
this state can be convincingly described by a @xmath8-@xmath9 state because 4 valence nucleons occupy almost parity - odd orbits with angular parts related to @xmath247 .
the property of a valence single - particle level of the four is shown in fig .
[ fig : ca - single](c ) .
the particle - hole nature is found also in the @xmath12-cluster states @xmath248 and @xmath249 obtained by the interaction(ii ) .
namely , these states have a @xmath8-@xmath9 feature as four valence neutrons occupy @xmath237-like orbits with 75% negative - parity component in the @xmath249 as seen in fig .
[ fig : ca - single](f ) .
thus , the single - particle levels in the intrinsic states disclose the many - particle many - hole characters in @xmath0ca . on the other hand ,
the cluster aspect has been found in prolate states which have parity - asymmetric shapes .
it raises a question how the cluster aspect links with the mean - field aspect .
here we examine the cluster aspect while paying attention to parity asymmetry of the single - particle orbits . in fig.[fig : single ] , we show the ratio of negative - parity component of the single - particle proton orbits in the prolate states @xmath118 .
the behavior of the neutron orbits is almost the same as that of the proton orbits . at almost the spherical @xmath97 region , 12 protons(@xmath250 )
occupy almost parity - even orbits and 6 protons(@xmath251 ) are in almost parity - odd orbits , which indicates the approximately @xmath29-shell closed configuration of this state . with the increase of @xmath57 ,
the parity - odd components in the highest four levels(@xmath252 ) increase and these four levels become almost pure parity - odd orbits at @xmath253 .
it means that four protons in the @xmath29-shell are excited into the @xmath237-shell . in the state @xmath254 , each of the higher ten levels @xmath255 has approximately good parity .
instead , the parity symmetry breaking in single - particle orbits is rather remarkable in the 11th and 12th levels(@xmath256 ) which are originally the lowest @xmath29-orbits as shown in fig.[fig : single ] ( b ) . in order to see the effect of the single - particle features on the collective shape of the total system @xmath254
, we show the density distribution summed up for the protons in the lower levels(@xmath257 ) and that for the protons in the higher levels(@xmath255 ) in fig.[fig : ca70-single ] .
the most striking thing is that 10 protons in the lower levels form a parity asymmetric shape , which reminds us of the @xmath12+@xmath31o clustering in @xmath81ne system .
these results imply that the parity asymmetry in this state dominantly originates not in the valence nucleons near the fermi surface but in the @xmath12-cluster - like correlation in the core part .
it is inconsistent with an mean - field picture that nucleons at the surface may contribute such an exotic shape as the octupole deformation .
we here remind the reader that this state(@xmath254 ) is the dominant component of the superdeformation and has the @xmath5c+@xmath6si - cluster structure(fig.[fig : ca70-single](c ) ) .
the @xmath5c+@xmath6si clustering is also seen in the aspect of molecular resonances .
this seems to be inconsistent with possible @xmath14s+2@xmath12 clustering , which may be naively expected from the @xmath3-@xmath4 nature of the superdeformed state . by the analysis of single - particle orbits
, we can interpret the appearance of the @xmath5c cluster as follows .
the origin of the @xmath5c cluster in the superdeformation is understood by the correlation within 12 nucleons ; 4 @xmath29-shell nucleons with the @xmath12 correlation and the 8 valence nucleons in the @xmath237-like orbits .
it leads to a new picture of the coexisting cluster and mean - field aspect .
namely , the deeply bound nucleons compose the parity - asymmetric core @xmath81ne with the @xmath12-cluster correlation , while the single - particle levels at the energy surface have particle - hole nature . in the further large @xmath57 region , the particle - hole nature disappears because of the spatial development of @xmath5c+@xmath6si clustering .
= 0.9 = 0.4 = 0.3
we studied deformed states in @xmath0ca with antisymmetrized molecular dynamics(amd ) by using effective nuclear interactions which contain finite - range 2-body and 3-body forces . in the framework of a constraint amd method
, we performed parity projection before energy variation , and total - angular - momentum projection after the energy variation .
the obtained amd wave functions are superposed to get better wave functions . by analyzing the calculated results of the intrinsic states , level scheme , and @xmath1 transition strengths
, it was found that the rotational bands with various deformations appear in @xmath0ca as well as the spherical ground state .
namely , above the spherical ground state , the rotational bands arise from the normal - deformed state and the superdeformed state as well as the oblate state . in the results with a weaker spin - orbit force
, we also obtained the @xmath12-cluster - like bands .
the present results of the superdeformed band reasonably reproduce the properties of the experimental superdeformed band build on the @xmath2(5.21 mev ) . on the other hand
, we could not obtain a satisfactory description of the experimental deformed band build on the @xmath7(3.35 mev ) .
possible assignment of this band is to the theoretical normal - deformed band , the @xmath12-cluster band , or admixture of them . in the anslysis of single - particle orbits
, we found the particle - hole nature of the superdeformed state , the oblate state and the @xmath12-cluster state , which are dominated by the @xmath3-@xmath4 , @xmath8-@xmath9 and @xmath8-@xmath9 configurations , respectively . on the other hand
, the normal deformation in the present results contains mixing of @xmath244-@xmath245 , @xmath8-@xmath9 and @xmath27-@xmath28 configurations .
one of new findings in the present study is that the superdeformed state has a parity asymmetric shape with @xmath5c+@xmath6si - like cluster structure .
from the point of view of a single - particle picture , we discussed the relation between the cluster aspect and the many - particle many - hole aspect .
we found that the origin of the @xmath5c clustering in the superdeformed state is understood by the correlation of 12 nucleons ; 4 @xmath29-shell nucleons with the @xmath12-cluster correlation and the 8 valence nucleons in the @xmath237-like orbits .
it reveals the coexistence of cluster and mean - field aspect in the superdeformed state .
namely , the @xmath12-cluster correlation in deeply bound nucleons composing the core @xmath81ne plays an important role in the parity asymmetric shape , while the particle - hole nature arises in the single - particle levels at the energy surface . the present result of the dominant @xmath3-@xmath4 feature in the superdeformation seems to be consistent with the strong population of the members in the superdeformed band observed in 8-nucleon - transfer reactions;@xmath14s(@xmath5c,@xmath12)@xmath0ca@xcite .
however , it is inconsistent with the gcm+hfbcs calculations@xcite , where neither the @xmath8-@xmath9 nor the @xmath3-@xmath4 hf solutions can be assigned to the bandheads of the normal - deformed band or superdeformed band
. one of the advantages of the present framework is the parity projection before energy variation , which is considered to be important in arising of the parity - asymmetric deformation .
we think that one of the key properties which stabilize the `` @xmath3-@xmath4 '' character in the superdeformed band is the @xmath5c+@xmath6si clustering which causes a parity asymmetric shape .
compared with the gcm+hfbcs calculations@xcite , the higher correlations are ignored in the present calculations .
for example , the number of the basis in the superposition is ten at most , which is much smaller than the case of the gcm+hfbcs calculations@xcite .
moreover , since we adopt the constraint on the total number of ho quanta in the energy variation , some basis are missed in terms of gcm with respect to the generator coordinate @xmath88 , which are often used in the gcm+hf calculations .
the pairing correlations are also ignored .
it is an remaining problem to check how the properties of the superdeformation are affected by introducing these effects . above the superdeformed band
, we found possible higher rotational bands based on @xmath5c+@xmath6si - clustering in the large prolate deformations .
it is predicted that @xmath5c+@xmath6si molecular bands may be built due to the excitation of the inter - cluster motion .
the present results suggest that the superdeformed band and these higher molecular bands are regarded as a series of @xmath5c+@xmath6si molecular bands .
this means that the @xmath5c+@xmath6si cluster aspect is rather prominent in the superdeformed state than @xmath14s+@xmath258 aspect , although the latter may be naively expected from the @xmath3-@xmath4 feature of this state .
we also suggest the negative - parity bands caused by the parity asymmetric deformation . in the present work ,
we use the stronger spin - orbit force than the nuclear interaction adopted in ref.@xcite to quantitatively reproduce the excitation energies of the superdeformed band and negative parity states . with the original weak spin - orbit force , the excitation energies of all the excited states in @xmath0ca are overestimated .
we would like to stress again that the intra - band properties such as energy spectra , @xmath1 transitions and intrinsic structure of the superdeformed band are not sensitive to choice of either interaction . in order to solve the remaining problems
, it may be necessary to introduce a suitable nuclear interaction and extended model wave functions .
the authors would like to thank prof .
h. horiuchi and dr .
taniguchi for many discussions .
they are also thankful to dr .
ideguchi and dr .
inakura for valuable comments .
the computational calculations in this work were supported by the supercomputer projects of high energy accelerator research organization(kek ) .
this work was supported by japan society for the promotion of science and a grant - in - aid for scientific research of the japan ministry of education , science and culture .
a part of the work was performed in the `` research project for study of unstable nuclei from nuclear cluster aspects '' sponsored by institute of physical and chemical research ( riken ) .
9 j. r. macdonald , d. h. wilkinson and d. e. alburger , phys .
c * 3 * , 219 ( 1971 ) , and references therein .
w. j. gerace and a. m. green , nucl .
a93 * , 110 ( 1967 ) .
w. j. gerace and a. m. green , nucl
. phys . * a123 * , 241 ( 1969 ) .
r. r. betts , h. t. fortune , j. n. bishop , m. n. i. al - jadir and r. middleton , nucl . phys . * a292 * , 281(1977 ) .
h. t. fortune , m. n. i. al - jadir , r. r. betts , j. n. bishop and r. middleton , phys .
c * 19 * , 756 ( 1979 ) .
t. yamaya _
et al . _ ,
a573 * , 154(1994 ) .
t. yamaya , k. katori , m. fujiwara , s. kato and s. ohkubo , prog .
. suppl . * 132 * , 73 ( 1998 ) .
s. ohkubo and k. umehara , prog .
theor . phys . * 80 * , 598 ( 1988 ) . t. sakuda and s. ohkubo , phys . rev .
c * 49 * , 149 ( 1994 ) .
t. sakuda and s. ohkubo , prog .
suppl . * 132 * , 103 ( 1994 ) .
r. middleton , j. d. garrett and h. t. fortune , phys .
lett . * 39b * , 339 ( 1972 ) .
e. ideguchi _
et al . _ ,
lett . * 87 * , 222501 ( 2001 ) c. j. chiara _
et al . _ , phys . rev .
c * 67 * , 041303(r ) ( 2003 ) .
d. c. zheng , d. berdichevsky and l. zamick , phys .
c * 38 * , 437 ( 1988 ) .
d. c. zheng , l. zamick and d. berdichevsky , phys .
c * 42 * , 1004 ( 1990 ) .
t. inakura , s. mizutori , m. yamagami and k. matsuyanagi , nucl .
phys . * a 710 * , 261 ( 2002 ) . g. leander and s. e. larsson , nucl
a239 * , 93 ( 1975 ) .
m. bender , h. flocard and p. -h .
heenen , phys . rev .
* c 68 * , 044321 ( 2003 ) .
a. poves , e. caurier , f. nowacki and a. zuker , eur .
j. a * 20 * , 119(2004 ) y. kanada - enyo , m. kimura and h. horiuchi , comptes rendus physique vol.4 , 497 ( 2003 ) .
m. kimura , phys .
c * 69 * , 044319 ( 2004 ) .
m. kimura and h. horiuchi , phys .
c * 69 * , 051304(r ) ( 2004 ) .
r. a. racca , p. a. deyoung , j. j. kolata and r. j. thornburg , phys . lett . *
129b * , 294 ( 1983 ) .
r. ost _ et al .
c * 19 * , 740 ( 1979 ) . y.kanada-enyo and h.horiuchi , prog .
. phys . * 93 * , 115 ( 1995 ) .
m. kimura , y. sugawa and h. horiuchi , prog .
* 106 * , 115 ( 2001 ) .
m.kimura , h.horiuchi prog . theor . phys . *
107 * , 33 ( 2002 ) .
y.kanada-enyo , phys.rev .
c * 71 * , 014303 ( 2005 ) .
y.taniguchi , m.kimura and h.horiuchi , prog .
phys . * 112 * , 475 ( 2004 ) .
y. kanada - enyo and y. akaishi , phys .
c * 69 * , 034306 ( 2004 ) .
y. kanada - enyo , h. horiuchi and a. ono , phys .
c * 52 * , 628 ( 1995 ) ; y. kanada - enyo and h. horiuchi , phys .
c * 52 * , 647 ( 1995 ) .
y. kanada - enyo and h. horiuchi , prog .
phys . suppl . * 142 * , 205 ( 2001 ) .
s. marcos , h. flocard , and p. h. heenen , nucl .
phys . * a 410 * , 125 ( 1983 ) .
n. yamaguchi , t. kasahara , s. nagata and y. akaishi , prog .
* 62 * , 1018 ( 1979 ) ; r. tamagaki , prog .
39 * , 91 ( 1968 ) .
a. bohr and b. r. mottelson , _ nuclear structure vol .
ii _ , ( w. a. benjamin , new york , 1975 ) . at .
. data tables * 23 * , 3 ( 1979 ) .
j. l. wood _ et al .
_ , phys . rep . * 215 * , 101 ( 1992 ) .
p. braun - muzinger and j. barrette , phys
. rep . * 87 * , 209 ( 1982 ) .
w. greiner , j. y. park and w. scheid , _
`` nuclear molecules '' _ , ( world scientific , singapore , 1995 ) .
d. baye and p .- h .
heenen , nucl . phys .
* a283 * , 176 ( 1977 ) .
k. kato , s. okabe and y. abe , prog .
theor . phys . *
74 * , 1053 ( 1985 ) .
y. kanada - enyo , _ proc . of frontiers of collective motions(cm2002 ) _ , aizu , japan(2002 ) , ed . by h. sagawa and h. iwasaki , 254 .
a.dte , h.horiuchi and y.kanada-enyo , phys .
c * 56 * , 1844 ( 1997 ) . | deformed states in @xmath0ca are investigated with a method of antisymmetrized molecular dynamics . above the spherical ground state ,
rotational bands arise from a normal deformation and a superdeformation as well as an oblate deformation .
the calculated energy spectra and @xmath1 transition strengths in the superdeformed band reasonably agree to the experimental data of the superdeformed band starting from the @xmath2 state at 5.213 mev . by the analysis of single - particle orbits
, it is found that the superdeformed state has particle - hole nature of an @xmath3-@xmath4 configuration .
one of new findings is parity asymmetric structure with @xmath5c+@xmath6si - like clustering in the superdeformed band .
we predict that @xmath5c+@xmath6si molecular bands may be built above the superdeformed band due to the excitation of inter - cluster motion .
they are considered to be higher nodal states of the superdeformed state .
we also suggest negative - parity bands caused by the parity asymmetric deformation . |
understanding the structure and kinematics of spiral galaxies , in particular explaining their rotation curves at large galactic radii , remains one of the pressing open questions in astrophysics .
optical observations of galactic rotation curves find that rather than falling off as one would expect from galaxy models where the mass traces the observed light , the rotational velocities remain constant at large radii .
these findings are further borne out by radio observations of the 21 cm line , from hi gas in the outer parts of the galactic disk .
an overview of rotation curves is provided in @xcite , @xcite and @xcite .
there is strong evidence from the cmb data for considerable amounts of non - baryonic dark matter on the cosmological scale @xcite .
while is it enticing to imagine that the dark matter problems on the galactic and cosmological scales have a common resolution , this need not necessarily be the case .
our focus in this paper will be on the galactic dark matter problem as manifested in the ubiquitous observation of flat rotation curves .
a number of ideas have been put forth to account for flat rotation curves .
all require some form of new physics .
these ideas include : 1 .
new particles : missing mass in the form of galactic dark matter , most likely non - baryonic @xcite .
2 . new interactions :
non - gravitational long - range couplings might exist , or gravitational physics might be subject to revisions over large distances .
new dynamics : scenarios such as modified newtonian dynamics ( mond ) in which gravity from visible matter is the only force acting but the system s response takes on new aspects .
the community consensus at present prefers the dark matter hypothesis , but galactic dark matter has thus far evaded all attempts to detect it .
we should strive to test , whenever possible , alternatives to the dark matter scenario .
motivated by the observed spiral galaxy rotation curves , @xcite proposed a modification of the dynamics of non - relativistic matter .
this modified behavior , termed mond for modification of newtonian dynamics , is conjectured to arise only in the regime of low accelerations .
mond is a proposed modification to an object s acceleration under an applied force , such that @xmath0 where @xmath1 is the acceleration expected under newtonian physics , and @xmath2 depends upon the mond acceleration scale @xmath3 , with @xmath4 .
a commonly adopted form is @xmath5 . in general a gravitating system s behavior under mond
can be described by taking the newtonian description and replacing @xmath6 , the coupling constant , by @xmath7 , with the understanding that the dynamics is being altered rather than the nature of the gravitational interaction . in this scenario
the mass of a galaxy resides in the ordinary astronomical components that we can detect by their emission or absorption of electromagnetic radiation , and the galaxy s light distribution traces out its mass distribution .
a review of mond as an alternative to dark matter is presented in @xcite . in the mond model
the response of a test particle to an applied force depends upon the magnitude of its absolute acceleration relative to a preferred frame , taken to be the local rest frame of the microwave background .
`` overacceleration '' at low values of @xmath8 then produces the observed rotation curves of spiral galaxies .
mond thereby eliminates the need for dark matter , at the expense of novel dynamics at low accelerations .
mond does a remarkably good job of fitting the rotation curves of galaxies across a wide range of surface brightness , with @xmath9 as the single free parameter @xcite .
the mond idea was recently placed on a more formal footing @xcite , but our approach will be cast in the original phenomenological framework , in terms of @xmath10 . from this standpoint , the formulation described above suggests an observational test for the self - consistency of mond . because @xmath10 depends on the ( scalar ) magnitude of a particle s total acceleration , comparing the vertical and rotational dynamics of test particles in the disk of a spiral galaxy provides a means to test for self - consistency .
this paper proposes a framework for carrying out such a test and illustrates the technique with recent data from m33 @xcite .
other recent efforts to investigate the viability of the mond hypothesis using kinematics include using galaxy clusters @xcite and globular clusters of stars @xcite .
our approach differs in that , as discussed in the following section , we are checking the self - consistency of mond rather than comparing the observed kinematics to a prediction .
we will adopt the common model for a galactic disk as a mass distribution with a volume matter density @xmath11 given by @xmath12 where @xmath13 and @xmath14 are characteristic scale lengths in the radial and vertical directions . in the mond scenario ,
the rotational speed of a thin galaxy with a mass structure described by equation ( 1 ) obeys ( adapting the expression for newtonian physics from @xcite ) @xmath15\ ] ] where @xmath16 and @xmath1 are the mond and newtonian accelerations respectively , @xmath6 is newton s gravitational constant , @xmath13 is the disk scale length , @xmath17 is the surface mass density at @xmath18 , @xmath19 is a dimensionless radius variable , @xmath20 and @xmath21 are @xmath22 order modified bessel functions , and @xmath23 gives rise to an object s modified response in the radial direction .
the vertical kinematics of the objects in the disk are described by @xmath24 , where @xmath25 is the local surface mass density , @xmath14 is the vertical scale height , and @xmath26 accounts for mondian behavior in the vertical direction .
given these two expressions and the exponential radial form for @xmath25 given in equation ( 1 ) , we can construct the dimensionless consistency parameter " ratio @xmath27 .
_ regardless of the local dynamical law that describes a star s response to feeble forces , @xmath28 should be unity , at all radii .
this is true even if the system makes a transition from the newtonian to the mond regime .
_ the expressions given above allow us to write @xmath28 as @xmath29.\ ] ] for a disk - only system , the combination of the observables on the right side should equal to unity at all radii in both the newtonian ( where @xmath30=1 ) and mondian ( where @xmath31 1 ) regimes .
@xmath28 can be understood as a parameter testing whether the tracer material is behaving in a self - consistent fashion .
note that this expression is independent of both the mass - to - light ratio of the disk material and of the central surface density of the disk .
ideally , one would obtain both face - on and edge - on observations of a single galaxy .
this would then allow the measurement of the rotation curve , of the vertical velocity dispersion , and of the scale lengths @xmath13 and @xmath14 . in practice , this is of course not possible .
there is , nevertheless , a realistic possibility of measuring the _ radial dependence _ of @xmath28 , using kinematic information alone .
a nearly face - on bulgeless spiral galaxy would provide a powerful testbed .
the apparent ( projected ) rotation curve would be suppressed by an unknown @xmath32 factor , where @xmath33 is the inclination angle , giving an observed @xmath34 .
this would simply rescale @xmath35 by an overall multiplicative factor , so that while still radius independent it will differ from unity by @xmath36 .
given that typical rotational velocities are a few 100 km / s while velocity dispersions are tens of km / s , a tenfold suppression of the rotation curve is quite tolerable .
this corresponds to using systems with inclination angles as small as a few degrees .
the advantage of using a face - on galaxy is that the line - of - sight velocity dispersion is a clean measure of vertical velocity dispersion , uncontaminated by other components of the velocity ellipsoid within the galactic disk .
the main observational challenge is obtaining high signal - to - noise measurements and then extracting both the circular velocity field and the vertical velocity dispersion .
this is a tractable problem .
the projected circular velocity produces a doppler - shifted centroid of a spectral feature .
the velocity dispersion can be determined from either the broadening of spectral lines , or by the width of the velocity distribution of a set of individual resolved objects .
we emphasize the fact that as long as the disk scale height @xmath14 is independent of @xmath37 , and the self - supporting stellar disk dominates @xmath11 , only velocity data are needed to test for variation in @xmath28 . by selecting nearly face - on galaxies for this test
, we lose the ability to measure their vertical scale height and must instead appeal to a statistical argument that invokes measurements of edge - on analogous systems .
measurements of the light distribution of edge - on galaxies in the near infrared , using images from the 2mass survey @xcite indicate that for typical galaxies the vertical scale height _ is _ independent of galactic radius , with typical values of @xmath38 varying between 0.1 and 0.4 .
we will therefore adopt the working hypothesis that @xmath14 in equation ( 3 ) is independent of @xmath37 .
there are a few instances where a galaxy disk s vertical velocity dispersion has been measured ( e.g. @xcite ) . a recent data set on the kinematics of m33 @xcite provides an interesting test case .
these authors obtained line - of - sight velocity data on 140 planetary nebulae in m33 .
this archival data set provides an opportunity for a concrete example of the @xmath35-violation test outlined above .
this local group galaxy is inclined at 56 degrees to the plane of the sky .
various determinations ( using multiple techniques ) yield a distance modulus of 24.8 @xmath39 0.1 _ mag_. the radial scale lengths for light are @xmath40 2.5 kpc and @xmath41 1.56 kpc in the v and k bands , respectively .
this implies ( taking the 2mass - derived typical values for @xmath42 ) a likely vertical scale height @xmath14 in m33 of a few hundred parsecs .
the m33 inclination angle of 56 degrees is not optimal for our purposes , but ciardullo et al ( 2004 ) provide their best estimates of m33 s rotational velocity and vertical velocity dispersion as a function of galactocentric distance .
their results are presented in table 1 .
the m33 circular velocity at @xmath37=10 kpc implies , in the mond scenario , that objects at that radius should experience a threefold increase in their radial ( and hence vertical ) acceleration , relative to the newtonian value .
the typical acceleration component normal to the disk is @xmath43 , well into in the mond regime .
it is therefore the radial component that determines the kinematics of the objects in the disk .
an extended rotation curve for m33 is presented in @xcite .
rrrrr 0.5 & 40 & 21 & 0.66 + 1 & 55 & 18.7 & 0.64 + 2 & 80 & 17 & 0.66 + 3 & 90 & 14 & 0.60 + 4 & 98 & 12.5 & 0.55 + 5 & 100 & 10.5 & 0.49 + 6 & 105 & 10 & 0.45 + 7 & 106 & 8 & 0.41 + 8 & 107 & 7.5 & 0.37 + 9 & 108 & 5.5 & 0.34 + 10 & 109 & 5 & 0.31 + [ tab : table1 ] the observational data in table 1 , in conjunction with equation ( 3 ) , provide us with the opportunity to map out the radial dependence of the consistency parameter , @xmath35 , across the face of m33 .
in the mondian view , the ordinary astronomical inventory of m33 , plus novel dynamics , produce the observed rotation curve .
the light distribution across m33 should then trace the galaxy s mass distribution .
we have evaluated the radial dependence of @xmath35 using values of @xmath13 obtained from visible and near - ir wavelengths .
the measurements of m33 s kinematics from table 1 were used in conjunction with equation ( 3 ) to determine @xmath28 , for different values of the structural parameters @xmath13 and @xmath14 .
one choice we made was to set @xmath44 1.56 kpc and @xmath45 0.4 kpc , corresponding to roughly the midpoint of the 2mass aspect ratio distribution .
the resulting @xmath28 values are shown in figure 1 . since for the m33 data
@xmath13 is better constrained than @xmath14 we also explored values of @xmath14 that produced the @xmath28 curve closer to unity , by fitting for @xmath14 while minimizing the sum @xmath46 for the radii listed in table 1 . with @xmath13
fixed at 1.56 and 2.5 kpc , the best - fit values of @xmath14 were 0.23 and 0.079 kpc , respectively , corresponding to values of @xmath38 of 0.14 and 0.03 .
the best - fit value of @xmath14 for the larger disk scale length is only 80 pc , and corresponds an unreasonably thin disk .
the corresponding @xmath28 curves for these cases are also shown in figure 1 .
the curves in figure 1 indicate that , as shown in equation ( 3 ) , the value of @xmath13 determines the _ shape _ of the @xmath28 curve , and the @xmath14 parameter only provides an overall multiplicative scaling that can be adjusted to drive the average @xmath35 value towards unity . in this mass - traces - light analysis ,
the @xmath35 parameter changes by a factor of 310 ( depending upon the fit used ) between the inner and outer regions of the m33 disk . in this illustrative example
the @xmath28 behavior appears inconsistent with the mond scenario .
the sense of the discrepancy , with @xmath35 values less than one , corresponds to @xmath47 , so that matter appears to be overaccelerating in the radial direction more than in the vertical direction .
the exponential scale length of the m33 disk emission depends upon the passband used to measure the surface brightness , ranging from 1.56 kpc in the k band to 2.5 kpc in the v band .
stepping back from the light - traces - mass approach , is interesting to explore what combinations of @xmath13 and @xmath14 would produce a @xmath28 closest to unity .
allowing both @xmath13 and @xmath14 as free parameters , with no constraints and with no assumption about the galaxy s mass - to - light ratio , the values that best match @xmath48 are @xmath49 kpc and @xmath50 kpc .
this corresponds to a remarkably thin disk , with a vertical scale height of only 35 pc .
the @xmath28 profile from this more general fit is shown in figure 2 .
this provides a better fit to the kinematic observations , but @xmath35 still varies by over a factor of two across the face of m33 . also , the value of @xmath51 is over an order of magnitude less than typical aspect ratios .
although we can achieve improved mond - inspired fits to the kinematic data , these models imply strong radial and vertical gradients in the galaxy s mass - to - light ratio .
this is at variance with the elegant what - you - see - is - all - there - is mond scenario .
our objective is to propose a general technique for testing the self - consistency of mond , using the existing m33 data as an illustrative example .
the vertical and circular motions of a galaxy can be jointly used for this test .
potential weaknesses in the argument presented above include i ) the assertion that the vertical scale height of galaxies is radius - independent , ii ) modeling the galaxy with the form shown in equation ( 1 ) , and iii ) the implicit assertion that either objects overaccelerate , or they do nt .
the first issue can be addressed with better observations and more statistics , and the second by a more comprehensive treatment of the system s kinematics . the third concern , namely the isotropy of mondian dynamics , is an interesting issue .
if mondian behavior arises from a modification of inertia @xcite , then this scalar quantity will determine an object s response to _ any _ applied force , and it will exhibit the same modified dynamics in all directions . on the other hand one might imagine that mond only applies component by component , with a modified response only to those forces that would give rise to accelerations below the @xmath9 threshold .
this could produce a difference in the radial and vertical dynamics and could perhaps account for a ratio of @xmath52 that differs from unity . in this circumstance
however a terrestrial cavendish experiment conducted at the north or south pole should see differing effective values of @xmath6 in different regimes of @xmath30 .
sensible next steps to obtaining observations that are optimally suited to the test we propose include 1 ) assessing the relative merits of planetary nebulae vs. integrated starlight as probes of vertical velocity dispersion , 2 ) selecting a favorable list of target galaxies , and 3 ) carrying out a set of appropriate observations .
it is sensible to include , as a control , examples of high surface brightness disk galaxies which should have their inner regions in the newtonian disk - dominated regime where @xmath30 = 1 , to verify that @xmath35 is constant and equal to unity for these systems .
h@xmath53 and 21 cm observations of the velocity field might also contribute to this technique .
we are grateful to j. beckenstein , g. bothun , j. dalcanton , and k. cook for interesting conversations about mond as an alternative to dark matter .
j. battat , a. miceli , d. sherman and the thoughtful students in harvard s fall 2005 freshman seminar on the hidden universe provided important encouragement .
we thank harvard university and the department of energy office of science for their support . | we propose a technique to test the idea that non - standard dynamics , rather than dark matter halos , might be responsible for the observed rotation curves of spiral galaxies . in the absence of non - luminous matter
, a galactic disk s rotational velocity and its vertical velocity dispersion can be used jointly to test the self - consistency of the galaxy s dynamics .
a specific illustrative example , using recent measurements of the disk kinematics of m33 , shows this to be a promising approach to assess the viability of modified newtonian dynamics ( mond ) . |
accretion onto supermassive black holes ( smbhs ) , and the corresponding release of gravitational potential energy , power active galactic nuclei ( agn ) .
this process requires that a significant amount of matter in the interstellar medium of a galaxy experience a loss of angular momentum sufficient to ultimately be captured by the smbh s accretion disk .
numerical simulations suggest that major mergers of galaxies are an effective mechanism for removing angular momentum @xcite .
observational evidence for this scenario includes bright quasi - stellar objects ( qsos ) that often show evidence of interactions or mergers at both obscured phases , such as dust - reddened qsos @xcite and ultra - luminous infrared galaxies ( ulirgs ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and in traditional qsos @xcite .
this scenario is also consistent with the hierarchical paradigm of galaxy evolution , in which massive stellar bulges are capable of fueling smbh growth through mergers of gas rich galaxies @xcite and suggests that smbh growth may be linked with merger events in order to maintain observed correlations between smbhs and their host galaxies .
for example , the masses of smbhs appear to be correlated with the stellar velocity dispersion @xcite and luminosities @xcite of the central stellar bulges , implying that the buildup of smbh mass is correlated with the buildup of stellar bulge mass @xcite . the structures of disk galaxies , on the other hand
, are thought to be shaped by the collapse of gas via energy dissipation and smooth accretion from cooling gas within the surrounding dark matter halo , with a previous phase of mergers shaping the central stellar bulge and stellar halo @xcite .
while subsequent major mergers will not necessarily destroy the disk if the gas supply of the progenitors is very high and the resulting bulge component is small @xcite , simulations have been successful at reproducing the observed properties of more massive bulges ( corresponding to substantial smbh growth ) through major mergers of spiral galaxies @xcite , implying that agn hosted by disk galaxies may be triggered via alternative routes .
such mechanisms include instabilities internal to the host galaxy @xcite and minor mergers @xcite .
however , the relative roles of internal instabilities , minor mergers and major mergers for growing smbhs is unclear .
for example , the majority of local agn reside in late - type galaxies unlikely to have experienced a major merger @xcite , though their luminosities do not extend to the more powerful regimes seen in high - redshift quasars .
these observables are explained in the model of @xcite where secular smbh growth dominates in the local universe but remains important only for low - luminosity agn triggering toward high - redshifts as the rate of gas - rich mergers increases .
however , finding direct and consistent evidence in support of this picture has been difficult .
numerous studies find that high luminosity agn show no preference for existing in galaxies with signs of merger activity when they are compared to a control sample of inactive galaxies ( @xcite , though see @xcite for a potential luminosity dependence ) . on the other hand
, several studies find that the fraction of agn in mergers , out of a parent agn sample , increases with increasing luminosity @xcite .
the above qualitative disagreements may be due to the variable conditions under which agn triggering may happen within a galaxy merger . for example , the stage of the merging galaxy system may be a particularly important parameter for the triggered accretion rate .
the tidal torques induced by the merger are predicted to funnel gas and dust toward the nuclear region of the merging system , and as the merger evolves the smbhs will lose angular momentum due to dynamical friction , thereby migrating toward the nuclear region as well .
therefore , at later merger stages , and thus smaller smbh separations , the supply of gas for accretion is greater , such that the accretion rate is likely to be higher @xcite , albeit with considerable uncertainty as to the timescales of agn duty cycles .
while several studies have examined the dependence on merger stage using nuclear separations as a proxy @xcite , spatial resolution limits have precluded systematic analyses from observing the small nuclear separations when the accretion rate is predicted by simulations to peak @xcite .
furthermore , simulations predict that the probability of triggering one versus two agn within a merger may be different and depend on properties of the host galaxies , such as the mass of their nuclear stellar cores @xcite . additionally , nuclear obscuration in mergers may hinder the establishment of a connection between mergers and agn at optical wavelengths .
indeed , @xcite found that the fraction of galaxies with disturbed morphologies increases with the level of nuclear obscuration , suggesting that this may be a key phase in the evolution of agn in galaxies but which is hidden from most observations . to understand the conditions under which agn triggering is correlated with galaxy mergers , uniform merger samples with well - understood selection biases are necessary .
while many galaxy merger candidates have been selected spectroscopically from velocity offset agn emission lines @xcite , the majority of them have been shown to host agn - driven outflows rather than dual smbhs based on follow - up observations @xcite . from imaging , several samples have been selected based on morphology , either visually @xcite or based on asymmetry @xcite .
however , selection by morphology is not necessarily capable of quantifying the merger stage accurately if two nuclei are not visible .
while samples based on galaxy pairs can measure separations , they are necessarily limited to earlier merger stages @xcite .
however , resolved separations can be significantly reduced by spatially constraining the relative locations of two individual smbhs in a merging system @xcite . in this study
, we exploit this concept by using a newly constructed sample of x - ray agn that are spatially offset from the nucleus of the host galaxy or a nearby companion galaxy from @xcite , hereafter referred to as paper i. the spatially offset agn sample can be used to quantify the merger stage based on separation , and was constructed using an astrometric registration procedure that detects offsets down to @xmath0 kpc .
additionally , the sample also allows us to compare the merger scenarios for single agn formation against those of dual agn formation .
this paper is organized as follows : in section [ sec : sample ] we described our sample , in section [ sec : analysis ] we analyze the effects of bolometric luminosity , nuclear separation , group environment and obscuration on the samples , in section [ sec : discussion ] we discuss the conditions that affect agn triggering in galaxy mergers and section [ sec : conclusions ] contains our conclusions .
we assume the cosmological parameters of @xmath2 km s@xmath3 mpc@xmath3 , @xmath4 and @xmath5 throughout .
in this section , we discuss the samples used in our analysis : an offset agn sample and two separate dual agn samples . the offset agn represent galaxy mergers hosting only a single agn that is off - nuclear ( section [ sec : offset_agn ] ) , and the dual agn represent galaxy mergers hosting two agn ( section [ sec : dual_agn ] ) . for our sample of galaxy mergers hosting only a single agn
, we use the spatially offset agn from paper i. full details of this sample can be found in paper i , though here we summarize the main properties .
the agn ( type 2 ) were originally selected from galaxies in the sdss seventh data release ( dr7 ) that are located in the agn regime of the baldwin - phillips - terlivich ( bpt ) diagram @xcite . from overlapping archival _
chandra _
coverage , we identified hard x - ray sources that are required to be within the sdss fiber and satisfy the x - ray criteria of an agn : unabsorbed hard x - ray luminosity of @xmath6 @xmath7 erg s@xmath3 , or a hardness ratio of @xmath8 @xmath9 where @xmath8 @xmath10 and @xmath11 and @xmath12 are the number of hard and soft x - ray counts , respectively .
the overlapping sdss and _ chandra _ images were astrometrically registered with our pipeline described in paper i , establishing the _ complete _ parent agn sample ( 48 ) .
spatially offset x - ray agn were then selected as those with significant spatial offsets ( @xmath13 in significance but @xmath14 kpc in projected physical separation ) from the nucleus of the host galaxy or a nearby companion galaxy , establishing the _ complete _ offset agn sample ( 18 ) . since the spatially offset agn detections require _ chandra _ imaging , the selection is biased toward agn that have been targeted by _
this bias will affect our analysis if the agn were targeted as potentially being in merging systems since it will not represent a random sampling of agn ( see section 7.2 of paper i ) .
this bias affects 8 offset agn from the _ complete _ sample .
removing these 8 agn leaves a sample of 40 parent agn and 10 spatially offset agn based on an unbiased sample of _ chandra _ observations .
these sources make up the final offset agn sample , and they are used in all of our subsequent analyses . below
we describe several unique characteristics of our offset agn sample and how we address them in our analyses : _ possible presence of dual agn : _ since our selection of offset agn in paper i imposed several conservative x - ray thresholds , we can not rule out the possibility that any of them are dual agn systems with the second agn either below the x - ray source detection threshold or below the x - ray agn thresholds ( see section 7.1 of paper i ) .
one particular example is the source sdssj110851.04 + 065901.5 ( sdssj1108 + 0659 ) which was selected as a spatially offset agn in paper i but was also previously discussed in @xcite who interpreted the system to be consistent with a dual agn . while sdssj1108 + 0659 is excluded from the sample because it was targeted by a _
chandra _ program as a dual agn candidate , secondary agn that are currently undetected may exist in some of our other offset agn systems . for uniformity , however , we adopt the definitions used in paper i and note that our conclusions can only be said to apply to x - ray bright agn .
we also acknowledge the possibility of two x - ray detected agn in one of our sources , sdssj111458.02 + 403611.41 ( sdssj1114 + 4036 ; see section 7.1 of paper i ) , though we note that using the secondary x - ray source ( sdssj1114 + 4036sw ) instead of the primary x - ray source ( sdssj1114 + 4036ne ) has a negligible effect on our results .
_ sdss fiber size : _ as described in paper i , due to the @xmath15 sdss fiber radius , spatial offsets between the x - ray agn and galaxy centers may be contained entirely within the fiber or not ( see figure 3 of paper i ) . since the x - ray agn is always constrained to be within the fiber radius , this distinction is made between cases in which the x - ray agn is offset from a galactic core that is inside the fiber ( ` in - fiber ` offset agn ) versus a galactic core that is outside the fiber ( ` out - fiber ` offset agn ) .
we emphasize that the distinction between the ` in - fiber ` and ` out - fiber ` subsamples is not a selection effect but merely an artifact of the sdss fiber size .
however , a side effect of this artifact is that spectroscopic coverage is available for the offset galactic cores of the ` in - fiber ` subsample ( smaller separation pairs with unresolved secondary nuclei in sdss imaging ) but not for the ` out - fiber ` subsample ( larger separation pairs with resolved secondary nuclei in sdss imaging ) . in principle
, this difference should not affect our results concerning the agn properties assuming the x - ray agn detection is associated with the optical agn detection from the fiber spectrum .
still , given that the fiber coverage of the overall systems is generally larger for the ` in - fiber ` subsample , in each analysis subsection we consider the effects of removing the ` out - fiber ` offset agn .
_ nature of the x - ray sources : _ since x - ray sources significantly in excess of @xmath6 @xmath16 erg s@xmath3 are known to be associated with accreting smbhs , the criterion of @xmath6 @xmath17 erg s@xmath3 is likely to rule out a non - agn contribution @xcite .
sources that only pass the @xmath8 @xmath18 criterion , however , may have smaller values of @xmath6 and therefore more ambiguous physical origins
. the x - ray luminosity function of off - nuclear x - ray sources does not extend far above @xmath19 erg s@xmath3 @xcite , with only a few brighter sources known ( e.g. * ? ? ?
* ; * ? ? ?
these objects , known as hyper luminous x - ray sources ( hlxs ) , are often associated with intermediate mass black holes ( imbhs ) .
such sources in our sample with @xmath6 @xmath20 erg s@xmath3 are likely associated with either a lower luminosity agn or otherwise accretion onto imbhs with high hardness ratios @xcite , an event that is likely the result of a minor galaxy merger . at lower luminosities ( @xmath6 @xmath21 erg s@xmath3 )
, however , we can not rule out the possibility of stellar - mass systems in star - forming regions ( ultra - luminous x - ray sources ) that are passing through phases of unusually hard x - ray spectra that can mimic the higher hardness ratios typically seen in agn @xcite . finally , we noted in paper i that the ionizing nature of agn optically classified as low - ionization nuclear emission regions ( liners ) is ambiguous and may not originate from accretion onto nuclear massive black holes @xcite .
therefore , in each analysis subsection we consider the effects of removing the subsample that does not pass the threshold of @xmath6 @xmath22 erg s@xmath3 or is optically classified as a liner .
our goal in this subsection is to create a sample of dual agn from the literature for comparison to our offset agn .
since no samples currently exist that satisfy both the optical and x - ray selection criteria from paper i , we have constructed two dual agn samples , one of which is optically - selected ( section [ sec : sdss_dual_agn ] ) and the other of which is x - ray - selected ( section [ sec : bat_dual_agn ] ) .
similar to the offset agn parent sample , the parent sample of the optically - selected dual agn is derived from the sdss dr7 spectroscopic agn @xcite .
since we require _ chandra _ detections to spatially isolate the agn relative positions , we have cross - matched the sdss spectroscopic agn with unique detections from the _ chandra _ source catalogue ( csc ; * ? ? ?
* ) within @xmath15 to create the final parent sample of the optically - selected agn .
from this parent sample , we use the previously identified dual agn systems found in the two studies that have selected dual agn starting from the sdss spectroscopic agn sample and for which @xmath23 imaging reveals spatially distinct x - ray agn @xcite .
the two samples are described below : there are two systems from @xcite that the authors classify as dual agn : sdssj1108 + 0659 , also in our _ complete _ offset agn sample ( section [ sec : offset_agn ] ) , and sdssj114642.47 + 511029.6 ( sdssj1146 + 5110 ) .
the x - ray agn in each pair are separated by @xmath13 and @xmath14 kpc , thereby satisfying the spatially offset criteria from paper i. however , in both systems one of the x - ray sources ( sdssj1108 + 0659nw and sdssj1146 + 5110sw ) passes the x - ray agn selection criteria from paper i , while the other x - ray source ( sdssj1108 + 0659se and sdssj1146 + 5110ne ) does not pass either of the @xmath6 or @xmath8 criteria .
however , @xcite find sdssj1108 + 0659se likely to be an agn based on a one - dimensional analysis of the psf profiles and x - ray luminosities ( soft and hard ) that are several times the expected contribution from star - formation , and they find sdssj1146 + 5110ne likely to be an agn because the soft x - ray luminosity is more than an order of magnitude larger than the expected contribution from star - formation . furthermore , slit spectroscopy of these two systems from @xcite suggest that two agn may be present in each system . while sdssj1108 + 0659 is also in our _ complete
_ offset agn sample , it is rejected from the ( unbiased ) offset agn sample and therefore only appears in the dual agn sample for our analyses .
there is one system from @xcite that is classified as a dual agn : sdssj112659.54 + 294442.8 ( sdssj1126 + 2944 ) . as with the dual agn from @xcite , both of the x - ray agn in sdssj1126
+ 2944 satisfy the spatially offset criteria from paper i ( @xmath13 and @xmath14 kpc ) . the brightest of the two x - ray agn ( sdssj1126 + 2944nw ) passes the @xmath6 criterion and the weaker source ( sdssj1126 + 2944se ) is consistent with the @xmath6 threshold when accounting for the uncertainty .
these x - ray agn detections are also consistent with the orientation and separation of agn photoionized double [ oiii]@xmath245007 components from slit spectroscopy presented in @xcite . for completeness
, we searched within the sdss agn - csc cross - matched sample for additional dual agn with separations of @xmath13 and @xmath14 kpc , and with velocity separations of less than 600 km s@xmath3 ( ( e.g . as in * ? ? ? * ) ) , finding none .
finally , we have omitted all sources from the comparison sample that are not within the range of redshifts ( @xmath25 @xmath26 @xmath27 ) , bolometric luminosities , @xmath28 ( @xmath29 @xmath28 @xmath30 erg s@xmath3 ) and projected physical separations , @xmath31 ( @xmath32 @xmath31 @xmath33 kpc ) of the parent sample of paper i. this leaves a sample of 69 optically - detected parent agn and 6 optically - detected agn in 3 dual agn systems .
we refer to this sample as the sdss dual agn .
_ two caveats about the sdss dual agn sample : _ first , the sdss dual agn were originally selected for follow - up imaging ( @xmath34 ) as dual agn candidates because of explicit double - peaks in the narrow agn emission lines of the sdss spectra and therefore do not represent an unbiased sample of agn .
second , the two dual agn systems from @xcite would only be selected as offset agn by the criteria from paper i , whereas we have adopted their dual agn interpretations ; in this sense , we can not claim the same x - ray properties for the dual agn as our offset agn sample .
this choice was made to increase the sample size from one to three systems .
@xmath35{wcs_sep_off.eps } & \hspace*{0.1 in } \includegraphics[width=3.4in]{wcs_sep_err_off.eps } \vspace*{-0.in } \end{array } $ ] the parent sample of the x - ray - detected dual agn consists of burst alert telescope ( bat ) detections from the 58 month survey @xcite that are crossmatched with agn @xcite .
the x - ray - selected dual agn systems we use are the subset of this sample that appear in the @xcite sample of dual agn ( based on the presence of galaxy companions also hosting an agn ) and with separations of @xmath14 kpc . as with the optically - selected dual agn ,
the spatial centroids of the agn host galaxies are separated by @xmath13 .
since the @xcite sample contains the full sample of bat agn in dual agn systems , we do not search for additional dual agn within the parent sample .
furthermore , all of the bat agn are within the redshift range of the parent sample .
this yields a parent sample of 246 bat - detected agn and 16 bat - detected agn in 8 dual agn systems ( ngc 6240 , mrk 739 , mrk 463 , iras 05589 + 2828 , eso 509-ig 066 , iras 03219 + 4031 , ngc 3227 and ngc 835 ) .
we refer to this sample as the bat dual agn .
we note that the agn classifications in the bat sample are not uniform and come from a variety of evidence including both optical and x - ray detections .
in this section , we use our offset agn sample and dual agn samples to constrain the conditions that affect agn triggering in galaxy mergers .
specifically , we examine the agn merger fractions ( section [ sec : frac ] ) , group environments ( section [ sec : environ_frac ] ) and level of nuclear obscuration ( section [ sec : loiii_lx ] ) .
[ sec : analysis ] in this section , we first derive corrections for the known selection biases in the offset agn sample ( section [ sec : bias ] ) .
then we investigate the agn merger fractions as a function of agn bolometric luminosity , @xmath28 ( section [ sec : lbol_frac ] ) and projected nuclear physical separation , @xmath31 ( section [ sec : phys_sep_frac ] ) . in each case
, we do so for both the offset agn fractions ( @xmath36 ) and the dual agn fractions ( @xmath37 and @xmath38 ) .
to reduce the statistical uncertainty , fractions are only shown in each bin if the number of parent agn is at least 2 ( which corresponds to @xmath39 confidence in the poisson count statistics but also allows for an adequate number of bins and dynamic range for our analysis ) .
the binomial distribution , defined by the size of the parent sample and the success rate of offset agn occurrences within the parent sample , is used to compute the lower and upper quantiles defining the @xmath40 confidence interval around all fraction values .
uncertainties associated with the functional parameterizations of fractions are the @xmath40 quantiles surrounding the median value of each parameter distribution obtained by adding simulated random uncertainties ( also drawn from the binomial distribution ) and refitting until the uncertainties converge .
the sample of offset agn was uniformly defined by requiring that the angular offset between the agn and galaxy core or secondary agn , @xmath41 , be three or more times its standard uncertainty , @xmath42 : @xmath41@xmath43@xmath42 ( see paper i for details ) . therefore , from the parent agn sample , selection of those with real spatial offsets depends directly on only the two parameters @xmath41 and @xmath42 .
figure [ fig : hist_bias ] shows the results of our selection process on the distributions of @xmath41 and @xmath42 for both the ` in - fiber ` and ` out - fiber ` subsamples , where the offset agn have a mean @xmath41 value greater than that of the parent sample , and a mean @xmath42 value less than that of the parent sample .
we have used a two - sample kolmogorov - smirnov ( ks ) test to determine the null hypothesis probability that the parent and offset agn sample values come from the same distribution ( @xmath44 ) . for @xmath41
, the small value of @xmath44 @xmath45 strongly suggests that the offset agn are biased toward large values of @xmath41 and that we have likely missed offset agn with small values of @xmath41 .
therefore , the selection has effectively introduced a minimum @xmath41 limit in the offset agn sample . for @xmath42
, on the other hand , the value of @xmath44 @xmath46 suggests that the selection is relatively insensitive to @xmath42 . to account for these direct selection effects ,
we have utilized monte - carlo simulations similar to those described in paper i. in short , the simulations produce offset nuclei with random projected physical separations ( @xmath47@xmath48@xmath47 @xmath14 kpc ) , redshifts ( @xmath49 @xmath50 @xmath51 ) and relative positional uncertainties ( @xmath52 @xmath53 @xmath54 ) that are drawn from uniform distributions .
projecting @xmath48 onto the sky based on random orientations and phases , and then scaling to @xmath50 yields a simulated angular offset , @xmath55 .
combined with @xmath53 , we selected offset nuclei using the same procedure as in paper i. we then calculated the recovered fraction of simulated offset nuclei as a function of @xmath55 and @xmath53 .
values of @xmath36 have been corrected for biases introduced by large values of @xmath41 and small values of @xmath42 based on the results of these simulations .
specifically , in each bin of @xmath28 or @xmath31 , we identified a simulated offset nucleus with values of @xmath55 and @xmath53 that most closely match @xmath41 and @xmath42 , respectively .
then , we divided the observed agn merger fraction by the average recovery fraction of the matched simulated nuclei .
the ranges of correction factors are quoted in sections [ sec : lbol_frac ] and [ sec : phys_sep_frac ] .
the simulated recovery fractions should not be taken as estimates of the absolute recovery fractions because the true distributions of the parent sample parameters are unlikely to be uniform as in our simulations .
therefore , they only provide estimates of the relative fractions as a function of @xmath55 and @xmath53 . since the corrections are relative , we have normalized them to unity ( i.e. no correction ) at the largest values of @xmath55 and @xmath53 .
we also caution that the corrections based on these simulations do not account for any potential indirect biases which are instead discussed individually in each subsection as footnotes .
we do not correct the values of @xmath37 and @xmath38 since the dual agn positional measurements and uncertainties are not uniformly measured ( section [ sec : dual_agn ] ) .
numerical simulations of galaxy mergers have predicted that the dependence on mergers for agn triggering is positively correlated with the agn bolometric luminosity @xcite , and observational evidence of morphological disturbances in the host galaxies of high luminosity agn supports these predictions @xcite .
however , other studies that have examined the morphological traits of agn and non - agn host galaxies find no statistical differences between samples at low and high agn luminosities @xcite .
these null - results imply that agn triggering is relatively independent of galaxy mergers such that internal instabilities play a comparable if not stronger role in smbh growth .
while these studies used different procedures and tests , their results are qualitatively in disagreement about the role that galaxy mergers play in triggering agn as a function of agn luminosity .
therefore , we use our systematically selected sample of galaxy mergers to address the role that agn bolometric luminosity has on the agn merger fraction .
@xmath56{lbol_oiii_off.eps } \\ \hspace*{-0.2 in } \includegraphics[width=0.5\textwidth]{lbol_dual_sdss_bat.eps } \vspace*{-0.1 in } \end{array } $ ] our estimates of @xmath28 for the parent sample of the offset agn and the parent sample of the sdss dual agn are calculated from the extinction - corrected [ ] @xmath57 luminosity ( @xmath58,corr.}}$ ] ; from @xcite based on balmer decrements ) since the original agn identifications of both samples were based on optical emission lines and are not affected by significant obscuration due to dust ( see section [ sec : loiii_lx ] ) .
we use a bolometric correction of @xmath28 @xmath59@xmath58,corr.}}$ ] @xcite .
since the companions of the bat dual agn were identified from non - uniform selection criteria , ultra - hard ( @xmath60 kev ) x - ray luminosities ( @xmath61 ) are available for only one agn in each system .
therefore , we have chosen to derive @xmath28 from @xmath61 using the bolometric correction of @xmath28 @xmath62@xmath61 @xcite .
while this choice means we can only use eight of the 16 agn in the bat systems , the use of ultra - hard x - rays mitigates the effect of nuclear obscuration @xcite .
figure [ fig : frac_lbol_oiii ] ( top ) shows @xmath36 , calculated as the number of offset agn ( @xmath63 ) out of the number of parent agn ( @xmath64 ) , as a function of @xmath28 .
figure [ fig : frac_lbol_oiii ] ( bottom ) shows @xmath37 and @xmath38 as a function of @xmath28 , where each is computed in the same manner as @xmath36 . to optimize the combination of signal and binning resolution , we have chosen bin sizes individually for each sample . due to
the multiple orders of magnitude spanned by @xmath28 , the bins have been given logarithmically uniform spacing with sizes of 0.5 dex for the offset agn sample and 1 dex for both dual agn samples . after implementing the threshold of @xmath652 parent agn in each bin , the resulting @xmath28 bin ranges are @xmath66 erg s@xmath3 for @xmath36 , @xmath67 erg s@xmath3 for @xmath37 and @xmath68 erg s@xmath3 for @xmath38 .
therefore , to examine both the offset and dual agn samples over the same @xmath28 values , we have chosen the plotting range of @xmath28 @xmath69 erg s@xmath3 . from @xmath28 @xmath70 erg s@xmath3 , @xmath36
is adequately fit by a power - law function ( @xmath71 ) with parameter values of @xmath72 @xmath73 and @xmath74 @xmath75 .
this fit corresponds to a slope with a @xmath76 significance from zero . for comparison ,
the correction factors ( section [ sec : bias ] ) range from 1.14 to 1.21 , and the best - fit power - law parameters for the uncorrected fractions are @xmath72 @xmath77 and @xmath74 @xmath78 , corresponding to a slope with a @xmath79 significance from zero . to test the implications of using the ` out - fiber `
subsample ( section [ sec : offset_agn ] ) , we remove the ` out - fiber ` sources and find that doing so does not change the qualitative result that no significant change in @xmath36 is seen as a function of @xmath28 .
we also find that removing sources with @xmath6 @xmath21 erg s@xmath3 or that are liners also has no effect on this qualitative result . from @xmath28 @xmath80 erg s@xmath3 , @xmath37
is adequately fit by a power - law function with parameter values of @xmath72 @xmath81 and @xmath74 @xmath82 .
compared to @xmath36 , this fit corresponds to a steeper slope with a @xmath83 significance from zero . from @xmath28
@xmath84 erg s@xmath3 , @xmath38 is also adequately fit by a power - law function with parameter values of @xmath72 @xmath85 and @xmath74 @xmath86 . compared to @xmath37 , this fit corresponds to a shallower slope , though it still has a @xmath87 significance from zero .
while the current data offers a null result for @xmath36 , a similar increase from low to high @xmath28 can not be ruled out due to the significant fit uncertainties . for comparison , in both panels of figure [ fig : frac_lbol_oiii ]
we show the best - fit linear function to the agn merger fraction from @xcite .
the parent sample used in @xcite consists of agn identified from x - ray , infrared and spectroscopic surveys , and the galaxy merger systems were taken from samples identified by visual classification . as a result ,
the number of agn in each system ( offset or dual agn ) is usually not possible to determine . in the top panel
, we see that the function from @xcite is consistent with @xmath36 below @xmath28 @xmath88 erg s@xmath3 while over - predicting @xmath36 at higher @xmath28 by @xmath89 .
this result provides tentative evidence that @xmath36 behaves differently from the @xcite function at high @xmath28 values , though the substantial uncertainties prohibit a firm conclusion .
the @xmath36 @xmath28 values plotted also do not extend down to the luminosities of @xmath90 erg s@xmath3 , and thus the behavior of @xmath36 compared to the lower @xmath28 end of the @xcite sample is not known .
the stronger positive correlations seen in the dual agn ( relative to the offset agn ) are in better qualitative agreement with the function from @xcite , though the bat - sample is generally over - predicted while the sdss sample is in agreement to within the @xmath91 uncertainties over nearly the full @xmath28 range plotted .
numerical simulations have predicted that , in an evolving galaxy merger , the probability of observing an agn increases with decreasing separation of the two smbhs from the progenitor galaxies @xcite .
observational evidence of rising agn merger fractions with decreasing nuclear separation supports these predictions @xcite .
however , the galaxy mergers from those samples were identified from galaxy pairs , thereby limiting the nuclear separations to larger values ( several kpc or greater ) so that individual galaxies can be distinguished .
simulations , on the other hand , predict that the agn observability continues to increase significantly below 1 kpc , and previous observational studies have not been able to probe the small separation regime where the dependence of the agn merger fraction on nuclear separation is predicted to peak .
therefore , we use our sample of offset agn with resolved x - ray agn offsets from @xmath92 to @xmath1 kpc to examine the agn merger fraction from early to late merger stages . we have adopted the physical separations presented in @xcite and @xcite for the sdss dual agn and those presented in @xcite for the bat dual agn .
@xmath93{physep_off.eps } \\ \hspace*{-0.2 in } \includegraphics[width=0.5\textwidth]{physep_dual_sdss_bat.eps } \vspace*{-0.1 in } \end{array } $ ] figure [ fig : frac_phys_sep ] ( top ) shows @xmath36 as a function of @xmath31 .
since values of @xmath31 are not available for the parent agn sample , we have computed the minimum projected physical separation that could potentially be resolvable ( by @xmath94 ) based on the @xmath42 and @xmath26 values for each parent agn : @xmath95 .
the values of @xmath36 in each @xmath31 bin are then the number of offset agn with @xmath31 within that bin ( @xmath63 ) out of the number of parent agn with @xmath95 less than or equal to the mean value of that bin ( @xmath96}}$ ] ) .
note that a parent agn may be in multiple bins using this approach .
figure [ fig : frac_phys_sep ] ( bottom ) shows @xmath37 and @xmath38 as a function of @xmath31 , where each is computed in the same manner as @xmath36 . as in section
[ sec : lbol_frac ] , we have chosen bin sizes individually for each sample to optimize the combination of signal and binning resolution .
the bins have been given linearly uniform spacing with sizes of 1.5 kpc for the offset agn sample , 3 kpc for the sdss dual agn sample , and 6 kpc for the bat dual agn sample .
after implementing the threshold of @xmath652 parent agn in each bin , the resulting @xmath31 bin ranges are @xmath97 kpc for @xmath36 and @xmath98 kpc for both @xmath37 and @xmath38 .
therefore , to examine both the offset and dual agn samples over the same @xmath31 values , we have chosen the plotting range of @xmath31 @xmath99 kpc . from @xmath31 @xmath99 kpc , @xmath36
is adequately fit by a power - law function with parameter values of @xmath72 @xmath100 and @xmath74 @xmath101 .
this fit corresponds to a slope with @xmath102 significance from zero .
the correction factors ( section [ sec : bias ] ) range from 0 for fractions in bins of large @xmath31 to 1.28 in the bin of smallest @xmath31 . for comparison ,
the best - fit power - law parameters for the uncorrected fractions are @xmath72 @xmath103 and @xmath74 @xmath104 , corresponding to a slope with @xmath105 significance from zero . as in section
[ sec : lbol_frac ] , we test the implications of including the ` out - fiber ` subsample by removing those sources , finding that doing so has no qualitative effect on the significant rise in @xmath36 at small physical separations since the ` out - fiber ` offset agn have physical separations larger than the small - separation regime in which the rapid change in @xmath36 is seen .
this test also confirms that the correlation is not due to the artifact introduced by the fiber size .
likewise , removal of the sources with @xmath6 @xmath21 erg s@xmath3 or those that are liners has no qualitative effect on the significant rise in @xmath36 at small physical separations . from @xmath31 @xmath99 kpc , @xmath37
is adequately fit by a power - law function with parameter values of @xmath72 @xmath106 and @xmath74 @xmath107 . compared to @xmath36 , this fit corresponds to a shallower slope , but it is still has a @xmath108 significance from zero . from @xmath31
@xmath109 kpc , @xmath38 is also adequately fit by a power - law function with parameter values of @xmath72 @xmath110 and @xmath74 @xmath111 . compared to @xmath37 , this fit is steeper ( though still shallower than @xmath36 ) and has a @xmath112 significance from zero .
we note that @xmath36 dramatically increases at small separations when compared to @xmath37 and @xmath38 ( as indicated by their respective magnitudes of @xmath113 ) .
this difference may indicate that the value of @xmath113 in the @xmath36 trend is strongly driven by the data point at @xmath31@xmath114 kpc which is lacking in the sdss and bat dual agn samples . for comparison , in both panels of figure [ fig : frac_phys_sep ]
we show a line representing the trend seen in the agn merger fraction as a function of projected pair separation from @xcite based on the _ wise _ color cut ( @xmath115 ) that has been empirically shown to select infrared bright agn with high reliability .
in particular , the line shows the slope made by the change in the agn merger fraction in their pre - coalescence sample from the largest separation data point to the smallest separation data point plotted in their figure 2 .
the @xcite sample utilizes galaxy mergers selected as galaxy pairs ( not distinguishing between the offset and dual agn scenarios ) that extends from a minimum separation of @xmath116 kpc ( based on the redshift range and the sdss fiber size ) out to @xmath31 @xmath117 kpc @xcite . above @xmath31@xmath118 kpc ,
@xmath36 , @xmath37 and @xmath38 are all consistent with the @xcite sample , while all three merger fractions would be under - predicted by a linear extrapolation of this slope at @xmath31@xmath119 kpc .
observationally , galaxies that exist in dense environments , such as groups or clusters , generally have redder colors ( due to suppressed star - formation ) and more elliptical morphologies compared to galaxies in less dense environments @xcite .
one route for dense environments to drive galaxies toward redder colors and elliptical morphologies is through interactions or mergers between galaxies .
while the probability of a direct merger is low among satellite galaxies in clusters due to their high relative velocities , galaxy groups are the environments most likely to contain merging systems @xcite .
therefore , we may expect to find a higher fraction of offset and dual agn in group environments .
we use our sample of offset and dual agn to test this prediction by quantifying the density of their environments .
we do so by first taking a catalogue of galaxy group members described in @xcite .
the catalogue was constructed by applying host dark matter halo and satellite dark matter halo mass functions @xcite to galaxies from the sdss dr7 .
based on the halo mass functions of each galaxy , the central most massive galaxy of a group and its satellites can be identified .
thus , every galaxy in the catalogue has an assigned number of group members , @xmath120 , with masses above @xmath121 m@xmath122 ( below this mass limit the method is not sensitive ) which we take as a parameterization of the group density .
we then matched the offset agn parent sample and the dual agn parent samples with the catalogue to assign them @xmath120 values .
since the redshift range of the galaxy group catalogue is limited to @xmath123 , only a subset of the agn in our samples will have @xmath120 values .
figure [ fig : hist_ngroup ] compares the @xmath120 distributions for the parent and offset agn .
while the mean value of the offset agn ( 14.0 ) is larger than that of the parent agn ( 5.6 ) , the difference is not at a significant level based on a two - sample ks test ( @xmath44 @xmath124 ) . as in sections
[ sec : lbol_frac ] and [ sec : phys_sep_frac ] , we test the implications of including the ` out - fiber ` subsample by removing those sources , finding that doing so does not change the qualitative result that no significant difference is seen between the two samples
. likewise , removal of the sources with @xmath6 @xmath21 erg s@xmath3 or those that are liners has no qualitative effect on this result .
too few of the sdss and bat agn are assigned @xmath120 values to statistically examine their environments .
while the number statistics are small , this result suggests that offset agn tend to be in environments similar to the general type 2 agn population , or at least that the difference is small enough to be undetectable in our sample .
this suggestion implies that the types of mergers leading to the spatially offset nature of an x - ray agn within the parent sample are not linked with over - dense environments compared to the parent sample .
in fact , the environments of both the parent and offset agn have relatively small densities of group members compared to cluster environments ( @xmath120 @xmath125 ) .
however , these densities only correspond to group members with masses above @xmath121 m@xmath122 , suggesting that mergers with lower - mass galaxies may play a role in producing the offset agn systems . the implications of this merger mass ratio effect are discussed in section [ sec : disc_single_dual ] . the nuclear regions of merging galaxy systems can potentially be heavily obscured as gas and dust will naturally settle toward the regions of largest gravitational potential . since the [ ] @xmath57 emission line originates far enough from the smbh to not be subject to nuclear obscuration , comparison with the x - ray luminosity , which does originate near the smbh accretion disk , can potentially reveal the presence of nuclear enshrouding material .
therefore , we have tested for this effect in figure [ fig : plot_lx_loiii ] by plotting @xmath6 against @xmath58,corr.}}$ ] for the parent agn sample and offset agn sample .
we have also shown in figure [ fig : plot_lx_loiii ] the mean value of a type 2 agn sample selected independently of mergers @xcite .
the effect of our @xmath6 and hardness ratio ( @xmath8 ) selection criteria can be seen in figure [ fig : plot_lx_loiii ] .
the @xmath6-selected subsample has @xmath6/@xmath58,corr.}}$ ] ratios similar to the merger - independent sample , whereas the @xmath8-selected subsample has systematically lower values of @xmath6 for a given value of @xmath58,corr.}}$ ] . since the @xmath6-selected subsample has an average column density that is larger by @xmath126 dex than that of the @xmath8-selected subsample , this difference between the two subsamples is consistent with underestimates of the column densities in the latter ( assuming that the distributions of intrinsic x - ray spectral slopes and intrinsic absorbing columns are the same ) .
indeed , the relatively simple absorbed power - law fit to the chandra spectra ( paper i ) may be under - estimating the column densities in these sources due to low counts .
@xmath127{loiii_lx_int_off_plot.eps } \end{array } $ ] while the mean @xmath6/@xmath58,corr.}}$ ] ratios of the offset agn are lower than those of the parent agn for the @xmath6-selected subsample ( @xmath128 versus @xmath129 ) and the @xmath8-selected subsample ( @xmath130 versus @xmath131 ) , they are in agreement to within the scatter . furthermore , the two - sample ks test results of @xmath44 @xmath132 and @xmath44 @xmath133 for the @xmath6-selected and @xmath8-selected subsamples , respectively , do not suggest a significant difference . therefore , for agn with a given value of @xmath6 and @xmath58,corr.}}$ ] , selection of x - ray sources that are spatially offset from the host galaxy core or a nearby galaxy core does not appear to introduce a bias toward obscuration .
instead , this result indicates that our selection of offset agn , from the parent agn sample , coincides with a tendency to select relatively unobscured systems .
the potential physical implications of this result are discussed in section [ sec : obscuration ] .
several studies have placed estimates on the fraction of agn hosted by galaxies in mergers or merger - remnants @xcite . however , these studies have not estimated the fractions for the specific scenarios of offset agn or dual agn because the methods by which those studies selected galaxy mergers and cross - matched with agn does not uniformly allow for a distinction between systems in which one or both galaxies hosts an agn . the distinction between the two scenarios is a crucial step toward understanding the physics that govern accretion onto smbhs within galaxy mergers . since our selection method requires the agn to be spatially isolated within the merger , we can measure the number of agn and constrain the conditions of offset versus dual agn formation . in this section
, we discuss the offset and dual agn scenarios in the context of agn luminosity ( section [ sec : disc_single_dual ] ) , merger stage ( section [ sec : disc_sep ] ) , the combined effect of agn luminosity and merger stage ( section [ sec : sep_lum ] ) , and finally nuclear obscuration ( section [ sec : obscuration ] ) . as mentioned in section [ sec : lbol_frac ] , some observational evidence suggests that high - luminosity agn are preferentially found in galaxies that are interacting , merging , or have merged in the past , while other studies find no evidence for agn hosts to have different morphologies from inactive galaxies . in section [ sec : lbol_frac ] and figure [ fig : frac_lbol_oiii ] , we used our agn merger samples ( selected independent of galaxy or merger morphology ) to address this discrepancy , producing the following results : based on their parameterizations ( section [ sec : lbol_frac ] ) , the offset and dual agn fractions behave differently as a function of @xmath28 .
the offset agn fraction displays no significant evolution over the agn bolometric luminosity range probed ( @xmath76 significance ) as shown in figure [ fig : frac_lbol_oiii ] ( top ) .
comparatively , the dual agn fractions show stronger evidence for a statistically significant increase at high agn bolometric luminosities ( @xmath87@xmath134@xmath83 significance ) as shown in figure [ fig : frac_lbol_oiii ] ( bottom ) .
these results can be broadly interpreted as in agreement with claims of a positive correlation between the agn merger fraction and bolometric luminosity .
specifically , observations of large - scale ( 10 - 100 kpc ) galaxy pairs have shown that the agn merger fraction is highest for major mergers @xcite , implying that major mergers are more efficient at removing angular momentum from gas and dust in their host galaxies .
additionally , @xcite find that their bat - selected dual agn are preferentially found in major mergers .
therefore , dual agn host systems may show a preference for major mergers so that enough fuel is available to power both agn .
for example , of the three sdss dual agn systems , two are hosted by major mergers @xcite .
since the sample from @xcite extends to high redshifts and was selected based on morphology , that sample may contain a relatively higher fraction of major mergers that allow for easier visual classification , possibly accounting for the agreement between that sample and the comparison sdss dual agn sample . by contrast , offset agn may show a preference to reside in minor mergers instead of major mergers .
for example , in section [ sec : environ_frac ] we examined the distribution of the number of group members for both the parent and offset agn , finding no significant evidence for a difference between the two samples ( figure [ fig : hist_ngroup ] ) and that both samples reside in relatively low - density environments .
however , since the procedure used for measuring group members is not sensitive to galaxies with masses @xmath135 m@xmath136 @xcite , we are likely missing many lower - mass group members . therefore , compared to galaxy mergers seen in dense environments of more massive galaxies , the offset agn are more likely to be undergoing mergers with galaxies of lower masses , corresponding to mass ratios that fall in the minor merger regime
. however , we caution that the offset agn slope is consistent within @xmath91 with those of both dual agn samples ( figure [ fig : frac_lbol_oiii ] ) so that the true difference between offset and dual agn evolution with agn bolometric luminosity is poorly constrained .
we also note that the effect of merger mass ratio on agn triggering is still poorly understood at small separations , particularly for offset agn , and may also depend on whether or not loss of angular momentum happens more efficiently in the major or minor galactic stellar core .
currently , theoretical work has provided ambiguous results , with one model suggesting that the more luminous agn likely resides in the more massive stellar bulge @xcite , while a recent simulation of galaxy mergers has suggested that the accretion rate is higher for the agn in the less massive galaxy @xcite . in paper iii ,
follow - up imaging of our offset agn sample with @xmath137 will put constraints on these predictions by allowing estimates of the merger mass ratios and smbh accretion rates . as mentioned in section [ sec : phys_sep_frac ] , numerical work predicts that the probability of agn triggering becomes strongest at separations below @xmath138 kpc as the two smbhs dynamically evolve toward the region of greatest gravitational potential along with a significant amount of gas and dust for accretion . in section [ sec : phys_sep_frac ] and figure [ fig : frac_phys_sep ] , we used our agn merger samples ( with measured physical separations reaching below @xmath138 kpc ) to test this prediction , producing the following results : evidence for a negative correlation with nuclear separations below @xmath92 kpc is seen at @xmath39 significance for both the offset agn fractions ( @xmath102 ) and the dual agn fractions ( @xmath112@xmath134@xmath108 ) as shown in figure [ fig : frac_phys_sep ] .
while the physical separations of the dual agn sample do not allow us to see this trend continue below @xmath139 kpc , we are able to do so for the offset agn sample due to our astrometric registration procedure .
indeed , the slope magnitude for the offset agn fractions is larger ( by @xmath140 ) than those of the dual agn fractions , an indication that the merger fraction rises most strongly at the smallest separations .
a similar , or larger , increase may be seen in the dual agn sample with a larger sample size or with data points at separations @xmath114 kpc .
our finding that the agn merger fraction rises fastest and peaks ( down to our resolution limits ) below 1 kpc is consistent with numerical predictions . at large separations , the offset agn fractions and the dual agn fractions are consistent with the slope of the agn merger sample presented in @xcite .
in fact , both the offset agn and sdss dual agn fractions are consistent with this slope over the full range shown ( @xmath141 kpc ) . at large separations ,
our results are also consistent with other studies @xcite .
when examining the absolute values of the agn merger fractions , we see that the offset agn and sdss dual agn fractions ( below @xmath142 kpc ) of @xmath143 and @xmath144 , respectively , are in agreement when accounting for their @xmath91 uncertainties . that the offset agn and sdss dual agn fractions are consistent at small separations may be indicating that , among similarly selected samples , dual agn triggering becomes more likely at smaller separations and comparable to single agn triggering .
this result is consistent with the expectation that there is an increased supply of gas for accretion at smaller nuclear separations .
the generally larger values of the bat dual fraction may indicate a stronger overall dependence of agn triggering on nuclear separation in that sample , though this conclusion is tenuous due to the substantial uncertainties ( @xmath145 below @xmath142 kpc ) and @xmath91 overlap with the offset agn and sdss dual agn . since the frequency of agn observability in mergers peaks at small separations due to the increased availability of fuel for accretion
, we may also expect that this regime coincides with enhanced accretion rates among agn . to investigate this prediction ,
we have examined the dependence of the offset agn fractions as a function of nuclear separation for two subsamples that are separated by bolometric luminosity : @xmath28 @xmath146 erg s@xmath3 and @xmath28 @xmath147 erg s@xmath3 ( figure [ fig : frac_phys_sep_lbollohi ] ) . due to the smaller numbers in each subsample compared to the full offset agn sample
, we have used bin sizes of @xmath148 kpc .
as shown in figure [ fig : frac_phys_sep_lbollohi ] , the fractions in the low- and high - luminosity bins are consistent within their uncertainties throughout the entire range of separations investigated .
therefore , we see no statistically significant evidence that the occurrence of offset agn is dependent on luminosity at any given nuclear separation . however , we note that the greatest divergence between the two fractions ( by @xmath89 ) occurs at the smallest separation . while this result is tenuous due to the large uncertainties , it may hint that while an overall dependence of the offset agn fraction with bolometric luminosity is not seen ( section [ sec : lbol_frac ] ) , it does appear at small nuclear separations when the supply of gas for accretion is larger .
this result is also qualitatively consistent with the results from @xcite in which the luminosities of dual agn increase at smaller separations .
we caution that the observability of agn in these systems depends not only on luminosity but also on the timescale of activity which we can not measure . still , these combined results are overall consistent with the numerical predictions that the bolometric luminosities of agn in merging systems peak at merger stages corresponding to small separations @xcite .
heavy obscuration of the nuclear regions of galaxies undergoing mergers is predicted by models of galaxy and quasar co - evolution in which mergers trigger enhanced levels of accretion onto agn but also pass through a stage of enhanced obscuration @xcite .
therefore , the continuum emission of agn that are hosted by on - going galaxy mergers may preferentially exhibit signs of obscuration compared to agn in non - merging systems .
extreme examples of agn obscuration are found in ulirgs that are rich in gas and dust @xcite , with most emission from the accretion disk obscured except for high energy photons such as hard x - rays .
heavily obscured ( compton - thick ) agn , in which x - ray emission is severely obscured , may constitute a significant population of agn in galaxy mergers @xcite and therefore understanding the role that x - ray absorption plays in the link between galaxy mergers and agn is crucial .
the effect of obscuration in optically - selected and x - ray detected dual or offset agn was noticed independently by @xcite and @xcite .
their samples of dual and offset type 2 agn systems show systematically lower observed hard x - ray to [ ] @xmath57 luminosity ratios compared to the optically - selected type 2 agn sample from @xcite . since the [ ] @xmath57 emission line originates far enough from the smbh to not be subject to nuclear obscuration , this result suggests that dual and offset agn are suffering heavier nuclear obscuration than the general population of type 2 agn .
in fact , @xcite have shown that even the absorption - corrected hard x - ray luminosities of their sample are still under - luminous compared to a general sample of type 2 agn that was cross - matched with the @xmath23 source catalogue @xcite .
this has led to the suggestion that the low counts and/or intrinsically high absorbing columns result in systematically underestimated column densities .
the analysis in section [ sec : loiii_lx ] showed that , while the offset agn sample has a lower mean hard x - ray to [ ] @xmath57 luminosity ratio than the parent agn , the difference is not at a significant level . therefore , no evidence of preferential nuclear obscuration is seen .
this result is opposite that seen in @xcite and @xcite , and may be a result of the spatially offset selection introducing a bias toward face - on systems ( as numerically predicted in paper i ) with shallower absorbing columns
. however , it may also be a result of the x - ray selection properties that target x - ray bright agn with intrinsically little nuclear obscuration . that the selection may preferentially target galaxy mergers with relatively dust - free nuclear regions means the sample is fundamentally different from the prototypical mergers seen in gas - rich systems such as ulirgs and
may be probing a specific subclass of galaxy mergers .
we have used our systematically constructed sample of spatially offset agn from paper i to constrain the parameters under which agn triggering is driven by galaxy mergers . due to the selection of galaxy mergers based on offset x - ray agn , our sample is not biased toward morphological disturbances or large projected physical separations , allowing us to investigate the agn merger fraction in major or minor mergers and at early or late merger stages .
we have investigated the fractions of offset agn , and those of similarly constructed dual agn samples , out of their respective parent samples , as functions of agn bolometric luminosity and projected nuclear separation .
additionally , we have examined their group environments and compared their x - ray to optical luminosity ratios to those of independent agn samples .
our conclusions are as follows : 1 .
the fraction of spatially offset agn shows no evidence for a dependence on agn bolometric luminosity , while the fractions of dual agn do show a positive dependence , increasing from @xmath149 at @xmath150 erg s@xmath3 to between @xmath151 and @xmath152 at @xmath153 erg s@xmath3 ( figure [ fig : frac_lbol_oiii ] ) .
these results suggest that agn triggering is indeed linked to mergers but that this dependence may only become strong in the specific scenarios of high bolometric luminosities , dual agn activation and possibly major mergers .
the offset agn group environments show no evidence for a difference from the parent agn sample ( figure [ fig : hist_ngroup ] ) , and both reside in environments with a low - density of massive galaxies . the lack of numerous massive companions may point toward a preference for minor mergers in the offset agn systems .
the fractions of spatially offset agn and dual agn show evidence for a negative dependence on projected physical nuclear separation , increasing from @xmath149 at 19 kpc to between @xmath154 and @xmath155 at @xmath119 kpc ( figure [ fig : frac_phys_sep ] ) .
the offset and dual agn fractions are similar at small separations , suggesting that the efficiency of dual agn triggering becomes similar to single agn triggering at late merger stages when significant material is available for accretion .
we can only trace the dual agn sample down to @xmath156 kpc , while the resolution of our offset agn sample allows it to be traced down to @xmath1 kpc where we see the most significant increase .
our sample of offset agn has allowed this analysis to be extended down to @xmath0 kpc for the first time .
we see tentative evidence that the inverse dependence of the agn merger fractions on separation become strongest when restricted to a high agn luminosity subsample ( figure [ fig : frac_phys_sep_lbollohi ] ) . if real , this result would be consistent with numerical predictions that agn triggering probabilities increase with decreasing nuclear separations , and that this late merger stage also corresponds with the stage of highest agn luminosity .
the hard x - ray to [ ] @xmath57 luminosity ratios of offset agn show no significant evidence for a difference from that of the parent agn ( figure [ fig : plot_lx_loiii ] ) , in contrast to the expectation from studies of many known merging galaxy systems .
while this similarity may reflect a tendency to select face - on systems , it may also point toward a selection of intrinsically unobscured systems that are fundamentally different from merging systems with coincident nuclear obscuration and on - going star - formation . in paper iii of this series
, we will present new and archival @xmath137 imaging for a subset of our offset agn sample to put constraints on the correlated evolution of smbhs and their host galaxies .
in particular , we will determine the effect of merger mass ratio on smbh growth , and we will put constraints on the correlated triggering of star - formation and agn .
+ the authors would like to thank an anonymous referee for detailed and insightful comments that greatly improved the quality of the paper .
support for this work was provided by nasa through chandra award number ar5 - 16010a issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of nasa under contract nas8 - 03060 .
the scientific results reported in this article are based in part on observations made by the _ chandra x - ray observatory _ , and this research has made use of data obtained from the _ chandra data archive _ and the _ chandra source catalog _ , and software provided by the _ chandra x - ray center ( cxc ) _ in the application packages _ ciao _ , _ chips _ , and _ sherpa_. | galaxy mergers are likely to play a role in triggering active galactic nuclei ( agn ) , but the conditions under which this process occurs are poorly understood . in paper
i , we constructed a sample of spatially offset x - ray agn that represent galaxy mergers hosting a single agn . in this paper , we use our offset agn sample to constrain the parameters that affect agn observability in galaxy mergers .
we also construct dual agn samples with similar selection properties for comparison .
we find that the offset agn fraction shows no evidence for a dependence on agn luminosity , while the dual agn fractions show stronger evidence for a positive dependence , suggesting that the merger events forming dual agn are more efficient at instigating accretion onto supermassive black holes than those forming offset agn .
we also find that the offset and dual agn fractions both have a negative dependence on nuclear separation and are similar in value at small physical scales .
this dependence may become stronger when restricted to high agn luminosities , though a larger sample is needed for confirmation .
these results indicate that the probability of agn triggering increases at later merger stages .
this study is the first to systematically probe down to nuclear separations of @xmath0 kpc ( @xmath1 kpc ) and is consistent with predictions from simulations that agn observability peaks in this regime .
we also find that the offset agn are not preferentially obscured compared to the parent agn sample , suggesting that our selection may be targeting galaxy mergers with relatively dust - free nuclear regions . |
in the past few years , social media services as well as the users who subscribe to them , have grown at a phenomenal rate .
this immense growth has been witnessed all over the world with millions of people of different backgrounds using these services on a daily basis to communicate , create and share content on an enormous scale . this widespread generation and consumption of content
has created an extremely complex and competitive online environment where different types of content compete with each other for the attention of users .
thus it is very interesting to study how certain types of content such as a viral video , a news article , or an illustrative picture , manage to attract more attention than others , thus bubbling to the top in terms of popularity . through their visibility ,
these popular items and topics contribute to the collective awareness reflecting what is considered important .
this can also be powerful enough to affect the public agenda of the community .
there have been studies on trends and trend - setters in western online social media @xcite @xcite . in this paper , we examine in detail a significantly less - studied but equally fascinating online environment : chinese social media , in particular , sina weibo : china s biggest microblogging network . over the years
there have been news reports on various internet phenomena in china , from the surfacing of certain viral videos to the spreading of rumors @xcite to the so called `` human flesh search engines '' @xcite in chinese online social networks .
these stories seem to suggest that many events happening in chinese online social networks are unique products of china s culture and social environment .
due to the vast global connectivity provided by social media , netizens all over the world are now connected to each other like never before .
this means that they can now share and exchange ideas with ease .
it could be argued that this means the manner in which the sharing occurs is similar across countries .
however , china s unique cultural and social environment suggests that the way individuals share ideas might be different than that in western societies . for example , the age of internet users in china is a lot younger , they may respond to different types of content than internet users in western societies .
the number of internet users in china is larger than that in the u.s , and the majority of users lives in large urban cities .
one would expect that the way these users share information can be even more chaotic .
an important question to ask is to what extent would topics have to compete with each other in order to capture users attention in this dynamic environment .
furthermore , it is known that the information shared between individuals in chinese social media is monitored @xcite .
hence another interesting question to ask is what types of content would netizens respond to and what kind of popular topics would emerge under such circumstances . given the above questions , we present an analysis on the evolution of trends in sina weibo . we have monitored the evolution of the top trending keywords in sina weibo for 30 days .
first , we analyze the model of growth of these trends and examine the persistance of these topics over time .
we investigate if topics initially ranked higher tend to stay in the top 50 trending list longer .
subsequently , by analyzing the timestamps of posts containing the keywords , we look at the propagation and decaying process of the trends in sina weibo and compare it to earlier observations on twitter @xcite . we establish that retweets play a greater role in sina weibo than on twitter , contributing more to the generation and persistance of trends .
when we examine the retweets in detail , we make an important discovery .
we observe that many of the trending keywords in sina weibo are heavily manipulated and controlled by certain fraudulent accounts that are setup for this purpose .
we found significant evidence suggesting that a large percentage of trends in sina weibo are actually due to artificial inflation by these fake users , thus making certain posts trend and be more visible to other users .
the users we identified as fraudulent were 1.08% of the total users sampled , but they were responsible for 49% of the total retweets ( 32% of the total tweets ) .
we evaluate some methods to identify these fake users and demonstrate that by removing the tweets associated with the fake users , the evolution of the tweets containing the trending keywords follow the same persistent and decaying process as the one in twitter .
the rest of the paper is organized as follows : in section 2 we give some related work . in section 3
we give some background information on the development of the internet and online social networks in china .
we analyze the trends and trend - setters on sina weibo in section 4 and finally in section 5 , we conclude and discuss our findings .
jin @xcite has studied the chinese online bulletin board systems ( bbs ) , and provided observations on the structure and interface of chinese bbs and the behavioral patterns of its users .
xin @xcite has conducted a survey on bbs s influence on the university students in china and their behavior on chinese bbs .
yu and king .
@xcite has looked at the adaptation of interests such as books , movies , music , events and discussion groups on douban , an online social network and media recommendation system frequently used by the youth in china .
there are various studies on trends on twitter @xcite @xcite @xcite @xcite .
recently , asur and others @xcite have examined the growth and persistence of trending topics on twitter .
they discovered that traditional media sources are important in causing trends on twitter .
many of the top retweeted articles that formed trends on twitter were found to arise from news sources such as the new york times . in this work ,
we evaluate how the trending topics in china relate to the news media .
yu et al . gave the first known study of trending topics in a chinese online microblogging social network ( sina weibo ) @xcite .
they discovered that there are vast differences between the content that is shared on sina weibo and that on twitter . in china , people tend to use microblogs to share jokes , pictures and videos and a significantly large percentage of posts are retweets .
the trends that are formed are almost entirely due to the retweeting of such media content .
this is contrary to what was observed on twitter , where the trending topics have more to do with current events and the effect of retweets is not as high .
the development of the internet industry in china over the past decade has been impressive . according to a survey from the china internet network information center ( cnnic ) , by july 2008 ,
the number of internet users in china has reached 253 million , surpassing the u.s . as the world s largest internet market .
furthermore , the number of internet users in china as of 2010 was reported to be 420 million . despite this
, the fractional internet penetration rate in china is still low .
the 2010 survey by cnnic on the internet development in china reports that the internet penetration rate in the rural areas of china is on average @xmath0 .
in contrast , the internet penetration rate in the urban cities of china is on average @xmath1 . in metropolitan cities such as beijing and shanghai ,
the internet penetration rate has reached over @xmath2 , with beijing being @xmath3 and shanghai being @xmath4 . according to the survey , china s cyberspace is dominated by young urban students between the age of 1830 .
the government plays an important role in fostering the advance of the internet industry in china . according to _ the internet in china
_ released by the information office of the state council of china : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the chinese government attaches great importance to protecting the safe flow of internet information , actively guides people to manage websites in accordance with the law and use the internet in a wholesome and correct way . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ online social networks are a major part of the chinese internet culture @xcite .
netizens in china organize themselves using forums , discussion groups , blogs , and other social networking platforms to engage in activities such as exchanging viewpoints and sharing information @xcite . according to _ the internet in china _ : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ vigorous online ideas exchange is a major characteristic of china s internet development , and the huge quantity of bbs posts and blog articles is far beyond that of any other country .
china s websites attach great importance to providing netizens with opinion expression services , with over 80% of them providing electronic bulletin service . in china , there are over a million bbss and some 220 million bloggers . according to a sample survey , each day people post over three million messages via bbs , news commentary sites , blogs , etc . , and over 66% of chinese netizens frequently place postings to discuss various topics , and to fully express their opinions and represent their interests .
the new applications and services on the internet have provided a broader scope for people to express their opinions .
the newly emerging online services , including blog , microblog , video sharing and social networking websites are developing rapidly in china and provide greater convenience for chinese citizens to communicate online . actively participating in online information communication and content creation ,
netizens have greatly enriched internet information and content .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ sina weibo was launched by the sina corporation , china s biggest web portal , in august 2009 . according to `` microblog revolutionizing china s social business development '' released by the sina corporation in october , 2011
, sina weibo now has 250 million registered accounts and generates 90 million posts per day . while both twitter and sina weibo enable users to post messages of up to 140 characters , there are some differences in terms of the functionalities offered .
we give a brief introduction on sina weibo s interface and functionalities .
similar to twitter , a user profile on sina weibo displays the user s name , a brief description of the user , the number of followers and followees the user has , and the number of tweets the user made .
a user profile also displays the user s recent tweets and retweets .
there are three types of user accounts on sina weibo , regular user accounts , verified user accounts , and the expert ( star ) user account .
a verified user account typically represents a well known public figure or organization in china .
sina has reported in the annual report that it has more than 60,000 verified accounts consisting of celebrities , sports stars , well known organizations ( both government and commercial ) and other vips .
regular users can register to become an expert ( star ) user , part of sina weibo s new user account verification program : `` daren '' ( expert in chinese ) .
this program offers incentives for users to verify their personal information such as passport numbers ( or i d numbers for chinese citizens ) and home addresses .
some criteria for a regular users to become an expert ( star ) user includes : 1 ) the user profile picture must be a photo of user himself / herself ; 2 ) the user must provide a registered mobile number ; 3 ) the user account must follow at least 100 people , among them at least 30% must be the followers of user ; 4 ) the user accounts must have at least 100 followers .
there is an important difference in the content of tweets between sina weibo and twitter . while twitter users can post tweets consisting of text and links , sina weibo users can post messages containing text , pictures , videos and links .
twitter users can address tweets to other users and can mention others in their tweets [ 13 ] .
a common practice on twitter is `` retweeting '' , or rebroadcasting someone else s messages to one s followers .
the equivalent of a retweet on sina weibo is instead shown as two amalgamated entries : the original entry and the current user s actual entry which is a commentary on the original entry .
figure [ retweet ] illustrates an example of a retweet on sina weibo .
sina weibo has another functionality absent from twitter : the comment .
when a weibo user makes a comment , it is not rebroadcasted to the user s followers . instead
, it can only be accessed under the original message .
sina weibo offers a list of 50 keywords that appear most frequently in users tweets .
they are ranked according to the frequency of appearances in the last hour .
this is similar to twitter , which also presents a constantly updated list of trending topics : keywords that are most frequently used in tweets over a period of time .
we extracted these keywords over a period of 30 days and retrieved all the corresponding tweets containing these keywords from sina weibo .
we first monitored the hourly evolution of the top 50 keywords in the trending list for 30 days .
we observed that the average time spent by each keyword in the hourly trending list is 6 hours . and
, the distribution for the number of hours each topic remains on the top 50 trending list follows the power law ( as shown in figure [ power ] a ) .
the distribution suggests that only a few topics exhibit long - term popularity .
another interesting observation is that a lot of the key words tend to disappear from the top 50 trending list after a certain hour and then later reappear .
we examined the distribution for the number of time keywords have reappeared in the top 50 trending list ( figure [ power ] b ) .
we observe that this distribution follows the power law as well .
both the above observations are very similar to our earlier study of trending topics on twitter @xcite although the average trending time is significantly higher on sina weibo ( on twitter it was 20 - 40 minutes ) .
[ ht ] following our observation that some keywords stay in the top 50 trending list longer than others , we wanted to investigate if topics that ranked higher initially tend to stay in the top 50 trending list longer .
we separated the top 50 trending keywords into two ranked sets of 25 each : the top 25 and the bottom 25 .
figure [ stay ] illustrates the plot for the percentage of topics that placed in the bottom 25 relating to the number of hours these topics stayed in the top 50 trending list .
we can observe that topics that do not last are usually the ones that are in the bottom 25 . on the other hand ,
the long - trending topics spend most of their time in the top 25 , which suggests that items that become very popular are likelier to stay longer in the top 50 .
this intuitively means that items that attract phenomenal attention initially are not likely to dissipate quickly from people s interests .
[ ht ] next , we want to investigate the process of persistence and decay for the trending topics on sina weibo . in particular , we want to measure the distribution for the time intervals between tweets containing the trending keywords .
we continuously monitored the keywords in the top 50 trending list . for each trending topic we retrieved all the tweets containing the keyword from the time the topic first appeared in the top 50 trending list until the time it disappeared .
accordingly , we monitored 811 topics over the course of 30 days . in total we collected 574,382 tweets from 463,231 users .
among the 574,382 tweets , 35% of the tweets ( 202,267 tweets ) are original tweets , and 65% of the tweets ( 372,115 tweets ) are retweets .
40.3% of the total users ( 187130 users ) retweeted at least once in our sample .
we measured the number of tweets that each topic gets in 10 minutes intervals , from the time the topic starts trending until the time it stops . from this
we can sum up the tweet counts over time to obtain the cumulative number of tweets @xmath5 of topic @xmath6 for any time frame @xmath7 , this is given as : @xmath8 where @xmath9 is the number of tweets on topic @xmath6 in time interval @xmath10 .
we then calculate the ratios @xmath11 for topic @xmath6 for time frames @xmath7 and @xmath12 .
figure [ ratio_general ] shows the distribution of @xmath13 s over all topics for two arbitrarily chosen pairs of time frames : ( 10 , 2 ) and ( 8 , 3 ) ( nevertheless such that @xmath14 , and @xmath7 is relatively large , and @xmath12 is small ) . [ ht ] s over all topics for two arbitrarily chosen pairs of time frames : ( 10 , 2 ) and ( 8 , 3),title="fig:",width=302,height=151 ] these figures suggest that the ratios @xmath13 are distributed according to the log - normal distributions .
we tested and confirmed that the distributions indeed follows the log - normal distributions .
this finding agrees with the result from a similar experiment on twitter trends .
asur et al @xcite argued that the log - normal distribution occurs due to the multiplicative process involved in the growth of trends which incorporates the decay of novelty as well as the rate of propagation .
the intuitive explanation is that at each time step the number of new tweets ( original tweets or retweets ) on a topic is multiplied over the tweets that we already have .
the number of past tweets , in turn , is a proxy for the number of users that are aware of the topic up to that point .
these users discuss the topic on different forums , including twitter , essentially creating an effective network through which the topic spreads .
as more users talk about a particular topic , many others are likely to learn about it , thus giving the multiplicative nature of the spreading . on the other hand , the monotically
decreasing decaying process characterizes the decay in timeliness and novelty of the topic as it slowly becomes obsolete .
however , while only 35% of the tweets in twitter are retweets @xcite , there is a much larger percentage of tweets that are retweets in sina weibo . from our sample
we observed that a high 65% of the tweets are retweets .
thus , we can hypothesize that the effect of retweets is much larger in sina weibo .
sina weibo users are more likely to learn about a particular topic through retweets .
we separate the tweets in sina weibo into original tweets and retweets and calculated the densities of ratios between cumulative retweets / original tweets counts measured in two respective time frames .
figure [ ratio_retweet ] and figure [ ratio_single ] show the distribution of original tweets / retweets ratio over all topics for the same two pairs of time frames : ( 10 , 2 ) and ( 8 , 3 ) [ ht ] [ ht ] we see from figure [ ratio_single ] that the distribution of original tweets ratios follows the log - normal distribution .
however , from figure [ ratio_retweet ] we observe that the distribution of retweets ratios seems to leaning more towards the left .
that is , there seems to be a large amount of of low retweet ratios in the distribution .
what s more , there are some extreme spikes in the lower ratios area of the distribution .
we tested and confirmed that the distributions in figure [ ratio_single ] indeed correspond to log - normal distribution .
however , we observe that the distribution of retweet ratios ( figure [ ratio_retweet ] ) does not satisfy all the properties of the log - normal distribution . from figure [ ratio_retweet ] in the previous section we observed that there is a high percentage of low ratios in the distribution of retweet ratios .
this means that for a lot of the topics , during the observed time duration the growth of retweets seems to be slower than usual .
we hypothesize that this is due to the activities of certain users in sina weibo : as these accounts post a tweet , they tend to set up many other fake accounts to continuously retweet this tweet , expecting that the high retweet numbers would propel the tweet to place in the sina weibo hourly trending list .
this would then cause other users to notice the tweet more after it has emerged as the top hourly , daily , or weekly trend setter .
figure [ example_spam ] illustrates the activity of a suspected spam account .
we observe that it tend to continuously retweets the same posts from the same users .
in fact , this particular account only retweets posts from one user . and , every post from that user gets retweeted multiple times .
[ ht ] we attempt to verify the above hypothesis empirically .
if the above hypothesis is true , we would expect that the distribution for the number of users retweets to be scale free , since an abnormally large number of retweets would be by the supposed spammers .
figure [ distribution_r_r ] a ) illustrates the distribution for the number of users and their corresponding number of retweets ( over all topics ) .
figure [ distribution_r_r ] b ) illustrates the distribution for the number of users and the numbers of topics that they caused to trend by their retweets .
we observe that both distributions in figure [ distribution_r_r ] follows the power law .
[ ht ] we define a spamming account as ones that are set up for the purpose of repeatedly retweeting certain messages , thus giving these messages artificially inflated popularity . according to our hypothesis in the previous section , the users who retweet abnormally high amounts are more likely to be spam accounts .
we test this hypothesis by manually checking the top 40 accounts who retweeted the most . to our surprise , after one month , 37 of these 40 accounts can no longer be accessed .
that is , after we enter the accounts ids , we retrieved a message from sina weibo stating that the account has been removed and can no longer be accessed .
according to the sina weibo frequently asked questions page , such error pages are displayed after sina weibo administrators delete a questionable account .
sina weibo users themselves can not delete or close down their own account . in order to do so
, they have to contact the administrators at sina weibo for help .
we assume that the only reason we can no longer retrieve an active account after one month is that this account was deleted by the administrator for performing malicious activities such as spamming , or spreading illegal information .
what s more , according to sina weibo s frequently asked question page , if a user sends a tweet containing illegal or sensitive information , such tweet will be immediately deleted by sina weibo s administrators , however , the users accounts will still be active . for the above reason
we assume that if an account was active one month ago and can no longer be reached , this account has very likely performed malicious activities such as spamming .
we inspect the user accounts with the most retweets in our sample and the number of accounts they retweeted .
we see that although the number of times these accounts retweeted was very high , they mostly only retweet messages from a few users .
we re - organize the users who retweeted by the ratio between the number of times he / she retweeted and the number of users he / she retweeted .
we refer to this as the user - retweet ratio .
table [ verified ] illustrates the top 10 users with the highest user - retweet ratios .
we note that for all these users , they each retweet posts from only one account .
we observe that this is true for the top 30 accounts with the highest user - retweet ratios .
.the top 10 accounts with the highest user - retweet ratios ( u - r ratio ) [ cols="^,^,^,^,^ " , ] for the 4665 accounts whose tweets were retweeted by suspected spam accounts , we used a crawler to automatically inspect the types of accounts they are .
we found that a high 79% of the accounts are verified accounts , 5% are expert accounts ( see section 3.2.1 for definitions of verified accounts and expert accounts ) .
we manually inspected 50 of the verified accounts and found that they are mostly accounts with commercial purposes ( such as the sina weibo accounts for travel agencies , super markets , fashion brands or hospitals ) .
we re - organized the users whose posts were retweeted by the number of times it was retweeted .
last_table ] illustrates the top 10 such users and the descriptions of their accounts .
we observe that only 2 out of the top 10 accounts were verified accounts .
9 out of 10 accounts seem to have a strong focus on collecting user - contributed jokes , movie trivia , quizzes , stories and so on .
these accounts seem to operate as discussion and sharing platforms .
the users who follow these accounts tend to contribute jokes or stories .
once they are posted , other followers tend to retweet them frequently .
this result verifies a similar finding from yu et al . @xcite
we have examined the tweets relating to the trending topics in sina weibo .
first we analyze the growth and persistance of trends .
when we looked at the distribution of tweets over time , we observed that there was a difference when contrasted with twitter .
the main reason for the difference was that the effect of retweets on sina weibo was significantly higher than on twitter .
we also found ( as our previous work @xcite suggests ) that many of the accounts that contribute to trends tend to operate as user contributed online magazines , sharing amusing pictures , stories and antidotes .
such posts tend to recieve a large amount of responses from users and thus retweets .
when we examined the retweets in more detail , we made an important discovery .
we found that 49% of the retweets in sina weibo containing trending keywords were actually associated with fraudulent accounts .
we observed that these accounts comprised of a small amount ( 1.08% of the total users ) of users but were responsible for a large percentage of the total retweets for the trending keywords .
these fake accounts are responsible for artificially inflating certain posts , thus creating fake trends in sina weibo .
we relate our finding to the questions we raised in the introduction .
there is a strong competition among content in online social media to become popular and trend and this gives motivation to users to artificially inflate topics to gain a competitive edge .
we hypothesize that certain accounts in sina weibo employ fake accounts to repeatedly repeat their tweets in order to propel them to the top trending list , thus gaining prominence as top trend setters ( and more visible to other users ) .
we found evidence suggesting that the accounts that do so tend to be verified accounts with commercial purposes .
, _ chinese bulletin board system s influence upon university students and ways to cope with it ( in chinese ) _ , journal of nanjing university of technology ( social science edition ) , 4 ( 2003 ) , pp .
100104 . , _ the evolution of friendships in chinese online social networks _ , social computing / ieee international conference on privacy , security , risk and trust , 2010 ieee international conference on , 0 ( 2010 ) , pp . | there has been a tremendous rise in the growth of online social networks all over the world in recent years .
this has facilitated users to generate a large amount of real - time content at an incessant rate , all competing with each other to attract enough attention and become trends .
while western online social networks such as twitter have been well studied , characteristics of the popular chinese microblogging network sina weibo have not been . in this paper , we analyze in detail the temporal aspect of trends and trend - setters in sina weibo , constrasting it with earlier observations on twitter .
first , we look at the formation , persistence and decay of trends and examine the key topics that trend in sina weibo . one of our key findings is that retweets are much more common in sina weibo and contribute a lot to creating trends . when we look closer , we observe that a large percentage of trends in sina weibo are due to the continuous retweets of a small amount of fraudulent accounts .
these fake accounts are set up to artificially inflate certain posts causing them to shoot up into sina weibo s trending list , which are in turn displayed as the most popular topics to users . |
figure 1 illustrates the sliding geometry of graphene layers in blg .
when the sliding vector @xmath2 is along @xmath3 , the blg gradually transforms from its pristine form of ab - stacking ( designated as ab - blg ) to aa - stacking ( @xmath4 ) in which all carbon atoms are right on top of each other ( aa - blg ) . for the sliding along @xmath5 ,
the ab - blg becomes upside - down ab - stacked blg ( ba - blg ) .
equivalent layer stackings are arranged by other sliding vectors along either @xmath6 or @xmath7 direction .
also combination of the sliding vectors is possible as drawn in fig .
. are vectors connecting the nearest neighbor carbon atoms in each layer . the sliding vector in the upper layer is denoted by @xmath2 . from the left to the right , top views of blg for @xmath8 , @xmath9 , @xmath10 , and @xmath11 .
( b ) and ( c ) , in - plane atomic displacements for @xmath12 and @xmath13 optical phonons , respectively . for each phonon mode , doubly - degenerate atomic motions are denoted by red ( lo - mode ) or blue ( to - mode ) arrows with carbon atoms of the same color moving in the same direction .
vertical dashed - lines are guides for eyes to the nearest interlayer carbon atoms ( a2 and b1 ) . ] among the phonon modes at @xmath14-point in the brillouin zone ( bz ) of ab - blg , high - frequency e@xmath15 [ fig .
1(b ) ] and e@xmath16 optical modes [ fig .
1(c ) ] are responsible for the raman g - band and ir - peak , respectively @xcite .
the in - plane atomic motions in the upper and lower layers [ figs .
1(b ) and ( c ) ] are in - phase and out - of - phase for e@xmath15 and e@xmath16 mode , respectively .
these modes are doubly degenerate as denoted by thick red arrows for longitudinal optical ( lo ) mode and by blue ones for transverse optical ( to ) mode as shown in figs .
1(b ) and ( c ) .
our calculated phonon frequencies of e@xmath16 and e@xmath15 modes in ab - blg are 1592 @xmath17 and 1585 @xmath17 , respectively ( fig .
2 ) , in good agreement with previous studies @xcite . as these two modes involve with the interlayer interaction as well as the crystal symmetry , the sliding motion of two graphene will imprint its effect on these modes .
- ( @xmath18 ) and @xmath19-stacking [ @xmath20 or @xmath21 geometries are shown in the second and third panels of fig .
1(a ) , respectively . ]
first , we find that the sliding in ab - blg lifts the degeneracy of optical phonon modes immediately with the splitting size depending on both the sliding geometries and the symmetry of the phonon modes [ fig . 2 ] .
the degeneracy is recovered when the sliding ends up in aa - blg [ fig .
this finding is readily understood from the crystal symmetry of ab- and aa - blg that guarantee the degeneracy of to and lo modes for e@xmath16 and e@xmath15 phonons .
however , e@xmath15 and e@xmath16 phonon modes exhibit completely different responses to the sliding because of their atomic motions ( one in - phase and the other out - of - phase ) .
the frequencies of ir - active e@xmath16 phonon modes are apparently split into two modes upon sliding with a strong dependence on the sliding vectors [ fig . 2 ] . on the other hand ,
the frequencies of raman - active e@xmath15 modes are seemingly unaffected at all by the sliding motion although the exact degeneracy does not hold any more as fig .
2 illustrates .
we observe that the splitting between to and lo phonons is as large as 20 @xmath17 for e@xmath16 mode but no more than 1 @xmath17 for e@xmath15 mode .
to understand the intriguing behavior of optical phonons upon the sliding , we construct a minimal model of the phonon modes in sliding blg . from our _
ab initio _
calculations for in - plane atomic displacements , we find that the interlayer ion - ion interaction energy changes very little ( less than 10% ) compared with the interlayer ion - electron interaction .
so , additional force constants that account for the renormalization of in - plane phonon frequencies by the interlayer electronic hopping are sufficient to include into the conventional force - constant model . within the harmonic approximation
, the phonon frequency of blg ( @xmath22 for @xmath23 mode ) at @xmath14-point can be expressed as @xmath24 , where @xmath25 is the frequency of @xmath26 phonon modes of slg and @xmath27 is the effective force constant for the @xmath23 phonon - renormalization owing to the interlayer electron hoppings in blg ( @xmath28 g is the mass of carbon atom ) @xcite .
the hamiltonian for blg with a sliding vector @xmath29 @xcite can be written as @xmath30 where the intralayer hamiltonian is @xmath31 and interlayer interaction @xmath32 . here
@xmath33 ( @xmath34 ) is an annihilation ( creation ) operator for electron at site @xmath35 ( @xmath36 ) in the upper and lower layers ( @xmath37 ) . @xmath38 and @xmath39 $ ] where @xmath40 ev is the nearest - neighbor ( _ nn _ ) intralayer - hopping constant , @xmath41 the intralayer _ nn _ distance , @xmath42 and @xmath43 the _ nn _ and the next _ nn _ interlayer - hoppings , respectively , and @xmath44 @xcite . with an intralayer electron - phonon coupling constant of @xmath45 ev / nm @xcite
, the hamiltonian with the phonons can be described by changing @xmath46 to @xmath47 for the lo mode and to @xmath48 for the to mode @xcite . here
@xmath49 is the amplitude of @xmath26 phonon modes in slg and then the atomic displacements at a and b sites in the layer @xmath50 are given as @xmath51 for the lo ( to ) mode [ fig .
1(b ) ] . with @xmath52
, @xmath27 is obtained from the energy variation with respect to the atomic displacement , @xmath53,\ ] ] where @xmath54^{1/2}$ ] and @xmath55 .
@xmath56 corresponds to the energy of sliding blg with @xmath12 and @xmath13 optical phonon modes , respectively , which can be obtained by direct diagonalization of @xmath57 .
@xmath58 are the valley and spin degeneracy , respectively , and @xmath59 is half the bz area accounting the valley degeneracy @xcite . in eq .
( 1 ) , the intralayer electron - phonon contribution [ @xmath60 of two graphene layers is subtracted from the energy of blg with phonons [ @xmath56 ] to obtain the renormalized interlayer force constant only since @xmath62 is assumed to be already renormalized intralyer phonon frequency .
the interlayer force - constant @xmath63 becomes negligible for @xmath12 phonon modes regardless of sliding , _
i.e. _ , @xmath64 .
this can be demonstrated easily by shifting @xmath65 to @xmath66 and @xmath67 for lo and to @xmath12 modes in eq .
( 1 ) , respectively .
so , this explains the reason why sliding does not change the frequency of @xmath12 modes .
in contrast , for @xmath13 modes of ab - blg without sliding , we find @xmath68 so that @xmath69 in a good agreement with our calculation in fig .
2 . with sliding along @xmath70 ( @xmath71 ) , we find that @xmath72 for the lo ( to ) @xmath13 mode ( here @xmath73 is a dimensionless constant ) , which shows the splitting of two degenerate ir - active modes upon sliding as shown in fig .
these calculations explain the different responses of raman - active and ir - active phonon modes in sliding blg .
having understood the origin of the anomalous optical phonon splittings upon sliding , we investigate their spectroscopic consequences .
our calculations readily indicate that the sliding systems have different ir reflectivity and raman spectrum when it is placed on insulating substrates with top or bottom gates .
the single fano - like ir spectra of gated blg @xcite will turn into the double - peak spectrum upon sliding because of the splitting in @xmath13 phonon frequencies . also , its anomalous splitting in raman g - band @xcite due to mixing between @xmath12 and @xmath13 modes will be affected by sliding .
the splitting of raman g - band ( or spectral transfer between opposite parity optical phonon modes ) of gated blg is known to originate from the inversion - symmetry breaking by the gate electric field @xcite .
therefore , we expect that the splitting in @xmath13 mode upon sliding will generate a more complex raman spectrum in the gated structure , _
e.g. _ , a three - peak structure . while the gated structure is the most straightforward way to observe the sliding - induced splitting of optical phonons , more generic ways are desirable of detecting tiny sliding on arbitrary substrates without the need of heavy doping using the field - effect transistor structure .
we present below conventional raman techniques with polarizers indeed can detect sub - angstrom atomic misalignment by sliding .
.calculated raman tensors for various sliding vectors ( @xmath2 ) .
@xmath74 is the schoenflies notation of the point - group symmetry for each sliding geometry . here , @xmath75 .
the components of all raman tensors expressed in the vibrational symmetry - group notations are normalized to ab - blg raman tensor . [ cols="^,^,^,^,^,^",options="header " , ] the position of raman g - band of sliding blg does not change at all for all sliding configurations , but the change in the crystal symmetry produces very interesting movement of the raman intensity .
in order to calculate the non - resonant stocks raman intensity of g - band , we used the placzek approximation for the raman intensity ( @xmath76 ) as @xmath77 is the raman tensor , @xmath78 and @xmath79 are the polarization of the incident and scattered lights , respectively @xcite .
@xmath80 associated with the doubly degenerate raman - active phonon modes is proportional to the raman susceptibility @xmath81 , which is defined as @xmath82 . here
@xmath83 is the @xmath84th phonon eigenvector of the @xmath76th atom along the @xmath85 direction and @xmath86 the electric polarizability tensor ( @xmath87 , as we neglect the irrelevant @xmath88 component ) @xcite .
we directly calculate the derivative of @xmath86 with respect to the atomic displacements corresponding to the @xmath84-th modes .
assuming an incident light with an energy of 2.41 ev , calculated raman tensors for various sliding geometries are summarized in table i together with corresponding crystal - symmetry groups and vibrational symmetries @xcite .
our calculations of raman tensors for ab - blg and aa - blg reproduce the well - known results with @xmath12 and @xmath26 symmetries @xcite . upon sliding ,
changes in the crystal symmetry produce correspondingly distinctive raman tensors . with @xmath89 , the polarized raman intensity for ab - blg
is given as @xmath90 so that the intensity does not depend on the polarization angle @xmath91 [ fig .
3(a ) ] @xcite .
this behavior is also the same for aa - blg while with a different intensity .
when the sliding direction ( @xmath92 ) is along @xmath93 , the polarized raman intensity is given as @xmath94 ( @xmath95 , and @xmath96 are given in table i and @xmath97 is a constant ) .
@xmath98 has an elliptic shape for @xmath99 or quadrupolar form otherwise as shown in fig .
3 . when @xmath2 deviates from the direction of @xmath100 , the polarized intensity tends to rotate as shown in fig .
in case that the overall shape of the polarized intensity looks similar , its magnitude has a strong dependence on the sliding size [ fig.3 ( b ) ] . , ( b ) @xmath101 , and ( c ) @xmath102 directions .
the panels in the left show the raman intensity normalized to that of ab - blg as a function of polarization angle ( @xmath91 ) and small rectangular pannels on the right show corresponding polar plots . ]
in conclusion , we have shown theoretically that tiny sliding of the layers in blgs can induce anomalous splitting of optical phonon that changes the spectroscopic features significantly .
we have shown that the frequencies of the degenerate in - phase optical phonons ( @xmath103 mode ) are hardly changed irrespective of sliding distance and direction .
on the other hand , the polarization - dependent raman intensity associated with @xmath103 mode is modified strongly so that the sub - angstrom misalignment between two graphene layers can be resolved by spectroscopic methods .
we expect that this study will provide essential information for spectroscopic measurements and for linking local atomic structures to novel electronic properties of stacked 2d atomic crystals .
we calculated electronic structures and phonon dispersions of the sliding blgs using first - principles methods with a plane - wave basis set @xcite .
the local density approximation is adopted for the exchange - correlation functional , and the phonon frequencies are calculated using the density functional perturbation theory @xcite .
computations are also repeated using the atomic - orbital basis set @xcite , the generalized gradient approximation ( gga ) @xcite , and frozen phonon method @xcite to find almost identical results . in calculating electronic structures and phonon dispersions with the atomic orbital basis set ,
the basis - set superposition errors were removed by including two ghost atoms in the unit cell @xcite .
a semi - empirical correction of van der waals ( vdw ) forces is added to all our calculations following grimme s proposal , which is essential to obtain the correct interlayer distance of sliding blgs @xcite .
we note that our methods accurately describe the phonon frequencies of graphitic systems when the appropriate interlayer - distance is provided @xcite
s .- m . c. thanks c .- h . park for discussions .
y .- w.s . is supported by the nrf of korea grant funded by mest ( qmmrc , no .
r11 - 2008 - 053 - 01002 - 0 and nano r&d program 2008 - 03670 ) .
s .- h.j . acknowledges the support from the nrf of korea grant funded by mest ( src program no .
2011 - 0030046 and wcu program no .
r31 - 2008 - 000 - 10059 ) .
computational resources have been provided by kisti supercomputing center ( project no .
ksc-2011-c1 - 21 ) and the cac of kias . | we study the variations of electron - phonon coupling and their spectroscopic consequences in response to sliding of two layers in bilayer graphene using first - principles calculations and a model hamiltonian .
our study shows that the long wave - length optical phonon modes change in a sensitive and unusual way depending on the symmetry as well as the parity of sliding atomic structures and that , accordingly , raman- and infrared - active optical phonon modes behave differently upon the direction and size of the sliding .
the renormalization of phonon modes by the interlayer electronic coupling is shown to be crucial to explain their anomalous behavior upon the sliding . also , we show that the crystal symmetry change due to the sliding affects the polarized stokes raman - scattering intensity , which can be utilized to detect tiny misalignment of graphene layers using spectroscopic tools
. successful isolation of graphene @xcite and subsequent experiments that reveal its special properties @xcite have generated excitement to explore the novel properties of the two dimensional ( 2d ) crystal from various disciplines .
advances in synthesis and experimental techniques enable the finding of other 2d crystals @xcite and the artificial fabrication of multiply - stacked structures @xcite . often , the stacking structures lead to very unusual electronic properties different from those of constituent 2d crystals depending on how they are piled up . among them , bilayer graphene ( blg ) , which is a stacked structure of two single - layer - graphene ( slg ) sheets , is unique in electronic structure and exhibits colorful variation in low energy states when its layer stacking is changed .
the in - plane three - fold rotational and mirror symmetries and the decoupling of strong @xmath0 and weak @xmath1 bondings of carbon atoms authorize the uniqueness of graphene systems . as such , the interlayer coupling in blg , while a weak van der waals type , produces interesting variations in low energy band structures upon changes of stacking geomerties . for example , its low - energy states in the bernal stacked pristine form @xcite has quadratic energy bands but change to have linear bands when it has a rotational staking fault @xcite . and , a sensitive electronic topological transition is also predicted for sliding systems @xcite .
interplay of the interlayer interaction , the electron - phonon coupling , and the stacking fault by layer slidings are thus expected to produce drastic changes in its low energy properties .
control of its electronic property is also enabled by manipulating such changes .
recently , an epitaxially grown blg on the vicinal surfaces of silicon carbide shows misalignment between two graphene layers exhibiting a complex nature of its electronic structures @xcite .
moroever , recent experiments on cvd bi- and trilayer graphene also reveal possible misaligned layers at the domain walls in between two ideally stacked graphene systems @xcite .
immediate questions are how such structural variations are reflected in spectroscopic features and whether the misalignment of layers can be detected or not .
raman and infrared ( ir ) spectroscopies have proven to be powerful non - destructive methods to study physical and chemical properties of 2d crystals @xcite .
physical properties of graphene under various conditions such as doping or mechanical strains have been verified using such tools @xcite . in consideration of the rapid progress of research in this field @xcite , comprehensive analysis of the spectroscopic features of sliding blg
will provide key information on the stacking geometry , the electron - phonon interactions , and low energy excitations . in this paper , we study long wave - length optical phonon modes responsible for raman and ir spectrum when two layers of blg slide each other .
we investigate how the tiny atomic misalignment between two layers changes its phonon frequencies and spectroscopic spectrum using first - principles calculations .
it is found that the degeneracy of ir - active modes is immediately lifted by the sliding and each mode changes separately depending on sliding geometries . unlike the behavior of ir modes , the frequencies of doubly degenerate raman - active optical phonons
do not change at all in all sliding circumstances .
such unconventional optical phonon splittings originate from the difference in their frequency renormalizations due to interlayer couplings , which is confirmed by our effective model hamiltonian calculations .
moreover , we demonstrate that , owing to its changes in the crystal - symmetry , non - resonant stokes raman - scattering intensities associated with raman - active phonons exhibit a strong polarization dependence so that they can be used to detect sub - angstrom misalignment between two layers in sliding blg . |
there are various reasons to suspect that baryon number is not a good symmetry of nature : ( i ) first is that nonperturbative effects of the standard model lead to @xmath6 , while keeping @xmath7 @xcite ; ( ii ) understanding the origin of matter in the universe requires @xmath8 @xcite and ( iii ) many theories beyond the standard model lead to interactions that violate baryon number @xcite .
if indeed such interactions are there , the important question is : can we observe them in experiments@xmath9 ?
two interesting baryon nonconserving processes of experimental interest are : ( a ) proton decay e.g. @xmath10 etc @xcite and ( b ) @xmath11 oscillation @xcite . these two classes of processes probe two different selection rules for baryon nonconservation : @xmath7 for proton decay and @xmath12 for @xmath11 oscillation and indicate totally different directions for unification beyond the standard model .
for example , observation of proton decay will point strongly towards a grand desert till about the scale of @xmath13 gev whereas @xmath11 oscillation will require new physics at an intermediate scale at or above the tev scale but much below the gut scale . while proton decay goes very naturally with the idea of eventual grand unification of forces and matter ,
recent discoveries of neutrino oscillations have made @xmath11 oscillation to be quite plausible theoretically if small neutrino masses are to be understood as a consequence of the seesaw mechanism @xcite .
this can be seen as follows : seesaw mechanism implies majorana neutrinos implying the existence of @xmath14 interactions . in the domain of baryons , it implies the existence of @xmath11 oscillation as noted many years ago @xcite .
an explicit model for @xmath11 oscillation was constructed in @xcite by implementing the seesaw mechanism within the framework of the pati - salam @xcite @xmath15 model , where quarks and leptons are unified .
it was shown that this process is mediated by the exchange of diquark higgs bosons with an amplitude ( @xmath16 ) which scales like @xmath17 . in the nonsupersymmetric version without fine tuning ,
one expects @xmath18 leading to @xmath19 .
so only if @xmath20 tev , the @xmath21 is in the range of @xmath22 sec and is accessible to experiments . on the other hand , in generic seesaw models for neutrinos ,
one expects @xmath23 gev depending on whether the third generation dirac mass for the neutrino is 1 - 100 gev .
an important question therefore is whether in realistic seesaw models , @xmath11 oscillation is at all observable .
another objection to the above nonsupersymmetric model for @xmath24 that was raised in the 80 s was that such interactions will erase any baryon asymmetry created at high scales
. it will be therefore important to overcome this objection . in this note
we point out that in a class of supersymmetric @xmath1 models ( called susy @xmath25 ) , an interesting combination of circumstances improves the @xmath26 dependence of the @xmath27 to @xmath28 instead of @xmath29 making @xmath11 oscillation observable .
further more the same model also allows us to overcome the difficulty with high scale baryogenesis .
an example of such a theory was presented in @xcite where it was shown that in the minimal version of the model , there exist accidental symmetries that imply that some of the @xmath30 s are in the tev range even though @xmath31 gev .
we discuss this class of theories in this letter .
the new results in this paper are : ( i ) a new diagram which enhances the @xmath11 oscillation amplitude ; ( ii ) a realistic example which is in agreement with the observed neutrino oscillation parameters while predicting observable @xmath11 oscillation amplitude and ( iii ) adequate leptogenesis mechanism @xcite for understanding the origin of matter via the quasi - degenerate leptogenesis @xcite with right - handed neutrinos at @xmath32 gev . at present , the best lower bound on @xmath33 comes from ill reactor experiment @xcite and is @xmath34 sec .
there are also comparable bounds from nucleon decay search experiments @xcite .
there are proposals to improve the precision of this search by at least two orders of magnitude @xcite .
we feel that the results of this paper should give new impetus to a search for neutron - antineutron oscillation .
the quarks and leptons in this model are unified and transform as @xmath35 representations of @xmath15 . for the higgs sector
, we choose , @xmath36 and @xmath37 to give mass to the fermions .
the @xmath38 to break the @xmath39 symmetry .
the diquarks mentioned above which lead to @xmath40 processes are contained in the @xmath41 multiplet .
we also add a @xmath39 neutral triplet @xmath42 which helps to reduce the number of light diquark states and a gauge singlet field @xmath43 .
the superpotential of this model is given by : @xmath44 where @xmath45 note that since we do not have parity symmetry in the model , the yukawa couplings @xmath46 and @xmath47 are not symmetric matrices .
when @xmath48 , this superpotential has an accidental global symmetry much larger than the gauge group@xcite ; as a result , vacuum breaking of the @xmath39 symmetry leads to the existence of light diquark states that mediate @xmath49 oscillation and enhance the amplitude .
in fact it was shown that for @xmath50 and @xmath51 and all vevs in the range of @xmath52 gev , the light states are those with quantum numbers : @xmath53 .
the symmetry argument behind is that @xcite for @xmath48 , the above superpotential is invariant under @xmath54 symmetry which breaks down to @xmath55 when @xmath56 .
this results in 21 complex massless states ; on the other hand these vevs also breaks the gauge symmetry down from @xmath57 to @xmath58 .
this allows nine of the above states to pick up masses of order @xmath59 leaving 12 massless complex states which are the six @xmath60 plus six @xmath61 states .
once @xmath62 , they pick up mass ( call @xmath63 ) of order of the elctroweak scale .
to discuss @xmath11 oscillation , we introduce a new term in the superpotential of the form : @xmath64 where the @xmath65 etc stand for @xmath66 indices and we have suppressed the @xmath67 indices .
we choose @xmath68 .
this does not affect the masses of the higgs fields .
when @xmath69 aquires a vev , @xmath70 interaction are induced from the superpotential , the contribution to neutron antineutron oscillation in this model comes from the diagram in fig .
1 , which gives @xmath71 which scales like @xmath72 .
there is also another pure higgs diagram that involves the vertex @xmath73 which also scales the same way and gives an identical contribution .
this graph is the supersymmetric partner of fig .
1 . note that for high scale seesaw models , these contributions to @xmath16 are considerably enhanced over that from the nonsupersymmetric theory@xcite which goes like @xmath29 .
oscillation.,width=226 ] in order to estimate the rate for @xmath11 oscillation , we need not only the different mass values for which we now have an order of magnitude , we also need the yukawa coupling @xmath74 .
now @xmath74 is a small number since its value is associated with the lightest right - handed neutrino mass .
however , in the calculation we need its value in the basis where quark masses are diagonal .
we note that the @xmath75 diagrams involve only the right - handed quarks , the rotation matrix need not be the ckm matrix .
the right - handed rotations need to be large e.g. in order to involve @xmath76 ( which is @xmath77 ) , we need @xmath78 to be large , where @xmath79 .
the left - handed rotation matrices @xmath80 contribute to the ckm matrix , but right - handed rotation matrices @xmath81 are unphysical in the standard model . in this model , however , we get to see its contribution since we have a left - right gauge symmetry as we will see later .
let us now estimate the time of oscillation .
when we start on a @xmath82-diagonal basis ( call the diagonal matrix @xmath83 ) , the majorana coupling @xmath74 in the diagonal basis of up- and down - type quark matrices can be written as @xmath84 .
now @xmath85 is @xmath77 and @xmath86 can be @xmath87 , so @xmath74 is about 0.4 in the diagonal basis of the quark matrices .
we use @xmath88 , @xmath63 @xmath89 gev and @xmath0 gev . due to the presence of the colored field
the couplings become non - perturbative very soon beyond the @xmath26 , we therefore take @xmath90 gev . the mass of @xmath91 i.e. @xmath92 is @xmath93 gev which is obtained from the vev of @xmath42 . putting all the above the numbers together ,
we get @xmath94 . along with the hadronic matrix element@xcite , the @xmath75 oscillation time is found to be about @xmath95 sec which is within the reach of possible next generation measurements .
we also note that as noted in @xcite the model is invariant under the hidden discrete symmetry under which a field @xmath96 , where @xmath97 is the baryon number of the field @xmath98 . as a result , proton is absolutely stable in the model .
since @xmath11 oscillation is a low intermediate scale phenomenon , it is necessary to see how it affects the baryogenesis discussion .
one could of course invoke weak scale baryogenesis , which would remain unaffected since it happen at the weak scale .
instead here we focus on leptogenesis mechanism for origin of matter .
the obvious problem here is that if leptogenesis takes place at a scale where the @xmath100 process is in thermal equilibrium , then the matter - antimatter asymmetry generated will be erased .
so two questions must be answered within our model : ( i ) what is the temperature @xmath101 below which the @xmath100 processes are out of equilibrium@xmath9 ? and
( ii ) what constraints must be obeyed by the right - handed neutrino spectrum in the theory so that it will be possible to generate the lepton asymmetry below @xmath101 ? clearly , @xmath101 must be above the weak scale so that the sphalerons can be effective in converting leptons to baryons . to answer the first question , we compare the @xmath100 reaction rate with the hubble expansion
. the diquark higgs field @xmath60 remains at weak scale in the thermal bath , while @xmath102 etc decouple at high scale .
the @xmath100 process surviving till lowest temperature is @xmath103 .
the coupling of the interaction can be estimated as @xmath104
. the out of equilibrium condition of this process would be : @xmath105 which gives @xmath106 gev .
this implies that for leptogenesis to work , at least one right - handed neutrino should be lighter than @xmath107 gev .
in fact in view of the low scale , we will need to invoke quasi - degenerate leptogenesis @xcite , in the case of thermal leptogenesis or one can also have the inflaton decaying into the two light right - handed neutrinos and the decay of the right - handed neutrinos would generate the required lepton asymmetry . in both these cases , the lighter right - handed neutrinos are almost degenerate to produce the correct amount of baryon asymmetry .
below , we give an example of the heavy right - handed neutrino as well as the dirac neutrino mass texture as an example that leads to the correct neutrino mixings as well as a correct amount of baryons .
we choose the majorana neutrino coupling to the @xmath108s as : @xmath109 where @xmath110 to have two right - handed neutrino around @xmath107 gev , where the heaviest neutrino mass is @xmath111 gev .
the degeneracy has to be broken , however , since we need @xmath112 , where @xmath113 and @xmath114 are masses and decay widths of right - handed neutrinos , respectively .
thus , we need small perturbation in 11 and 22 elements .
the formula for lepton asymmetry is : @xmath115 where the neutrino dirac yukawa coupling @xmath116 in this expression is given in the basis where the right - handed majorana neutrino mass matrix is diagonal .
we choose the following dirac neutrino coupling ( which generates the correct neutrino masses using type i seesaw ) @xmath117 where we have omitted phases of order one .
the asymmetry generated by this texture needs to be multiplied by the efficiency factor @xmath118 . for quasi degenerate right - handed neutrinos
, one gets @xmath119 . as a result
, there is an enhancement factor for the lepton asymmetry depending on the mass difference which is a parameter in this model .
the decay parameter @xmath120 , where @xmath121 is the decay width of the heavy majorana neutrino , is large ; for example , @xmath122 for above choice of @xmath123 and @xmath124 around @xmath107 gev , @xmath125 or smaller .
our model has several interesting phenomenological consequences both for colliders and rare decays for up - type quarks .
for instance , since our diquark higgs bosons and their superpartners have masses in the few hundred gev range , they should be produced at lhc , possibly at tevatron .
the production can happen singly or in pairs via parton processes such as @xmath126 , @xmath127 etc .
these new higgs boson and its superpartners can decay into to two charm , up quarks , top quarks should provide a good signal . at the lhc ,
the signal from a pair production may contain the following final states : 4 charm quarks , 2 top-2 charm quarks , 4 top quarks etc .
the fermionic partner of these diquark higgs , however , would produce missing energy in the final states . the single production will produce a change in the top quark pair production .
a detailed study is underway .
a second interesting consequence is that in the mass eigenstate basis for quarks , the light @xmath128 will mediate flavor changing processes such as @xmath129 , @xmath130 etc .
the present bounds on charm changing processes is consistent with the value of the diquark higgs boson mass that has been used in the paper .
in conclusion , we have presented a quark - lepton unified model where despite the high seesaw ( @xmath26 ) scale in the range of @xmath131 gev , the @xmath75 oscillation time is around @xmath132 sec . due to the presence of a new supersymmetric graph and accidental symmetries of the higgs potential ( also connected to supersymmetry ) .
no unatural fine tuning of diquark masses is needed to get such an enhanced effect .
our predicted oscillation time is within the reach of possible future experiments .
we believe that this work should provide a new motivation to conduct a new round of search for @xmath75 oscillation . to show that this is a completely realistic model , we note that ( i ) the proton is absolutely stable in this model ; ( ii ) the model can provide a realistic description of known quark and lepton physics as well as ( iii ) a way to generate an adequate lepton asymmetry via the mechanism of quasi - degenerate leptogenesis .
p. minkowski , phys .
b * 67 * , 421 ( 1977 ) ; m. gell - mann , p. ramond , and r. slansky , _ supergravity _ ( p. van nieuwenhuizen et al
. eds . ) , north holland , amsterdam , 1979 , p. 315 ; t. yanagida , in _ proceedings of the workshop on the unified theory and the baryon number in the universe _ ( o. sawada and a. sugamoto , eds . ) , kek , tsukuba , japan , 1979 , p. 95 ; s. l. glashow , _ the future of elementary particle physics _ , in _ proceedings of the 1979 cargse summer institute on quarks and leptons _
( m. lvy et al
. eds . ) , plenum press , new york , 1980 , p. 687
; r. n. mohapatra and g. senjanovi , phys .
lett . * 44 * , 912 ( 1980 ) .
s. rao and r. shrock , phys .
b * 116 * , 238 ( 1982 ) ; j. pasupathy , phys .
b * 114 * , 172 ( 1982 ) ; riazuddin , phys .
d * 25 * , 885 ( 1982 ) ; s. p. misra and u. sarkar , phys .
d * 28 * , 249 ( 1983 ) . | we discuss a realistic high scale ( @xmath0 gev ) supersymmetric seesaw model based on the gauge group @xmath1 where neutron - anti - neutron oscillation can be in the observable range without fine tuning of parameters .
this is contrary to the naive dimensional arguments which say that @xmath2 and should therefore be unobservable for seesaw scale @xmath3 gev .
two reasons for this enhancement are : ( i ) accidental symmetries which keep some of the diquark higgs masses at the weak scale and ( ii ) a new supersymmetric contribution from a lower dimensional operator .
the net result is that @xmath4 rather than @xmath5 .
the model also can explain the origin of matter via the leptogenesis mechanism and predicts light diquark states which can be produced at lhc . |
it is widely believed , as suggested by a host of independent cosmological and astrophysical observations , that the most part of the matter in the universe is not visible , revealing its existence only through gravitational effects .
in particular , data on the rotational curves of galaxies indicate that the galactic visible parts are surrounded by approximately spherical dark halos which extend up to several times the size of the luminous components .
the best candidates to provide dark matter in galaxies are weakly interacting massive particles ( wimp ) .
several wimp direct detection experiments are operating @xcite , with the goal of measuring the nuclear recoil energy ( in the kev range ) expected to be deposited in solid , liquid or gaseous targets by the scattering of the non - relativistic dark halo wimps .
unfortunately , expected rates are small and the exponential decay of the wimp recoil spectrum resembles that of the background at low energies .
however , a specific signature can be exploited in order to disentangle a wimp signal from the background : the annual modulation of the rate @xcite .
this effect , expected to be of the order of a few per cent , is induced by the rotation of the earth around the sun . due to its smallness
, the annual modulation signature requires large mass detectors with high statistics in order to overcome background fluctuations and be unambiguously detected .
the annual modulation effect has been experimentally investigated by the dama collaboration , which has indeed reported a positive evidence by using a 100 kg sodium iodide detector @xcite .
one of the most important sources of uncertainty in the calculation of wimp direct detection rates is the modeling of the velocity distribution function ( df ) of the particles populating the dark halo . in the literature , a simple isothermal sphere model is usually adopted , i.e. a wimp gas described by an isotropic maxwellian with r.m.s velocity of the order of 300 km s@xmath0
this leads to a sinusoidal time dependence of the expected signal with maximum ( or minimum ) around june 2@xmath1 , i.e. with the same ( or opposite ) phase as the relative velocity between the earth and the halo rest frame . however , the actual form of the wimp velocity df is unknown , and many different models , alternative to the isothermal sphere , are compatible with observations @xcite .
the goal of the present letter is to show that the main features of the annual modulation effect ( the sinusoidal dependence with time , the occurrence of maxima and minima during the year and , under some circumstances , the one year period ) may be affected by anisotropies in the velocity df .
the most relevant effect is a distortion of the sinusoidal time
behaviour at low recoil energies .
these energies , though below the current detector thresholds , might be reached in the future .
the observation of the effects discussed in this letter could provide informations on the phase space distribution of our galactic halo , especially on the degree of its anisotropy .
due to the rotation of the disk around the galactic center , the solar system moves through the wimp halo , assumed to be at rest in the galactic rest frame . in
the following , we will assume a right handed system of orthogonal coordinates : the @xmath2 axis in the galactic plane , pointing radially outward ; the @xmath3 axis in the galactic plane , pointing in the direction of the disk rotation ; the @xmath4 axis directed upward , perpendicular to the galactic plane .
notice that our system differs from standard `` galactic coordinates '' by the different choice of the @xmath2 axis , which for us is directed outward .
the relative velocity between the wimp halo and the detector is given by the earth velocity @xmath5 , as seen in the galactic rest frame .
it is the sum of three components : the galactic rotational velocity @xmath6 km s@xmath0 ( we will assume : @xmath7 km s@xmath0 ) , the sun proper motion @xmath8 km s@xmath0@xcite and the earth orbital motion @xmath9@xcite : @xmath10 \label{eq : vearthx } , \\
v^e_y & = & v^g_y + v^s_y + u^e(\lambda ) \cos\beta_y \cos[\omega ( t - t_y ) ] \label{eq : vearthy } , \\
v^e_z & = & v^g_z + v^s_z + u^e(\lambda ) \cos\beta_z \cos[\omega ( t - t_z ) ] \label{eq : vearthz},\end{aligned}\ ] ] where @xmath11 is the ecliptic longitude , which is function of time
. we can express @xmath11 as @xcite : @xmath12 where @xmath13 and @xmath14 and @xmath15 denotes the time expressed in days relative to ut noon on december 31 .
in eqs.([eq : vearthx][eq : vearthz ] ) @xmath16 $ ] is the modulus of the earth rotational velocity , which slightly changes with time due to the small ellipticity @xmath17 of the earth orbit ( @xmath18 km s@xmath0 , @xmath19 and @xmath20 ) @xcite . in eqs.([eq : vearthx][eq : vearthz ] ) the @xmath21 denote the ecliptic latitudes and the @xmath22 are the phases of the three velocity components : @xmath23 , @xmath24 , @xmath25 and @xmath26 day , @xmath27 day , @xmath28 day .
the angular velocity has a period of 1 year , and is given by : @xmath29 .
with the numbers given above , the modulus of the earth velocity changes in time as : @xmath30 $ ] ( in km s@xmath0 ) , where @xmath31 days , i.e. june 2@xmath32 .
notice the slight offset between @xmath33 and @xmath34 , due to the composition of the velocity components .
as a good approximation , @xmath35 is usually taken as : @xmath36 $ ] .
the direct detection differential rate @xmath37 is proportional to the integral : @xmath38 where @xmath39 and @xmath40 are the wimp velocity df and the wimp velocity , respectively , in the earth s rest frame ; @xmath41 is the minimum value of @xmath40 for a given recoil energy @xmath42 , wimp mass @xmath43 and nuclear target mass @xmath44 and is given by : @xmath45 . by indicating with @xmath46 and
@xmath47 the wimp velocity df and the wimp velocity in the galactic reference frame , the following transformations hold : @xmath48 which imply that @xmath49 , and so @xmath37 , develops a time dependence induced by @xmath50 .
time dependence of @xmath49 for an isotropic isothermal sphere .
the different curves refer to values of @xmath41 ranging from 0 ( lower ) to 400 ( upper ) km s@xmath0 , in steps of 20 km s@xmath0 ( the dashed curves correspond to @xmath51 , 200 and 300 km s@xmath0 ) .
the vertical dashed line denotes @xmath52 days . ] in the case of the isothermal sphere model the df is given by a truncated isotropic maxwellian which depends only on @xmath53 : @xmath54 where @xmath55 is the wimp r.m.s .
velocity , given by : @xmath56 .
it is clear that , through the change of reference frame of eq .
( [ eq : transformation ] ) , @xmath49 depends on time only through @xmath35 . since the relative change of @xmath35 during the year is of the order of a few per cent , we can approximate @xmath49 with its first
order expansion in the small parameter @xmath57 , around its mean value @xmath58 : @xmath59 $ ] .
the well known result is then obtained that the wimp rate has a sinusoidal time dependence with the same phase ( @xmath60 june 2@xmath1 ) as @xmath35 , for all values of @xmath41 .
this is shown in fig.[fig : ivmin_isotropic ] , where @xmath49 is plotted as a function of time for various values of @xmath41 .
the simplest generalization of the isothermal sphere model is given by a triaxial system described by a multivariate gaussian : @xmath61 where @xmath62 is the normalization constant .
for @xmath63 , then @xmath64^{-1/2}$ ] .
the usual isothermal sphere is the spherical limit of eq.([eq : triaxialgs ] ) , obtained with : @xmath65 . in order to discuss the effect of anisotropy at fixed wimp mean kinetic energy ,
we will fix @xmath66 as in the isothermal case ( @xmath56 ) and discuss our results in terms of the two independent parameters : @xmath67 and @xmath68 . at variance with the isothermal sphere ,
now @xmath49 depends in general on all the three components of @xmath69 , and not simply on @xmath35 .
we can write : @xmath70 , \label{eq : triaxiales}\ ] ] where we have defined the reduced ( dimensionless ) variables : @xmath71 and @xmath72 $ ] ( @xmath73 ) . in the isotropic case ( isothermal sphere )
one has @xmath74 , @xmath75 and @xmath76 .
the presence of the order one parameter @xmath77 and of the small oscillation amplitudes @xmath78 allows the taylor expansion of @xmath49 in terms of @xmath79 parameters .
a straightforward calculation shows that the conclusions of the previous section are recovered .
the same as in fig .
[ fig : ivmin_isotropic ] , for an anisotropic model with @xmath80 and @xmath81 . ] on the contrary , allowing anisotropies such that @xmath82 and/or @xmath83 , the values of the parameters @xmath84 and @xmath85 are enhanced , and the time dependence of @xmath86 and @xmath87 in @xmath49 may become important .
an example of this situation is shown in fig .
[ fig : ivmin_tangential ] , where , the time evolution of @xmath49 is plotted for @xmath88 and @xmath89 ( i.e. : @xmath90 km s@xmath0 . )
this choice of @xmath91 , @xmath92 , which refers to a tangential anisotropy , corresponds to triaxial models discussed , for instance , in refs .
@xcite . a distortion of the curves of fig .
[ fig : ivmin_tangential ] , as compared to the familiar sinusoidal time dependence , appears : this effect may be explained by the fact that now @xmath93 , a taylor expansion of the type used in the isothermal sphere case breaks down and a full numerical calculation of the integral of eq.([eq : ivmin ] ) is required .
the final result is not sinusoidal .
this peculiar behaviour is more pronounced at low values of @xmath41 ( i.e. low recoil energies ) namely for @xmath94 80 km s@xmath0 , for which the distortion is strong and the maxima ( in absolute value ) of the rate are shifted as compared to the standard case @xcite . for larger values of @xmath41
the distortion is less pronounced , and it dies away when @xmath95 km s@xmath0 .
this may be explained by the fact that , as @xmath41 grows , the integral of eq.([eq : ivmin ] ) becomes less sensitive to the parameter @xmath96 since it gets increasingly dominated by wimps with velocities along the @xmath3 axis , which is the one along which the boost due to the galactic rotation is directed .
we notice that for values of @xmath41 around ( @xmath97 ) km s@xmath0 a distortion is present , but the amplitude of the modulation is suppressed and therefore difficult to detect . the same as in fig .
[ fig : ivmin_isotropic ] , for an anisotropic model with @xmath98 and @xmath99 .
the curves refer to : @xmath100 ( lower ) , 180 , 190 , 200 ( dashed ) , 210 , 220 , 230 , 240 , 250 ( upper ) km s@xmath0 . ] as a second example , in fig.[fig : ivmin_radial ] we plot @xmath49 as a function of time for the case @xmath101 , @xmath102 ( i.e. : @xmath103 km s@xmath0 .
this situation is representative of a radial anisotropy . in this case
, we have further enhanced the contribution of @xmath104 over @xmath85 , so that one should expect to draw the same conclusions as in the isothermal sphere case , with the usual sinusoidal time dependence of the rate and a phase close to @xmath105 .
this is indeed the case , except for a narrow interval of values of @xmath41 around @xmath106 210 km s@xmath0 . in this range , @xmath49 develops two maxima because , for that particular choice of @xmath41 , there is an exact cancellation in the first term of the expansion in @xmath107 , so that the term @xmath108 proportional to @xmath109 $ ] sets in
. this particular cancellation of the first term in the taylor expansion of @xmath49 happens also for the isothermal sphere model , but in that case the size of the quadratic term is strongly suppressed because @xmath110 is much smaller . in a wimp direct detection experiment
this effect would show up in a very peculiar way : a halving of the modulation period of the rate in a narrow range of recoil energies .
note , however , that in order to have some realistic chance to detect this effect , it should show up in one of the experimental energy bins just above threshold , where the highest signal / background ratio is usually attained .
values of @xmath41 as a function of the quenched nuclear recoil energies @xmath111 , for : nai scintillators @xcite , ge ionization detectors @xcite and te bolometers @xcite . for each panel ,
the different curves refer to wimp masses of : 20 gev ( solid ) , 50 gev ( dotted ) , 100 gev ( dashed ) , 200 gev ( long dashed ) and 1 tev ( dot dashed ) .
the dashed vertical lines denote the current energy thresholds . ] in order to establish a link between our discussion and wimp direct detection experiments @xcite , in fig .
[ fig : vmin ] we plot @xmath41 as a function of the quenched nuclear recoil energy @xmath112 ( @xmath113 is the quenching factor ) for the target nuclei : na , i , ge , te and for different wimp masses .
the vertical dashed lines show current energy thresholds achieved by each type of detector .
[ fig : vmin ] shows that values of @xmath94 80 km s@xmath0 , i.e. sufficiently low to observe a sizeable distortion effect as the one discussed for tangential anisotropy , correspond to wimp recoil energies below the threshold of present direct detection experiments , and that the effect would be more easily detected at higher wimp masses .
however , a foreseeable lowering of the threshold , down to 0.51 kev , could be enough to observe the distortion .
on the other hand , for radial anisotropy , we can conclude that the recoil energy corresponding to a halving of the modulation period can actually coincide to the experimental thresholds within the reach of present day detectors for 20 gev @xmath114 100 gev , depending on the particular target nucleus . in this respect ,
we note that the properties of the annual modulation effect observed by the dama / nai experiment @xcite ( a one year
period sinusoidal behaviour in the 26 kev energy bins @xcite ) implies that the dama / nai experiment is already able to set constraint on strong radial anisotropies .
in the present letter we have shown that the main features of the annual modulation of the signal of wimp direct searches , i.e. the sinusoidal dependence of the rate with time , the position of its maxima and minima during the year and even the period , may be affected by relaxing the isothermal sphere hypothesis in the description of the wimp velocity phase space .
we have considered a multivariate gaussian and found that different situations may occur , depending on the pattern of anisotropy : tangential anisotropies induce a departure at low energies from the usual sinusoidal time dependence , along with a shift in the position of the maximum of the signal during the year , while radial anisotropies may produce a halving of the modulation period in a particular energy bin .
the former effect turns out to be relevant at low recoil energies , actually below the threshold of present day experiments , while the latter should be already within the reach of current detectors .
in particular , the properties of the annual modulation effect observed by the dama / nai experiment @xcite may already indicate that strong radial anisotropies are excluded . | we show that the main features of the annual modulation of the signal expected in a wimp direct detection experiment , i.e. its sinusoidal dependence with time , the occurrence of its maxima and minima during the year and ( under some circumstances ) even the one year period , may be affected by relaxing the isothermal sphere hypothesis in the description of the wimp velocity phase space .
the most relevant effect is a distortion of the time
behaviour at low recoil energies for anisotropic galactic halos .
while some of these effects turn out to be relevant at recoil energies below the current detector thresholds , some others could already be measurable , although some degree of tuning between the wimp mass and the experimental parameters would be required .
either the observation or non observation of these effects could provide clues on the phase space distribution of our galactic halo . |
exotic pentaquark baryons has been a hot issue , since the leps collaboration @xcite announced the new finding of the @xmath7 baryon @xmath8 which was soon confirmed by a number of other experiments @xcite , together with an observation of exotic @xmath9 states by the na49 experiment at cern @xcite , though it is still under debate .
those experiments searching for the pentaquark states was stimulated by diakonov et al .
@xcite : masses and decay widths of exotic baryon antidecuplet were predicted within the chiral quark - soliton model .
the discoveries of the pentaquark baryon @xmath1 and possibly of @xmath10 have triggered intensive theoretical investigations ( see , for example , refs.@xcite ) .
the production mechanism of the @xmath1 has been discussed in refs.@xcite . in particular , it is of great interest to understand the photoproduction of the @xmath1 theoretically , since the leps and clas collaborations used photons as a probe to measure the @xmath1 . in order to describe the mechanism of the pentaquark photoproduction , we have to know the magnetic moment of the @xmath1 and its strong coupling constants .
however , information on the static properties such as antidecuplet magnetic moments and their strong coupling constants is absent to date , so we need to estimate them theoretically .
recently , two of the present authors calculated the magnetic moments of the exotic pentaquarks , within the framework of the chiral quark - soliton model @xcite in the chiral limit .
since we were not able to fix all the parameters for the magnetic moments in the chiral limit , we had to rely on the explicit model calculations @xcite .
a very recent work @xcite extended the analysis for the magnetic moments of the baryons , taking into account the effect of su(3 ) symmetry breaking so that the necessary parameters are fixed by the magnetic moments of the baryon octet . in the present talk
, we would like to present the main results of ref .
the collective hamiltonian describing baryons in the su(3 ) chiral quark - soliton model takes the following form @xcite : @xmath11 with the symmetry breaking term given by:@xmath12 where parameters @xmath13 , @xmath14 and @xmath15 are of order @xmath16 and are given as functions of the @xmath17 sigma term @xcite . @xmath18
denote su(3 ) wigner rotation matrices and @xmath19 is a collective spin operator .
the hamiltonian given in eq.([hsplit ] ) acts on the space of baryon wave functions : @xmath20 here , @xmath21 stands for the allowed irreducible representations of the su(3 ) flavor group , _
i.e. _ @xmath22 and @xmath23 are the corresponding hypercharge , isospin , and its third component , respectively .
right hypercharge @xmath24 is constrained to be unity for the physical spin states for which @xmath25 and @xmath26 are spin and its third component . the _ model - independent approach _ consists now in using eqs .
( [ eq;1 ] ) and ( [ hsplit ] ) ( and/or possibly analogous equations for other observables ) and determining model parameters such as @xmath27 from experimental data
. the symmetry - breaking term ( [ hsplit ] ) of the collective hamiltonian mixes different su(3 ) representations as follows : @xcite@xmath28 where @xmath29 denotes the state which reduces to the su(3 ) representation @xmath21 in the formal limit @xmath30 and the spin index @xmath26 has been suppressed .
all relevant expressions for the mixing coefficients @xmath31 , @xmath32 , and @xmath33 can be found in ref .
the collective operator for the magnetic moments can be parameterized by six constants by definition in the _ model - independent approach _
they are treated as free @xcite:@xmath34 the parameters @xmath35 are of order @xmath36 , while @xmath37 are of order @xmath16 , @xmath38 being regarded as a small parameter .
the full expression for the magnetic moments can be decomposed as follows : @xmath39 where the @xmath40 is given by the matrix element of the @xmath41 between the purely symmetric states @xmath42 , and the @xmath43 is given as the matrix element of the @xmath44 between the symmetry states as well .
the wave function correction @xmath45 is given as a sum of the interference matrix elements of the @xmath40 between purely symmetric states and admixtures displayed in eq.([admix ] ) .
these matrix elements were calculated for octet and decuplet baryons in ref.@xcite .
the measurement of the @xmath1 mass constrains the parameter space of the model .
recent phenomenological analyzes indicate that our previous assumption on @xmath15 , i.e. @xmath46 , has to be most likely abandoned .
therefore , our previous results for the magnetic moments of @xmath47 , @xmath48 and @xmath49 have to be reanalyzed .
now , we show that a _ model - independent _ analysis with this new phenomenological input yields @xmath50 much larger than initially assumed , which causes @xmath51 for realistic values of @xmath52 to be negative and rather small . our previous results for the decuplet magnetic moments turn out to hold within the accuracy of the model .
the octet and decuplet magnetic moments were calculated in refs.@xcite . for the antidecuplet @xmath53 can be found in ref.@xcite . in order to calculate the @xmath45 , several off - diagonal matrix elements of the @xmath41 are required .
these have been calculated in ref.@xcite in the context of the hadronic decay widths of the baryon antidecuplet .
denoting the set of the model parameters by @xmath54 the model formulae for the set of the magnetic moments in representation @xmath21 ( of dimension @xmath55)@xmath56 can be conveniently cast into the form of the matrix equations:@xmath57\cdot\vec{w},%\ ] ] where rectangular matrices @xmath58 , @xmath59 , and @xmath60 can be found in refs.@xcite .
note their dependence on the pion - nucleon @xmath52 term .
in order to find the set of parameters @xmath61 $ ] , we minimize the mean square deviation for the octet magnetic moments : @xmath62-\mu_{b,\,exp}^{8}\right ) ^{2}},\ ] ] where the sum extends over all octet magnetic moments , but the @xmath63 .
the value @xmath64 is in practice independent of the @xmath52 in the physically interesting range @xmath65 @xmath66 @xmath5 mev .
the values of the @xmath67 $ ] are independent of @xmath52 . similarly , the value of the nucleon strange magnetic moment is independent of @xmath52 and reads @xmath68 in fair agreement with our previous analysis of ref.@xcite .
parameters @xmath69 , however , do depend on @xmath52 .
this is shown in table.[tablewi ] : [ c]|c|cccccc| & @xmath70 & @xmath50 & @xmath71 & @xmath72 & @xmath73 & @xmath74 + @xmath65 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 & @xmath80 + @xmath81 & @xmath82 & @xmath83 & @xmath77 & @xmath84 & @xmath79 & @xmath80 + @xmath5 & @xmath85 & @xmath86 & @xmath77 & @xmath87 & @xmath79 & @xmath80 + note that parameters @xmath88 are formally @xmath89 with respect to @xmath70 . for smaller
@xmath52 , this @xmath90 counting is not borne by explicit fits .
the @xmath91 can be parametrized by the following two parameters @xmath92 and @xmath93 : @xmath94{ccrcc}% v & = & \left ( 2\mu_{\mathrm{n}}-\mu_{\mathrm{p}}+3\mu_{\xi^{0}}+\mu_{\xi ^{-}}-2\mu_{\sigma^{-}}-3\mu_{\sigma^{+}}\right ) /60 & = & -0.268,\\ w & = & \left ( 3\mu_{\mathrm{p}}+4\mu_{\mathrm{n}}+\mu_{\xi^{0}}-3\mu _ { \xi^{-}}-4\mu_{\sigma^{-}}-\mu_{\sigma^{+}}\right ) /60 & = & 0.060 .
\end{array } \label{eq : mean}%\ ] ] which are free of linear @xmath38 corrections @xcite .
this is a remarkable feature of the present fit , since when the @xmath38 corrections are included , the @xmath38-independent parameters need not be refitted .
this property will be used in the following when we restore the linear dependence of the @xmath95 on @xmath38 .
the magnetic moments of the baryon decuplet and antidecuplet depend on the @xmath52 .
however , the dependence of the decuplet is very weak , which as summarized in table [ table10 ] , [ c]|c|cccc|@xmath52 [ mev ] & @xmath96 & @xmath97 & @xmath98 & @xmath99 + @xmath65 & @xmath100 & @xmath101 & @xmath102 & @xmath103 + @xmath81 & @xmath104 & @xmath105 & @xmath106 & @xmath107 + @xmath5 & @xmath104 & @xmath105 & @xmath108 & @xmath109 + & @xmath110 & @xmath111 & @xmath112 & @xmath113 + [ c]|c|cccccc|@xmath52 [ mev ] & @xmath114 & @xmath115 & @xmath116 & @xmath117 & @xmath118 & @xmath119 + @xmath65 & @xmath120 & @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 + @xmath81 & @xmath120 & @xmath121 & @xmath126 & @xmath123 & @xmath127 & @xmath125 + @xmath5 & @xmath128 & @xmath121 & @xmath129 & @xmath130 & @xmath127 & @xmath125 + & @xmath131 & @xmath132 & @xmath133 & @xmath134 & @xmath135 & @xmath136 + where we also display the theoretical predictions from ref . for @xmath137 .
let us note that the @xmath38 corrections are not large for the decuplet and the approximate proportionality of the @xmath138 to the baryon charge @xmath139 still holds . finally , for antidecuplet we have a strong dependence on @xmath52 , yielding the numbers of table [ tab10b ] .
[ c]|c|cccccc|@xmath52 [ mev ] & @xmath1 & @xmath140 & @xmath141 & @xmath142 & @xmath143 & @xmath144 + @xmath65 & @xmath145 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & @xmath150 + @xmath81 & @xmath151 & @xmath152 & @xmath153 & @xmath154 & @xmath155 & @xmath156 + @xmath5 & @xmath157 & @xmath158 & @xmath159 & @xmath160 & @xmath161 & @xmath162 + [ c]|c|cccc|@xmath52 [ mev ] & @xmath163 & @xmath164 & @xmath165 & @xmath166 + @xmath65 & @xmath167 & @xmath168 & @xmath169 & @xmath170 + @xmath81 & @xmath171 & @xmath172 & @xmath173 & @xmath174 + @xmath5 & @xmath175 & @xmath176 & @xmath177 & @xmath178 + the results listed in table [ tab10b ] are further depicted in fig.[fig:10bar ] . .
] the wave function corrections cancel for the non - exotic baryons and add constructively for the baryon antidecuplet . in particular
, for @xmath179 mev we have large admixture coefficient of 27-plet : @xmath180 tends to dominate otherwise small magnetic moments of antidecuplet . at this point
, the reliability of the perturbative expansion for the antidecuplet magnetic moments may be questioned . on the other hand ,
as remarked above , the @xmath90 counting for the @xmath69 coefficients works much better for large @xmath52 .
one notices for reasonable values of @xmath52 some interesting facts , which were partially reported already in ref.@xcite : the magnetic moments of the antidecuplet baryons are rather small in absolute value . for @xmath8 and @xmath181
one obtains negative values although the charges are positive . for @xmath182 and @xmath183
one obtains positive values although the signs of the charges are negative .
our present analysis shows that @xmath184 , although the magnitude depends strongly on the model parameters .
the measurement of @xmath185 could therefore discriminate between different models .
this also may add to reduce the ambiguities in the pion - nucleon sigma term @xmath52 . in the present work , we determined the magnetic moments of the baryon antidecuplet in the _ model - independent _ analysis within the chiral quark - soliton model , i.e. using the rigid - rotor quantization with the linear @xmath38 corrections included .
starting from the collective operators with dynamical parameters fixed by experimental data , we obtained the magnetic moments of the baryon antidecuplet .
the expression for the magnetic moments of the baryon antidecuplet is different from those of the baryon decuplet .
we found that the magnetic moment @xmath186 is negative and rather strongly dependent on the value of the @xmath52 .
indeed , the @xmath187 ranges from @xmath2 to @xmath3 for @xmath188 and @xmath5 mev , respectively .
h .- ch.k and g .- s .
y are very much thankful to the organizers of the workshop pentaquark 04 , in particular , t. nakano and a. hosaka for their hospitality .
ch.k is grateful to j.k .
ahn , s.i .
nam , m.v .
polyakov , and i.k .
yoo for valuable discussions .
the present work is supported by korea research foundation grant : krf-2003 - 041-c20067 ( h .- ch.k . ) and by the polish state committee for scientific research under grant 2 p03b 043 24 ( m.p . ) and by korean - german and polish - german grants of the deutsche forschungsgemeinschaft .
10 t. nakano _ et al .
_ [ leps collaboration ] , phys . rev .
lett . * 91 * , 012002 ( 2003 ) [ arxiv : hep - ex/0301020 ] .
s. stepanyan _ et al . _ [ clas collaboration ] , phys .
* 91 * ( 2003 ) 252001 [ arxiv : hep - ex/0307018 ] ; phys . rev .
* 92 * ( 2004 ) 032001 [ erratum - ibid .
* 92 * ( 2004 ) 049902 ] [ arxiv : hep - ex/0311046 ] ; v. v. barmin _ et al . _
[ diana collaboration ] , phys .
* 66 * ( 2003 )
1715 [ yad . fiz .
* 66 * ( 2003 ) 1763 ] [ arxiv : hep - ex/0304040 ] ; j. barth [ saphir collaboration ] , hep - ex/0307083 ; a. e. asratyan , a. g. dolgolenko and m. a. kubantsev , phys .
* 67 * ( 2004 ) 682 [ yad .
* 67 * ( 2004 ) 704 ] [ arxiv : hep - ex/0309042 ] ; a. airapetian _ et al . _
[ hermes collaboration ] , phys .
b * 585 * ( 2004 ) 213 [ arxiv : hep - ex/0312044 ] ; a. aleev _ et al .
_ [ svd collaboration ] , arxiv : hep - ex/0401024 ; m. abdel - bary _ et al . _ [ cosy - tof collaboration ] , arxiv : hep - ex/0403011 ; p. z. aslanyan , v. n. emelyanenko and g. g. rikhkvitzkaya , arxiv : hep - ex/0403044 ; s. chekanov _ et al . _
[ zeus collaboration ] , phys .
b * 591 * ( 2004 ) 7 [ arxiv : hep - ex/0403051 ] ; c. alt _ et al . _ [ na49 collaboration ] , phys .
* 92 * ( 2004 ) 042003 [ arxiv : hep - ex/0310014 ] ; d. diakonov , v. petrov and m. v. polyakov , z. phys .
* a359 * ( 1997 ) 305 [ hep - ph/9703373 ] . b. k. jennings and k. maltman , phys .
d * 69 * , 094020 ( 2004 ) [ arxiv : hep - ph/0308286 ] and references therein .
s. l. zhu , arxiv : hep - ph/0406204 and references therein .
k. goeke , h .- ch .
kim , m. praszaowicz , and gh .- s .
yang , to appear in prog .
( 2004 ) .
h .- ch .
kim , m. praszaowicz and k. goeke , phys .
* d57 * ( 1998 ) 2859 [ hep - ph/9706531 ] .
kim , m. praszaowicz , m. v. polyakov and k. goeke , phys .
* d58 * ( 1998 ) 114027 [ hep - ph/9801295 ] .
g. s. yang , h.- ch .
kim , m. praszalowicz and k. goeke , arxiv : hep - ph/0410042 .
a. blotz , k. goeke , n. w. park , d. diakonov , v. petrov and p. v. pobylitsa , phys .
b * 287 * ( 1992 ) 29 ; a. blotz , d. diakonov , k. goeke , n. w. park , v. petrov and p. v. pobylitsa , nucl .
a * 555 * ( 1993 ) 765 .
j. r. ellis , m. karliner and m. praszaowicz , jhep * 0405 * ( 2004 ) 002 [ arxiv : hep - ph/0401127 ] ; m. praszaowicz , acta phys .
b * 35 * ( 2004 ) 1625 [ arxiv : hep - ph/0402038 ] . | we present in this talk a recent analysis for the magnetic moments of the baryon antidecuplet within the framework of the chiral quark - soliton model with linear @xmath0 corrections considered .
we take into account the mixing of higher representations to the collective magnetic moment operator , which comes from the su(3 ) symmetry breaking .
dynamical parameters of the model are fixed by experimental data for the magnetic moments of the baryon octet as well as by the masses of the octet , decuplet and of @xmath1 .
the magnetic moment of @xmath1 is rather sensitive to the pion - nucleon sigma term and ranges from @xmath2 to @xmath3 as the sigma term is varied from @xmath4 to @xmath5 mev , respectively . on top of them
, we obtained that the strange magnetic moment of the nucleon has the value of @xmath6 n.m . within this scheme and
turns out to be almost independent of the sigma term . |
clusters of galaxies distort the images of background galaxies due to the gravitational lensing effect .
the statistics of such lensed arcs have been recognized as a powerful probe of the density profile of lens clusters @xcite . in particular , combined statistics of tangential and radial arcs are useful in determining the density profile of clusters @xcite .
a knowledge of the density profile is important because of recent indications that the cold dark matter scenario predicts a cuspy profile ( e.g. , * ? ? ?
* ; * ? ? ?
* ) while observations seem to prefer the existence of flat density cores ( e.g. , * ? ? ?
most of the previous analytic work assume the infinitesimally small size of source galaxies .
as pointed out by @xcite and @xcite , however , the cross section for radial arcs may be severely affected by the finite source size . moreover
, finite size sources can produce the fold image , thus the cross section may be somewhat different from the one obtained by assuming the infinitesimal source size . therefore , we analytically calculate the cross section for radial arcs including the source size effect .
we also derive the cross section using monte carlo method and compare this with theoretical predictions .
another systematic effect is due to the asymmetry in the mass distribution of the lens cluster @xcite .
@xcite claimed that the numerically modeled clusters produce long arcs about two orders of magnitude more frequently than spherically symmetric cluster models .
therefore , we also study the effects of asymmetries of lens objects .
especially we concentrate on the difference of asymmetry effects between tangential and radial arcs . as for the source ellipticity
, @xcite analytically calculated the cross section including the source ellipticity and found that the effect of the source ellipticity is small compared with the effect of different density profiles .
nevertheless we also examine the effect of the source ellipticity in order to check the theoretical predictions and to see the effect of different threshold axis ratios of arcs .
this paper is organized as follows . in [ sec : ana ] , we analytically derive the cross section for radial arcs including the finite source size .
section [ sec : sim ] describes the simulation method , and [ sec : result ] presents the cross section derived by numerical simulations .
the validity of our selection criterion for tangential and radial arcs is discussed in [ sec : select ] .
finally , we summarize the results in [ sec : summary ] . throughout this paper
, we assume the lambda - dominated cosmology @xmath2 .
the hubble constant in units of @xmath3 is denoted by @xmath4 .
the previous analytic work often assumed the infinitesimal size of source galaxies . on the other hand ,
it has been pointed out that the finite source size affects the cross section particularly for radial arcs @xcite .
moreover , sources with a finite size have a possibility to produce the fold image which is formed by merging two images near the radial critical curve , even in the case of the spherical symmetric lens .
therefore , the cross section for radial arcs including the finite source size effect may be somewhat different from the one calculated by neglecting the source size . in this section
, we derive an analytic expression of the cross section for radial arcs taking account of the finite size of source galaxies assuming spherical lenses .
the image position @xmath5 in the lens plane corresponding to the source at @xmath6 in the source plane is determined by the lens equation ( e.g. , * ? ? ?
* ) : @xmath7 where @xmath8 , @xmath9 , and @xmath10 and @xmath11 denote the angular diameter distances from the observer to the lens and the source planes , respectively .
the normalization length @xmath12 can be arbitrary .
the deflection angle @xmath13 is related to the mass distribution of the lens object . in the case of spherical lenses ,
a source at @xmath14 is stretched by the factor @xmath15 along the tangential direction and @xmath16 along the radial direction .
then for the infinitesimal source , the cross sections for tangential and radial arcs are simply given by the areas in the source plane in which the inequalities : @xmath17 are satisfied .
for the threshold axis ratio , e.g. , @xcite adopted @xmath18 .
since the assumption that the source size is sufficiently small seems to be valid for usual tangential arcs ( see * ? ? ?
* ) , in this paper we use equation ( [ tth ] ) to predict the cross section for tangential arcs . consider the cross section for radial arcs when the source has the finite size .
first we consider the fold images .
these images are produced only when sources cross the radial caustic : @xmath19 where @xmath20 is the radius of the radial caustic and @xmath21 is the dimensionless source radius .
in this situation , we approximate the criterion for radial arcs as ( see figure [ fig : radial ] ) @xmath22 where @xmath23 is the radius of the radial critical line and @xmath24 denotes two farthest points from the radial critical line . the latter is related to the center of the source @xmath25 as @xmath26 with @xmath27 being the radius corresponding to @xmath28 . the effective radius of the source @xmath29 in equation ( [ rf ] )
is defined by @xmath30 we calculate the cross section as the area in the source plane in which the condition ( [ rf ] ) is satisfied at both sides of the radial critical line in the image plane , that is , the area in which the following condition is satisfied : @xmath31 where @xmath32 are two radii corresponding to solutions of the equation @xmath33 at both sides of the radial critical line .
next consider images which do not touch the radial critical line .
these images are possible only for @xmath34 except for this condition , the cross section can be calculated similarly as the one with infinitesimal source size , and is explicitly expressed as @xmath35 where @xmath36 are two radii corresponding to solutions of the equation @xmath37 at both sides of the radial critical line . combining equations ( [ fold ] ) , ( [ foldlim ] ) , ( [ sep ] ) , and ( [ notfoldlim ] ) , the cross section for radial arcs @xmath38 is written as @xmath39 where @xmath40 is defined by @xmath41 the radii @xmath42 , @xmath43 , @xmath44 , and @xmath45 are defined by equations ( [ fold ] ) , ( [ fold ] ) , ( [ foldlim ] ) , and ( [ notfoldlim ] ) , respectively .
figure [ fig : plotline ] shows the dependence of each radius defined in the above on the source radius @xmath21 .
this figure indicates that the moderate amount of source size _ enhances _ the cross section , while the larger source size significantly decreases the cross section .
in this section , we briefly describe the mapping and detecting algorithm of arcs . most of our method described in [ sec : map ] and [ sec : rec ] follows the work by @xcite and @xcite . for density profiles of clusters , we adopt the generalization of those proposed by @xcite , the generalized nfw profile @xcite : @xmath46 for this profile , we choose the normalization of the lens equation as @xmath47 .
the concentration parameter @xmath48 , where @xmath49 is the virial radius of dark halo , depends on the halo mass @xmath50 and the redshift @xmath51 .
we calculate @xmath52 using the fitting form derived by @xcite : @xmath53 for @xmath54 , and we generalize it to @xmath55 by multiplying @xmath56 ( see * ? ? ?
ellipticities of the lens and source are included as follows .
for the lens ellipticity , we simply substitute @xmath57 for @xmath14 in the axially symmetric lens potential @xmath58 : @xmath59 where @xmath60 . with this substitution ,
the ellipticity is @xmath61 , where @xmath62 and @xmath63 are the semi - major and semi - minor axes .
then the deflection angle @xmath64 is obtained by @xmath65 where @xmath66 is the deflection angle for the axially symmetric lens .
note that this elliptical model is same as the one adopted by @xcite ( @xcite ; see also @xcite ) .
the source ellipticity is included in the detection of images ( eq . [ [ withinsource ] ] ) . to begin with ,
we choose a sufficiently large region ( @xmath67 ) in the lens plane in which all the arcs exist . in this region
we prepare regular grids . each grid point
is denoted by @xmath68 , where integers @xmath69 and @xmath70 are restricted in @xmath71 . in the practical calculations , we adopt @xmath72 throughout the paper . given the deflection angle @xmath73 , for all grid points we can calculate the source point @xmath74 which corresponds to @xmath68 by using the lens equation .
next we consider the source with center @xmath75 and dimensionless radius @xmath21 and ellipticity @xmath76 .
we regard the grid point @xmath68 is a part of lensed images if the following condition is satisfied : @xmath77 ^ 2}{r_{\rm s}^2/(1-e_{\rm s})}+\frac{\left[y_2(i , j)-y_{\rm 2c}\right]^2}{r_{\rm s}^2(1-e_{\rm s})}\leq 1 .
\label{withinsource}\ ] ] for all the grids we check this condition and obtain the pattern of lensed images . in general , multiple images can be generated by gravitational lensing .
thus we search each image grid and recognize neighboring image grids as the same image .
the magnification of the image is then proportional to the number of grid points it contains . to analyze the lens properties ,
we calculate the magnification @xmath78 , length @xmath79 , width @xmath80 , orientation @xmath81 of the image . for the multiply lensed system , we calculate these quantities for each image . first , the magnification is easily calculated from the number of grid points @xmath82 which are recognized as the image ; @xmath83 , where @xmath84 is the dimensionless interval of grids .
this gives fairly accurate values because an image contains many grid points in our calculations ; typically @xmath85 .
next we calculate the length as follows .
first we search the center of the image c as the point at which the value of the left hand side in equation ( [ withinsource ] ) becomes smallest .
then we find the point a in the image which is farthest away from the center .
then we find the point b , also along the image , which is farthest away from the point a. we calculate the length of the image by @xmath86 . the width is simply taken such that @xmath87 .
we define the orientation of the image as the angle between the normal of the segment joining a and b and the segment joining the origin and the center of the image .
the orientation @xmath81 takes the values in the range of @xmath88 . using the above quantities
, we define the tangential and radial arcs as @xmath89 where @xmath90 is the threshold axis ratio .
the validity of this selection criterion is discussed in [ sec : select ] .
ll parameters & values + number of grids & @xmath91 + lens mass & @xmath92 + lens redshift & 0.3 + source redshift & 1.2 + source diameter & @xmath93 + a canonical parameter set used in this paper is summarized in table [ table : arcmap_cond ] . with this number of grids ,
a single mapped image contains typically @xmath94 grids , so it is sufficient to resolve the shape of images .
we examine the shapes of the images of @xmath95 sources and then estimate the cross sections .
we also estimate errorbars of numerically calculated cross sections simply by statistical errors .
= 13.0 cm we numerically calculate the cross sections for tangential and radial arcs using the method described in [ sec : sim ] .
first we show examples of simulated images with and without ellipticities in figure [ fig : map_mono ] .
these plots clearly show that the effects of ellipticities are indeed large and can not be neglected .
therefore , in this section , we numerically study the effects of source and lens ellipticities as well as the finite source size effect . in figure
[ fig : zs ] , to check simulated cross sections , we plot the lensing cross sections for tangential and radial arcs against the source redshift @xmath96 , in the case of no ellipticities .
the density profile is the generalized nfw profile ( eq . [ [ generalizednfw ] ] ) with @xmath97 . in what follows ,
we consider three criteria of axis ratio for the detection of arcs ; @xmath98 , @xmath99 , and @xmath100 .
theoretical predictions , which are calculated from equation ( [ tth ] ) for tangential arcs and equation ( [ cs_rad ] ) for radial arcs , are also plotted in thin lines .
these plots make sure that the cross sections for tangential and radial arcs show good agreement between theoretical predictions and simulations . in figure
[ fig : rs ] , we show the effect of the finite source size .
we also plot the theoretical prediction of cross sections for radial arcs ( eq . [ [ cs_rad ] ] ) and find that our analytic calculations show fairly good agreement with simulations .
these plots clearly indicate that the source size severely affects the cross section for radial arcs . the longer arcs ( larger @xmath101 )
are more strongly affected by the source size than shorter arcs .
therefore , we need to know the source size and properly take account of the finite source size effect in calculating the number of radial arcs . on the other hand , the number of tangential arcs does not change so much within the range we examined .
the typical angular diameter of sources ( faint galaxies , @xmath102 ) is @xmath103 ( e.g. , * ? ? ?
* ) , and this is small compared with the cross sectional region for tangential arcs .
this implies that we can estimate the lower limit of inner slope of density profile , @xmath64 , from observations of radial arcs if we use the maximum cross section for radial arcs , because the density profiles with larger @xmath64 usually produce the larger number of radial arcs .
we also found that the size of cutoff of cross sections adopted in @xcite is not so adequate , which corresponds to the width of the cross sectional region and is described by arrows in figure [ fig : rs ] ; the cross sections begin to decrease when the source diameter ( @xmath104 ) is two times the cutoff width . from this figure ,
the cross section for radial arcs have a steep cutoff against the source size while the cross section for tangential arcs seems to decrease mildly as the source becomes larger .
we conclude that the radial arcs are more sensitive to the source size effect also from this point .
figure [ fig : sel ] shows the effect of the source ellipticity . as pointed out by @xcite , the source ellipticity increases the number of both tangential and radial arcs , but decreases the number ratio of radial to tangential arcs . although @xcite analytically derived the cross section including the source ellipticity assuming the infinitesimal source size and showed the above dependence , we confirmed this by the numerical simulations where sources have the finite size . @xcite only considered @xmath98 case , but we also consider @xmath105 and @xmath100 and find that arcs with the smaller threshold values are more sensitively affected by the source ellipticities . in conclusion , except for the @xmath18 case , the number ratio of radial to tangential arcs seems not to be severely affected by source ellipticity ; the difference between @xmath97 and @xmath106 is much larger than the change due to the source ellipticity .
the effect of the lens ellipticity is shown in figure [ fig : el ] .
as previously indicated by @xcite , lens ellipticity drastically changes the total number of arcs ; more than two orders of magnitude from @xmath107 to @xmath108 .
the lens ellipticity amplifies the numbers of both tangential and radial arcs .
in particular , @xmath97 profile and @xmath106 profile can produce almost the same number of tangential arcs when @xmath108 .
nevertheless , the number ratio of radial to tangential arcs is not affected much by the lens ellipticity . even in the case of @xmath109 , however , the number ratio of radial to tangential arcs changes less than one order of magnitude compared with the spherical lens and is still about one order of magnitude different between @xmath97 and @xmath106 .
therefore , we conclude that the number ratio remains a good indicator for halo density profile which does not so affected by the uncertainty of the lens ellipticity as to hide the difference caused by the density profile . in the above examples , we consider the massive halo with mass @xmath110 only . while this is quite typical mass of lensing clusters , it is important to check whether our qualitative results described above remain valid for less massive halos .
therefore in this subsection we see the effects of source and lens ellipticities for halos with different mass .
the results are shown in figure [ fig : mass ] .
we plot the cross sections for three halos with different mass ; @xmath110 , @xmath111 , and @xmath112 . in this plot
we fix the threshold axis ratio as @xmath105 and examine only extreme cases , @xmath113 and @xmath108 , as well as the case of no ellipticities .
this figure clearly shows that our qualitative results are not affected at all by changing the mass of the lens object ; the values of cross sections show quite similar behavior for halos with different mass
. therefore our results are generic and applicable to a wide range of the lens mass .
moreover , it seems that the number ratio of radial to tangential arcs is highly insensitive to the mass of halos , especially for the cases including ellipticities .
this means that the uncertainty arising from the lens mass in determining the density profile is quite small .
in equations ( [ detecttan ] ) and ( [ detectrad ] ) , we discriminated tangential and radial arcs by the orientation @xmath81 .
we regard arcs with @xmath114 as tangential arcs and with @xmath115 as radial arcs .
there is , however , no obvious reason to adopt this selection criterion .
it may be possible that tangential and radial arcs are misinterpreted with each other .
therefore , we check the validity of our selection criterion in the extreme cases we examined : @xmath109 and @xmath113 .
figure [ fig : incpos ] shows the distances of arcs from the center of the lens against the orientations of arcs defined in [ sec : rec ] .
this figure suggests that our selection criterion distinguishes tangential and radial arcs quite well for @xmath113 cases . in the @xmath108 cases
, however , there seems to be mixing between both arcs
. in particular , the case with @xmath108 and @xmath106 produces `` radial '' arcs which have @xmath116 but seem to be connected with tangential arcs in the @xmath81-@xmath117 plane .
the fraction of such arcs , however , is small for longer arcs ( @xmath118 ) .
therefore , it is safe to use our selection criterion of tangential ( eq . [ [ detecttan ] ] ) and radial arcs ( eq . [ [ detectrad ] ] ) , especially for @xmath119 .
we have studied several systematic effects which may change the number of tangential and radial arcs .
more specifically , we have examined the effect of the finite source size , source ellipticity , and lens ellipticity .
first we derived an expression of the cross section for radial arcs ( eq . [ [ cs_rad ] ] ) taking account of the finite source size , assuming spherical lenses .
we found that the cross section is enhanced by the moderate amount of the source size , while larger source size ( source diameter @xmath120 ) rapidly decreases the probability of producing the radial arcs .
we also derived the cross section for tangential and radial arcs numerically and confirmed that radial arcs are more sensitive to the finite source size effect than tangential arcs .
it has been also found that our analytic prediction of the cross section for radial arcs shows fairly good agreement with the numerical simulations .
therefore , one can accurately predict the number of radial arcs with an arbitrary source size .
in addition to the finite source size effect , we studied effects of lens and source ellipticities .
we found that both ellipticities can change the number of both tangential and radial arcs significantly ; the enhancement can become as large as two orders of magnitude .
the number ratio of radial to tangential arcs is , however , not so affected by ellipticities . even in the extremely elliptical cases ,
the number ratio changes less than one order of magnitude on average , and still remains the difference between @xmath97 and @xmath54 profiles .
the exception is the arcs with @xmath18 .
such short arcs are more severely affected by the source ellipticity ; the number ratio becomes much smaller compared with no ellipticity case .
although the these results are obtained for one specific halo with mass @xmath121 , we also examined the effects of ellipticities for several halos with different mass and confirmed that our qualitative results remain valid .
as a conclusion , number ratio of radial and tangential arcs with high threshold axis ratio @xmath122 is quite robust for the uncertainties of ellipticities . in summary , the number ratio of radial to tangential arcs which have the large axis ratio ( @xmath1 ) becomes good statistics which can probe the density profile of the lens object , if the source size effect is correctly taken into account .
if we use the maximum cross section for radial arcs for the theoretical prediction , observations of tangential and radial arcs mainly give the lower limit of the inner slope @xmath64 because the effects of ellipticities as well as the source size decrease the fraction of radial arcs . on the other hand ,
the total number of tangential arcs , which is also highly sensitive to the density profile of the lens object @xcite , should be carefully calibrated with the ellipticities of the lens and source . since the detailed non - spherical modeling of dark halos is now being attempted @xcite
, it would become an important task to take account of the effect of the lens ellipticity systematically . of course , there are other effects of simplifications we should consider .
first , we neglect the irregularity of mass distribution in clusters .
this effect , however , seems to be small enough @xcite .
secondly , we neglect the effect of central cd galaxies .
radial arcs occur very close to cluster center , thus it is possible that radial arcs are hidden by the light of cd galaxies .
multicolor imaging , however , may help the identification of radial arcs because arcs have rather different color compared with those of member galaxies @xcite .
more importantly , the gravitational potential of cd galaxies may have serious effects on arcs , especially for radial arcs @xcite .
discriminating the dark halo profile from the cd galaxy profile in the central region is in general difficult because the gravitational lensing can probe only the sum of these profiles .
therefore , it may be needed to use the sample of clusters without cd galaxies to probe the density profile of dark halos .
bartelmann , m. 1995 , , 299 , 11 bartelmann , m. 1996 , , 313 , 697 bartelmann , m. , steinmetz , m. , & weiss , a. 1995 , , 297 , 1 bartelmann , m. , & weiss , a. 1994 , , 287 , 1 bullock , j. s. , kolatt , t. s. , sigad , y. , somerville , r. s. , kravtsov , a. v. , klypin , a. a. , primack , j. r. , & dekel , a. 2001 , , 321 , 559 flores , r. a. , maller , a. h. , & primack , j. r. 2000 , , 535 , 555 golse , g. , & kneib , j. p. 2002
, , submitted ( astro - ph/0112138 ) hattori , m. , watanabe , k. , & yamashita , k. 1997 , , 319 , 764 jing , y. p. , & suto , y. 2000 , , 529 , l69 jing , y. p. , & suto , y. 2002 , , in press ( astro - ph/0202064 ) keeton , c. r. 2001 , , 562 , 160 keeton , c. r. , & madau , p. 2001
, , 549 , l25 lilly , s. j. , cowie , l. l. , & gardner , j. p. 1991
, , 369 , 79 meneghetti , m. , bolzonella , m. , bartelmann , m. , moscardini , l. , & tormen , g. 2000 , , 314 , 338 meneghetti , m. , yoshida , n. , bartelmann , m. , moscardini , l. , springel , v. , tormen , g. , & white s. d. m. 2001 , , 325 , 435 meneghetti , m. , bartelmann , m. , & moscardini , l. 2002 , , submitted ( astro - ph/0201501 ) miralda - escud , j. 1993a , , 403 , 497 miralda - escud , j. 1993b , , 403 , 509 miralda - escud , j. 1995 , , 438 , 514 molikawa , k. , & hattori , m. 2001 , , 559 , 544 molikawa , k. , hattori , m. , kneib , j. p. , & yamashita , k. 1999 , , 351 , 413 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1996 , , 462 , 563 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , , 490 , 493 oguri , m. , taruya , a. , & suto , y. 2001 , , 559 , 572 schneider , p. , ehlers , j. , & falco , e. e. 1992 , gravitational lenses ( new york : springer ) tyson , j. a. , kochanski , g. p. , & dellantonio i. p. 1998
, , 498 , l107 williams , l. l. r. , navarro , j. f. , & bartelmann , m. 1999 , , 527 , 535 wu , x .- p . , & hammer , f. 1993 , , 262 , 187 | it has been recognized that the arc statistics of gravitational lensing is a useful probe of the density profile of clusters of galaxies .
we examine several systematic effects which are important in predicting the number of arcs , with particular attention to the difference between tangential and radial arcs .
first we derive an analytic expression of the cross section for radial arcs taking account of the source size and find that the moderate source size enhances the cross section for radial arcs while larger source size ( @xmath0 in our example ) suppresses the number of radial arcs . on the other hand , tangential arcs are much less sensitive to the source size .
next we numerically calculate the cross section for arcs considering the lens and source ellipticities .
we find that the numbers of both tangential and radial arcs are highly enhanced by both ellipticities , by one or two orders of magnitude .
the number ratio of radial to tangential arcs is , however , not so affected if the threshold axis ratio of arcs is large ( @xmath1 ) .
the number ratio therefore still remains good statistics which probe the density profile of the lens objects , if the source size effect is correctly taken into account . |
neutrinos are very intriguing objects in particle physics .
they interact very weakly and their masses are much smaller than those of the other fundamental fermions ( charged leptons and quarks ) . in the standard model ( sm ) , neutrinos are massless and have only weak interactions .
however , the observation of neutrino oscillations by many experiments implies that neutrinos are massive and mixed .
therefore , the sm must be extended to account for neutrino masses . in many extensions of the sm
, neutrinos also acquire electromagnetic properties through quantum loop effects ( see refs .
@xcite for detail ) .
hence , the theoretical and experimental study of neutrino electromagnetic interactions is a promising tool to search for the fundamental theory beyond the sm ( bsm ) .
the most theoretically studied electromagnetic properties of neutrinos are the dipole magnetic and electric moments .
the neutrino magnetic moments expected in the minimally extended sm are very small and proportional to the neutrino masses : @xmath0 with @xmath1 being the electron bohr magneton ( in units @xmath2 ) , and @xmath3 is the electron mass .
any larger value of @xmath4 can arise only from the bsm physics @xcite .
current direct experimental searches for a magnetic moment of the electron antineutrinos from reactors have lowered the upper limit on its value down to @xmath5 @xcite .
these ultra low background experiments use germanium crystal detectors exposed to the neutrino flux from a reactor and search for scattering events by measuring the energy deposited by the neutrino scattering in the detector .
their sensitivity to @xmath4 crucially depends on lowering the threshold for the energy transfer @xmath6 .
this is because the electromagnetic contribution to the inclusive differential cross section for the neutrino scattering on a free electron ( fe ) at small energy transfer ( @xmath7 ) behaves as @xmath8 @xcite , while that induced by weak interaction , @xmath9 , is practically constant in @xmath6 @xcite .
the current experiments using germanium detectors have reached threshold values of @xmath6 as low as few kev , where one can expect modifications of the fe formulas due to the binding of electrons in the germanium atoms . our theoretical analysis ( see ref .
@xcite and references therein ) , involving the wkb and thomas - fermi models , has shown that the so - called stepping approximation , introduced in @xcite from an interpretation of numerical data , works with a very good accuracy .
the stepping approximation treats the process as scattering on independent electrons occupying atomic orbitals and suggests that the cross sections @xmath10 and @xmath11 follow the fe behaviors down to the ionization threshold for the orbital ; and below that energy the electron on the corresponding orbital is `` inactive '' , thus producing a sharp `` step '' in the @xmath6 dependence of the cross section .
to the best of our knowledge , the issue of the center - of - mass atomic motion has remained practically unadressed so far in theoretical studies devoted to the ionization channel of neutrino - atom collisions .
usually the recoil effects are neglected under the assumption that the atomic nucleus due to its large mass stays at rest during the ionization process . at the same time
, it appears that in the case of reactor antineutrinos the recoil energy can be comparable to atomic binding energies .
thus , in this contribution , we analyze the role that can play the center - of - mass atomic motion in the discussed processes .
in particular , we inspect how it can affect the validity of the stepping approximation .
we specify the incident neutrino energy and momentum by @xmath12 and @xmath13 , respectively .
the atomic target is supposed to be initially at rest , unpolarized and in its ground state @xmath14 with the corresponding energy @xmath15 .
we treat the initial and final electronic systems nonrelativistically under conditions @xmath16 and @xmath17 , where @xmath18 is the atomic number and @xmath19 is the fine - structure constant .
the incident and final neutrino states are described by the dirac spinors assuming @xmath20 .
first we consider the experimentally measured single - differential inclusive cross section @xmath21 of atomic ionization by neutrino impact , without accounting for atomic recoil in this process .
the standard electroweak contribution to the cross section is given by ( see e.g. in @xcite ) @xmath22 where @xmath23 is the fermi constant , @xmath24 is the weinberg angle , and @xmath25 is the dynamical structure factor that can be presented as follows ( @xmath26 ) : @xmath27 with @xmath28 being the fourier transform of the electron density operator and the @xmath29 sum running over all the atomic states @xmath30 with energies @xmath31 of the electron system .
the @xmath4 contribution to the cross section can be similarly expressed in terms of the same factor ( [ dsf ] ) as @xcite @xmath32 where the function @xmath25 is integrated over @xmath33 with @xmath34 , rather than a unit weight as in eq .
( [ d1sw ] ) .
the kinematical limits for @xmath33 are explicitly indicated in eqs .
( [ d1sw ] ) and ( [ d1s ] ) . at large @xmath35 ,
typical for the reactor antineutrinos ( @xmath36mev ) , the upper limit can in fact be extended to infinity , since in the discussed here nonrelativistic limit the range of momenta @xmath37 is indistinguishable from infinity on the atomic scale .
the lower limit can be shifted to @xmath38 , since the contribution from the region of @xmath39 can be expressed in terms of the photoelectric cross section @xcite and is negligibly small ( at the level of below one percent in the considered range of @xmath6 ) .
for this reason one can discuss , without loss of accuracy , the momentum - transfer integrals in eqs .
( [ d1sw ] ) and ( [ d1s ] ) running from @xmath38 to @xmath40 . in the one - electron approximation and in the semiclassical limit
, the dynamical structure factor ( [ dsf ] ) acquires the form @xcite @xmath41 where @xmath42 is an average momentum of the electron bound in the atomic orbital from which ionization takes place , @xmath43 is the ionization potential for this orbital , and @xmath44 it is straightforward to show ( see ref .
@xcite for detail ) that using the semiclassical approximation ( [ dsfwkb ] ) in eqs .
( [ d1sw ] ) and ( [ d1s ] ) one arrives at the fe expressions @xcite @xmath45 let us take into account the atomic recoil .
this means that the energy transfer @xmath6 is partly deposited in the center - of - mass atomic motion .
the energy of this motion , that is , the recoil energy , is @xmath46 , where @xmath47 is the atomic mass .
hence , the first argument of the function @xmath25 in ( [ d1sw ] ) and ( [ d1s ] ) must be replaced with @xmath48 substituting @xmath49 in eq .
( [ dsfwkb ] ) , we obtain @xmath50 where @xmath51 . from eq .
( [ dsfwkb1 ] ) it follows that the atomic recoil has no effect if @xmath52 .
indeed , in the latter case the fe results ( [ d1swfe ] ) and ( [ d1sfe ] ) for the cross sections ( [ d1sw ] ) and ( [ d1s ] ) , respectively , remain unaltered .
in contrast , when @xmath53 in eq .
( [ dsfwkb1 ] ) , the cross sections become suppressed relative to the respective fe values and even vanish at @xmath54 .
clearly , the dynamical structure factor of a real atom exhibits more complex dependencies on @xmath6 and @xmath55 as compared to its semiclassical approximation ( [ dsfwkb1 ] ) .
nevertheless , this approximation mimics the main qualitative features pertinent to the exact function @xmath56 , for example , such as `` spread and shift '' of the free - electron @xmath57 peak @xcite .
thus , the criterion for the atomic - recoil effects to come into play can be formulated for a real atom as @xmath58 or , accordingly , @xmath59 . at such energy - transfer values
the cross sections ( [ d1sw ] ) and ( [ d1s ] ) become suppressed and vanish when @xmath60 .
general numerical estimates can be obtained within the thomas - fermi model of manyelectron atoms .
the scale for the average electron momentum @xmath42 in this model is determined by @xmath61 .
it gives the following estimate for the energy range where the atomic - reoil effects are important : @xmath62 where @xmath63ev is the hartree energy .
for germanium ( @xmath64 ) we obtain @xmath65ev .
this energy scale is insignificant for the experiments searching for magnetic moments of reactor antineutrinos @xcite . as mentioned in the introduction
, these experiments have reached threshold values of @xmath6 as low as few kev .
such values are already below the ionization threshold for @xmath66 electrons in germanium ( @xmath67kev ) .
these atomic electrons are most strongly bound , and the thomas - fermi average @xmath42 value substantially underestimates their average momentum .
the latter can be evaluated using the virial theorem as @xmath68 .
the corresponding energy scale is @xmath69ev .
it appears to be about an order of magnitude larger than that given by the thomas - fermi model , but it is still insignificant for the discussed experiments .
in summary , we have inspected how the center - of - mass atomic motion can affect ionization of an atom by neutrino impact . employing the semiclassical approximation , we have derived the criterion that defines the energy - transfer range where the atomic - recoil effects are important .
our numerical estimates have shown that these effects play no appreciable role in current experiments searching for neutrino magnetic moments of electron antineutrinos from reactors with ge detectors .
one of the authors ( a.s . ) is thankful to arcadi santamara , salvador mart and juan fuster for the kind invitation to participate at the ichep 2014 conference and to all of the organizers for their hospitality in valencia .
this work was supported by the russian foundation for basic research ( grants no .
14 - 01 - 00420-a and no .
14 - 22 - 03043-ofi - m ) . | we consider theoretically ionization of an atom by neutrino impact taking into account electromagnetic interactions predicted for massive neutrinos by theories beyond the standard model .
the effects of atomic recoil in this process are estimated using the one - electron and semiclassical approximations and are found to be unimportant unless the energy transfer is very close to the ionization threshold . we show that the energy scale where these effects become important is insignificant for current experiments searching for magnetic moments of reactor antineutrinos . neutrino - impact atomic ionization , atomic recoil , neutrino magnetic moments , bsm physics |
the interaction of charged particles with matter has been a subject of extensive research over many decades .
these studies provide information for many technological applications such as nuclear safety , applied material science , medical physics and fusion and fission applications @xcite . among the measurable quantities associated to the interaction between ions and solids ,
the stopping power @xmath4 @xcite has received much attention ; it provides information regarding the energy transfer between the incoming projectile and the solid target . when a fast ion moves through a material , it loses most of its kinetic energy due to the excitations of the target electrons along its trajectory in what constitutes a fundamentally non - adiabatic process .
this energy - loss phenomenon plays an important role in many experimental studies involving radiation in solids , surfaces , and nanostructures @xcite .
various models and theories have been proposed to calculate stopping cross sections due to electrons . employing the first born approximation ,
bethe @xcite has introduced the first calculations of inelastic and ionization cross section .
the bloch correction @xcite provides a convenient link between the bohr and the bethe scheme .
fermi and teller @xcite using electron gas models had reported electronic stopping for various targets .
the bethe formula for stopping has been studied in details by lindhard and winther @xcite on the basis of the generalized linear - response theory . in particular , the condensed matter community has introduced sophisticated numerical computer simulation techniques for this fundamentally non - adiabatic problem as spearheaded by echenique _
et al . _
@xcite aimed to overcome limitations of historical approaches @xcite
. a unified _ ab initio _ theoretical approach suitable for different projectiles and energies is in its developing stages @xcite .
a review on the topic can be found in ref .
@xcite and references therein . using a kohn - sham ( ks ) scheme of time - dependent density functional theory ( tddft ) where the ks wave functions are expanded in a basis set of spherical harmonics , quijada _
_ @xcite have studied the energy loss of protons and anti - protons moving inside metallic spherical @xmath5 ( jellium ) clusters and obtained good results for the projectile - target energy transfer over a restricted energy range .
recently uddin _
_ @xcite and haque _ et al .
_ @xcite have calculated stopping cross sections for various media using atomic density functions from dirac - hartree - fock - slater wave functions in the lindhard - schraff theory @xcite with fitted parameters and obtained close agreement with the experimental and srim data .
srim @xcite provides both a fitted model for electronic stopping as well as a large set of experimental points . in studying the role of ion - solid interactions of @xmath6 in @xmath5 ,
@xcite have shown that the electronic excitations affect the interatomic forces relative to the adiabatic outcome .
recently , schleife _
@xcite have calculated the electronic stopping power ( @xmath7 ) of @xmath8 and @xmath9 projectiles including tddft non - adiabatic electron dynamics and found that off - channeling trajectories _ along with
_ the inclusion of semicore electrons enhance @xmath7 , resulting in much better agreement with the srim experimental and modeled data @xcite in a wide range of energies . in this case
we concentrate in a metal with a richer electronic band structure around the fermi energy , such as @xmath10 .
the recent measurements by cantero _
_ @xcite and by markin _
_ @xcite of slow ( @xmath11 ) @xmath6 in @xmath10 give a glimpse of the interesting extreme low velocity limit .
although disagreeing with each other in absolute scale by @xmath12 ( fig . [
fig : stopping_power ] ) , both reveal the stopping due to conduction band electronic excitations at lower velocity , evidenced as a change in slope near @xmath13 or @xmath14 respectively .
the change of slope was deduced qualitatively to be caused by the participation of @xmath15-electrons @xcite . in this paper
we will address the problem of theoretical calculation of @xmath7 of protons in crystalline @xmath10 for a wide range of available experimental velocities ( @xmath16 ) .
we perform our calculations by directly simulating the process of a proton traversing a crystal of @xmath10 atoms , producing individual and collective electronic excitations within the tddft framework @xcite including ehrenfest molecular dynamics ( emd ) @xcite .
this method is used to calculate most microscopic quantities along the process ( forces , electronic density , charges , etc ) ; in particular , we concentrate here in the calculation of @xmath7 . a quantitative explanation and interpretation of our results
are furnished along with a detailed experimental comparison .
during the course of the simulation , we monitored the energy transferred to the electrons of the target due to a constant velocity moving proton . for simplicity , and since the eletronic stopping is a velocity - resolved quantity the proton is constrained to move at constant velocity , hence the total energy of the system is not conserved .
the excess in total energy is instead used as a measure of the stopping power as a function of the proton velocity . as the proton moves , the time - dependent kohn - sham ( tdks ) equation @xcite describes the evolution of the electronic density and energy of the system , due to the dynamics of effective single - particle states under the external potential generated by the proton and the crystal of @xmath10 nuclei .
these states are evolved in time with a self - consistent hamiltonian that is a functional of the density : @xmath17(\mathbf{r } , t)\right\}\psi_i(\mathbf{r } , t ) \label{eq : tdks1}\ ] ] the ks effective potential @xmath18(\mathbf{r } , t)$ ] is given by @xmath19 = v_\text{ext}[\{\mathbf{r}_j(t)\}_j ] + v_\text{h}[n ] + v_\text{xc}[n ] \label{eq : tdks3 } \end{aligned}\ ] ] where the external potential is @xmath20(\mathbf{r } , t)$ ] due to ionic core potential ( with ions at positions @xmath21 ) , @xmath22(\mathbf{r } , t)$ ] is the hartree potential comprising the classical electrostatic interactions between electrons and @xmath23(\mathbf{r } , t)$ ] denotes the exchange - correlation ( xc ) potential .
the spatial and time coordinates are represented by @xmath24 and @xmath25 respectively . at time
@xmath25 the instantaneous density is given by the sum of individual electron probabilities @xmath26 .
the xc potential used in this study is due to perdew - burke - ernzerhof ( pbe ) @xcite , and we used norm - conserving troullier - martins pseudopotential to represent @xmath27 , with @xmath28 explicit electrons per @xmath10 atom ( not necessarily all 17 electrons participate in the process as we will discuss later ) .
each simulation of the ion - solid collision consists of a well - defined trajectory of the projectile in the fcc metallic bulk sample with experimental density .
the calculations were done using the code qbox @xcite with time - dependent modifications @xcite@xcite .
the ks orbitals are expanded in a supercell plane - wave basis .
the advantages of using the plane - wave approach is that it systematically deals with basis - size effects , which was a drawback for earlier approaches @xcite .
periodic boundary conditions along with ewald summation @xcite are used throughout this study .
the supercell size was chosen so as to reduce the specious size effects while maintaining controllable computational demands . since the larger size effects are negligible this calculation used @xmath29 conventional cells containing @xmath30 host @xmath10 atoms and @xmath6 .
to integrate the brillouin zone a single @xmath31-point ( @xmath32 ) was used , except for test cases .
the screening length of @xmath10 is close to the interatomic spacing , which reduces the range of coulomb interactions and makes it controllable in a periodic representation .
finite size effects are studied between 108 and 256 atoms in a supercell of @xmath29 and @xmath33 respectively in this simulation , the errors remain within 3% in conformity with the earlier observation @xcite .
the plane - wave basis set is sampled accurately with a @xmath34 energy cutoff .
we also tested for k - point convergence in a @xmath29-monkhorst - pack grid ( for the cubic 108-supercell ) , which would be equivalent to a 2916 ( @xmath35 simulation cell of an hypothetical periodic system , including replicas of the proton ) , for selected velocities with negligible differences within 0.08% .
the projectile @xmath6 is initially placed in the crystal and a time-_independent _ dft calculation was completed to obtain the converged ground state results that are required as the initial condition for subsequent evolution with the moving projectile .
we then perform tddft calculations on the electronic system with the moving proton in the channeling and off - channeling geometries . following the method introduced by pruneda _
_ @xcite the projectile is put in motion with a constant velocity in a straight along a @xmath36 channeling trajectory ( also called hyper - channeling ) which minimizes the collision of the projectile with the host atoms @xcite . in the off - channeling case
the projectile takes random trajectory directions through the host crystal yielding occasionally stronger interaction between the projectile and the tightly bound electrons of the host atom .
the use of off - channeling trajectories was introduced in ref .
@xcite . the fourth - order runge - kutta scheme ( rk4 ) @xcite
is used to propagate the electronic orbitals in time , following the scheme of ref .
@xcite the tdks equation ( see eq . [ eq : tdks1 ] ) with a time step of , at most , @xmath37 ( which is within the numerical stability limit time - integration scheme for this basis set ) .
high velocity points were simulated with smaller time steps .
the wavefunctions were then propagated for up to tens of femtoseconds .
the total electronic energy ( @xmath38 ) of the system changes as a function of the projectile position ( @xmath39 ) since the projectile ( forced to maintain its velocity ) deposits energy into the electronic system as it moves through the host atoms .
the increase of @xmath38 as a function of projectile displacement @xmath39 enables us to extract the electronic stopping power as a time - averaged quantity for each velocity , @xmath40 @xmath7 has the dimension of a force ( e.g. @xmath41 ) and it has the interpretation of a drag force acting on the projectile .
( color online ) . total energy increase as a function of position for a proton in crystalline @xmath10 .
the lattice atoms are kept at their equilibrium positions while the projectile passes in a @xmath36 channeling trajectory at velocity @xmath42 . ]
our calculations of @xmath43 system in the range of velocities between @xmath44 and @xmath45 are presented in details in this section .
[ fig : energy_distance ] shows the total electronic energy of the @xmath46 system as a function of position for various projectile velocities for the hyper - channeling case . at a lower velocities regime ,
the energy transfer is smaller , approaching the adiabatic behavior , while at higher velocities ( aside oscillations of the total energy with the position of the projectile ) the total energy of the system increases linearly with time .
after the projectile travels some short distance in the crystals ( @xmath47 ) the increase in total energy of the system stabilizes to a steady rate . at that steady state , the @xmath7 is then extracted from the average slope of the total energy vs. projectile displacement ; it represents the rate of energy gained by the target and loss by the projectile .
( color online ) .
the increase of @xmath38 as a function of proton position at @xmath48 ( green dashed line ) along @xmath36 channeling trajectory and the adiabatic energy ( magenta dot - dashed line ) .
the adiabatic curve would correspond to a proton moving infinitely slowly , where there is no transfer of energy , just oscillations of the total energy with an overall zero slope .
the oscillations in the curves reflect the periodicity of the @xmath10 lattice .
the red solid line shows the energy difference ( subtraction of adiabatic energy from the @xmath48 curves ) .
the blue dashed and black dotted lines shows the slope of linear and oscillatory fits of the red solid line from @xmath49 ( after the transient ) to a given maximum position @xmath39 as a function of this maximum position respectively . for visualization purposes
the black line has been shifted vertically .
a linear fit , @xmath50 ( blue line ) yields a slope of @xmath51 with an error of @xmath52 .
we then proceed for a linear fit in addition to an oscillatory function @xmath53 ( black dotted line ) to capture any remnant oscillation .
this oscillatory fit generates a slope of @xmath54 with an error of @xmath55 , that is a minimal fitting error is obtained in the _ channeling _ trajectory . ] in the low velocity cases extracting a slope becomes more challenging , first because a longer time simulation is required to sample the crystalline structure and second because the natural oscillations associated with the crystal periodicity becomes relatively larger . for the channeling cases , we subtract the adiabatic energy from time - dependent energy and we perform an oscillatory fit to the resulting curve . fig .
[ fig : fit_graph ] explains how the slopes of fig .
[ fig : energy_distance ] are evaluated in these cases . for higher velocities ,
a linear fit alone is enough to obtain reasonable values with small relative errors .
( color online ) the total energy increase as a function of proton position at velocities @xmath56 ( magenta line ) , @xmath57 ( blue line ) and @xmath58 ( red line ) along the off - channeling direction [ @xmath59 trajectory with the corresponding linear fits ( black solid lines ) . ] for the off - channeling trajectory , the procedure for computing the @xmath7 is depicted in fig .
[ fig : fit_off_channel ] .
we used two directions , approximately [ @xmath60 and [ @xmath59 ( given normalized here ) .
the first direction was chosen by visual inspection of the supercell in order to _ not _ match any simple channel but also avoid an immediate head on collision . the second direction is the normalized version of @xmath61 $ ] , where @xmath62 is the golden ratio ( @xmath63 ) , which guarantees a trajectory most incommensurate with the cell due to its mathematical properties as an irrational number .
it is important to note here an interesting geometrical fact that , for a direction incommensurate with the crystal directions , all available densities and impact parameters ( distances of closest approach to host atom ) are probed ( averaged ) eventually for a long enough trajectory .
our simulations are limited in space ( and time ) but it is clear that the trajectories explore a wide range of impact parameters and therefore densities .
the viability and the necessity of considering this geometrical averaging method was shown earlier in @xcite . in fig .
[ fig : fit_off_channel ] , the sharp peaks show when the proton is in the vicinity of a host @xmath10 atom during an off - channeling trajectory , while the smaller peaks and flatter regions indicate that the proton is not very close to any host atom . to obtain the @xmath7 we compute the slopes of the curves by a linear fit of the form @xmath50 ( black solid lines ) using our data from @xmath64 ( to eliminate the transient region ) up to a given maximum position of @xmath39 determined by minimizing reentrancy in the periodic supercell near the initial position .
the slope ( @xmath65 ) gives the electronic stopping for this off - channeling case .
( color online ) .
the @xmath7 vs. projectile velocity @xmath42 : for a channeling trajectory ( red lines ) and for off - channeling trajectory ( blue crosses ) .
the black continuous line refers to the nominal tabulated model from srim , based on its database for electronic stopping power @xcite .
the gray dots refer to the srim experimental database contained therein @xcite .
the error bars of the off - channeling trajectory are shown by the blue vertical lines , while error bars in channeling trajectories are smaller than the points .
open circles and open squares show data of cantero _ et al .
_ @xcite and markin _ et al . _
@xcite respectively . ] fig .
[ fig : stopping_power ] shows a comparison of our calculated @xmath7 with srim - based model and experimental data . in
the _ channeling _ case , the maximum of our calculated stopping is lower in value and velocity compared to the srim database and the _ off - channeling _ case , and it decreases faster after the maximum . for the off - channeling case , there is a better agreement between our @xmath7 results with the srim data in most of the range . in experiments , where trajectories are not necessarily finely controlled , the projectile does indeed explore core regions of the host atoms , and that is presumably why off - channeling simulations are a better representation for the most common experiments @xcite . at higher velocities ( @xmath66 )
further disagreement stems from combined effect of the lack of explicit deeper core electrons in the simulation and also size effects , as excitations of long wavelength plasma oscillations are artificially constrained by the simulation supercell @xcite .
it is clear that a larger cell and eventually the inclusion of more core electrons would be necessary to obtain better agreement in this high velocity region that is out the of scope of this article .
although experimental values have considerable vertical spread , our calculated stopping power is on the low side for most of the points and also below the fitted by srim model @xcite .
while this was partially explained by taking into account off - channeling trajectories near the maximum of stopping , there are other possible intrinsic limitations of the approximations to the density functional theory used . along this line , we would like to note that more sophisticated approaches , based on the dielectric and current - density response but including the exact many - body and dynamic exchange - correlation treatment , are available in the literature @xcite . this type of advanced approaches which are beyond the current scope contain explicitly additional channels of dissipation not taken into account in our adiabatic exchange and correlation functionals , which can be relatively important . given the aforementioned limitations of the orbital based method and the exchange and correlation used it is still reassuring to see agreement up to a few times the velocity of the maximum stopping and gives us confidence to make predictions in the lower velocity regime . at low velocity
we observe that the off - channeling and channeling simulated results collapse into a common curve , this effect has been seen in the simulations before @xcite .
the effect is that the computed quantity less sensitive to the precise geometry of the trajectory , as the geometric cross section increases .
we speculate that this is because the effective binary cross section increases beyond the interatomic separation making the energy loss less sensitive the precise geometry of the environment .
a logarithmic version of the findings of fig . [ fig : stopping_power ] is depicted in fig . [
fig : log_stopping_power ] , where we have observed that the resulting curve is not as particularly simple . in order to interpret the results we also calculated the linear response stopping @xmath67 @xcite based on a simple lindhard rpa dielectric function @xmath68 for different effective densities @xmath69 of the homogeneous electron gas @xcite @xmath70 ( which assumes a proton effective charge of @xmath71 ) . as shown in fig .
[ fig : log_stopping_power ] , for @xmath72 the response of the effective electron gas with one electron per @xmath10 mimics the tddft results .
while more sophisticated dielectric models can be used @xcite , we use the minimal model that can explain the simulation in the different regimes . the resulting curves in fig .
[ fig : log_stopping_power ] shows that for @xmath73 at least the @xmath74 electrons per atom ( full valence ) participates in the stopping electron gas within linear response .
we observe a @xmath7 kink around @xmath75 due to a mixture of @xmath15-band in the electronic density of states .
similarly , according to this analysis of our new results , for @xmath76 , are primarily due to @xmath77-band electrons within linear response . in the simulation
we directly show a crossover region between the two linear regimes , and we find that the friction is in direct relation to the velocity with a power law with exponent @xmath78 .
( in linear scale kinks are represented by changes of curvature , here logarithmic scale is more appropriate to discuss the physics . )
the kink we found at @xmath79 can be explained by conservation laws in the effective homogeneous electron gas and general properties of electronic density of states in crystalline @xmath10 .
the minimum energy loss with maximum momentum transfer from an electron to an ion moving with velocity @xmath42 are respectively @xmath80 and @xmath81 ( plus small corrections of order @xmath82 ) . due to pauli exclusion , only electrons in the energy range @xmath83
can participate in the stopping process .
taking into account that dft band structure predicts that the @xmath15-band edge is @xmath84 below the fermi energy ( see for example , fig .
3(a ) in ref .
@xcite ) , that electron ( band ) effective mass close to @xmath85 and that @xmath86 for the effective homogeneous electron gas of @xmath10 @xmath77-electrons @xcite , we can derive an approximate value of @xmath87 caused by the participation of @xmath15-electrons . based in this dft ground state density of states plus conservation laws ,
we obtain an estimate of @xmath88 in near agreement with our tddft prediction . in reality
, the @xmath15-band is about @xmath89 below the fermi energy as indicated by arpes @xcite , which means that both the dft - based estimate and the full tddft result should be giving an underestimation of 25% of the kink location .
the second ( negative ) kink at @xmath90 is more difficult to explain precisely as the qualitative description in terms of @xmath91 ( as in the homogeneous electron gas ) becomes more ambiguous , but it is related to the point at which the whole conduction band ( @xmath74 ` @xmath92 ' electrons ) starts participating in the process .
( color online ) the average @xmath7 vs. projectile velocity @xmath42 : for a channeling trajectory ( red circles ) and for off - channeling trajectory ( blue crosses ) .
the solid black and solid magenta lines refer to the nominal tabulated results from the srim database for electronic and ionic stopping powers respectively @xcite .
open circles and open squares show data of cantero _ et al .
_ @xcite and markin _ et al .
@xcite respectively .
dash ( green ) line and dash - dot ( brown ) line corresponds to a linear rpa calculation for a free - electron gas with @xmath93 ( @xmath85e per @xmath10 atom ) and @xmath94 ( @xmath74e per @xmath10 atom ) .
the thin solid black line , obtained by a linear fit to our simulated results , is the polynomial curve @xmath95 to fit the crossover region ( with a power of @xmath96 ) . ] at low velocity , our results show good agreement with the experiments of markin _ et al . _
@xcite but in relative disagreement with the measurements of cantero _ et al .
_ @xcite and the srim model .
the difference between experiments could be due to a simple experimental scaling issue related to the difference between measuring relative and absolute stopping power at low velocities @xcite .
below @xmath97 , the lack of experimental points precludes a direct comparison , but we find linear behavior at least down to @xmath98 .
below @xmath98 the direct real time integration becomes less efficient and the accuracy is compromised by the quality of the numerical time integrator and the number of steps necessary to complete a calculation @xcite .
probing this regime experimentally would be rather difficult , especially to disentangle it from nuclear stopping effects , but if this is confirmed it would be an unexpected new regime . in any case
, the combination of experimental and theoretical results shows that the limit @xmath3 is intricate for metals as it is for insulators @xcite where analogous band and gap threshold effects have been found .
finally , we point out that the investigation of the low velocity limit of stopping power is important for the understanding of the non - adiabatic coupling between ions and electrons @xcite and also for modeling dissipative molecular dynamics @xcite where electrons act both as a thermal bath and a friction media . in simulations of radiation events the final state is precisely controlled by dissipation in the late stages when ions move slowly but still non - adibatically @xcite .
( color online ) .
snapshots of density change produced by a @xmath6 moving in @xmath10 atoms with @xmath57 ( kinetic energy of @xmath99 ) along the @xmath36 channeling trajectory .
the brown balls represents the @xmath10 atoms and the single gray ball with light - blue iso - density contours represents host electron density affected by the presence of @xmath6 . from initial condition ( @xmath100 ) to a representative steady condition after @xmath101 ( see fig . [
fig : energy_distance ] ) .
visualization produced with visit @xcite ] a snapshot of the redistribution of the charge density as a function of time for the simulation is shown in fig .
[ fig : snapshot_copper ] . at @xmath102 ( see fig . [
fig : snapshot_copper]a ) the @xmath6 interacts with the neighboring @xmath10 atoms charge density where some electron density is acquired .
elongated wake tails are characteristic of projectiles traveling at or above the fermi velocity @xcite , in turn this affects the forces exerted on the neighboring atoms .
( color online ) .
radial force exerted on a @xmath10 atom as a function of parallel distance to projectile at different projectile velocities @xmath42 .
the force is _ negative _ radial , which means that adiabatically ( red curve ) the proton attracts the neighbour @xmath10 atom , but as the electron wake develops at higher velocities ( @xmath103 and @xmath104 the force becomes repulsive after passing . for visualization purposes the non - adiabatic curves
have been shifted vertically upwards . ]
[ fig : force_on_neighbor ] shows the radial force exerted on a neighboring @xmath10 atom closest to @xmath6 trajectory as a function of parallel distance to the projectile at different projectile velocities along the @xmath36 channeling trajectory .
the forces on the nuclei are evaluated using the time - dependent electron density , @xmath105 .
the adiabatic force is recovered for @xmath106 with no oscillations as expected .
the maximum value for the force is obtained at the closest distance between the @xmath6 and a neighbor @xmath10 atom . as the proton moves further from the @xmath10 atom , the force decreases and eventually reduces to zero . as the velocity increases the position of the maximum value of the force first shifts and later results in persistent oscillations .
the existence of plasma oscillations is detected in our simulations by persistent charge motion above a certain threshold velocity of @xmath107 .
these plasma oscillations affect the components forces over individual @xmath10 atoms near the trajectory of the passing hydrogen ( fig .
[ fig : force_on_neighbor ] ) . these forces persist ( and oscillate ) even after the proton has passed .
in this paper we have reported the @xmath7 of protons in copper in a very wide range of velocities .
tddft - based electron dynamics is capable of capturing most of the physics in the different ranges , starting from non - linear screening effects , electron - hole excitations and production of plasmons .
we disentangled channeling and off - channeling effects and observe a collapse of the two curves at low velocities ; and identified four regimes i ) the linear @xmath77-only ( @xmath108 ) , ii ) linear @xmath109 ( @xmath110 ) , iii ) crossover with @xmath1-power law ( @xmath111 ) and iv ) plasma - like ( @xmath112 ) .
this further illustrates that the electronic stopping in general does not possess a simple behavior in the limit @xmath3 , and that band and bound effects dominate this behavior .
we would like to thank professor c. a. weatherford , dr .
e. r. schwegler and dr .
k. j. reed for their keen interests , constant encouragements and useful suggestions throughout the progress of this work .
this work was performed under the auspices of the u.s .
department of energy by lawrence livermore national laboratory under contract de - ac52 - 07na27344 .
computing support for this work came from the lawrence livermore national laboratory institutional computing grand challenge program .
this is a collaborative project between the department of physics , florida a&m university and lawrence livermore national laboratory .
e. e. quashie and b. c. saha thankfully acknowledge the support by national nuclear security administration ( nnsa ) , award number(s ) de - na0002630 .
63ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1007/s10891 - 013 - 0976-y [ * * , ( ) ] link:\doibase 10.1103/physrevlett.91.125004 [ * * ( ) , 10.1103/physrevlett.91.125004 ] link:\doibase 10.1142/s1793626809000235 [ * * , ( ) ] in link:\doibase 10.1007/1 - 4020 - 3286 - 2_50 [ _ _ ] ( , ) pp .
link:\doibase 10.1103/physrevb.16.115 [ * * , ( ) ] link:\doibase 10.1016/j.nimb.2006.03.023 [ * * , ( ) ] link:\doibase 10.1103/physreva.75.019905 [ * * ( ) , 10.1103/physreva.75.019905 ] link:\doibase 10.1103/physrevb.78.195122 [ * * ( ) , 10.1103/physrevb.78.195122 ] in link:\doibase 10.1007/978 - 3 - 642 - 46025 - 8_10 [ _ _ ] ( , ) pp .
link:\doibase 10.1063/1.2994975 [ * * , ( ) ] link:\doibase 10.1007/978 - 3 - 319 - 05564 - 0 [ _ _ ] ( , ) in link:\doibase 10.1017/cbo9780511565007.006 [ _ _ ] ( , ) pp . @noop * * ( ) link:\doibase 10.1002/andp.19334080303 [ * * , ( ) ] link:\doibase 10.1103/physrev.72.399 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1002/qua.560360854 [ * * , ( ) ] link:\doibase 10.1002/andp.19303970303 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.235501 [ * * ( ) , 10.1103/physrevlett.99.235501 ] link:\doibase 10.1103/physrevb.91.014306 [ * * ( ) , 10.1103/physrevb.91.014306 ] link:\doibase 10.1103/physrevb.91.125203 [ * * ( ) , 10.1103/physrevb.91.125203 ] link:\doibase 10.1088/0034 - 4885/73/11/116501 [ * * , ( ) ] link:\doibase 10.1103/physreva.75.042902 [ * * ( ) , 10.1103/physreva.75.042902 ] link:\doibase 10.1140/epjd / e2013 - 40164 - 8 [ * * ( ) , 10.1140/epjd / e2013 - 40164 - 8 ] link:\doibase 10.1080/00268976.2015.1074743 [ ( ) ] @noop * * , ( ) link:\doibase 10.1016/j.nimb.2010.02.091 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.069901 [ * * ( ) , 10.1103/physrevlett.109.069901 ] link:\doibase 10.1103/physreva.80.032904 [ * * ( ) , 10.1103/physreva.80.032904 ] link:\doibase 10.1103/physrevb.80.205105 [ * * ( ) , 10.1103/physrevb.80.205105 ] link:\doibase 10.1103/physreva.87.062903 [ * * ( ) , 10.1103/physreva.87.062903 ] link:\doibase 10.1063/1.4758792 [ * * , ( ) ] link:\doibase 10.1109/mcse.2014.55 [ * * , ( ) ] in link:\doibase 10.1007/bfb0016643 [ _ _ ] ( , ) pp .
link:\doibase 10.1016/s0370 - 1573(00)00043 - 0 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/19/43/436209 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.096403 [ * * ( ) , 10.1103/physrevlett.101.096403 ] link:\doibase 10.1021/ct800518j [ * * , ( ) ] link:\doibase 10.1103/physrevlett.52.997 [ * * , ( ) ] link:\doibase 10.1103/physrevb.45.13244 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.77.3865 [ * * , ( ) ] link:\doibase 10.1147/rd.521.0137 [ * * , ( ) ] in link:\doibase 10.1109/ipdps.2016.46 [ _ _ ] ( , ) \doibase 10.1002/1096 - 987x(200009)21:12<1075::aid - jcc4>3.0.co;2-l [ * * , ( ) ] in link:\doibase 10.1533/9780857099426.77 [ _ _ ] ( , ) pp .
link:\doibase 10.1103/physrevb.71.094113 [ * * ( ) , 10.1103/physrevb.71.094113 ] \doibase
http://www.srim.org/srim/srimpics/\stop01/stop0129.gif [ http://www.srim.org/srim/srimpics/ + stop01/stop0129.gif ]
link:\doibase 10.1103/physreva.47.3062 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.205103 [ * * ( ) , 10.1103/physrevb.76.205103 ] https://books.google.com/books?id=8kk_tgaacaaj[__ ] , mathematisk - fysiske meddelelser ( , ) in link:\doibase 10.1017/cbo9780511619915.006 [ _ _ ] ( , ) pp .
link:\doibase 10.1103/physreve.54.4134 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.075133 [ * * ( ) , 10.1103/physrevb.77.075133 ] @noop _ _
( , , ) link:\doibase 10.1103/physrevb.19.4952 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/19/27/275211
[ * * , ( ) ] link:\doibase 10.1103/physrevlett.116.043201 [ * * ( ) , 10.1103/physrevlett.116.043201 ] link:\doibase 10.1103/physrevb.92.144309 [ * * ( ) , 10.1103/physrevb.92.144309 ] link:\doibase 10.1103/physreva.40.2287 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/19/1/016207 [ * * , ( ) ] link:\doibase 10.1088/0953 - 8984/26/8/085401 [ * * , ( ) ] in link:\doibase 10.1201/b12985 - 21 [ _ _ ] ( , ) link:\doibase 10.1016/0029 - 554x(80)91020 - 4 [ * * , ( ) ] | we present an ab initio study of the electronic stopping power of protons in copper over a wide range of proton velocities @xmath0 where we take into account non - linear effects .
time - dependent density functional theory coupled with molecular dynamics is used to study electronic excitations produced by energetic protons .
a plane - wave pseudopotential scheme is employed to solve the time - dependent kohn - sham equations for a moving ion in a periodic crystal .
the electronic excitations and the band structure determine the stopping power of the material and alter the interatomic forces for both channeling and off - channeling trajectories .
our off - channeling results are in quantitative agreement with experiments , and at low velocity they unveil a crossover region of superlinear velocity dependence ( with a power of @xmath1 ) in the velocity range @xmath2 , which we associate to the copper crystalline electronic band structure .
the results are rationalized by simple band models connecting two separate regimes .
we find that the limit of electronic stopping @xmath3 is not as simple as phenomenological models suggest and it plagued by band - structure effects . |
recently , in a very interesting paper , zheng _ et al . , _ @xcite have studied the collective directed transport of symmetrically coupled lattices in symmetric periodic potentials .
they show that under the action of an external wave , that breaks the spatiotemporal symmetry and introduces inhomogeneities of the lattice , a net unidirectional current can be observed .
apparently the current originates from the collaboration of the lattice and the wave ( amplitude , frequency , and phase shifts ) .
the study of directed transport of particles is very important in the physics of molecular motors @xcite , vortex dynamics in superconductors @xcite , josephson junction lattices @xcite , nanotechnology @xcite and many other systems .
many studies have been dedicated to directed transport in spatiotemporal systems @xcite .
these include systems with deterministic ac drives and stochastic forces @xcite . very important
are models composed of a symmetrically coupled lattice in a symmetric potential field , which is driven by an external wave @xcite .
the equation studied by zheng _
et al . , _
@xcite is the following : @xmath1 this is a frenkel kontorova model driven by the wave @xmath2 .
the average current introduced by them @xmath3 was shown to be nonzero for certain values of the parameters of the system .
we wish to re - state this problem in terms of the directed motion of solitons .
solitons are considered as particles in many systems . moreover , in many systems , the solitons are charge carriers @xcite .
so it is very important to study the directed transport of these objects . in the present paper
we investigate the generalized klein
gordon equation : @xmath4 where @xmath5 , @xmath6 is a symmetric potential with , at least , two minima at @xmath7 , @xmath8 , and a maximum at @xmath9 ( @xmath10 ) ; @xmath11 describes an external wave moving with velocity @xmath12 .
both the sine gordon and @xmath13 equations are particular cases of equation ( [ eq2:kg ] ) .
the sine gordon equation is a continuous relative of eq .
( [ eq1:fk ] ) .
these equations possess important applications in condensed matter physics .
for instance , in solid sate physics , they describe domain walls in ferromagnets and ferroelectric materials , dislocations in crystals , charge density waves , interphase boundaries in metal alloys , fluxons in long josephson junctions and josephson transmission lines , etc @xcite
. our initial condition will be a kink soliton whose center of mass is situated at rest in the point @xmath14 .
we will show that the transport of the soliton by the wave depends on the shape of the wave and the parameters of the system .
we will also investigate the stabilization and transport of bubbles using external perturbations in @xmath0systems as the following @xmath15 we will also present the results of real experiments with laser induced bubbles in liquids . the bubbles can be trapped , stabilized and transported by a laser beam of relatively low power .
we will show that the directed transport of solitons and bubbles are related .
for both phenomena some conditions for the external perturbations must be satisfied .
before discussing equation ( [ eq2:kg ] ) we would like to present some results about the dynamics of solitons in the presence of time
independent inhomogeneous perturbations @xcite as in the following equation : @xmath16 suppose we are interested in the stability of a soliton situated in equilibrium positions created by the inhomogeneous force @xmath17 .
we construct a solution @xmath18 with the topological and general properties of a kink .
then , we solve an inverse problem such that @xmath17 possesses the properties of the physical system we are studying @xcite .
now we can investigate the stability of the solution . for this we should solve the spectral problem @xmath19 where @xmath20 , @xmath21 , @xmath22 .
the results obtained with this function can be generalized qualitatively to other systems topologically equivalent to the exactly solvable systems @xcite .
let us see the following example : @xmath23 where @xmath24 .
the exact stationary solution of eq .
( [ eq6:ex ] ) is @xmath25 $ ] .
this solution represents a kink soliton equilibrated on the position @xmath14 . the stability problem ( [ eq5:sp ] )
can be solved exactly .
the eigenvalues of the discrete spectrum are given by the formula @xmath26 where @xmath27 .
the integer part of @xmath28 , i.e. , @xmath29 $ ] , yields the number of eigenvalues in the discrete spectrum , which correspond to the soliton modes ( this includes the translational mode @xmath30 , and the internal or shape modes @xmath31 ( @xmath32 ) ) .
everything that happens with this solution near point @xmath33 can be obtained from eq .
( [ eq7:ds ] ) .
now let us return to the general case ( [ eq4:tiifs ] ) and a review of the obtained results about the dynamics of solitons in eq . ( [ eq4:tiifs ] ) . wherever @xmath17 is positive ,
the kink soliton will be accelerated to the `` left '' . wherever @xmath17 is negative , the kink soliton will be accelerated to the `` right '' .
the zeroes of @xmath17 are candidates for equilibrium positions for the kink soliton @xcite .
if @xmath17 possesses only one zero @xmath34 ( @xmath35 ) , then it is a stable position for the soliton if @xmath36 .
otherwise , the position is an unstable equilibrium .
the opposite is true for the antikink soliton .
a soliton can be trapped by a @xmath17 with a zero that represents a stable position , and it can move away from an unstable position . now we will discuss what happens in the eq .
( [ eq2:kg ] ) with different waveforms .
a ) the soliton feels a moving deep stable potential well .
b ) the soliton feels an unstable equilibrium position .
c ) the soliton feels a potential well with finite walls .
d ) the soliton feels a potential with a finite maximum.,width=529 ] fig .
[ fig : fig1 ] shows different waveforms for @xmath37 . in case
a ) the kink will move trapped inside the potential well created by @xmath37
. the effective potential @xmath38 of this force @xmath17 possesses the property that @xmath39 when @xmath40 .
thus , the kink will move inside a potential equivalent to a harmonic oscillator .
so a wave like this will always transport the soliton to the `` right '' . in the case
b ) , the wave will `` push '' the soliton to the `` right '' as long as the center of mass of the soliton is situated righter than the zero of @xmath17 .
in fact , @xmath38 has an absolute maximum .
however , there is a limit for the wave velocity .
if the wave velocity exceeds this limit , the center of mass of the soliton can reach the maximum of the potential @xmath38 , and the center of mass of the soliton will be behind the maximum of the potential .
in this situation the wave can no longer push the soliton to the `` right '' .
we will here introduce the concept of `` controlled transport '' when the soliton can be transported by the external wave and the soliton velocity can be approximately equal to the wave velocity . in the case
a ) of fig .
[ fig : fig1 ] , we have a controlled transport . in the case
b ) , the soliton is pushed by the wave but the soliton will move ahead of the wave in an `` uncontrolled '' fashion . the waveform shown in fig .
[ fig : fig1 ] c represents a force that creates a stable equilibrium for the soliton .
so , as in case a ) , this wave can carry the soliton .
however , the effective potential corresponding to this force is finite for @xmath41 ( see fig . [
fig : fig2 ] a ) .
( c ) and ( d ) respectively.,width=529 ] so here there is also a limit for the wave velocity .
the wave represented in fig .
[ fig : fig1 ] d is an unstable equilibrium for the soliton as in case b ) .
this wave can push the soliton to the `` right '' provided its velocity is smaller that certain finite value .
[ fig : fig3 ] shows some different localized wave forms .
we would think that these waves could also be used to transport a soliton in a controlled fashion because there is a stable equilibrium involved ( provided the wave velocity is smaller than certain limit ) .
however , here there are some additional limitations in the shape of the wave . investigating different functions in ( [ eq4:tiifs ] ) and the stability problem ( [ eq5:sp ] ) , we have found that if @xmath17 possesses several zeroes and the distance between them is smaller than the soliton width , surprising phenomena can occur .
for instance , in the presence of forces as that shown in fig .
[ fig : fig3 ] a , if the zeroes of @xmath17 are too close , the center zero is not a stable point anymore . in practice , the soliton does not `` feel '' a minimum there .
it feels an unstable equilibrium point . in this case
, this wave would not be able to transport the soliton in a controlled fashion , but it can push the soliton to the right under certain conditions . on the other hand , the force shown in fig .
[ fig : fig3 ] c can carry the soliton if the zeroes are sufficiently separated .
otherwise , the soliton would not `` feel '' the left maximum of the potential @xmath38 ( fig .
[ fig : fig3 ] d ) .
so , using this wave the soliton can not move trapped in the minimum
. even localized forces without a zero like that represented in fig .
[ fig : fig4 ] can `` push '' the soliton to the `` right '' .
note that , considering the effective potential @xmath38 , this wave `` acts '' on the soliton as a moving `` wall '' .
but once and again , there is a limit for the velocity of the wave .
later we will present several examples of systems of type ( [ eq2:kg ] ) for which we can calculate all these limit velocities .
before , we should discuss the most ubiquitous kind of waves : the sinusoidal wave ( see fig .
[ fig : fig5 ] ) : @xmath42 .
% \label{eq8:sw}\ ] ] the previous discussion of waves with the shape shown in fig .
[ fig : fig3 ] leads us to think that the parameters @xmath43 , @xmath44 and @xmath12 should satisfy some conditions for a soliton to be transported by wave ( [ eq8:sw ] ) .
in fact , if the velocity @xmath12 is larger than some critical value , the soliton will be left behind .
if @xmath44 is too large ( and @xmath43 is small ) , the distance between the zeroes of @xmath17 will be very small and the potential wells , where the soliton can be trapped , would be so shallow that , already with certain small velocity @xmath12 , the soliton will be left behind .
thus there are waves of type ( [ eq8:sw ] ) that can not carry the soliton .
let us present here some concrete examples .
consider the following perturbed @xmath13 equation : @xmath45 where @xmath46 $ ] , @xmath47 , @xmath48 .
as can be expected from our discussion , this external wave is able to transport a soliton to the right in a controlled fashion .
this can be observed in fig .
[ fig : fig6 ] .
$ ] ( see eq .
( [ eq9:phi4])).,width=264,height=302 ] later we will see that @xmath49 can be in practice not only a wave but a controlling external perturbation .
if this external perturbation can be manipulated at wish in such a way that the velocity @xmath50 can be any function of time that we desire , then we can move the soliton and place it wherever we wish .
suppose now that in eq .
( [ eq9:phi4 ] ) @xmath51 $ ] , @xmath47 , @xmath48 . in this case , the soliton is pushed to the `` right '' while it is in the `` zone '' where @xmath52 ( i.e. , while the center of mass of the soliton is situated righter than the zero of @xmath49 . )
let us define @xmath53 as the roots of the algebraic equation @xmath54 .
the maximum velocity @xmath55 that the solution can reach in its motion to the right @xcite can be calculated using the equation @xmath56 if the external wave velocity is such that @xmath57 , then the soliton eventually will start to move backward .
now we will analyze the following wave @xmath58 , where @xmath59 : @xmath60 where @xmath61 $ ] , @xmath62 is the point where @xmath63 has a local maximum ; and @xmath64 .
this problem is equivalent to that shown in fig .
[ fig : fig3 ] c and [ fig : fig3 ] d provided @xmath65 , @xmath66 .
the effective potential possesses a local minimum and a local maximum at the left side of the minimum . using the results of papers @xcite we can solve exactly the dynamical problem of the soliton in this potential .
our analysis shows that if @xmath67 , then the soliton feels the barrier in the point @xmath68 .
so , if the wave velocity @xmath12 is small , the soliton can be carried to the right .
on the other hand , if @xmath69 ( although the waveform is still as that shown in fig .
[ fig : fig3 ] c such that @xmath58 has two zeroes ) , the soliton will move to the left , crossing the barrier even if its center of mass is placed in the minimum of the potential and its initial velocity is zero .
the soliton does not feel the barrier .
even with an infinitesimal wave velocity , the soliton will not move to the right dragged by the wave .
we have checked this phenomenon numerically . the effect of a wave shaped as that in fig .
[ fig : fig4 ] can be seen in the following example with exact solution : @xmath70}. % \label{eq13:10}\ ] ] when @xmath71 , the following travelling soliton is an exact solution @xmath72 $ ] .
note that a wave shaped like that in fig .
[ fig : fig4 ] can make the soliton move to the right .
however , for a fixed @xmath43 in eq .
( [ eq13:10 ] ) , the wave velocity should not exceed a limit .
otherwise , the soliton can be left behind . in the case of a sinusoidal wave acting in the framework of the @xmath13 equation : @xmath73 , % \label{eq14:11}\ ] ] we can write the following conditions for the soliton to be dragged by the wave to the right : @xmath74 , @xmath75 , where @xmath55 is defined by the equation @xmath76 , and @xmath77 is the `` middle '' root to the algebraic equation @xmath78 as in eq .
( [ eq7:ds ] ) .
we have experimented with other waveforms that illustrate the different behaviors discussed above . the force @xmath79 $ ] , with @xmath80 has the properties of the function shown in fig .
[ fig : fig1]b .
$ ] , @xmath81 $ ] , @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87,width=283 ] fig .
[ fig : fige1 ] shows that the soliton can be pushed by this external perturbation , provided the velocity @xmath12 is not too large . in fig .
[ fig : fige2 ] we present the behavior of the soliton under the action of the external perturbation @xmath88}{\cosh^3[b(x - wt)]}$ ] , which is of the kind shown in fig .
[ fig : fig1]c .
/\cosh^3[b(x - wt)]$ ] , @xmath89 , @xmath82 , @xmath83 , @xmath90 , @xmath91 , @xmath92,width=283 ] when @xmath93 , the soliton is transported in a controlled way by the external wave . however , already when @xmath94 , the soliton is abandoned behind .
this can be observed in fig .
[ fig : fige3 ] . .
however , the wave velocity is larger , and the soliton is not transported by the wave .
( @xmath95/\cosh^3[b(x - wt)]$ ] , @xmath89 , @xmath82 , @xmath83 , @xmath96 , @xmath91 , @xmath97),width=283 ] a periodic wave like @xmath98 $ ] is able to carry the soliton as is shown in fig .
[ fig : fige4 ] where @xmath99 . the values of the other parameter can be found in the figure caption of fig .
[ fig : fige4 ] .
$ ] , @xmath89 , @xmath82 , @xmath83 , @xmath99 , @xmath100 , @xmath101 @xmath92),width=283 ] if the wave velocity is too large ( say @xmath102 ) , then the soliton can not be carried by the external wave ( see fig . [
fig : fige5 ] ) . $ ] , @xmath89 , @xmath82 , @xmath83 , @xmath103 , @xmath100 , @xmath101 @xmath92),width=283 ]
phase transitions @xcite can be described by an equation similar to ( [ eq2:kg ] ) : @xmath104 recall that @xmath6 possesses two minima ( @xmath105 and @xmath106 ) and a maximum ( @xmath53 ) .
suppose for the moment that @xmath107 .
the main phenomenon for @xmath108 is that of `` nucleation '' where drops or bubbles grow or disappear depending on their gibbs energies @xcite .
the relevant issue is the transition from the metastable state @xmath106 to the state @xmath105 . in phase
transition theory , researchers consider the existence of a critical `` germ '' for the transition .
field configurations @xmath109 with a `` radius '' larger than that of the critical germ should grow in order to produce the transition to the state @xmath105 .
let us first discuss this situation in the context of the one dimensional equation @xmath110 for @xmath107 , it is possible to show ( see refs .
@xcite and @xcite ) that there exists an unstable stationary critical solution that plays the role of a critical germ .
this solution is represented in fig .
[ fig : fig7 ] . ) , width=264 ] this solution can be interpreted as a kink antikink pair @xcite .
in general , a kink and an antikink attract each other @xcite . on the other hand , a constant external force acts on a kink and an antikink in different directions .
so the solution represented in fig .
[ fig : fig7 ] is possible when the kink and the antikink are at some critical distance where the attraction force and the external force balance each other @xcite .
the critical solution is unstable @xcite .
perturbations of the critical solution can lead to the growth of the critical `` bubble '' initiating the transition to the state @xmath7 .
`` bubbles '' with a radius smaller than the radius of the critical solution will disappear , whereas `` bubbles '' with a radius larger than the radius of the critical solution will grow until the whole system is in the state @xmath7 . an example can be studied in the equation @xmath111 where @xmath112 .
note that eq .
( [ eq17:14 ] ) can be written in the form ( [ eq16:13 ] ) , where @xmath113 if @xmath114 , then @xmath115 .
( [ eq17:14 ] ) possesses a bell shaped stationary solution as that shown in fig .
[ fig : fig7 ] .
this solution is unstable @xcite . similar but
different bell shaped initial states will evolve either to the state @xmath116 or the state @xmath8 . if the kink and the antikink that form the one dimensional `` bubble '' are too close , then the bubble will collapse .
if the distance between the kink and the antikink is larger than the `` width '' of the critical germ , then the `` bubble '' will grow .
if @xmath117 , all the initial bubbles will collapse .
can we stabilize the bubbles of phase @xmath7 inside the `` sea '' of pase @xmath118 using external perturbations ?
let us analyze what happens when a kink antikink pair as that shown in fig .
[ fig : fig7 ] is in the presence of an external perturbation like that shown in fig .
[ fig : fig8 ] ) , width=264 ] note that these graphs look similar but they represent very different physical quantities . if the kink of fig .
[ fig : fig7 ] is close to the `` right '' zero of function @xmath17 , it will `` feel '' an effective potential well ( see fig .
[ fig : fig9 ] ) .
the same will happen to the antikink if it is close to the `` left '' zero of @xmath17 in fig .
[ fig : fig8 ] ( see the potential in fig .
( [ fig : fig10 ] ) . ) ) in the presence of perturbation shown in fig .
( [ fig : fig8]),width=264 ] ) in the presence of perturbation shown in fig .
( [ fig : fig8]),width=264 ] if the potential barrier is sufficiently high , the attraction force between the kink and the antikink will not lead to the `` bubble '' collapse .
the attraction force decays rapidly ( exponentially ) with the distance , so for large `` bubbles '' practically `` any '' @xmath17 with the shape of fig .
[ fig : fig8 ] is good for stabilizing the `` bubble '' . on the other hand
, even small bubbles can be stabilized using an external perturbation @xmath17 with two zeroes @xmath119 and @xmath120 ( as in fig .
[ fig : fig8 ] ) such that @xmath121 is sufficiently large .
that is , the change of sign in @xmath17 should be sufficiently sharp .
we have checked numerically the existence of the stable `` bubbles '' .
let us experiment with the following equation @xmath122 where @xmath123 - \tanh\left [ b(x+d ) \right ] + \epsilon \right\ } % \label{eq20:17}\ ] ] for instance , we have established that the perturbations ( [ eq20:17 ] ) , with the parameters @xmath124 , @xmath125 , @xmath126 , @xmath127 , can sustain a stable bubble . an initial configuration for @xmath128 where the kink and the antikink are not exactly in the zeroes of @xmath17
will , anyway , tend to the asymptotically stable bubble .
that is , there is a bubble which is a spatiotemporal attractor .
all close initial configurations will tend ( for @xmath129 ) , to the asymptotically stable bubble .
this phenomenon occurs in the case that a bubble is considered as an initial configuration .
that is , a bubble had been created before the `` force '' @xmath17 is applied .
however , it is interesting to remark that for sufficiently large @xmath43 and @xmath130 , the perturbation itself can create the bubble . that is
, there is no need for a preexisting bubble .
the bubble is created by the perturbations and , then , the perturbation can sustain the stabilized bubble .
if @xmath43 is too small , it is possible that @xmath17 can not sustain a bubble .
an example is produced with the following parameters @xmath131 , @xmath132 , @xmath133 , @xmath134 .
all initial conditions lead to the bubble collapse .
once we have an external perturbation @xmath17 able to sustain a stable bubble , we can transport the bubble if we can move @xmath17 as a whole .
this phenomenon is in the same spirit of all the theory discussed above concerning the soliton transport . in the case of the bubble , as in many of the discussed situations with solitons , there is a limit for the velocity of the perturbation as a whole . when the shape represented in fig .
[ fig : fig8 ] is moved too fast , the bubble can be left behind , and then it will collapse .
what about the bubbles in three dimensions ?
let us concentrate on solutions with radial symmetry .
the stationary solutions of eq .
( [ eq15:12 ] ) ( @xmath135 ) , satisfy the equation @xmath136 it is possible to prove @xcite that ( for @xmath107 ) there exists a solution with a minimum in @xmath137 and that asymptotically it tends to @xmath106 as @xmath138 .
this solution corresponds to the critical germ discussed above .
again as before , this critical solution is unstable . in general , exact solutions for the three dimensional case of the nonlinear equation ( [ eq21:18 ] )
are very difficult to find @xcite .
nevertheless , we have developed an inverse problem approach that allows us to find exact solutions to a class of topologically equivalent systems .
for instance , a wealth of work @xcite has been dedicated to find a solution to equations of type @xmath139 where @xmath140 possesses only terms of higher order than @xmath141 .
this equation is important in field theory .
we have found that function @xmath142 is the exact solution to equation @xmath143 where @xmath140 contains only terms of order higher than @xmath141 .
likewise , it can be proved that function @xmath144 is the critical germ solution of an equation of type ( [ eq21:18 ] ) , where @xmath6 possesses all the properties that we need : two minima and a maximum , @xmath107 , etc . as in the one dimensional case ,
small bubbles will collapse , and bubbles with a radius larger than that of the critical germ will grow . in the context of the three dimensional equation : @xmath145 all the bubbles are non stationary .
however , we should say that very large bubbles are quasi
stationary because the interaction between the walls is so small that ( although they should collapse eventually ) their life time is very large .
what about inhomogeneous perturbations as in eq.([eq15:12 ] ) ?
suppose we have an initial configuration @xmath146 with the shape of a spherical bubble of phase @xmath7 inside a `` sea '' of phase @xmath106 .
this means that there is a spherical spatial sector where @xmath147 , and , outside this sector , @xmath148 .
space zones where @xmath149 will `` try '' to expand the bubble .
space zones where @xmath150 will `` try '' to make the bubble collapse .
in fact , as in the one dimensional case , we can stabilize a spherical bubble using an inhomogeneous external perturbation @xmath151 such that @xmath149 for points @xmath152 that are inside the bubble and @xmath153 for points @xmath152 outside the bubble . to ensure the stability of the bubble ,
the sign change of @xmath151 should be sharp .
we have an exact solution .
let us define the equation @xmath154 where @xmath155 \right.\nonumber\\ % & & \left.- \tanh\left [ \frac{a}{2}(r+r ) \right ] \right\}\nonumber\\ % & & + \frac{3}{2}\left\ { \left ( \phi_1 + \phi_2 \right)\left [ \phi_1\phi_2 + a\left ( \phi_1 + \phi_2 \right ) + a^2 \right ] \right\}\nonumber\\ & & -\frac{2}{r}\left [ \phi_{1r } + \phi_{2r } \right ] - \frac{1}{2}a + \frac{1}{2}a^3,\nonumber\end{aligned}\ ] ] @xmath156 , @xmath157 $ ] , @xmath158 $ ] ; @xmath43 , @xmath130 and @xmath159 are real parameters . here
@xmath160 has the desired property that @xmath161 in a spherical sector around @xmath137 , and @xmath162 outside this spherical sector .
there is an exact bubble solution to eq .
( [ eq26:23 ] ) : @xmath163 - \tanh\left[\frac{a}{2}\left ( r+r \right)\right ] + 1 \right\}. % \label{eq27:24}\ ] ] but we can say more . for @xmath164 , @xmath165 ,
this bubble solution is stable . manipulating @xmath160 we can control this bubble .
in fact , we can move this bubble to another point in space .
many works have been dedicated to the nucleation process in phase transitions @xcite . here
we wish to focus on the works based on the theory of relaxation of metastable states @xcite . in refs.@xcite , the nucleation in the metastable phase near the critical point of a thermodynamic system is described as a relaxation of a metastable state of the order parameter field .
a metastable state of matter can be produced in a first order phase transition .
its relaxation into a thermodynamically stable state is facilitated by the formation of a critical nucleus of a new phase .
the system is described by a scalar field @xmath109 for which a relaxation equation is obtained : @xmath166 where @xmath167 $ ] is the system free energy which is taken in the landau form : @xmath168 = \frac{1}{2}\int \left\ { c\left ( \nabla\phi \right)^2 + \mu\phi^2 + \frac{1}{2}g\phi^4 - 2h\phi \right\}\mathrm{d}\mathbf{x}. % \label{eq29:26}\ ] ] here @xmath31 is a kinetic coefficient and @xmath169 is an external perturbation . at @xmath170 and @xmath171
, there is a critical point in the system .
the line @xmath170 ( @xmath172 ) is a line of first order phase transition .
( [ eq28:25 ] ) is an overdamped version of the @xmath13 equation discussed above .
this equation can be written in dimensionless units as our equations ( [ eq2:kg ] ) .
the solution we have been investigating describes a nucleus of phase @xmath173 in phase @xmath174 . at @xmath175 , the nuclei with small radius attenuate , while the nuclei with large radius increase .
the phase @xmath174 at @xmath175 is metastable unstable to formation of a nucleus with a large radius .
the parameters of the system ( [ eq28:25][eq29:26 ] ) for the liquid vapor first
order phase transition depend on the thermodynamics quantities , e.g. , pressure and temperature @xcite .
if an external perturbation ( say a laser beam ) creates the thermodynamics conditions for a first order transition ( e.g. , @xmath176 in the context of system ( [ eq28:25][eq29:26 ] ) ) in a small zone of a liquid mass , then a vapor bubble could be generated and sustained .
we should say here that in the framework of equation @xmath177 in order to create a bubble able to increase to a macroscopic radius , @xmath151 must take negative values in a limited spatial zone in such a way that @xmath178 . on the other hand , once the bubble has been generated ,
the perturbation @xmath151 necessary to sustain the bubble should be `` negative '' inside the bubble and positive `` outside '' , but the absolute value of @xmath151 in the `` negative '' part should not be so high . in other words , the `` intensity '' of the perturbation needed for generating the bubble is larger than the needed `` intensity '' for stabilizing the bubble .
now we will discuss several experiments about bubble formation in a weak absorbing liquid and its subsequent trapping under the action of a cw laser beam of relatively low power .
the generation of bubbles by intense light pulses was studied in the papers @xcite . on the other hand ,
the pioneering works about laser bubble trapping can be found in refs .
@xcite . here
we present a review of previous @xcite and very recent experiments concerning new effects related to laser bubble trapping , in the light of the theory we are developing .
high intensity light induces a local first
order liquid vapor phase transition .
if light power is reduced after the bubble formation , the bubble remains trapped .
if the intensity is very low then the bubble collapses .
the experimental setup is shown in fig .
[ fig : figsetup ] . in figure
[ fig : figsetup ] we show the experimental set - up used for the laser bubble generation and subsequent trapping . a cw 200 mw diode pumped green laser ( l=532 nm )
is directed vertically down toward a horizontally located sample cell using the mirror m1 .
a 15-cm focal length lens l focuses this light onto the sample . by moving this lens
we can change the radius of the focal spot .
the sample is a 4-mm thick glass cell containing an iodine ethanol solution of high concentration ( 1 to 10 ml ) .
the focused beam heats this solution , generating locally a small bubble at power levels larger than 60 mw . following the light absorption a distribution of temperatures
is induced in the focal region and the bubble surface . because the surface tension depends upon the temperature , a spatial distribution of surface tension is generated on the bubble .
the generated gradient of surface tension induces a force that traps the bubble at the point of maximal temperature ( focal point ) .
this generated force together with the floatation archimedes force make the bubble sticks to the upper surface of the cell . by moving the laser beam using the mirror m1
, the bubble follows the beam spot demonstrating the actual bubble trapping by the laser beam .
the bubble can be observed using the mirror m2 and a ccd camera connected to a computer .
the used liquids were ethanol , benzene , and ethylene glycol .
the liquids were colored with iodine or a dye in order to increase absorption .
we will focus here on the bubble behavior .
there are several interference phenomena that will not be discussed here . in fig .
[ fig : figbub1][fig : figbub3 ] pictures of the trapped bubbles are shown .
mw ( left ) , @xmath179 mw ( center ) and @xmath180 mw ( right ) .
the bubble s diameters depend on the laser power.,title="fig:",width=151 ] mw ( left ) , @xmath179 mw ( center ) and @xmath180 mw ( right ) .
the bubble s diameters depend on the laser power.,title="fig:",width=151 ] mw ( left ) , @xmath179 mw ( center ) and @xmath180 mw ( right ) .
the bubble s diameters depend on the laser power.,title="fig:",width=151 ] in one of the experiments the bubble is generated at a laser power of @xmath181 mw .
the bubble size decreases with decreasing power .
these phenomena can be described in terms of thermal gradients and forces @xcite .
however , we wish to stress the universal character of the phenomena related to bubble formation and transport . these effects can occur in many other physical systems with phase transitions , solitons , domain walls and bubbles of one phase inside other phase , the bubble can be controlled by the laser .
if after creating the bubble , the light beam is defocused in such a way that its spot size becomes much larger than the radius of the generated bubble , then changes of the bubble radius occur until this radius is again stabilized .
this can be explained using the interpretation discussed above where the bubble is described as trapped inside a stable equilibrium position or potential well .
the bubbles are created with a laser power of @xmath181 mw .
then the power is decreased about @xmath182 times . in fig .
[ fig : figbub1 ] we can observe a trapped bubble with a diameter @xmath183 mm using a laser power @xmath184 mw . when the laser power is @xmath180 mw the bubble diameter is @xmath185 mm .
[ fig : figbub2 ] shows the transport of the bubble using the laser beam . with the help of a mirror
the light beam is moved .
the trapped bubble can be moved as the hight beam is moved slowly .
if the motion of the light beam is too fast , then the bubble is left behind and then it collapses . in fig .
[ fig : figbub2 ] ( left ) , the bubble is shown at is initial position .
then it is moved to the left superior corner using the laser beam .
[ fig : figbub3 ] shows the collapse of the bubble while the light power is maintained in @xmath186 mw .
we have investigated generalized nonlinear klein
gordon equations perturbed by moving inhomogeneous external perturbations .
we have shown that a kink soliton can be transported by an external wave , provided the shape of the wave and its velocity satisfy certain conditions . for instance , a moving external perturbation @xmath37 , where function @xmath58 possesses a zero @xmath187 such that @xmath188 , can carry a kink soliton provided its velocity does not exceed certain limit .
figures [ fig : fig1 ] , [ fig : fig3 ] and [ fig : fig5 ] show examples of these transporting inhomogeneous perturbations .
sinusoidal waves as that shown in fig .
[ fig : fig5 ] can also be successful in transporting a kink soliton under certain conditions .
the investigated equations are very general .
besides , the concepts and results that lead to the existence of the reported phenomena are also universal .
so we believe that these phenomena can be observed in many other physical systems .
for instance , the stabilization and transport of bubbles using an external perturbation can occur in other systems undergoing a phase transition .
so the mentioned bubbles ( or drops ) can be superconducting , ferromagnetic , superfluid , etc
. on the other hand , the discussed transport of solitons and kink antikink bubbles can be very important in spin systems , molecular motors , charge
density waves , ferroelectric materials and josephson junctions . in particular , in long josephson junctions the kink
solitons are fluxons .
these fluxons can be controlled using external currents @xcite which can be very relevant in new communication and computer technologies where the fluxons are used as very stable information units . moreover , the kinks are examples of topological defects .
many results about the kinks can be generalized to other topological defects as vortices and spiral waves which are being studied intensively in countless systems nowadays @xcite .
z. zheng , m. c. cross and g. hu , phys .
. lett . * 89 * , 154102 ( 2002 ) .
r. d. astumian , science * 276 * , 917 ( 1997 ) .
f. julicher _ et al . , _ rev
phys . * 69 * , 1269 ( 1997 ) . c. s. lee _
et al . , _ nature * 400 * , 337 ( 1999 ) .
c. j. olson _ et al .
lett . * 87 * , 177002 ( 2001 ) .
et al . , _ phys .
e * 61 * , 2257 ( 2000 ) .
j. rousselet _
_ nature * 370 * , 446 ( 1994 ) .
a. v. oudenaarden _
et al . , _ science * 285 * , 1046 ( 1999 ) .
p. reimann , phys .
rep . * 361 * , 57 ( 2002 ) .
i , derenyi and t. vicsek , phys .
lett . * 75 * , 374 ( 1995 ) .
z. csahok _
et al . , _ phys .
e * 55 * , 5179 ( 1997 ) .
a. igarashi _
et al . , _ phys .
e * 64 * , 051908 ( 2001 ) .
z. zheng _ et al .
lett . * 86 * , 2273 ( 2001 ) .
z. zheng and x. li , commun .
. phys . * 36 * , 151 ( 2001 ) .
et al . , _ phys .
lett . * 76 * , 3436 ( 1996 ) .
j. l. mateos , phys .
lett . * 84 * , 258 ( 2000 ) .
et al . , _ phys . rev .
lett . * 84 * , 2358 ( 2000 ) .
o. yevtushenko _ et al .
, _ europhys .
lett . * 54 * , 141 ( 2001 ) .
s. a. gredeskul and y. s. kivshar , phys . rep . *
216 * , 1 ( 1992 ) .
a. j. heeger , s. kirelson , j. r. schriefer and w. p .
phys . * 60 * , 781 ( 1988 ) .
a. s. davidov , _ solitons in molecular systems _ ( naukova dumka , kiev , 1984 ) .
m. a. rice , phys .
a * 71 * , 152 ( 1979 ) , _ ibid _ * 73 * , 153 ( 1979 ) .
y. lu , _ solitons and polarons in conducting polymers _
( world scientific , singapore , 1988 ) .
, edited by k. lonngren and a. c. scott ( academic , london , 1978 ) .
a. r. bishop , j. a. krumhanst and s. e. trullinger , physica d * 1 * , 1 ( 1980 ) .
s. kivshar and b. a. malomed , rev .
phys . * 61 * , 763 ( 1989 ) .
j. a. gonzlez , b. a. mello , l. i. reyes and l. e. guerrero , phys .
lett . * 80 * 1361 ( 1998 ) . j. a. gonzlez and j. a. hoyst , phys .
b * 45 * 10338 ( 1992 ) .
j. a. gonzlez and b. a. mello , phys .
a * 219 * 226 ( 1996 ) .
j. a. gonzlez , l. e. guerrero and a. bellorn , phys .
e * 54 * 1265 ( 1996 ) .
j. a. gonzlez , a. bellorn and l. e. guerrero , phys .
e * 60 * r37 ( 1999 ) .
j. a. gonzlez , a. bellorn and l. e. guerrero , phys .
e * 65 * 065601(r ) ( 2002 ) .
j. a. gonzlez and f. a. oliveira , phys .
b * 59 * , 6100 ( 1999 )
. j. s. langer , ann .
( n.y . ) * 41 * , 108 ( 1967 ) .
m. bttiker and r. landauer , phys .
a * 23 * , 1397 ( 1981 ) . j. s. aubry , j. chem .
phys . * 62 * , 3217 ( 1975 ) .
s. coleman , phys .
d * 15 * , 2929 ( 1977 ) ; _ ibid .
_ * 16 * , 1248(e ) ( 1977 ) . c. g. callan and s. coleman , phys
d * 16 * , 1762 ( 1977 ) .
m. a. collins , a. blumen , j. f. currie and j. ross , phys .
b * 19 * , 3630 ( 1979 ) d. a. gorukhov and g. blatter , phys .
b * 58 * , 5486 ( 1998 ) .
a. e. filippov , yu .
e. kuzovlev and t. k. sobolova , phys .
a * 165 * , 159 ( 1992 ) .
a. z. patashinskii and b. i. shumilo , sov .
jetp * 50 * , 712 ( 1979 ) .
a. z. patashinskii and v. l. pokrovskii , _ fluctuation theory of phase transitions _
( navka , moscow , 1975 ) .
a. z. patashinskii , _ statistical description of metastable states _
preprint inp 7844 , novosibirsk , 1978 ( unpublished ) .
j. a. gonzlez and j. a. hoyst , phys .
b * 35 * 3643 ( 1987 )
. j. a. gonzlez and j. estrada
sarlabous , phys .
a * 140 * 189 ( 1989 ) .
b. a. mello , j. a. gonzlez , l. e. guerrero and e. lpez
atencio , phys .
a * 244 * 277 ( 1998 ) .
w. lauterborn , acustica * 31 * , 51 ( 1974 ) .
p. giovanneschi and d. dufresne , j. app . phys .
* 58 * , 651 ( 1985 ) .
a. marcano , app . opt . * 31 * , 2757 ( 1992 ) .
a. marcano and l. aranguren , app .
b * 56 * , 343 ( 1995 ) .
m. c. cross and p. c. hehenberg , rev .
phys . * 65 * , 851 ( 1993 ) .
i. s. aranson and l. kramer , rev .
* 74 * , 99 ( 2002 ) .
m. br and m. or guil , phys .
lett . * 82 * , 1160 ( 1999 ) .
l. q. zhou and q. ouyang , phys .
85 * , 1650 ( 2000 ) . | we investigate generalized soliton bearing systems in the presence of external perturbations .
we show the possibility of the transport of solitons using external waves , provided the waveform and its velocity satisfy certain conditions .
we also investigate the stabilization and transport of bubbles using external perturbations in @xmath0systems .
we also present the results of real experiments with laser induced vapor bubbles in liquids . |
wireless communications are inherently insecure , due to the broadcast nature of the medium , which makes security a pivotal design issue in the implementation and operation of current and future wireless networks . compared to the traditional key - based cryptographic techniques that are applied to upper layers , physical layer security can enhance the secrecy of wireless communications without using secret keys and complex encryption / decryption algorithms , and thus has been recognized as an alternative for cryptographic techniques .
the key idea of physical layer security is to exploit the randomness of wireless channels to offer secure data transmissions @xcite . in early studies , e.g. , @xcite , the principle of physical layer security
was established in a single - input single - output wiretap channel .
subsequently , physical layer security in multi - input multi - output ( mimo ) communication systems has been intensively investigated @xcite , due to the benefits of mimo techniques such as high data rate and high link reliability .
most recently , physical layer security in large - scale wireless networks , such as mobile ad hoc and sensor networks , has been receiving considerable attention @xcite .
a key property of large - scale wireless networks is that the node locations in the network were spatially randomly distributed . as such , stochastic geometry and random geometric graphs
are used for modeling the locations of spatially random - distributed nodes . in @xcite ,
the throughput of large - scale decentralized wireless networks with physical layer security constraints was investigated .
considering the path loss as the sole factor influencing the received signal - to - noise ratio ( snr ) at the receiver , @xcite examined the secrecy rate in cellular networks . in @xcite
the secrecy rate achieved by linear precoding was analyzed in the broadcast channel with spatially random external eavesdroppers , and in @xcite the secrecy rate achieved by linear precoding in cellular networks was analyzed .
we note that @xcite only considered point - to - point transmissions .
this leaves physical layer security with cooperative relays in large - scale networks as an open problem .
since the relay channel efficiently improves the coverage and reliability in wireless networks @xcite , it is of practical significance to investigate the secrecy performance of such channels . particularly important in this context would be extensions of previous work on relay wiretap channels which focused on scenarios where the location of an eavesdropper(s ) is fixed and known at the source ( e.g. , @xcite ) . in this work we propose a secure transmission scheme for a relay wiretap channel , where a source transmits to a destination via a decode - and - forward ( df ) relay in the presence of multiple spatially random - distributed eavesdroppers .
the source is equipped with multiple antennas , while the relay , the destination , and the eavesdropper are equipped with a single antenna each .
in addition to information signals , we assume that the source transmit artificial noise signals in order to confuse the eavesdroppers . aiming at maximizing the secrecy throughput , while satisfying a secrecy outage probability constraint , we determine both the optimal power allocation factor between an signals and information signals and the optimal wiretap code rates . compared to current studies on physical layer security in relay networks @xcite , our contributions are threefold .
first , we consider a more practical scenario where the locations of the eavesdroppers are spatially randomly distributed and not known at the source .
second , we propose a new secure transmission scheme that maximizes the secrecy throughput in such a scenario .
third , we derive explicit expressions for the transmission outage probability and the secrecy outage probability , which enable us to analytically characterize the secrecy throughput of the relay wiretap channel .
we note that these expressions are independent of realizations of the main channel and the eavesdropper channel .
the rest of the paper is organized as follows .
section [ sec : system_model ] describes the system model and presents the proposed scheme . in section
[ sec : analysis ] , we analyze and optimize the secrecy performance achieved by the proposed scheme . numerical results and related discussions are presented in section [ sec : numerical ] . finally , section [ sec : conclusion ] draws conclusions .
we consider a relay wiretap channel , as depicted in fig . [ system_model ] , where a source ( s ) communicates with a destination ( d ) with the aid of a relay ( r ) in the presence of multiple spatially random eavesdroppers .
we assume that the source is equipped with @xmath0 antennas , while the destination , the relay , and the eavesdroppers are equipped with a single antenna each .
we also assume that there is no direct link between the source and the destination .
we denote @xmath1 and @xmath2 as the distance between the source and the relay and the distance between the relay and the destination , respectively , and denote @xmath3 as the path loss exponent . as in @xcite ,
the eavesdroppers are modeled as a homogeneous poisson point process ( ppp ) @xmath4 with density @xmath5 .
this model is practical and representative for decentralized networks where each node is randomly distributed @xcite .
the source , the relay , and the destination do not belong to @xmath4 .
we now detail the proposed transmission scheme .
the proposed scheme utilizes two time slots . in the first time slot ,
the source transmits information signals together with an signals to the relay . in this time slot ,
the transmitted signals from the source are overheard by the eavesdroppers .
the role of the an signals is to confuse the eavesdroppers . in the second time slot
, we assume that the source is silent .
while it is clear that this will lead to a sub - optimal solution , such an assumption does allow for analytical tractability in a special case . removing this assumption means
more power can be added to the signal in the first time slot .
our other work @xcite , in which we consider multiple antennae at all nodes , quantifies the advantage of adding source noise in the second time slot .
we also assume the relay transmits the received signals to the destination using the df protocol . in this time slot
, the broadcast signals are also overheard by the eavesdroppers .
we assume that all the channels are subject to independent and identically distributed rayleigh fading .
we also assume a quasi - static block fading environment in which all the channel coefficients remain the same within one time slot .
we denote @xmath6 as the @xmath7 channel vector from the source to the relay and denote @xmath8 as the channel coefficient from the relay to the destination .
we now express the transmitted and received signals in two time slots separately . in the first time slot , the transmitted signal at the source is given by @xmath9 where @xmath10 denotes the @xmath11 beamforming matrix at the source and @xmath12 denotes the combination of the information signal and the an signal . to perform a such transmission , we first design @xmath10 as @xmath13 where @xmath14 is used to transmit the information signal and @xmath15
is used to transmit the an signal .
the aim of @xmath10 is to degrade the quality of the received signals at the eavesdroppers . by transmitting an signals into the null space of @xmath6 through @xmath10
, the source ensures that the quality of the received signals at the relay is free from an interference . in designing @xmath10 ,
we choose @xmath14 as the eigenvector corresponding to the largest eigenvalue of @xmath16 .
we then choose @xmath15 as the remaining @xmath17 eigenvectors of @xmath16 .
such a design ensures that @xmath10 is a unitary matrix .
we then design @xmath12 as @xmath18 where @xmath19 denotes the information signal and @xmath20 is an @xmath21 vector of the an signal .
we define @xmath22 , @xmath23 , as the fraction of the power allocated to the information signal . as such
, we obtain @xmath24=\beta$ ] and @xmath25=\frac{1-\beta}{n-1}\mathbf{i}_{n-1}$ ] , where @xmath26 $ ] is expectation and @xmath27 is the @xmath28 identity matrix .
moreover , we confirm that @xmath29=\mathbf{i}_n$ ] . based on , , and
, the received signal at the relay in the first time slot is expressed as @xmath30 where @xmath31 denotes the transmit power at the source , and @xmath32 denotes the thermal noise at the relay , which is assumed to be a zero mean complex gaussian random variable with variance @xmath33 , i.e. , @xmath34 .
we next express the received signal at a typical eavesdropper located at @xmath35 , @xmath36 , in the first time slot as @xmath37 where @xmath38 denotes the @xmath7 channel vector from the source to the typical eavesdropper located at @xmath35 , @xmath39 denotes the distance between the source and the typical eavesdropper located at @xmath35 , and @xmath40 denotes the thermal noise at the typical eavesdropper located at @xmath35 , which is assumed to be a zero mean complex gaussian random variable with variance @xmath41 , i.e. , @xmath42 . in the second time slot
, the relay adopts the df protocol to forward signals to the destination .
specifically , the relay first decodes the transmitted signals from the source and then broadcasts them after re - encoding .
therefore , we express the received signal at the destination as @xmath43 where @xmath44 denotes the transmit power at the relay , @xmath8 denotes the channel coefficient from the relay to the destination , @xmath45 denotes the transmitted signal of the relay with @xmath46=1 $ ] , and @xmath47 denotes the thermal noise at the destination , which is assumed to be a zero mean complex random variable with variance @xmath48 , i.e. , @xmath49 .
we next express the received signal in the second time slot at the typical eavesdropper located at @xmath35 is expressed as @xmath50 where @xmath51 denotes the channel coefficient from the relay to the typical eavesdropper located at @xmath35 , @xmath52 denotes the distance between the relay and the typical eavesdropper located at @xmath35 , and @xmath53 denotes the thermal noise at the the typical eavesdropper located at @xmath35 , which is assumed to be a zero mean complex random variable with variance @xmath54 , i.e. , @xmath55 .
we express the instantaneous snr at the destination as @xmath56 where @xmath57 and @xmath58 . for the instantaneous snr at the eavesdroppers ,
we assume that the eavesdroppers are non - colluding , which means that each eavesdropper decodes her own received signals from the source and the relay , without cooperating with other eavesdroppers .
thus , we express the instantaneous snr at the eavesdroppers as @xmath59 where @xmath60 and @xmath61 . in order to evaluate and optimize the secrecy performance achieved by our proposed scheme ,
we apply the performance metric proposed in @xcite , which is given by @xmath62 where the factor @xmath63 is due to the two time slots used in the transmission , @xmath64 denotes a parameter pair of the wiretap code used by the source , @xmath65 denotes the transmission rate of the wiretap code , @xmath66 denotes the cost of preventing the transmitted wiretap code from eavesdropping , and @xmath67 denotes the transmission outage probability ( @xmath67 is defined as the probability that the instantaneous snr at the destination is less than @xmath68 ) .
mathematically , @xmath67 is formulated as @xmath69 henceforth , we refer to @xmath70 as the secrecy throughput . the goal of this work is to maximize the secrecy throughput of the relay wiretap channel with spatially random eavesdroppers under secrecy constraints . to achieve this goal
, we formulate the design problem as @xmath71 where @xmath72 denotes the secrecy outage probability .
@xmath72 is defined as the probability that @xmath73 is larger than @xmath74 .
mathematically , @xmath72 is formulated as @xmath75
in this section , we first analyze the secrecy performance by deriving explicit expressions for the transmission probability and the secrecy outage probability , respectively . based on these results
we determine the optimal parameters , e.g. , @xmath65 , @xmath66 , and @xmath22 , that achieve the optimal secrecy performance of the relay wiretap channel .
notably , the determined optimal secrecy performance is independent of realizations of the main channel and the eavesdroppers channels . in this subsection
we focus on the transmission outage probability , @xmath67 .
we first obtain the cumulative distribution functions ( cdfs ) of @xmath76 and @xmath77 as @xmath78 with @xmath79 , and @xmath80 with @xmath81 . based on , , and
, we re - express the transmission outage probability in as @xmath82 we now focus on the secrecy outage probability , @xmath72 . to derive @xmath72 , we first express the cdfs of @xmath83 and @xmath84 as @xmath85 with @xmath86 , and @xmath87 with @xmath88 .
based on , and , we present the secrecy outage probability in the following theorem .
[ t1 ] the secrecy outage of the relay wiretap channel is derived as @xmath89 where @xmath90 @xmath91 @xmath92 and @xmath93 .
see appendix [ app_t1 ] .
we find that provides an easy - to - compute expression for the secrecy outage probability . despite that @xmath94 and @xmath95 for general @xmath3
can not be obtained in closed - form , they can be easily evaluated since only a double integral is involved in @xmath94 and @xmath95 .
we next present simplified closed - form expressions for the special case where @xmath96 . for this special case , we first simplify @xmath94 , which yields a closed - form expression given by ( next page ) , @xmath97 ' '' '' where @xmath98 follows by applying the expression of the zero - order modified bessel function @xmath99
, @xmath100 follows by applying the series representation of @xmath99 , @xmath101 follows by using @xmath102 and noting that @xmath103 and @xmath104 follows by applying the series expression representation of the exponential function .
similarly , for the special case where @xmath96 , we simplify @xmath95 , which yields a closed - form expression given by ( next page ) . the simplified expressions in and offer us a computationally efficient way to calculate the secrecy outage probability for the special case where @xmath96 .
@xmath105 ' '' '' in this subsection we determine the optimal parameters that maximize the secrecy throughput , @xmath70 , for general @xmath3 . specifically , we first determine the optimal wiretap code rates pair , @xmath106 , that maximizes the secrecy throughput for a given power allocation factor @xmath22 .
we then determine the joint optimal power allocation factor and wiretap code rates , @xmath107 , that maximizes @xmath70 .
the optimal wiretap code rates pair , @xmath106 , that maximizes @xmath70 for a given @xmath22 is determined as @xmath108 taking the first - order derivative of @xmath72 with respect to @xmath66 , we confirm that @xmath109 , which states that @xmath72 monotonically decreases as @xmath66 increases . as such , the value of @xmath110 satisfying is the value of @xmath110 that satisfies the secrecy outage probability constraint , i.e. , @xmath111 .
we then confirm that @xmath112 is first positive then negative as @xmath65 increases .
this demonstrates that the value of @xmath113 satisfying is unique .
although a closed - form solution for @xmath106 is mathematically intractable , we are able to determine the values of @xmath106 numerically . accordingly , the maximal secrecy throughput based on the values of @xmath106 for a given @xmath22 is defined as @xmath114 . the joint optimal power allocation factor and wiretap code rates which maximizes @xmath70 in , @xmath107 ,
is determined as @xmath115 using and , we are able to solve numerically . specifically , we first determine the value of @xmath106 using for each value of @xmath22 .
this leads to the secrecy throughput with @xmath106 , denoted by @xmath116 .
we then determine the value of @xmath22 that maximizes @xmath114 , denoted by @xmath117 .
accordingly , the value of @xmath106 associated with @xmath117 is determined as @xmath118 .
finally , the maximal secrecy throughput based on the values of @xmath117 , @xmath119 , and @xmath120 is defined as @xmath121 .
in this section we present numerical results to validate our analysis of the outage probabilities and examine the benefits of the proposed scheme . for illustrative purpose , throughout this section we concentrate on the practical example of a highly shadowed urban area with @xmath122 .
in addition , we adopt @xmath123 . versus @xmath124 for different values of @xmath125 with @xmath126 , @xmath127 , and @xmath128.,title="fig:",width=288,height=220 ] we first verify the accuracy of the transmission outage probability and the secrecy outage probability using monte carlo simulations . in fig .
[ fig_side_a ] , we plot @xmath67 versus @xmath124 for different values of @xmath125 with @xmath127 and @xmath129 . in this figure , we consider @xmath130 .
we first see that the analytical curves , generated from , precisely match the simulation points , which demonstrates the correctness of our expression for @xmath67 in .
second , we see that @xmath67 increases monotonically as @xmath124 increases for a given @xmath125 , which implies that the transmission outage probability increases when the transmission rate of the wiretap code increases .
we further see that @xmath67 decreases as @xmath125 increases for a given @xmath124 .
this reveals that the transmission outage probability reduces when the source and the relay use more power to transmit under a fixed @xmath124 . versus @xmath131 for different values of @xmath132 with @xmath126 , @xmath127 , and @xmath128.,title="fig:",width=288,height=220 ] in fig .
[ fig_side_b ] , we plot @xmath72 versus @xmath131 for different values of @xmath132 with @xmath127 and @xmath133 . in this figure
we consider @xmath134 .
we see an excellent match between the analytical curves generated from and the simulation points , demonstrating the correctness of our expression for @xmath72 in .
we then see that @xmath72 decreases monotonically as @xmath131 increases for a given @xmath132 , which shows that the secrecy outage probability decreases when the redundancy rate of the wiretap code increases .
we further observe that @xmath72 increases as @xmath132 increases .
this is due to the fact the eavesdroppers receive signals from _ both _ the source and the relay .
it follows that increasing the transmit power at the source and the relay leads to an improved received snr at the eavesdroppers . versus @xmath65 for different values of @xmath0 with @xmath126 , @xmath133 , @xmath135 , @xmath136 db , and @xmath137.,title="fig:",width=288,height=220 ] we now examine the impact of the system parameters @xmath65 and @xmath22 on the secrecy throughput . in fig .
[ fig_side_c ] , we plot @xmath70 versus @xmath65 for different values of @xmath0 with a fixed @xmath22 and the optimal @xmath66 .
we first observe that there exists a unique @xmath113 that maximizes @xmath70 for a given @xmath22 .
we also observe that the maximal @xmath70 for a given @xmath22 , i.e. , @xmath114 , increases as @xmath0 increases .
this shows that adding extra transmit antennas at the source significantly enhances the secrecy performance of the relay wiretap channel . versus @xmath22 for different values of @xmath0 with @xmath126 , @xmath135 , @xmath136 db , and @xmath138.,title="fig:",width=288,height=220 ] in fig .
[ fig_side_d ] , we plot @xmath114 versus @xmath22 for different values of @xmath0 . for each point of @xmath114
, we choose the @xmath106 pair that maximizes @xmath70 for the corresponding @xmath22 .
we first observe that there exists a unique @xmath117 that maximizes @xmath114 .
we then observe that the value of @xmath117 that maximizes @xmath114 is around @xmath139 , which demonstrates that the secrecy throughput is improved if approximately 10% of the total transmit power at the source is allocated to an signals .
we also observe that the maximal @xmath114 , i.e. , @xmath121 , increases as @xmath0 increases .
furthermore , we observe that the value of @xmath117 slightly decreases as @xmath0 increases , which shows that in order to maintain the optimal secrecy throughput , the power allocated to an must increase as the source antenna number increases .
in this work we proposed a secure transmission scheme for a relay wiretap channel , in which an @xmath0-antenna source transmits both information signals and an signals in the presence of multiple spatially random single - antenna eavesdroppers .
conditioned on the use of a decode - and - forward protocol at the relay , we determined the optimal parameters @xmath107 that maximizes the secrecy throughput , based on our derived expressions for the transmission outage probability and the secrecy outage probability .
in addition , we examined the impact of @xmath0 on the secrecy throughput , showing how the maximal secrecy throughput increases with @xmath0 .
the work reported here provides valuable insights into the design of new physical layer security schemes in which the locations of the eavesdroppers are randomly distributed and not known at the source .
the work of j. yuan and r. malaney was funded by the australian research council discovery project dp120102607 .
the work of n. yang was funded by the australian research council discovery project dp150103905 , and the ian potter foundation s travel grant 20160251 .
according to , and , we re - express as @xmath140\notag\\ & = 1-\mathbb{e}_{\phi}\left[\prod_{i\in\phi } \left(1-\left(1+\frac{\left(1-\beta\right)\tau_e}{\beta\left(n-1\right)}\right)^{-\left(n-1\right)}\right.\right.\notag\\ & ~~\left.\left.\times\exp\left(-\frac{\tau_e}{\beta\overline{\gamma}_{si}}\right)\right)\left(1-\exp\left(-\frac{\tau_e}{\overline{\gamma}_{ri}}\right)\right)\right]\notag\\ & \overset{(a)}{=}1-\exp\left(-2\lambda\left(\mathcal{j}_1+\mathcal{j}_2-\mathcal{j}_3\right)\right ) . \ ] ] where @xmath141 @xmath142 and @xmath143 and @xmath98 follows by applying the probability generating functional ( pgfl ) for the ppp @xmath4 , given by @xcite @xmath144=\exp\left\{-\int_{\mathbb{r}^2}\left[1-f\left(x\right)\right]\lambda dx\right\},\end{aligned}\ ] ] and by changing to polar coordinates .
to proceed , we first derive @xmath145 as @xmath146 where in @xmath100 we have used @xmath147 , in @xmath101 we have used @xmath148 , and @xmath104 follows from the definition of the gamma function .
we then derive @xmath94 as @xmath149 we further derive @xmath95 as @xmath150 n. yang , p. l. yeoh , m. elkashlan , r. schober , and i. b. collings , `` transmit antenna selection for security enhancement in mimo wiretap channels , '' _ ieee trans .
1 , pp . 144154 , jan .
2013 .
n. yang , p. l. yeoh , m. elkashlan , r. schober , and j. yuan , `` mimo wiretap channels : a secure transmission using transmit antenna selection and receive generalized selection combining , '' _ ieee commun . lett .
17541757 , sep . 2013 .
n. yang , g. geraci , j. yuan , and r. malaney , `` confidential broadcasting via linear precoding in non - homogeneous mimo multiuser networks , '' _ ieee trans .
62 , no . 7 , pp .
25152530 , july 2014 .
s. yan , n. yang , r. malaney , and j. yuan , `` transmit antenna selection with alamouti coding and power allocation in mimo wiretap channels , '' _ ieee trans .
wireless commun .
3 , pp . 16561667 , mar
. 2014 .
x. zhou , r. k. ganti , j. g. andrews , and a. hjrungnes , `` on the throughput cost of physical layer security in decentralized wireless networks , '' _ ieee trans .
wireless commun .
10 , no . 8 , pp .
27642775 , aug .
2011 .
g. geraci , s. singh , j. g. andrews , j. yuan , and i. b. collings , `` secrecy rates in broadcast channels with confidential messages and external eavesdroppers,''_ieee trans .
wireless commun .
_ , vol . 13 , no .
29312943 , may 2014 .
g. geraci , h. s. dhillon , j. g. andrews , j. yuan , and i. b. collings , `` physical layer security in downlink multi - antenna cellular networks , '' _ ieee trans .
20062021 , jun . 2014 .
r. bassily , e. ekrem , x. he , e. tekin , j. xie , m. r. bloch , s. ulukus , and a. yener , `` cooperative security at the physical layer : a summary of recent advances , '' , vol .
5 , pp . 1628 , sep . 2013 . | we propose a secure transmission scheme for a relay wiretap channel , where a source communicates with a destination via a decode - and - forward relay in the presence of spatially random - distributed eavesdroppers .
we assume that the source is equipped with multiple antennas , whereas the relay , the destination , and the eavesdroppers are equipped with a single antenna each . in the proposed scheme , in addition to information signals ,
the source transmits artificial noise signals in order to confuse the eavesdroppers . with the target of maximizing the secrecy throughput of the relay wiretap channel
, we derive a closed - form expression for the transmission outage probability and an easy - to - compute expression for the secrecy outage probability . using these expressions
, we determine the optimal power allocation factor and wiretap code rates that guarantee the maximum secrecy throughput , while satisfying a secrecy outage probability constraint . furthermore , we examine the impact of source antenna number on the secrecy throughput , showing that adding extra transmit antennas at the source brings about a significant increase in the secrecy throughput . |
the war ended on august 15 , 1945 .
although large areas of tokyo had been reduced to ashes by the intense bombing , tokyo imperial university ( todai ) was luckily spared from destruction .
however , life in the tokyo area had become very difficult , because of the extreme shortage of housing and the breakdown of the food - rationing system , established during the war , in the postwar confusion .
todai was closed until the end of 1945 , and students went back to their homes in the countryside .
todai reopened in january 1946 .
i returned to todai to resume my study of physics .
tokyo was still extremely short of housing .
some students and staff were living in their offices or laboratories . fortunately
a cousin of my father offered me a room to live .
sometime in the spring of 1946 yoichiro nambu showed up in the physics department of todai , still wearing his army uniform .
he had graduated from todai in september 1942 .
then he served in the army until the end of the war .
but he was not discharged from the army for few more months , since he was in charge of dismantling the army research facility to which he was assigned .
i had enrolled in the physics department of todai in september 1944 . because of the war - time measure of accelerated curriculum our class took all of our first - year courses during the fall semester of 1944 and all of our second - year courses during the spring semester of 1945 .
thus when i returned to todai in january 1946 , there were no more formal courses to take .
i was able to work for more than one and a half years on research projects without being bothered by course work or exams . before describing the early post - war period ,
i should like to mention an important work of s. tomonaga published in 1943 @xcite . during the war most physicists
were mobilized to the war effort .
tomonaga was assigned to develop radar .
i understand that he developed an elegant theory of radar based on the idea of the s - matrix , to which he was exposed while visiting heisenberg before the war . what is more relevant for the development of physics in post - war japan , tomonaga found time to work on the super - many - time reformulation of quantum field theory @xcite .
this was an attempt to reformulate the non - relativistic form of the heisenberg - pauli theory @xcite into a fully relativistic form .
i was interested in particle physics , or quantum field theory in general .
but in 1946 todai had no professor of particle physics . looking for an advisor
, i recalled that i was strongly impressed by the extreme clarity of the lectures on mathematical physics professor kunihiko kodaira gave during the fall term of 1944 .
although he was not a particle physicist , i asked him to supervise my study of quantum field theory .
he kindly agreed to supervise me and my classmate yoichi fujimoto in the study of particle physics . actually , although todai had no professor of particle physics at that time , there was no shortage of people who were interested in particle physics .
discussion of particle physics had been going on all the time .
particularly active were ziro koba , satio hayakawa , yoichiro nambu , gyo takeda , yoshio yamaguchi , and yoichi fujimoto . when prof .
kodaira asked us what we wanted to study , we said that we wanted to begin with the latest papers available on quantum field theory .
( this is because we had studied by ourselves many of the early papers on quantum field theory . ) during the war research papers from abroad stopped coming to japan .
but we did find some fragments of journals published during 1941 - 1944 in the library of the todai physics department .
my memory is not clear but i think we chose as the first topic of our weekly seminar dirac s classical theory of the electron @xcite or his theory of the indefinite metric @xcite . in a footnote of one of these papers
, we found that pauli mentioned that the s - matrix theory of heisenberg is a picture frame without a picture " @xcite . from this footnote
we learned that heisenberg wrote two papers on the s - matrix , and we began looking for them . according to my classmate y. yamaguchi zeitschrift fr physik journals containing these articles
were brought to japan during the war by a german submarine , and some copies were kept in the office of professor tomonaga at tokyo education university .
yamaguchi went to prof .
tomonaga , and brought copies back to todai with his permission . in view of the top secret " stamp on the covers of these copies , their existence must have been known to only a few people around prof .
we chose these papers @xcite and @xcite as the subject of our next seminar and studied them intensely .
quantum electrodynamics ( qed ) was formulated around 1929 by heisenberg , pauli , dirac , and others . however , for many years qed suffered from a curious disease : although it agreed well with experiments in the lowest order of perturbation theory , attempts to improve agreement further by calculating higher - order corrections always ran into divergent results .
clearly something was wrong with qed . since the 1930 s many people had tried to solve this puzzle .
heisenberg s s - matrix papers may be regarded as one of these efforts .
his first paper is an attempt to realize his philosophy that the divergence difficulty of quantum field theory must be solved by describing physical phenomena in terms of observable quantities only .
specifically , he was trying to remove the hamiltonian from the theory , because he regarded it as unobservable .
the first paper @xcite presents only the framework of a theory , as pauli had observed .
the second paper @xcite is an attempt to put a
toy picture " in it .
heisenberg was trying to understand the mechanism of cosmic - ray showers , to see whether they are multiple production ( many particles are produced in one collision ) or plural production ( particles are produced in several successive collisions ) , which was then a hot topic among cosmic - ray physicists .
this model is not a successful theory as far as physics is concerned .
however , heisenberg explored various consequences of this model in great detail .
his calculation was rather difficult for our shallow knowledge of mathematics .
kodaira showed us how to treat these problems . at some point
it was necessary for us to study the behavior of a function with two complex variables .
we were ignorant of the theory of functions of several complex variables .
we were fascinated by prof .
kodaira s description of the riemann surface of 2 complex variables as a picture of 4-dimensional space of real variables .
kodaira became interested in heisenberg s s - matrix theory not as a theory of physics but because of its mathematical structure .
he realized that heisenberg s method is actually related to the eigenvalue problem of the jacobi matrix .
thus we began to study the jacobi matrix .
since the last chapter of m. h. stone s book linear transformation in hilbert space " has an introduction to the jacobi matrix , we students studied this chapter . but to understand this chapter it was necessary to understand the preceding chapter , and then further earlier chapters . by the time we arrived at the chapter on 2nd - order ordinary differential equations , prof .
kodaira had discovered a general formula for eigenfunction expansion of 2nd - order ordinary differential equations , and we students were prepared to understand his new theory .
tokyo in 1946 was still burned out and short of food .
circumstances were far from ideal for studying physics .
however , we worked hard to understand the heisenberg s theory step by step .
the seminar started right after lunch .
occasionally it lasted past 8.00 pm without a supper break . for a short rest at around 3.00 pm prof .
kodaira took out a lemon , partly covered by mold , from the drawer of his desk and put a small slice in a potful of black tea .
i still remember its taste .
kodaira summarized his work in a preprint , the eigenvalue problem for ordinary differential equation of the second order and heisenberg s theory of s - matrix " , and asked prof .
yukawa , who happened to be going to the institute for advanced study in princeton , to give it to h. weyl . shortly afterwards
, kodaira received a letter from weyl informing him that titchmarsh had obtained the same result by an entirely different method . furthermore
, weyl kindly sent titchmarsh s book to kodaira .
we found that titchmarsh s method was somewhat pedestrian , in contrast to kodaira s very elegant method which used weyl s limit circle " method effectively and displayed a deep understanding of the hilbert space .
my memory is hazy whether it was in late 1946 or early 1947 , but it became widely known around tokyo that heisenberg s s - matrix theory was being studied by the kodaira group , and fujimoto and i were asked to give an open seminar on heisenberg s s - matrix theory .
we were surprised to find that the lecture room was overflowing with people .
we managed to present the seminar without embarrassing mistakes . according to s. hayakawa ( 2 years senior to our class )
, professor tomonaga attended this seminar .
but i do not recall seeing him in the audience .
probably i was too nervous in our first public lecture .
shortly after the end of the war prof .
tomonaga assembled several physicists from tokyo education university ( predecessor of today s tsukuba university ) and todai , and began to extend his super - many - time theory to qed and other field theories .
his seminar was held in a building standing in the middle of the burned - out ruins of an army facility at okubo in northern tokyo , because his own university had been destroyed by bombing .
sometime toward the end of 1946 or early 1947 y. yamaguchi took y. fujimoto and me to visit tomonaga s lab in okubo , which was about one hour s trip away from todai .
tomonaga agreed to let us join his weekly seminar .
tomonaga s seminar became very popular and attended by more than 30 people from the tokyo area .
the main texts used in his seminar were heitler s quantum theory of radiation " and bethe s review article on nuclear physics published in the reviews of modern physics .
besides discussing these texts , we often went back to the original articles to clarify questions .
i recall in particular the article of bloch and nordsieck @xcite concerning the infrared ( ir ) catastrophe , and the article of pauli and fierz @xcite analyzing the ultraviolet ( uv ) divergence . in many countries research in physics , interrupted by the war , was resumed and new discoveries began to be published . new literature was not yet available in the library of todai s physics department .
however , the latest issues of the physical review , as well as magazines such as newsweek and time , began to arrive at the cie library ( of the occupation force ) at hibiya , not far from the tokyo central railroad station .
we used to visit this library early in the morning so that we did not miss the latest journals and magazines .
this is where we first learned about the shelter island conference held in the summer of 1947 where the discoveries of the lamb shift @xcite and the anomalous magnetic moment of the electron @xcite were reported .
when the exciting news from the u. s. arrived , the tomonaga group was ready to tackle these new developments . within few weeks tomonaga and
koba discovered the theory of renormalization @xcite , which removed the major difficulty of uv divergence of quantum electrodynamics ( qed ) and opened a way for new developments of quantum field theories .
i recall that i had difficulty understanding weisskopf s calculation of the electron s self - energy " @xcite quoted in heitler s text , and asked prof .
tomonaga about it . instead of answering my question
, he told me to give two or three seminars on weisskopf s paper .
weisskopf calculated the second - order radiative correction to the self - energy of the electron by exactly following the method of dirac s hole theory .
verifying his calculation step by step with great care , i found that my questions were all based on my misunderstanding and weisskopf s result was in fact correct .
this experience was very helpful when i read dyson s article later @xcite .
the self - energy of an electron is infinitely large and was regarded as one of the difficulties of qed .
what weisskopf found is that , the divergence is not as strong as linear or quadratic but is actually a very mild logarithmic divergence .
if the divergence were stronger than logarithmic , the renormalization method would not have succeeded .
in other words weisskopf s result was a crucial step towards the success of renormalization theory . in the u. s. ,
schwinger @xcite discovered the renormalization of qed and applied it in particular to the calculation of the anomalous magnetic moment of the electron @xcite .
nambu worked on the electron anomalous magnetic moment , too .
unfortunately , he did not publish his work since schwinger published his paper first .
bethe @xcite solved the lamb shift problem , applying the renormalization idea to the non - relativistic version of qed .
in august 1948 , oppenheimer invited yukawa to the institute for advanced study in princeton .
yukawa used this opportunity to transmit by numerous letters the exciting new developments in the u. s. to physicists in japan .
in particular , we were inspired by the activity of young physicists such as feynman and dyson .
at the beginning of 1948 there were two competing methods to deal with the ultraviolet divergence of the electron self - energy : one is tomonaga s renormalization method and the other is the c - meson theory proposed by prof .
sakata of nagoya university @xcite , which cancels the uv divergence due to the radiative correction provided that @xmath0 where @xmath1 is the coupling constant of the c - meson to the electron .
naturally the question was raised as how to decide experimentally which one was correct .
i asked prof .
tomonaga for his opinion .
he told me that z. koba was working on this problem right now , so that it would be a good idea to work with him . in collaboration with koba
, i began to study the difference between the predictions of c - meson theory and renormalization theory .
what we found may be summarized as follows : * the predictions of tomonaga s theory and the form of sakata s theory can be distinguished by examining the elastic scattering of an electron by a fixed potential for an incident electron energy of up to several hundred mev @xcite . * shortly after publication of that result , however , it occurred to us that the renormalization must be applied to the c - meson theory , too . when renormalization is applied to the c - meson theory ,
the two theories become indistinguishable within the experimental precision of the elastic scattering up to several hundred mev ( highest energy then available ) @xcite .
several years later i proved that c - meson theory does not remove the divergence of self energy in the 4th order .
this was the end of c - meson theory @xcite .
after this collaboration koba moved to the yukawa institute of theoretical physics at kyoto university .
thus i did not have any further occasion to work with him .
he later moved to poland , and then to the bohr institute in copenhagen .
unfortunately , he passed away prematurely in 1973 .
bloch and nordsieck treated the infrared problem by a non - perturbative method assuming that all infrared photons have zero energy @xcite .
however , their method can not be applied directly to the emission and absorption of real photons of non - zero energies . in order to deal with this problem
i developed a method to handle the infrared problem within the perturbative framework of feynman - dyson theory @xcite .
my insight was not deep enough , however , and i overlooked a subtlety in the infrared problem .
this became the source of an error in the calculation of radiative correction to the @xmath2-e decay , as is described in detail in sec .
in 1948 i was trying to see whether the renormalization theory of tomonaga could be extended to the interaction of a charged vector meson with the electromagnetic field .
i found out that nambu was also interested in the same problem .
thus we started working together .
while our work was in progress , we received from prof .
yukawa a brand new preprint of dyson s paper @xcite , which proved the equivalence of the tomonaga - schwinger theory and feynman s theory , in spite of their very different appearance .
( actually at that time we did not know what feynman s theory looked like , since feynman did not publish his theory until a year later @xcite .
thus we had to guess what feynman s theory looked like by studying dyson s paper . )
the feynman - dyson theory opened up a new method of computation , that was much more transparent and systematic than the perturbation method based on the non - covariant formulation of heisenberg and pauli @xcite . recognizing the power of feynman - dyson theory , nambu and i quickly reorganized our theory a l dyson .
also , instead of the conventional formalism of vector mesons , we adopted the duffin - kemmer equation @xcite as the starting point to take advantage of its close structural similarity with the dirac equation . in april of 1949 nambu
, together with s. hayakawa , y. yamaguchi , and k.nishijima , moved to the newly established physics department of osaka city university .
thus my collaboration with nambu came to an end . in spite of this exodus
the particle physics group at todai remained very active with younger members such as h. miyazawa and m. koshiba . in the summer of 1949
both tomonaga and kodaira were invited to the institute for advanced study in princeton .
tomonaga returned to japan next year so that the tomonaga seminar continued uninterrupted during his absence .
the kodaira seminar was terminated , however , since he stayed in the u. s. for many years .
nambu and i wrote up papers on our work @xcite , and sent them to were invited to the institute for advanced study for the academic year 1952 - 1953 ( extended to 1954 ) .
this opened up a possibility to collaborate with nambu again .
i left japan from nagasaki in early august 1952 by a cargo boat , and arrived at seattle about 17 days later .
then i moved to berkeley by greyhound bus , and visited berkeley , caltech , and ucla while waiting for nambu s arrival .
nambu had left from yokohama by a cargo boat and arrived at san francisco about two weeks later .
we took a train named
california zephyr " to cross the rockies , stopping at denver to visit the cosmic - ray laboratory at mt .
evans , elevation 14,000 ft .
the next stop was at chicago to visit the university of chicago and talk with the people who found the 3 - 3 resonance of pion - nucleon scattering with their proton cyclotron .
then we stopped at rochester to visit r. marshak at the university of rochester , and finally arrived at princeton in late august . at the institute for advanced study
nambu and i were given desks in the same room on the second floor of the physics building .
this room was farthest from the office occupied by lee and yang .
but when they started arguing we could hear them very loudly from our office .
we tried to develop a general theory of hartree fields , a collective description of many - particle systems @xcite . at the end of the spring term of 1954 nambu
was invited to the university of chicago , where he used our paper @xcite as the starting point for his theory of superconductivity , which was developed further to the theory of spontaneous symmetry breaking . in may of 1954
i became puzzled by the sizable discrepancy between theory and measurement of the ionization energy of the he atom , and began examining hylleraas well - known calculation .
this work continued for several years at columbia ( 1954 - 1955 ) and cornell ( 1955 - ) .
at last i found that the discrepancy was caused by a subtle error in hylleraas s numerical work @xcite , which did not strictly follow the rules of the variational method .
i was told that fermi was intrigued by this problem and conjectured that the he atom might have some new force acting between electron and nucleus . to my disappointment
my calculation disproved his conjecture .
i learned a lot from nambu in our collaboration .
but we pursued different paths afterwards .
the year 1956 was a very exciting year .
after an exhaustive examination of the so - called @xmath4-@xmath5 puzzle , t. d. lee and c. n. yang suggested that the parity symmetry is broken in the weak interaction and proposed ways to test it experimentally @xcite .
experimental verification followed soon afterwards @xcite .
the two - component neutrino theory became the favored theory .
a detailed comparison of the two - component theory with experiment would require inclusion of radiative corrections .
thus i started working on the radiative correction to the parity - violating part of the @xmath3 decay .
it turned out that albert sirlin had worked on the radiative correction to the parity - conserving muon decay before coming to cornell as a graduate student @xcite . in collaboration with him
i calculated the radiative correction to the parity - violating part of the muon decay . as expected , the infrared divergent terms of real and virtual emissions of photons canceled out so that the radiative correction to the decay lifetime of muon was finite , although it was very large : @xcite @xmath6 where @xmath7 . we presented our result at the rochester conference in the spring of 1957 .
then , one day in 1957 or 1958 , we received a preprint from s. berman stating that the paper @xcite was wrong because of an inconsistent treatment of real and virtual emissions of photons .
( see next paragraph for more details . ) correcting this error he obtained an answer that is linear in @xmath8 . when we read his preprint , however , we suspected immediately that his result must also be wrong since the mechanism to remove the @xmath9 term is likely to remove the @xmath8 term , too .
alberto and i worked hard for a week or two and found that our intuition was in fact correct : no term proportional to @xmath8 or @xmath9 is present in the muon life - time @xcite .
berman agreed with us and revised his paper accordingly @xcite .
the error in @xcite was caused by the incorrect treatment of real photons .
recall that , in a covariant calculation , the virtual photon is treated as a vector meson of mass @xmath10 with the understanding that @xmath11 in the physical limit . to be consistent with this , the real photon
must also be regarded as a vector meson of mass @xmath10 .
this means in particular that the real photon has longitudinal and time - like polarizations in addition to two transverse polarizations .
our mistake was to consider only the contribution of transverse polarizations .
the contribution of longitudinal and time - like polarizations does not vanish if the photon has infrared divergence , even in the limit @xmath11 .
as a matter of fact this extra contribution has an @xmath8 dependence that cancels exactly the @xmath8 dependence from transverse photons .
feynman visited cornell for three months in the fall semester of 1958 .
he explained to me how he and berman made exactly the same mistake .
feynman had asked berman to check our calculation for his thesis work .
actually , feynman told me that he himself was doing this calculation independently in order to check berman s work . at the end they compared notes and were satisfied that their results agreed . when confronted with our new result , they checked notes once again and found that they made the same mistake in copying the bottom equation of a page to the top of next page .
feynman was so disturbed by this mistake that he told me how sorry he was more than a few times while at cornell .
see @xcite for more detailed discussion of this section .
feynman brought to cornell a new preprint by berman on the one - loop radiative correction to @xmath12 decay .
berman s result was of the form @xcite @xmath14 where @xmath15 is the uncorrected ratio of @xmath16 and @xmath17 decay rates in the v - a theory .
the value of this @xmath18 looked strange since @xmath19 diverges for @xmath20 .
feynman and i were so puzzled by this result that we decided to check it with a fresh calculation .
for the next two months we worked hard , totally independent of each other , except that we agreed to start from the same effective lagrangian @xmath21 where @xmath22 represents the physical mass of muon or electron and @xmath23 . the most time - consuming part of the calculation is integration over the final states consisting of an electron , a neutrino , and a photon .
following the conventional method it took for me about two months to evaluate this integral .
the calculation occupied over 30 sheets of papers .
although feynman spent about the same amount of time , he was looking for a more efficient method and found one .
his approach is to maintain lorentz invariance throughout the intermediate steps of calculation , which allowed him to choose a frame of reference in which angular integration becomes trivial . to my great amazement
this enabled him to write the complete derivation in just two sheets of paper .
when we finished the calculation and compared the notes we found that we agreed with each other ; in particular , our results did not contain the @xmath24 term .
we thought for a while that berman s calculation was wrong .
but , after scrutinizing our calculation very closely , we realized that it was our effective lagrangian ( [ eq5 ] ) that was not justifiable .
berman started from an effective lagrangian containing a derivative coupling : @xmath25 this term is to be evaluated with respect to the states which include radiative corrections up to the order @xmath26 . to compare ( [ eq6 ] ) with ( [ eq5 ] )
turn it into a non - derivative form @xmath27 by integration by part , and using the equation of motion @xmath28 where @xmath29 and @xmath30 are * bare * masses , @xmath31 is the physical mass , and @xmath32 is the counterterm to the radiative correction .
the relation @xmath33 was used in eq .
( [ eq7 ] ) .
it is seen that when the radiative correction ( to order @xmath26 ) is included in the same way as in the ordinary qed , eq .
( [ eq7 ] ) depends on the bare mass @xmath29 instead of the physical mass .
the appearance of bare mass in a similar context had been noticed by ruderman @xcite . on the other hand
, the lagrangian ( [ eq5 ] ) was obtained from the equivalent lagrangian ( [ eq7 ] ) assuming that @xmath34 and @xmath29 is the physical lepton mass .
the discussion above shows that it was this assumption that was not justifiable .
( [ eq7 ] ) shows that berman s lagrangian may be written as ( assuming @xmath34 ) @xmath35 namely , our result can be turned into berman s result by simply replacing the physical mass in ( [ eq5 ] ) by the bare mass . at first sight
it would look strange that the @xmath36 term appears in eq .
( [ eq3 ] ) in view of absence of such a term in the total probability according to the mass singularity theorem @xcite .
actually eq .
( [ eq3 ] ) is consistent with the theorem since it is multiplied by @xmath37 as is seen from eq .
( [ eq4 ] ) and thus vanishes in the limit @xmath38 .
although feynman was fully involved in this calculation , he refused to put his name on the paper @xcite .
the calculation in @xcite and @xcite relied on implicit untested assumption that the effective lagrangians ( [ eq5 ] ) and ( [ eq6 ] ) are viable and that the uv cut - off is common for both @xmath16 and @xmath39 decays
. justification of these assumptions had to wait for the emergence of the standard model calculation @xcite , which favors berman s result .
taking the hadronic effect into account @xcite , the value of @xmath18 becomes @xmath40 this is in good agreement with the experiments @xcite : @xmath41 typos in some equations in @xcite are corrected in this article .
this work is supported in part by the u. s. national science foundation under the grant no .
nsf - phy-1316222 . | this article gives a sketch of teachers and colleagues who have had strong influence on my becoming a particle theorist . |
comet nuclei formed beyond the protoplanetary disk frost line ( heliocentric distances , @xmath4 5 au ; * ? ? ?
* ) , among the giant planets and were scattered into the kuiper belt and beyond into the oort cloud ( oc ) . since their formation
, the interiors and surfaces of most comets have remained at temperatures below 140 k while in `` cold storage '' in the kuiper belt or the oc @xcite .
moreover , most nucleus surfaces have remained below 400 k even during perihelion passage . at such low temperatures ,
dust mineralogy remains stable and each comet nucleus retains a record of the minerals , ices , and volatiles extant in the comet agglomeration zones in the early solar system .
furthermore , comet nuclei may retain their primordial compositional inhomogeneities so that different topographic regions would have different compositions ( e.g. , * ? ? ?
* ) , leading to variations in volatile production rates as a function of nucleus rotation .
nucleus heterogeneities are apparent in the fly - by imagery of ecliptic comet 9p / tempel 1
. regions of distinct topography @xcite , and spatially distinct sites of water and co@xmath8 release @xcite are evident , as well as heterogeneities in surface and subsurface composition @xcite .
there are two general dynamical families of comets , classified by derived orbital elements ( i.e. , tisserand parameter , @xmath14 ) based on current observations .
nearly - isotropic comets ( nics ; @xmath15 ) have orbits that are approximately uniformly distributed in inclination , and are derived from the oc .
ecliptic comets ( ec ) have orbits that are confined to inclinations near the ecliptic plane , @xmath16 , and originate in the kuiper belt . resulting from frequent perihelion passage over the 4.0 4.5 gyr period since their formation , ec
comets have become noticeably less active than oc comets , characterized by lower gas and dust production rates .
multi - epoch spectral energy distributions ( seds ) of ecs , from which comae dust properties can be constrained , are needed to assess the possible interrelationships between their reduced activity levels and dust properties .
the study of the physical properties of cometary nuclei and comae both are equally important to our understanding of the outer solar system environment during the era of icy planetesimal formation , and complement efforts to discern conditions extant in early protoplanetary disks during the epoch of planetesimal formation .
we present new optical and contemporaneous _ spitzer
_ observations of comet 21p / giacobini - zinner obtained during its 2005 apparition obtained as part of a larger survey of both ecs and nics ( e.g. , * ? ? ? * ) .
ground - based optical observations enabled us to obtain precise optical photometry to asses variations in dust productivity with heliocentric distance and to study the the comet s near - nucleus structures , including its jets and coma . _
spitzer _
images at mid - infrared wavelengths enable the study of the spatial distributions of volatiles and dust in the coma , as well as facilitating investigation of comet trail properties .
we describe our observations in [ obs ] , and discuss analysis of our optical and infrared ( ir ) observations in [ disc ] , while our conclusions are summarized in [ concl ] .
comet 21p / giacobini - zinner was discovered by the french astronomer michel giacobini in 1900 and rediscovered two apparitions later , in 1913 , by the german astronomer ernst zinner .
21p , the parent body of the draconids ( also know as the giacobinids ) meteor shower @xcite , is a jupiter - family ( i.e. , ec with a short - period ) comet with an orbital period of 6.61 years and an aphelion distance just exterior to jupiter s orbit . in 1933
a spectacular giacobini meteor storm was visible across europe .
the 1946 apparition of 21p was especially noteworthy as the comet passed only 0.26 au from earth in late september with a visible magnitude @xmath17 .
in early october , an unexpected outburst of activity caused the comet to brighten to sixth magnitude .
every alternate apparition of comet 21p presents favorable observing geometries for observers on earth , with integrated coma visual magnitudes as high as 7 .
21p was also the first comet visited by a spacecraft when the international cometary explorer ( ice ) flew past at distance of 7,800 km on 1985 september 11 @xcite .
optical images of comet 21p / giacobini - zinner were obtained on the university of hawaii ( uh ) 2.2 m telescope on mauna kea , using a tektronix @xmath18 ccd camera at the @xmath19 focus of the telescope ( image scale of 0.219 pixel@xmath10 ) through a kron - cousins filters set ( @xmath20 : @xmath21 , @xmath22 = 836 ; @xmath23 : @xmath24 , @xmath22 = 1245 ; @xmath25 : @xmath26 , @xmath22 = 1888 ) , during multiple observing runs from 2004 june , 2005 october and december , and 2006 march .
the camera read - noise was 6.0 @xmath27 with a gain of 1.74 @xmath27 per adu .
non - sidereal guiding at the cometary rates of motion , derived from the jpl / horizons ephemeris , was used during all image observations .
specific observational details , including orbital geometry for 21p , filters , number of exposures , individual exposure times , and sky conditions are summarized in table [ table : tb_obslog ] .
the optical ccd frames were reduced with standard iraf routines .
all the images were reduced using flat field images taken during the evening and morning twilight sky and cleaned of bad pixels and cosmic rays .
the frames were calibrated with the standard stars of @xcite , which were observed on each photometric night .
observations of typically 20 standard stars were obtained over a range of air - masses , and with a wide dispersion of color to fit for both extinction and color terms .
apparent magnitudes of twenty to thirty stars field stars of equal or greater brightness to the comet on each frame were measured in order to do relative photometry . after correcting the measured magnitudes for extinction
, we used the deviations of the field star magnitudes in each frame from their nightly average values to correct for frame - to - frame extinction in the comet s measured signal .
figure [ fig : opt_4panel ] shows selected individual optical images for each ground - based observing run .
we observed comet 21p with the infrared array camera ( irac ; * ? ? ? * ) and the multiband imaging photometer for _ spitzer _ ( mips ; * ? ? ?
* ) on the _ spitzer space telescope _
@xcite , post - perihelion as summarized in table [ table : tb_obslog ] .
the mips astronomical observation request ( aor ) used the 24 array in mapping mode to assess the existence of the comet s dust trail ( aor key , @xmath28 map ) . a duplicate mips observation was taken 25 hr later to observe the sky background after the comet had moved out of the frame ( aor key ) . the irac aor utilized the 4.5 and 8.0 arrays to obtain complementary deep photometry of the coma ( aor key , @xmath29 map ) .
the irac and mips observations were calibrated with _ spitzer _
pipelines s13.2.0 and s13.0.1 , respectively .
the images were mosaicked in the rest frame of the comet with the mopex software @xcite at the native irac and mips pixel scales ( 1.22 pixel@xmath10 for irac , 2.5 pixel@xmath10 for mips 24 ) .
the mips background observation was mosaicked with the same size , scale , and orientation as the primary observation , then subtracted from the primary . before mosaicking ,
all images were masked to remove cosmic rays and bad pixels .
no additional corrections were necessary .
our irac image of comet 21p was obtained at a post - perihelion distance of @xmath30 au .
analysis of the secular optical light curves of 21p by @xcite suggests that the nucleus turn - on time ( marking the onset of steady coma activity ) occurs pre - perihelion near 3.7 au , whereas onset of nucleus turn - off may not occur until 21p reaches 5.4 au .
indeed @xcite argue that 21p is active at 4.5 au .
the irac 4.5 image , fig .
[ fig:21p - irac ] , shows an extended coma surrounding the nucleus .
discussion of this extended emission is presented in [ coco2 ] .
the _ spitzer _
mips 24 image is presented in fig .
[ fig:21p - mips ] .
the photometry derived from our optical images enable investigation of cometary activity with heliocentric distance , estimation of dust production rates ( @xmath31 ; * ? ? ? * ) , determination of the coma color , and construction of rotational light curves .
the stellar and comet fluxes were extracted from the optical ccd frames using the iraf photometry routine `` phot '' with a circular aperture .
this routine automatically finds the centroid of the image within the user - specified photometry aperture , with the sky background determined in an annulus lying immediately outside the photometry aperture ( for stellar images ) , or selected from an average of many sky positions outside the extent of any coma .
photometric apertures between 1 5 were used for the comet photometry with the average sky background determined using sky annuli with an inner radii of 8.0 10 , with widths of 4.0 5.0 . averaging many sky background positions adjacent to
the comet also enabled rejection of any bad pixels or field stars found in individual sky annuli circumscribing the aperture used to determine the comet surface brightness .
the smallest possible aperture that minimized contamination from extended surface brightness of the coma while including most of the flux from the nucleus ( with respect to the full - width half - maximum seeing determined from stellar point sources on the frame ) was 3 .
a 4 aperture was also used for the field stars used for differential photometry . after correcting the measured magnitudes for extinction
, we used the deviations of the field stars from their nightly average magnitudes in each frame to correct for frame - to - frame extinction variations in the comet s measured signal .
for the non - photometric data ( table [ table : tb_obslog ] ) from the nights of 2004 june 21 and 22 , and 2006 march 05 , the comet fields were re - imaged under photometric conditions on 2005 may 18 and 2006 june 03 for `` boot - strap '' calibration .
multiple ( @xmath32 ) field stars were measured on both the calibration images and the non - photometric images and differential photometry was used to compute the photometrically - calibrated brightness of the comet .
this technique works well for up to @xmath33 mag of extinction .
the most direct indicator of nucleus activity is the appearance of visible coma around a comet nucleus .
our first observations of 21p ( fig .
[ fig : opt_4panel]a ) were obtained on 2004 june ( @xmath34 days before perihelion ) when the comet was at a large heliocentric distance , @xmath35 au . no coma was evident on individual images , nor on the composite image created by mosaicking the 2 to 3 images obtained on each night to increase the signal - to - noise ratio of any diffuse coma emission .
absence of a detectable coma @xmath34 days before perihelion is consistent with 1991 apparition observations of @xcite who found no coma at @xmath36 days before perihelion , yet a clearly detectable coma at @xmath37 days .
our observations indicate that the heliocentric turn - on point for 21p nucleus activity this apparition was similar to the behavior observed in the 1991 .
the average @xmath23-band magnitude measured for 21p on a given night are summarized in table [ table : tb_mags ] , column [ 3 ] , as is the reduced magnitude , column [ 6 ] .
the reduced magnitude , @xmath38 , i.e. , the observed magnitude normalized to unit heliocentric distance , @xmath4 , and geocentric distance , @xmath39 , for zero phase angle , @xmath40 , was derived from the relationship for active comets @xmath41 where @xmath42 is the @xmath23-band photometric magnitude measured in a 3 circular aperture , @xmath43 the linear phase coefficient = 0.035 mag deg@xmath10 , and @xmath44 and @xmath45 are constants , taken to be unity and 2 respectively .
nominally , values for these two latter constants ( @xmath44 being dependent on dust grain production , coma shape and size ) are derived from modeling a given comet at a variety of heliocentric distances if there are sufficient observational data .
equation [ eq : redmags ] also is appropriate for expressing the brightness of an active comet with a spherical coma around the nucleus ( measured in a circular aperture , as described by * ? ? ?
* ) , as opposed to the relation used by @xcite which is the proscription for the brightness of bare nuclei ( non - active stage ) .
computation of the reduced magnitude is a standard method that enables of photometric data of an individual comet obtained at a variety of heliocentric distances ( different activity level ) to be intercompared . for 21p
, we obtained observations of the comet when there was no visible coma , when the comet exhibited a small coma around the nucleus , and when the comet was highly active displaying a significant , spatially extended coma with an elongated tail .
to properly intercompare variations in the integrated aperture surface brightness , the @xmath23-band photometric magnitudes were determined by measuring the comet s brightness in a small 3 circular aperture centered on the nucleus .
use of this small nucleo - centric aperture minimized the coma contribution to the measured flux density , and allowed estimation of the nucleus brightness consistent with the uniform azimuthally - averaged surface brightness profile .
our observations of 21p span a range of heliocentric distance , both pre- and post - perihelion ( table [ table : tb_obslog ] ) , thus we can examine trends in @xmath38 with @xmath4 .
figure [ fig : rm_rh ] shows the variation in the reduced magnitude as a function of @xmath4 derived from @xmath23-band photometry , where the dotted horizontal lines represent the likely brightness range for a bare nucleus , calculated from our photometry data when no coma was seen .
evidently , cometary activity commenced near @xmath0 au this apparition of 21p , and the majority of our optical observations and all of our _ spitzer _
images were obtained while the comet was in an active state .
average coma color differences ( @xmath46 , @xmath47 ) are also listed in table [ table : tb_mags ] for dates when multi - wavelength observations were conducted .
the average colors of comet 21p , @xmath48 , @xmath49 are redder than the respective solar colors , 0.36 and 0.28 , in the krons - cousins system , transforming the johnson solar colors @xcite using the relation of @xcite .
band emission from gaseous molecular species , such as c@xmath8 ( @xmath50 ) , nh@xmath8 ( @xmath51 ) , and cn ( @xmath52 ) , or atomic o i ( @xmath53 @xmath54d ) potentially could contribute significantly to the flux density of comet comae observed within our @xmath55 broadband filters .
however , optical spectra of @xcite indicates that band and line emission from these species is relatively weak in the coma of 21p .
21p also lies along the @xmath56 versus @xmath57 color - color trend line of jupiter - family comet nuclei attributed by @xcite to albedo variation ( in [ oir - albedo ] we derive an dust albedo estimate for 21p ) , although we can not quote a precise value for the nucleus color of 21p due to coma contamination .
broad - band colors of cometary nuclei are related to surface properties of the nucleus @xcite and provide a metric to compare cometary nuclei to other small solar system bodies and to assess which parent body populations give rise to jupiter - family comets .
the bluer color of 21p and other jupiter - family comets , as compared to the mean color of kuiper - belt object / trans - neptunian object populations , @xmath58 @xcite , suggests that cometary activity and surface processing in the inner solar system have altered the primitive surfaces of these nuclei agglomerated from materials in the kuiper belt ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . the quantity @xmath31 of a comet , a proxy describing the dust production rate @xcite , is often used to describe nucleus activity vigor , and to distinguish between dusty and inactive comet taxonomies .
@xmath31 is given by @xmath59 where @xmath60 is the wavelength - dependent albedo of dust particles , @xmath61 is the filling factor of grains ( total cross - section ) , @xmath62 is the linear radius of the aperture at the comet , @xmath4 and @xmath39 are heliocentric and geocentric distance ( au ) , @xmath63 is the angular diameter of the field of view ( in arcseconds ) , @xmath64 is the observed cometary flux and @xmath65 is the solar flux .
the @xmath31 parameter is typically expressed in centimeters . by providing a metric to quantify a comet s level of dust output ( large @xmath31 values indicate higher activity )
, one can ascertain the dust contribution to a set of photometric measurements and assess the heliocentric dependence of dust production rates in a given comet .
@xmath23-band values of @xmath31 for 21p derived from our photometry are summarized in column [ 7 ] of table [ table : tb_mags ] and are plotted as a function of @xmath4 in fig .
[ fig : afrho_rh ] .
peak dust production rates of 131 @xmath2 occurred just post - perihelion .
we fit the post - perihelion @xmath31 values with a function of the form @xmath66 using the least absolute deviation method @xcite .
our best - fit values are @xmath67 cm , @xmath68 .
the pre - perihelion observations do not span a large enough range of @xmath4 to derive a slope .
our post - perihelion @xmath31 slope ( @xmath69 ) is comparable to the pre - perihelion value of @xmath70 derived by @xcite based on observations obtained when 21p was at a @xmath71 au .
however , @xcite find a significantly larger value ( @xmath72 ) when the 21p is interior to 1.86 au post - perihelion suggesting that the comet s activity peaks just after perihelion passage .
we measured the flux of comet 21p in the mips and irac images using the aper.pro procedure of the idl astronomy user s library @xcite .
the 2005 december irac , mips , and optical photometry are all measured at different heliocentric distances ( @xmath73 au ) , observing geometries ( @xmath74 au ) , and spatial resolutions .
however , differences in observational circumstances are small such that simple scaling laws can be applied to the photometry to create a composite 0.524 sed of comet 21p . for reasons discussed below ,
we elect to correct the ir photometry ( [ obs - ir ] ) to match the observing conditions of the optical measurements ( [ obs - optuh ] ) .
first , we correct the ir photometry for the limited resolution of the _ spitzer _ telescope , as compared to the optical data .
the irac and mips images are diffraction limited and stellar sources have full - width half maximum @xmath75 , 2.0 , and 6 at 4.5 , 8.0 , and 24 @xcite .
the 24 point - spread - function ( psf ) is too broad to compare to the 3 @xmath23-band photometry in table [ table : tb_mags ] ( column [ 3 ] ) , and the irac psf is marginally appropriate . to derive the ir photometry in an aperture size of 3741 km ( corresponding to the optical photometry , 3 at 1.72 au ) , we measure the comet s ir coma with increasingly larger apertures and use a curve - of - growth analysis to estimate the fluxes at apertures comparable to or smaller than the resolution of the _ spitzer _ instruments .
we examined the logarithmic profile of the optical data to verify that the coma slope at moderate sized apertures ( @xmath76 ) extends into the inner - coma ( @xmath77 ) .
the photometry at 10 and the coma logarithmic slopes are presented in table [ table : ir - photom ] .
next we correct the irac and mips photometry for the differences in observer - comet distance .
the correction is simply @xmath78 : 0.93 for irac , and 1.18 for mips . finally , we scale the irac and mips photometry to match the heliocentric distance ( @xmath79 au ) of the optical observations .
there are two effects that depend on @xmath4 : the temperature of the dust , and the dust production of the comet .
to correct for changes in the dust production , we scale the photometry using the @xmath31 heliocentric distance trend derived for 21p , @xmath80 ( [ dust_afr ] ) .
the mips photometry is scaled by a factor of 0.86 and the irac photometry scaled by 1.08 . to correct for the dust temperature , we scale the photometry by @xmath81 , where @xmath82 is the planck function , and
@xmath83 and @xmath84 are the dust temperatures at the optical and ir heliocentric distances , which we set to the blackbody temperature , @xmath85 k. the scale factors are : 1.37 for @xmath86 , 1.20 for @xmath87 , and 0.88 for @xmath88 for the irac and mips data respectively .
our measured fluxes , azimuthally averaged surface brightness profile logarithmic slopes , and the corrected photometry is presented in table [ table : ir - photom ] .
the total sed of the comet is presented in fig .
[ fig : sed ] .
the optical magnitudes of table [ table : tb_mags ] were converted to flux units using the zero magnitude flux densities of 3600 jy ( @xmath20-band ) , 3090 jy ( @xmath23-band ) , and 2550 jy ( @xmath25-band ) .
we fit the sed with a scaled , reddened solar spectrum to represent the scattered light at the optical wavelengths , plus a scaled planck function , to represent the thermal emission from the ir wavelengths .
we exclude the gas contaminated 4.5 data point ( [ coco2 ] ) from the fit .
solar spectrum reddening ( the spectral slope ) is typically expressed as percent per 0.1 ( e.g. , * ? ? ?
a negative spectral slope represents a bluing of the spectrum .
our observations constrain the scattered light reddening between the @xmath20- and @xmath23-bands , and the @xmath23- and @xmath25-bands . following @xcite
, we express the reddening with the linear slope @xmath89 expressed as % per 0.1 . at optical wavelengths ,
our computed slopes for comet 21p ( see table [ table : tb_mags ] ) are @xmath90% per 0.1 , and @xmath91% per 0.1 .
the former slope of the reflectivity gradient is in good agreement with the average value of @xmath92% per 0.1 derived by @xcite for comets , while the latter @xmath93 slope value agrees with the general trend of bluer scattering at longer wavelengths ( see fig . 4 of * ?
indeed , @xcite observed 21p in the near - ir and found @xmath94% per 0.1 and @xmath95% per 0.1 .
the fit has no degrees of freedom , so we do not present errors or a @xmath96 analysis .
the best - fit dust temperature is 206 k ( @xmath97% warmer than a blackbody ) .
we derive a bolometric albedo for the dust of comet 21p using eqs . 6 and 7 from @xcite .
the bolometric albedo of 21p is @xmath13 , slightly lower than values derived for other comets at the same phase angle ( @xmath98 , * ? ? ?
* ) , but not unusual .
the size of the equivalent spherical nucleus , @xmath99 , can be estimated from the apparent mean magnitude of comet 21p measured at heliocentric distances when the comet is relatively inactive .
pre - perihelion , @xmath100 ( table [ table : tb_mags ] ) and @xmath99 can be derived from the standard relation @xmath101 } \label{eq : nucsiz}\ ] ] where @xmath102 is the geometric albedo , @xmath4 ( au ) the heliocentric distance , @xmath39 ( au ) the geocentric distance , @xmath103 the @xmath23-band magnitude of the sun @xcite , @xmath43 ( mag deg@xmath10 ) the linear phase coefficient , and @xmath40 is the phase angle . for 21p , we adopt value of @xmath104 @xcite and @xmath105 mag deg@xmath10 which is generally used for most comets with unknown @xmath43 ; however , @xcite argue that a @xmath106 may be more appropriate based on the observed phase coefficient for 10 comets .
we find @xmath107 km , commensurate with other estimates nucleus size ranging from 1.0 km derived by @xcite or @xmath108 km value of derived from the snapshot survey of @xcite , assuming @xmath104 . in principle
, variations in the photometric brightness derived from the @xmath23-band individual images obtained over @xmath109 hr intervals during the period of 2005 december 21 through 22 ( table [ table : tb_mags ] ) can be used to generate a composite light curve yielding estimates of nucleus rotational periods . however , our composite light curve of 21p at this epoch , @xmath79 au exhibits little periodic variation and we are unable to determine a rotational period .
21p was also very active during this epoch , with a significant coma contribution which could mask subtle variations ( @xmath110 mag ) in nucleus flux .
a rotational period of @xmath111 hr for 21p has been reported by @xcite .
volatiles are frozen as ices or trapped as gases in amorphous water ice in the nuclei of comets @xcite .
cometary activity occurs when gases are released through sublimation or through the exoergic crystallization of amorphous water ice .
water sublimation primarily drives vigorous activity at @xmath112 au @xcite , but other volatiles can still be important drivers . at 1.5 au
, @xcite found that co@xmath8 gas was correlated with comet 9p / tempel s southern hemisphere the location of the strongest dust jet @xcite yet water vapor was more strongly associated with the northern hemisphere .
some molecular species ( e.g. , co ) are released not only from the nucleus ( native source ) , but also from within the coma ( `` distributed '' or `` extended '' source ) . in
the extensively studied , bright oc comet c/1995 o1 ( hale - bopp ) , there was evidence for distributed co sources , possibly arising from icy grains in the coma @xcite at large @xmath4 , whereas at @xmath113 au the strong distributed sources probably arise from the desorption of an organic grain component @xcite .
although icy grains may be the distributed source of the co @xcite , pure - co ice grains do not survive transport into the coma because of their low sublimation temperatures , 20100 k @xcite .
the irac 4.5 bandpass encompasses the co@xmath8 @xmath114-band and the co 10 fundamental band at 4.26 and 4.67 respectively .
the co and co@xmath8 bands are observed as the dominate components at 34 in excess of the dust continuum emission in _ iso _ spectra of comets hale - bopp and 103p / hartley @xcite and in the ambient coma of 9p / tempel @xcite . after removing the dust continuum ,
the irac images can be used to map the co and co@xmath8 spatial distributions in the comet coma . in principle
, the emission from the two molecules can be decoupled when we consider that co and co@xmath8 have different lifetimes in the coma and therefore produce different radial distributions . with a measured radial distribution
, we can also test for evidence of an extended source of co or co@xmath8 .
emission from other prominent volatiles , such as water @xcite , lie outside the bandpass of the irac 4.5 filter .
the slope of the co @xmath115 co@xmath8 emission in the _ spitzer _ images at @xmath116 pixels ( corresponding to projected cometocentric distances of @xmath117 km to @xmath118 km ) can be determined by subtracting a scaled irac 8.0 image from the irac 4.5 image ( fig .
[ fig:21p - irac ] ) .
the irac 8.0 image is dominated by thermal emission from dust , whereas the 4.5 image is a combination of emission from gas , dust thermal emission , and sunlight scattered by dust .
the color of sunlight scattered by comet dust varies with wavelength @xcite , and we can not constrain the spectral slope of the scattered light at 4.5 with our data . rather than estimating the 4.5 dust flux from our best - fit sed ( [ oir - albedo ] ) , we subtract a scaled 8.0 image from the 4.5 image to yield an image of the gas emission ( fig . [ fig:21p - irac - gas ] ) .
our best scale factor is @xmath119 , which we derived from the ratio of the 4.5 and 8.0 images .
we measured the ratio along a 3 pixel wide rectangle positioned on the dust tail from 5 to 40 pixels from the nucleus .
the error in the scale factor is the standard deviation of the pixels in the rectangle .
we fit the residual surface brightness profile ( emission from the gas ) in 24 azimuthal steps .
the mean surface brightness logarithmic slope is @xmath120 ( median @xmath121 , error in the mean @xmath122 ) .
the derived gas profile is consistent with a long - lived species ejected in a constant outflow from the nucleus , i.e , it shows no evidence for an extended source .
the lifetimes of co and co@xmath8 at 2.4 au are co @xmath123 days and co@xmath124 days @xcite .
for an outflow velocity of 1.0 km s@xmath10 , the characteristic length scales are @xmath125 km ( co ) and @xmath126 km ( co@xmath8 ) . in our dust
subtracted image , there appears to be emission out to 60 pixels from the nucleus ( @xmath127 km ) , but background stars make it difficult to measure the logarithmic slope beyond 20 pixels .
both species are long lived on scale lengths @xmath128 pixels even if the outflow velocity is 0.1 km s@xmath10 . from our analysis , both co and co@xmath8 are equally viable candidates for the 4.5 excess in the difference image , fig .
[ fig:21p - irac - gas ] . for comparison , in comet hale - bopp @xcite
, co@xmath8 was the dominant emission species at 2.9 au ( comet 21p was observed at 2.4 au ) . we can measure the intensity of the gaseous emission and derive upper - limits to the co@xmath8 and co contributions .
we have attempted to remove the dust from the 4.5 image ; however , emission from the nucleus has not been removed .
although we derived an effective radius from our optical observations ( [ com - lcs ] ) , the orientation , shape , and temperature distribution of the nucleus is unknown at the epoch of the irac images . rather than measuring the flux inside a circular aperture centered on the nucleus ,
we avoid contamination from the nucleus by measuring the gas coma inside an annulus with an inner - radius of 6 pixels , and an outer - radius of 8 pixels .
we convert the measured annular flux density , @xmath129 erg s@xmath10 @xmath130 hz@xmath10 , into line emission fluxes following the irac prescriptions discussed by the @xcite , where we perform a linear interpolation of values given in the irac spectral response tables ] near the wavelength position of our lines , assuming the residual emission within the effective bandpass of the irac 4.5 filter is either entirely from either co@xmath8 ( @xmath131 ) or co ( @xmath132 ) .
we find @xmath133 erg s@xmath10 @xmath130 , and @xmath134 erg s@xmath10 @xmath135 .
the errors include the irac absolute calibration uncertainty of 3% , and a dust subtraction uncertainty of 11% .
the gaseous coma has an azimuthally average profile surface brightness slope of @xmath136 ( see table [ table : ir - photom ] ) , thus we can convert the flux at 68 pixels to a flux inside a circular aperture of projected radius 7 pixels , @xmath137 , where @xmath138 pixels , @xmath139 is the flux measured in the annulus , and @xmath60 is the area of the annulus in square pixels ( e.g. , * ? ? ?
* ) . the total emission inside this 7 pixel aperture yields an average column density , @xmath140 , in @xmath130 , assuming optically thin emission from the expression @xmath141 where @xmath39 is the observer - comet distance in cm , @xmath142 ( ) is the wavelength of the emission , @xmath143 is planck s constant ( erg s )
, @xmath144 is the speed of light ( s@xmath10 ) , @xmath4 is the heliocentric distance in au , @xmath145 is the solar fluorescence @xmath146-factor at 1 au in units of s@xmath10 , and @xmath62 is the projected radius of the aperture , here in units of cm . we have adopted @xmath146-factors given by @xcite of @xmath147 s@xmath10 for the co@xmath8 @xmath148 band , and @xmath149 s@xmath10 for the co 10 band , noting that other cited literature values for these @xmath146-factors differ by @xmath15010% . from eqn .
[ eqn : moleden ] , we derive average column densities of : @xmath151 @xmath130 , if the emission is solely from co@xmath8 and @xmath152 @xmath130 , if the emission is solely from co. @xcite measured oh radio emission profiles in spectra of several comets as a function of @xmath4 , including comet 21p , to derive the expansion velocity of that species . for most low production rate comets ( @xmath153 )
the expansion velocities are constant for @xmath154 au .
radio measurements of the oh production rates of 21p during the 1985 apparition indicate that @xmath155 molecules s@xmath10 , with little change in the derived expansion velocity , @xmath156 km s@xmath10 ( see table 4 ; * ? ? ?
* ) over a range of @xmath4 .
given the lack of a strong @xmath4 dependency of @xmath157 at distance greater than 1.4 au , we will adopt @xmath158 km s@xmath10 for the parent expansion velocity to estimate the expansion velocities of co@xmath8 and co for 21p at 2.4 au from the sun . with this velocity
, we can derive the production rates , @xmath159 , in molecules s@xmath10 , for each species : @xmath160 where @xmath62 is measured in km , and @xmath161 in km s@xmath10 .
the production rates are : @xmath162 molecules s@xmath10 , and @xmath163 molecules s@xmath10 .
we note that these are upper - limits to the production rates for each species as we can not distinguish between the two .
@xcite detected co at 4.67 in the coma of comet 21p at @xmath164 au ( pre - perihelion ) in october 1998 the first ir detection of this volatile in a jupiter - family comet .
however , two weeks later @xcite did not detect co at the same wavelengths .
@xcite suggest that comet 21p has a chemically heterogeneous nucleus , and that the co was limited to a vent that was not illuminated during the observations conducted by @xcite .
this explanation seems reasonable as the comet was near perihelion when seasonal variations of insolation are the greatest .
our irac images of comet 21p were obtained at a post - perihelion distance of @xmath165 au .
if the vent responsible for the decrease in co production remains inactive out to 2.4 au , co@xmath8 would remain the favored molecule responsible for the irac 4.5 excess .
determining the cometary dust fraction of the zodiacal dust complex is becoming increasingly important .
spitzer _ discoveries of dust surrounding white dwarfs @xcite suggest that mass - loss from cometary bodies ( in kuiper belt or oort cloud analogs ) or tidal disruption of asteroidal objects are feeding their circumstellar dust complexes @xcite .
comets lose a large fraction of their mass in dust grains @xmath166 @xcite .
these grains weakly respond to solar radiation pressure and form distinct dynamical structures known as dust trails @xcite .
trails typically appear to follow their parent nucleus along the comet s projected orbit , although the `` trail '' may also lead the comet .
dust trails are ubiquitous in jupiter - family comets , but remain unobserved in oc comets .
the existence of oc dust trails is inferred from meteor stream studies of halley - type comets , which are derived from the oc .
for example , the halley - type comet 55p / tempel - tuttle is the parent body of the leonid meteor stream and at least 80% of jfcs
have dust trails @xcite .
the existence of dust trails in comets with perihelia near 1 - 2 au has interesting astrobiological significance .
@xcite demonstrate that comets have trail mass production rates that are : ( 1 ) similar to water mass production rates , and ( 2 ) greater than the small grain ( radii @xmath167 ) mass production rates .
if the composition of the comet s coma approximates the comet interior , then comet nuclei are mostly refractory in nature ( dust - to - ice mass ratio @xmath168 ) , which limits the amount of water a comet could deliver to the surface of a young terrestrial planet .
comet 21p is the parent body of the draconid meteor stream @xcite , which can produce exceptionally active meteor showers ( @xmath169 meteors per hour ; * ? ? ?
the outburst activity is strongly associated with years when the earth and 21p closely approach each other , suggesting that the comet is presently ejecting meteoroids . thermal radiation from large grains in a comet trail
is easily detected at ir wavelengths from space - based telescopes @xcite , where the emission from warm ( @xmath170 k ) dust peaks .
we examined our mips 24 image of 21p ( obtained at 2.4 au ) for dust trail emission along the projected velocity vector of the comet ( fig .
[ fig:16apr08trail ] ) .
no trail is detected with a @xmath9 surface brightness upper - limit of 0.3 mjy sr@xmath10 pixel@xmath10 . assuming a typical trail grain temperature of @xmath171 k @xcite , the latter surface brightness corresponds to an ir optical depth of @xmath172 .
our upper - limit is larger than the typical comet trail observed in other _
mips images @xcite , thus it is possible that 21p still has a dust trail .
@xcite , using deep @xmath23-band imaging of 21p , do not detect the presence of brightness enhancements at the expected position of the dust trail , and derive a upper - limit to the number density of trail grains to be @xmath173 m@xmath12 . using similar assumptions ( typical grain size of 1 mm ) , and assuming a typical trail thickness of @xmath174 km @xcite
, our mips optical depth limit suggests a number density @xmath175 m@xmath12 slightly lower than the optically determined value .
@xcite also estimate the dust trail number density from meteor shower activities .
their lower - limit ( @xmath176 m@xmath12 ) is about 4 orders of magnitude smaller than our upper - limit . for comparison ,
the trail grain number density for 81p / wild is @xmath177 m@xmath12 @xcite , and for 67p / churyumov - gerasimenko is @xmath178 m@xmath12 @xcite .
altogether , the evidence suggests comet 21p has a dust trail fainter than the detectable limits of the available observations .
we have presented new optical and _ spitzer _ infrared observations of comet 21p / giacobini - zinner obtained during its 2005 apparition .
analysis of optical imagery indicates that 21p was dusty ( peak @xmath1 @xmath2 ) and active out to heliocentric distances @xmath3 au following a logarithmic slope with @xmath4 of @xmath5 .
onset of nucleus activity occurred at a pre - perihelion distance @xmath0 au ( @xmath34 days pre - perihelion ) , similar in behavior to that observed in the 1991 apparition .
the derived average coma colors , @xmath179 are slightly redder than solar , comparable to colors derived for other jupiter - family comets . pre - perihelion observations during quiescence yields a nucleus radius of @xmath7 km . _
spitzer _ irac images obtained at @xmath180 au , post - perihelion exhibit an extensive coma with a prominent dust tail , where excess emission ( over the dust continuum ) in the 4.5 image at cometocentric distances of @xmath181 km likely arises from co@xmath8 , although a distributed source of co can not be discounted .
the upper limits to the production rates are @xmath182 molecules s@xmath10 and @xmath163 molecules s@xmath10 .
the surface brightness of the gas emission is observed to peak along the sun angle , while the dust tail peaks near the anti - sunward angle .
a search for dust trail emission along the projected velocity vector of comet 21p using our mips 24 image ( @xmath180 au ) , yielded no trail ( @xmath9 surface brightness upper - limit of 0.3 mjy sr@xmath10 pixel@xmath10 ) , suggesting that the number density of trail particles ( typical particle size @xmath183 mm ) is @xmath11 m@xmath12 .
the bolometric albedo of 21p derived from the contemporaneous optical and _ spitzer _ observations is @xmath13 , slightly lower than values derived for other comets at the same phase angle .
this work is based on observations made with the _ spitzer _ space telescope , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
support for this work was provided by nasa through an award issued by jpl / caltech .
support for this work also was provided by nasa through contracts 123741 , 127835 , and 1256406 issued by jpl / caltech to the university of minnesota .
c.e.w . also acknowledges support from the national science foundation grant ast-0706980 .
m.s.k . acknowledges support from the university of minnesota doctoral dissertation fellowship .
jp was supported , in part , by the nasa planetary astronomy program , and by nasa through contract 1278383 issued by jpl / caltech to the university of hawaii and the scientific grant agency vega of the slovak academy of sciences , 2/4002/04 .
the authors wish to thank dr .
karen meech for her insightful discussions , as well as an anonymous referee whose comments and suggestions improved the clarity of this manuscript .
kadono , t. , sugita , s. , sako , s. , ootsubo , t. , honda , m. , kawakita , h. , miyata , t. , furusho , r. , watanabe , j. 2007 , , 661 , l89 kolokolova , l. , hanner , m. s. , levasseur - regourd , a .- c . , & gustafson , b. . s. 2005 , in comets ii , eds . m. festou , h. u. keller , and h. a. weaver , ( university of arizona press : tucson ) , p.577 meech , k. & svore@xmath184 , j. 2004 , in comets ii , eds . m. festou , h. u. keller , and h. a. weaver , ( university of arizona press : tucson ) , p.317 miura , n. , ishiguro , m. , sarugaku , y. , usui , f. , & ueno , m. 2007 , dust in planetary systems , 643 , 23 tancredi , g. , fernndez , j. a. , rickman , h. , & licandro , j. 2000 , , 146 , 73 weaver , h. a. , chin , g. , bockele - morvan , d. , crovisier , j. , brooke , t. y. , cruikshank , d. p. , geballe , t. r. , kim , s. j. , & meier , r. 1999 , icarus , 142 , 482 lllcccccc 2004 jun 21&uh2.2 m & 3.80 & 3.21 & 13.61 & 0.80 & @xmath185 & 1200 & c + 2004 jun 22&uh2.2
m & 3.79 & 3.21 & 13.76 & 0.80 & @xmath186 & 300 & c + 2005 oct 20&uh2.2 m & 1.76 & 1.74 & 33.02 & 0.60 & @xmath187 & 30 & p + 2005 oct 22&uh2.2 m&1.78 & 1.74 & 32.80 & 0.76 & @xmath188 , @xmath186&90&p + & & & & & & @xmath189 + 2005 dec 02&spitzer & 2.14 & 1.94 & 28.19 & @xmath190 & mips24@xmath191 & 10 & @xmath190 + 2005 dec 21&uh2.2 m&2.32 & 1.72 & 22.56 & 0.71&@xmath192 , @xmath189&200&p + 2005 dec 22&uh2.2 m&2.32 & 1.72 & 22.32 & 0.65&@xmath193 , @xmath194&150&p + & & & & & & @xmath195 + 2005 dec 31&spitzer & 2.40 & 1.89 & 23.61 & @xmath190&irac4.5@xmath196 & 30 & @xmath190 + 2005 dec 31&spitzer & 2.40 & 1.89 & 23.61 & @xmath190&irac8.0@xmath196 & 30 & @xmath190 + 2006 mar 05&uh2.2 m&2.91 & 2.24 & 16.37 & 0.84&@xmath197 , @xmath198&500&c + 2006 mar 06&uh2.2 m&2.92 & 2.25 & 16.45 & 0.73&@xmath199 , @xmath200&400&p + & & & & & & @xmath198 + lcccccc 2004 jun 21&3.80&[email protected] & @xmath190 & @xmath190 & 17.19 & 4.80 + 2004 jun 22&3.79&[email protected] & @xmath190 & @xmath190 & 17.24 & 4.55 + 2005 oct 20&1.76&[email protected] & @xmath190 & @xmath190 & 12.82 & 130.66 + 2005 oct 22&1.78&[email protected]&[email protected]&[email protected]&12.89&124.47 + 2005 dec 21&2.32 & [email protected] & @xmath190 & [email protected] & 13.74 & 83.00 + 2005 dec 22&2.32&[email protected]&[email protected]&[email protected]&13.76 & 82.07 + 2006 mar 05&2.91 & [email protected] & @xmath190 & [email protected] & 14.80 & 39.06 + 2006 mar 06&2.92 & [email protected] & @xmath190 & @xmath190 & 14.61 & 46.50 + | we present ground - based optical and _ spitzer space telescope _
infrared imaging observations of the ecliptic ( jupiter - family ) comet 21p / giacobini - zinner , the parent body of the draconid meteor stream , during its 2005 apparition . onset of nucleus activity occurred at a pre - perihelion heliocentric distance , @xmath0 au , while post - perihelion 21p was dusty ( peak @xmath1 @xmath2 ) and active out to heliocentric distances @xmath3 au following a logarithmic slope with @xmath4 of @xmath5 .
coma colors , @xmath6 are redder than solar , yet comparable to colors derived for other jupiter - family comets . a nucleus radius of @xmath7 km is derived from photometry at quiescence . _
spitzer _ images post - perihelion exhibit an extensive coma with a prominent dust tail , where excess emission ( over the dust continuum ) in the 4.5 irac image arises from volatile gaseous co and/or co@xmath8 .
no dust trail was detected ( @xmath9 surface brightness upper - limit of 0.3 mjy sr@xmath10 pixel@xmath10 ) along the projected velocity vector of comet 21p in the mips 24 image suggesting that the number density of trail particles is @xmath11 m@xmath12 .
the bolometric albedo of 21p derived from the contemporaneous optical and _ spitzer _ observations is @xmath13 , slightly lower than values derived for other comets at the same phase angle . |
thermoelectric ( te ) materials have huge potential for power generation , heat pumping , and refrigeration . however , their practical application is restricted because of the low performance of the te devices compared to traditional fossil fuel power generators and compressor - based refrigerators @xcite .
a high performance te material has to be a good electrical conductor as well as a poor thermal conductor and , at the same time , possess a large seebeck coefficient@xcite .
quantitatively the efficiency of te devices is expressed by the dimensionless figure of merit @xmath4 , @xmath5 where @xmath6 , @xmath7 , @xmath8 are the seebeck coefficient , electrical conductivity and absolute temperature , and @xmath9 , @xmath10 are phonon and electron contributions to the total thermal conductivity , respectively .
bismuth and antimony tellurides , bi@xmath0te@xmath1 , sb@xmath0te@xmath1 , and the alloys based on these materials play a significant role for thermoelectric technology .
already early studies of the ( bi@xmath11,sb@xmath12)te@xmath1 compounds in the late 1950s reported @xmath13 at room temperature @xcite , confirmed by further experiments @xcite .
this value remains , even to date , the maximal one available at room temperature for bulk materials .
current progress in nanostructure fabrication , in particular , epitaxial growth of high - quality superlattices @xcite , encourages the possibility of significant performance improvement of te devices .
et al . _ reported @xmath14 and @xmath15 at 300k for @xmath2-type and @xmath3-type bi@xmath0te@xmath1/sb@xmath0te@xmath1 superlattices , respectively@xcite .
these experimental advances motivated extensive theoretical studies of the electronic structure of the bulk bismuth and antimony telluride aimed to understand the possible origin of the increased thermoelectic performance in the multilayered structures .
while in the previous years only few _ ab initio _ band - structure calculations of the bulk bismuth telluride @xcite could be mentioned , in the last decade various aspects of the electronic structure of both pure and doped bulk bi@xmath0te@xmath1 and sb@xmath0te@xmath1 as well as their transport properties were discussed in ref .
_ ab initio _ studies of the electronic structure and the transport properties of bi@xmath0te@xmath1/sb@xmath0te@xmath1 superlattices were also reported@xcite .
an explanation of directional anisotropy of the transport properties in bi@xmath0te@xmath1/sb@xmath0te@xmath1 superlattices could play a crucial role for the understanding of their increased figure of merit .
et al._@xcite found a strong dependence of the anisotropy of the carrier mobility on both the superlattice period and the relative thickness of the constituents .
the enhancement of the electrical conductivity parallel to the epitaxial growth direction , i.e. the trigonal axis of the rhombohedral lattice of bismuth and antimony tellurides , together with the possibility to suppress the lattice thermal conductivity @xmath9 along this direction could provide the desirable zt enhancement .
in this study we concentrate on the anisotropy of the transport properties in the bulk bi@xmath0te@xmath1 and sb@xmath0te@xmath1 as a first step on a way of understanding the corresponding properties of the bi@xmath0te@xmath1/sb@xmath0te@xmath1 superlattices . since epitaxial growth always implies lattice distortions we included , as discussed below , the effect of the lattice relaxation in our study .
both bismuth and antimony telluride possess the rhombohedral crystal structure with five atoms , i.e. one formula unit , per unit cell belonging to the space group @xmath16 ( @xmath17 ) .
experimental lattice parameters@xcite are @xmath18 , @xmath19 , and @xmath20 , @xmath21 , were @xmath22 is the angle between the rhombohedral basis vectors of the length @xmath23 . in order to emphasize the layered character of this structure it is convenient to rearrange it into the hexagonal unit cell built up by three formula units , as shown in fig.[structure ] .
the hexagonal cell contains 15 atoms grouped in the 3 quintuple layers , te1-bi(sb)-te2-bi(sb)-te1 , where te1 and te2 are two different crystal kinds of tellurium atoms .
the hexagonal lattice parameters are @xmath24 , @xmath25 , and @xmath26 , @xmath27 , for bi@xmath0te@xmath1 and sb@xmath0te@xmath1 , respectively . in bismuth telluride the nearest interatomic distances between the individual monolayers inside the quintuple blocks are 3.07 for te1-bi and 3.25 for bi - te2 .
two adjacent quintuple layers in bi@xmath0te@xmath1 are separated by somewhat longer distance , 3.63 for te1-te1 . in the antimony
telluride these distances are 2.98 for te1-sb , 3.17 for sb - te2 inside the quintuple blocks , and 3.74 for te1-te1 between the blocks .
the rhombohedral unit cell superimposed with the hexagonal one to emphasize the layered character of the material . ] in the bi@xmath0te@xmath1/sb@xmath0te@xmath1 multilayers atoms change their bulk positions due to the mismatch of the lattice parameters .
the description of the realistic crystal structure of the multilayers is out of the scope of this study .
nonetheless , we modeled bi@xmath0te@xmath1 with the experimental lattice parameters and interatomic distances of sb@xmath0te@xmath1 , and vice versa .
since both materials have very similar lattice parameter @xmath28 along @xmath29-axis this variation is essentially compression and extension of the lattice in ( @xmath30 ) plane for bismuth and antimony telluride , respectively .
we assume that one could estimate the effect of the lattice relaxation on the electronic and transport properties in the bi@xmath0te@xmath1/sb@xmath0te@xmath1 heterostructures from these two limiting cases .
calculations of the electronic structures were performed by means of the screened korringa - kohn - rostoker green s function method @xcite in the atomic sphere approximation ( asa ) within the local density approximation of the density functional theory in the parameterization of vosko _
et al._@xcite .
it is generally recognized that the effects of spin - orbit coupling are mandatory for the correct treatment of the band structure in these materials .
therefore we used a fully relativistic version of the method based on the dirac equation@xcite .
the obtained self - consistent band structures were used for the calculations of the transport distribution @xmath31 within the boltzmann formalism assuming a constant relaxation time @xmath32 @xcite , @xmath33 where @xmath34 is the _
j_-th band energy at the * k*-point of the brillouin zone ( bz ) , @xmath6 and @xmath35 denote cartesian coordinates .
we assume the relaxation time to be isotropic . in this approximation
the transport anisotropy ratio @xmath36 is independent on the relaxation time , and we do not have to specify it . the * k*-space integration over the isoenergetic surfaces was performed with the tetrahedron method on the blchl mesh@xcite of 96@xmath3796@xmath3796 points in the whole bz . in the energy intervals of the width
about 0.1 ev in the vicinity of both band edges the integration was refined by means of an adaptive mesh up to 4@xmath374@xmath374 times as dense as the original one . as a convergence criterium we used the values of the transport anisotropy ratio calculated from the effective mass approximation at the band extrema @xcite .
the electronic band structures for bi@xmath0te@xmath1 and sb@xmath0te@xmath1 for both experimental and strained lattices are shown in fig.[bands ] .
the positions of the high - symmetry points in the bz of the rhombohedral structure are denoted in fig.[bzone ] .
our results at the experimental lattice parameters agree well with the previous _ ab initio _ studies of mishra _
et al._@xcite , larson _ et al._@xcite for bi@xmath0te@xmath1 , and of eremeev _ et al . _
@xcite both for bi@xmath0te@xmath1 and sb@xmath0te@xmath1 . at the same time the calculations made with the full - potential linearized augmented plane wave method ( flapw ) result in slightly different band structures for the bismuth and antimony tellurides at both the experimental@xcite and strained @xcite lattices .
band structures of bi@xmath0te@xmath1 and sb@xmath0te@xmath1 along symmetry lines for both experimental ( solid ) and strained ( dashed ) lattices .
energies are given relative to the vbm.,title="fig : " ] band structures of bi@xmath0te@xmath1 and sb@xmath0te@xmath1 along symmetry lines for both experimental ( solid ) and strained ( dashed ) lattices .
energies are given relative to the vbm.,title="fig : " ] the key question of the band structure of bi@xmath0te@xmath1 and sb@xmath0te@xmath1 is the position of the valence band maximum ( vbm ) and conduction band minimum ( cbm ) in the bz .
the calculations of ref .
[ ] result in a six - valley both vbm and cbm located in the symmetry plane ( @xmath38zu ) in agreement with experiments for both the bismuth@xcite and antimony@xcite tellurides . unlike these results , in our case the cbm of bi@xmath0te@xmath1 , at both lattice parameters , @xmath39 and @xmath40 , is a two - valley minimum located on the symmetry line @xmath38z , similarly to ref .
[ ] . for sb@xmath0te@xmath1 at the experimental lattice parameters we found a direct band gap located at the center of the bz , while at the larger in - plane lattice constant both six - valley vbm and cbm lie in the symmetry plane ( @xmath38zu ) .
in contrast to our results thonhauser@xcite found that the increase of the lattice parameters in sb@xmath0te@xmath1 led to the formation of a direct band gap at the @xmath38 point . on the other hand , the negative hydrostatic pressure discussed in ref .
[ ] implies an increase of both in - plane and out - of - plane lattice parameters in comparable degree , while in our study essentially the first one is included .
additionally , calculations in ref .
[ ] were performed for optimized atomic positions with respect to the total energy , which can affect the band structure of antimony telluride@xcite .
brillouin zone of the rhombohedral lattice . ]
as already discussed @xcite , these differences in the band structures are probably due to the non - spherical part of the potential , which is not included in the asa . at the same time , as discussed below , these differences have no significant impact on the transport distribution @xmath41 .
the details of the band structures for all four systems are compiled in table i. lattice parameters & & + gap(ev ) & & + extremum & vbm & cbm & vbm & cbm + position & 0.517 0.366 0.366 & 0.173 0.173 0.173 & 0.405 0.405 0.335 & 0.151 0.151 0.151 + effective masses & & & & + @xmath42 & -0.024 & 0.178 & -0.039 & 0.154 + @xmath43 & -0.134 & 0.178 & -0.077 & 0.154 + @xmath44 & -1.921 & 0.835 & -0.207 & 1.370 + principal axes & & & & + * e*@xmath45 & 0.500 -0.867 0.000 & 1.000 0.000 0.000 & 0.866 0.499 -0.024 & 1.000 0.000 0.000 + * e*@xmath0 & 0.600 0.346 0.721 & 0.000 1.000 0.000 & 0.500 -0.867 0.000 & 0.000 1.000 0.000 + * e*@xmath1 & 0.625 0.361 -0.693 & 0.000 0.000 1.000 & 0.021 0.012 0.999 & 0.000 0.000 1.000 + @xmath36 & 5.452 & 4.700 & 4.020 & 9.013 + + latt .
param . & & + gap(ev ) & & + extremum & vbm & cbm & vbm & cbm + position & 0.000 0.000 0.000 & 0.000 0.000 0.000 & 0.547 0.392 0.392 & 0.004 0.020 0.020 + effective masses & & & & + @xmath42 & -0.054 & 0.045 & -0.039 & 1.124 + @xmath43 & -0.054 & 0.045 & -0.083 & 1.774 + @xmath44 & -0.102 & 0.114 & -2.046 & 6.861 + principal axes & & & & + * e*@xmath45 & 1.000 0.000 0.000 & 1.000 0.000 0.000 & 0.500 -0.867 0.000 & -0.316 -0.183 0.931 + * e*@xmath0 & 0.000 1.000 0.000 & 0.000 1.000 0.000 & 0.594 0.343 0.727 & 0.500 -0.867 0.000 + * e*@xmath1 & 0.000 0.000 1.000 & 0.000 0.000 1.000 & 0.630 0.363 -0.686 & 0.806
0.465 0.365 + @xmath36 & 1.889 & 2.507 & 2.397 & 2.080 + the calculated transport distribution of bi@xmath0te@xmath1 at both the experimental , @xmath46 , and compressed , @xmath47 , lattice parameters are shown in fig.[condbite ] together with the anisotropy ratio @xmath36 . in terms of the rigid band model the energies below and above the band gap
simulate @xmath2- and @xmath3-doping respectively . while for @xmath2-doping close the vbm @xmath36 varies smoothly approaching the limiting value , the ratio increases drastically and forms a prominent two - peak structure for @xmath3-doping case .
transport distribution @xmath48 and @xmath49 and the transport anisotropy @xmath36 for bi@xmath0te@xmath1 at the experimental and compressed lattice parameters .
crosses at the band edges mark the @xmath36 ratio derived from the effective mass model @xcite using the parameters of table i. , title="fig : " ] transport distribution @xmath48 and @xmath49 and the transport anisotropy @xmath36 for bi@xmath0te@xmath1 at the experimental and compressed lattice parameters .
crosses at the band edges mark the @xmath36 ratio derived from the effective mass model @xcite using the parameters of table i. , title="fig : " ] this structure originates from the two topological transformations of the constant energy surfaces in the conduction band .
fig.[iso](a ) shows the contour plot of @xmath50 for bi@xmath0te@xmath1 at @xmath46 in the plane ( @xmath38zu ) for energies 0 to 0.19ev relative to the conduction band edge .
the main features of the band structure are the global cbm on the line @xmath38z , the local conduction band minimum ( lcbm ) at ( 0.666,0.602,0.602 ) , and two saddle points @xmath51 at ( 0.722 , 0.667 , 0.667 ) and @xmath52 at ( 0.493,0.461,0.461 ) , in crystallographic coordinates .
the saddle point @xmath51 occurs at @xmath53ev and causes the first peak of @xmath36 , while @xmath52 appears at 0.17ev and forms the second peak . at room temperature the chemical potential
would be located in the saddle point @xmath51 or @xmath52 , for an electron carrier concentration of about @xmath54 or @xmath55 , respectively .
contour plots of @xmath50 at the experimental lattice constants in the plane ( @xmath38zu ) + ( a ) bi@xmath0te@xmath1 , 10 isolines for ( @xmath56 ) at 0 to 0.19ev with a constant increment(dotted ) , additionally , 2 isolines at @xmath53ev and @xmath57ev with the saddle points @xmath51 and @xmath52 , respectively ( bold ) , the positions of the cbm and the lcbm are marked with crosses .
+ ( b ) sb@xmath0te@xmath1 , 10 isolines for ( @xmath58 ) at -0.24 ev to 0 with a constant increment ( dotted ) , additionally , 1 isoline with the saddle point @xmath59 ( bold ) , the position of the vbm is marked with a cross.,title="fig : " ] contour plots of @xmath50 at the experimental lattice constants in the plane ( @xmath38zu ) + ( a ) bi@xmath0te@xmath1 , 10 isolines for ( @xmath56 ) at 0 to 0.19ev with a constant increment(dotted ) , additionally , 2 isolines at @xmath53ev and @xmath57ev with the saddle points @xmath51 and @xmath52 , respectively ( bold ) , the positions of the cbm and the lcbm are marked with crosses .
+ ( b ) sb@xmath0te@xmath1 , 10 isolines for ( @xmath58 ) at -0.24 ev to 0 with a constant increment ( dotted ) , additionally , 1 isoline with the saddle point @xmath59 ( bold ) , the position of the vbm is marked with a cross.,title="fig : " ] the location of the lcbm in our band structure calculation is close to the position of the six - valley cbm reported in ref .
[ ] , and the energy difference between these two minima,@xmath60=0.04ev , is quite small . at the same time
the saddle point @xmath52 close to the local band maximum at z was found as well in agreement with these calculations . a slight shift of the energies at the cbm and the lcbm would bring these two band structures in accordance .
since the saddle point @xmath51 lies close to the line connecting the cbm and the lcbm this modification would not affect remarkably the band structure topology .
moreover , the transport anisotropy @xmath36 at the lcbm from the effective mass approximation is 4.95 , which is fairly close to 4.7 at the cbm .
this indicates that the transport anisotropy of the bismuth telluride at the experimental lattice parameters is stable with respect to small modifications of the band structures with a two- and six - valley cbm , respectively .
in bi@xmath0te@xmath1 the in - plane compression of the lattice parameters from @xmath46 to @xmath47 increases the transport anisotropy ratio at the conduction band edge remarkably . within the effective mass approximation
this can be explained by the enhancement of the ratio @xmath61 due to the expansion of the bz in @xmath30 plane . at the same time
@xmath36 decreases at the valence band edge . in this case
the compression of the lattice results in a re - orientation of the longest axis of the effective mass ellipsoid near the vbm closer to the @xmath29 axis , which lead to an increase of the transport anisotropy ratio , and at the same time decreases the anisotropy of the dominating effective masses @xmath43 and @xmath44 ( see table i ) . the last effect prevail over the enhancement of @xmath36 due to the rotation of the effective mass ellipsoid .
transport distribution and the transport anisotropy for sb@xmath0te@xmath1 at the experimental and expanded lattice parameters .
crosses at the band edges mark the @xmath36 derived from the effective mass model @xcite using the parameters of table i.,title="fig : " ] transport distribution and the transport anisotropy for sb@xmath0te@xmath1 at the experimental and expanded lattice parameters .
crosses at the band edges mark the @xmath36 derived from the effective mass model @xcite using the parameters of table i.,title="fig : " ] fig.[condsbte ] shows the transport distribution and the anisotropy ratio of sb@xmath0te@xmath1 at both @xmath47 and @xmath46 .
the kink of the anisotropy ratio at the experimental lattice parameters is induced by the saddle point @xmath62(0.831 , 0.784 , 0.784 ) at @xmath63 ev , which corresponds to a hole carrier concentration of @xmath64 at 300k .
this topology is illustrated in fig.[iso](b ) , which shows the contour plot of @xmath50 in the valence band of the antimony telluride for energies -0.24 ev to 0 relative to @xmath65 .
the in - plane expansion of the lattice parameter increases the density of the occupied states near the valence band edge and suppresses the kink . the transport anisotropy ratio increases with the in - plane compression of the bz at the valence band edge , and , at the same time , decreases at the conduction band edge due to the larger angle between the @xmath29 axis and the largest axis of the effective mass ellipsoid , similarly to the discussed changes in bismuth telluride .
on the basis of _ ab initio _ electronic structures obtained with the fully relativistic kkr method we studied the anisotropy of the transport properties of the bismuth and antimony tellurides in the constant relaxation time approximation within the boltzmann formalism .
in addition to the systems with the experimental lattice parameters we modeled bismuth telluride within the lattice of sb@xmath0te@xmath1 , and vice versa .
we found that a decrease of the in - plane lattice parameters increases the transport anisotropy for the @xmath3-doping and , at the same time , decreases the anisotropy for the @xmath2-doped case .
this effect can be understood within the effective mass approximation at the valence band maximum and conduction band minimum , respectively .
j. p. fleurial , l. gailliard , r. triboulet , h. scherrer , and s. scherrer , j. phys .
solids * 49 * , 1237 ( 1988 ) , + t. caillat , m. carle , p. pierrat , h. scherrer , and s. scherrer , j. phys .
solids * 53 * , 1121 ( 1992 ) o. madelung , m. schulz , h. weiss ( eds . ) landolt - brnstein , new series , group iii , vol .
17f , ( springer , new york , 1983 ) + r. w. g. wyckoff , crystal structures 2 , j. ( wiley and sons , new york , 1964 ) + th .
l. anderson , h. krause , h. brigitte : acta crystallogr .
b 30 , 1307 ( 1974 ) j. r. drabble , r. d. groves , and r. wolfe , proc . phys .
london * 71 * , 430 ( 1958 ) + r. b. mallinson , j. a. rayne , and r. w. ure , jr .
lett * 19 * , 545 ( 1965 ) h. schwarz , g. bjrck , and o. beckman , soli .
state commun .
* 5 * , 905 ( 1967 ) + v. a. kulbachinskii , z. m. dashevskii , m. inoue , m. sasaki , h. negishi , w. x. gao , p. lostak , j. horak , and a. de visser , phys .
b * 52 * , 10915 ( 1995 ) | on the basis of detailed _ ab initio _ studies the influence of strain on the anisotropy of the transport distribution of the thermoelectrics bi@xmath0te@xmath1 and sb@xmath0te@xmath1 was investigated .
both tellurides were studied in their own , as well as in their co - partners lattice structure to gain insight to the electrical transport in epitaxial heterostructures composed of both materials .
it is shown , that the anisotropy of the transport distribution overestimates the experimental findings for bi@xmath0te@xmath1 , implying anisotropic scattering effects .
an increase of the in - plane lattice constant leads to an enhancement of the transport anisotropy for @xmath2-doping , whereas the opposite occurs for @xmath3-doping .
the recent findings and special features of the transport distribution are discussed in detail in relation to the topology of the band structures . |
this research was supported in part by the u.s .
department of energy under grants no . da - ac02 - 76-er02289 task b and no .
de - fg02 - 95er40896 and in part by the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation .
the research of ts is supported in part by the u.s .
department of energy under grant no .
de - fg02 - 91er40626 .
one of us ( mmb ) would like to thank the aspen center for physics for its hospitality during the preparation of this manuscript .
m. m. block _ et al .
_ , photon - proton and photon - photon scattering from nucleon - nucleon forward amplitudes , e - print archive : hep - ph/9809403 , phys . rev .
d,*60 * 054024 , 1999 .
r. m. baltrusaitis _
lett . * 52 * , 1380 , 1984 .
et al . _ ,
lett . * 70 * , 525 , 1993 . m. aglietta _
et al _ , proc 25th icrc ( durban ) * 6 * , 37 , 1997 .
r. s. fletcher _
et al . _ ,
rev . d*50 * , 5710 , 1994 .
see http://sunshine.chpc.utah.edu/research/cosmic/hires/ the pierre auger project design report , fermilab report ( feb . 1997 ) . | -.35 in we use the high energy predictions of a qcd - inspired parameterization of all accelerator data on forward proton - proton and antiproton - proton scattering amplitudes , along with glauber theory , to predict proton
air cross sections at energies near @xmath0 30 tev .
the parameterization of the proton - proton cross section incorporates analyticity and unitarity , and demands that the asymptotic proton is a black disk of soft partons . by comparing with the p - air cosmic ray measurements , our analysis results in a constraint on the inclusive particle production cross section .
@=11
@=12 .65 in cosmic ray experiments measure the penetration in the atmosphere of particles with energies in excess of those accelerated by existing machines interestingly , their energy range covers the energy of the large hadron collider ( lhc ) and extends beyond it .
however , extracting proton proton cross sections from cosmic ray observations is far from straightforward @xcite . by a variety of experimental techniques , cosmic ray experiments
map the atmospheric depth at which cosmic ray initiated showers develop .
the measured shower attenuation length ( @xmath1 ) is not only sensitive to the interaction length of the protons in the atmosphere ( @xmath2 ) , with @xmath3 but also depends on the rate at which the energy of the primary proton is dissipated into electromagnetic shower energy observed in the experiment .
the latter effect is parameterized in eq.([eq : lambda_m ] ) by the parameter @xmath4 ; @xmath5 is the proton mass and @xmath6 the inelastic proton - air cross section .
the value of @xmath4 depends on the inclusive particle production cross section in nucleon and meson interactions on the light nuclear target of the atmosphere and its energy dependence .
we here ignored the fact that particles in the cosmic ray `` beam '' may be nuclei , not just protons .
experiments allow for this by omitting from their analysis showers which dissipate their energy high in the atmosphere , a signature that the initial energy is distributed over the constituents of a nucleus .
the extraction of the pp cross section from the cosmic ray data is a two step process .
first , one calculates the @xmath7-air total cross section from the measured inelastic cross section @xmath8 next , the glauber method@xcite is used to transform the measured value of @xmath6 into a proton proton total cross section @xmath9 ; all the necessary steps are calculable in the theory . in eq.([eq : spa ] ) the measured cross section for particle production is supplemented with @xmath10 and @xmath11 , the elastic and quasi - elastic cross section , respectively , as calculated by the glauber theory , to obtain the total cross section @xmath12 .
the subsequent relation between @xmath6 and @xmath9 involves the slope of the forward scattering amplitude for elastic @xmath13 scattering , @xmath14 , @xmath15_{t=0 } \,,\ ] ] and is shown in fig.[fig : p - air ] , which plots @xmath16 against @xmath9 , for 5 curves of different values of @xmath6 .
this summarizes the reduction procedure from @xmath6 to @xmath9 @xcite .
also plotted in fig.[fig : p - air ] is a curve of @xmath16 _ vs. _ @xmath9 which will be discussed later .
a significant drawback of the method is that one needs a model of proton air interactions to complete the loop between the measured attenuation length @xmath1 and the cross section @xmath6 , _
i.e. , _ the value of @xmath4 in eq .
( [ eq : lambda_m ] ) . a proposal to minimize the impact of theory is the topic of this letter .
we have constructed a qcd - inspired parameterization of the forward proton proton and proton antiproton scattering amplitudes@xcite which is analytic , unitary and fits all data of @xmath17 , @xmath16 and @xmath18 , the ratio of the real - to - imaginary part of the forward scattering amplitude ; see fig.[fig : ppcurves ] .
we emphasize that all 3 quantities are simultaneously fitted .
using vector meson dominance and the additive quark models , it accommodates a wealth of data on photon - proton and photon - photon interactions without the introduction of new parameters@xcite .
because the model is both unitary and analytic , it has high energy predictions that are essentially theory independent . in particular
, it also _ simultaneously _ fits @xmath9 and @xmath16 , forcing a relationship between the two .
specifically , the @xmath16 _ vs. _ @xmath9 prediction of the model is shown as the dashed curve in fig.[fig : p - air ] .
the dot corresponds to our prediction of @xmath9 and @xmath16 at @xmath19 = 30 tev .
it is seen to be slightly below the curve for 490 mb , the lower limit of the fly s eye measurement , which was made at @xmath20 30 tev .
the percentage error in the prediction of @xmath9 at @xmath21 tev is @xmath22% , due to the statistical error in the fitting parameters ( see references @xcite,@xcite ) . in fig.[fig :
sigpp_p - air ] , we have plotted the values of @xmath9 _ vs. _ @xmath6 that are deduced from the intersections of the @xmath16-@xmath9 curve with the @xmath6 curves of fig.[fig : p - air ] .
figure [ fig : sigpp_p - air ] allows the conversion of the measured @xmath6 to @xmath9 .
the percentage error in @xmath6 is @xmath23 % near @xmath24 mb , due to the error in @xmath9 from the model parameter uncertainties .
our prediction for the total cross section @xmath9 as a function of energy is confronted with all of the accelerator and cosmic ray measurements@xcite in fig.[fig : sigtodorpp ] . for inclusion in fig.[fig : sigtodorpp ] , we have calculated the cosmic ray values of @xmath9 from the _ published _ experimental values of @xmath6 , using the results of fig.[fig : sigpp_p - air ] .
we note the systematic underestimate of the cosmic ray points , roughly about the level of one standard deviation .
it is at this point important to recall eq.([eq : lambda_m ] ) and consider the fact that the extraction of @xmath6 from the measurement of @xmath1 requires a determination of the parameter @xmath4 .
the measured depth @xmath25 at which a shower reaches maximum development in the atmosphere , which is the basis of the cross section measurement in ref .
@xcite , is a combined measure of the depth of the first interaction , which is determined by the inelastic cross section , and of the subsequent shower development , which has to be corrected for .
the position of @xmath25 also directly affects the rate of shower attenuation with atmospheric depth which is the alternative procedure for extracting @xmath6
. the model dependent rate of shower development and its fluctuations are the origin of the deviation of @xmath4 from unity in eq.([eq : lambda_m ] ) .
its values range from 1.5 for a model where the inclusive cross section exhibits feynman scaling , to 1.1 for models with large scaling violations@xcite .
the comparison between data and experiment in fig.[fig : sigtodorpp ] is further confused by the fact that the agasa@xcite and fly s eye@xcite experiments used different values of @xmath4 in the analysis of their data , _
i.e. , _
agasa used @xmath26 and fly s eye used @xmath27 .
we therefore decided to match the data to our prediction and extract a common value for @xmath4 .
this neglects the possibility that @xmath4 may show a weak energy dependence over the range measured . by combining the results of fig.[fig : ppcurves](a ) and fig.[fig : sigpp_p - air ] , we can plot our prediction of @xmath6 _ vs. _ @xmath19 . to obtain @xmath4 ,
we leave it as a free parameter and make a @xmath28 fit to the rescaled high energy cosmic ray data for @xmath6 . in fig.[fig : p - aircorrected2 ] we have replotted the published high energy data for @xmath6 , against our prediction of @xmath6 _ vs. _ @xmath19 , using the common value of @xmath29 , obtained from a @xmath28 fit .
the error of @xmath30 is the statistical error of the @xmath28 fit , whereas the error of @xmath31 is the systematic error due to the error in the prediction of @xmath6 .
clearly , we have an excellent fit , with good agreement between agasa and fly s eye .
the analysis gives @xmath32 for 6 degrees of freedom ( the low @xmath28 is probably due to overestimates of experimental errors ) .
of course , the improved fit of the cosmic ray data has no effect on the fit to the accelerator data .
our result for @xmath4 is interesting it is close to the value of @xmath33 obtained using the sibyll simulation@xcite for inclusive particle production .
this represents a consistency check in the sense that our model for forward scattering amplitudes and sybill share the same underlying physics .
the increase of the total cross section with energy to a black disk of soft partons is the shadow of increased particle production which is modeled by the production of ( mini)-jets in qcd .
the difference between the @xmath4 values of 1.20 and 1.33 could be understood because the experimental measurement integrates showers in a relatively wide energy range , which tends to increase the value of @xmath4 .
we predict @xmath34 mb for the total cross section at lhc energy ( 14 tev ) , where the error is due to the statistical errors of the fitting parameters . in the near term
, we look forward to the possibility of repeating this analysis with the higher statistics of the hires @xcite cosmic ray experiment that is currently in progress and the auger @xcite observatory . |
the study of oxide heterostructures is rapidly developing into one of the most promising areas of research in strongly correlated electronic systems .
the current excitement in this field was in part triggered by the recent discovery of conducting interfaces , with a substantial high carrier mobility , between two insulating perovskites @xcite .
these results were obtained by growing abrupt layers of the insulators lati(3+)o@xmath1 and srti(4+)o@xmath1 .
when the spatial distribution of the extra electron was observed with an atomic - scale electron beam , it was found to correspond to a metallic state at the interface @xcite .
theoretical investigations of these systems @xcite , using hartree - fock and dmft techniques , concluded that the leakage of charge from one layer to the other explains the results .
thus , charge transfer between materials in heterostructures can be used to stabilize interface states that otherwise would only be obtained via the chemical doping of the parent oxide compound .
this last procedure has the concomitant effect of introducing quenched disorder into the sample .
then , for the preparation of doped materials without the extra complication of disorder , for the realization of two - dimensional versions of perovskites , and for the exploration of `` oxide electronics '' devices and functionalities , the experimental and theoretical analysis of oxide heterostructures is a promising and vast field , and its study is only now starting @xcite . before the recent developments mentioned above , ferromagnetic / superconducting ( fm / sc ) heterostructures had already received considerable attention .
the analysis of spin injection from a ferromagnetic material to an electrode is of much importance for further progress in the area of spintronics .
other properties of interest in fm / sc heterostructures include : ( i ) spin - mixing effects that can induce spin - triplet pairing correlations at the interface @xcite ; ( ii ) josephson couplings between two singlet - superconducting layers separated by a half - metallic ferromagnet @xcite ; ( iii ) oscillations in the singlet pairing amplitude detected in sc / fm / sc geometries @xcite ; and ( iv ) magnetic exchange coupling in fm / sc / fm geometries @xcite
. since transition metal oxides have similar lattice spacings , the combination of colossal magnetoresistance ( cmr ) manganites @xcite and high-@xmath0 cuprates has been specially investigated @xcite .
experimental results suggest a strong fm / sc interplay resulting in the injection of spin polarized carriers into the sc layers @xcite .
of particular importance for the purposes of our investigations , the transfer of charge from the la@xmath2ca@xmath3mno@xmath1 ( lcmo ) ferromagnet to the yba@xmath4cu@xmath1o@xmath5 ( ybco ) superconductor has been observed in recent experiments using eels techniques @xcite . in this case , the addition of electrons , namely the suppression of holes , in the ybco component led to the concomitant suppression of superconductivity and a transition to an insulating state .
similar effects are also believed to occur in grain boundaries of superconductors @xcite . in other lcmo / ybco superlattice setups ,
the superconducting critical temperature was also found to be suppressed @xcite .
this may be caused by the influence of spin - polarized carriers moving from lcmo to ybco , breaking singlet cooper pairs , or by the reduction of the number of hole carriers @xcite , or both .
it is the main purpose of the investigation reported here to analyze whether the previously observed electronic charge transfer at interfaces from manganites , such as lcmo or la@xmath2sr@xmath3mno@xmath1 ( lsmo ) , to hole - doped cuprates , such as ybco , can be extended to the case where the cuprate is in an undoped state . if this still occurs , namely if estimations of work functions lead us to believe that the fermi level of the manganite remains above the first empty states of the undoped cuprate , then this cu oxide will receive an excess of electrons , leading to the @xmath6 @xmath7 of an antiferromagnetic ( af ) state at heterostructure interfaces @xcite .
this mixing of cuprates and manganites could lead either to ( i ) a metallic spin - polarized state at the interface , if the manganite is ferromagnetic , or to ( ii ) an electron - doped superconductor at low enough temperatures , if the spin polarization is not strong enough to destroy cooper pairs or if the manganite used is not ferromagnetic .
this mechanism would allow for the electronic doping of a parent cuprate compound without the complication of the coulombic and structural disorder introduced by chemical doping .
the fact that this is electron doping , as opposed to the already investigated interfacial hole doping , is particularly interesting since a variety of material issues has prevented the electron - doped branch of high-@xmath0 superconductors to develop as vastly as those that are hole doped .
once again , note that if the manganite is ferromagnetic , then the charge transfer to the undoped cuprate would occur via polarized electrons , which suppress spin singlet superconductivity .
thus , there are two @xmath8 tendencies : adding electrons favors superconductivity , but its spin polarization suppresses it .
the outcome is difficult to predict .
recent investigations using realistic microscopic models focusing on a manganite / insulator interface predict that a complex pattern of phase separation between fm and af regions could occur @xcite .
thus , even if superconductivity is not induced , the resulting state would be interesting , due to the interplay between ferro and antiferromagnetic tendencies . more importantly ,
note that the manganite involved in the heterostructure does @xmath9 need to be ferromagnetic : low - bandwidth mn oxides present a wide variety of spin / charge / orbital order states , other than ferromagnetism @xcite .
even large - bandwidth manganites have non - fm states at small and large enough hole concentration .
thus , inducing a sc state is a real possibility with the proper choice of the manganite partner to the cuprate . a qualitative representation of the transfer of charge is shown in fig . [ systems ] for the cases of a doped cuprate , such as discussed in @xcite , and for an undoped cuprate
, the focus of our investigation reported below . in case ( a )
, sc is destroyed , while in the proposed case ( b ) , sc is induced ( if the manganite is not ferromagnetic ) .
our main result is that the novel scenario ( b ) appears possible .
our conclusion is based on ( i ) a detailed analysis of the work functions of the materials of relevance for these investigations , mainly involving the study of photoemission and diffusion voltage published data , as well as ( ii ) rough calculations at the mean - field and dynamical mean - field levels , showing that there is no fundamental problem preventing superconductivity from happening at the interface .
however , it must be clearly expressed that our contribution should be considered as only the first theoretical steps toward realizing electron doping at cuprate interfaces , and considerable more work remains to be carried out .
for instance , issues such as lattice reconstructions and polarity of the involved interfaces have @xmath9 been analyzed here ( only electronic reconstructions at idealized perfect interfaces were investigated ) .
vacancies and the most optimal location of oxygens at the interface have not been investigated as well . however , even with these caveats we believe the conclusions described below are sufficiently interesting that they deserve experimental efforts to attempt their realization in real heterostructures .
the organization of the manuscript is as follows : first , we carry out a detailed analysis of available experimental information to judge if the fermi level of manganites is above the lowest - energy empty state of several undoped cuprates .
this is described in detail , because it provides a systematic procedure to address other oxide heterostructures in the future .
second , a model calculation in the mean - field approximation is carried out , both for static and dynamical cases .
no fundamental problem is found with the proposal of having superconductivity induced at the manganite / cuprate interface .
the paper concludes with a discussion and summary .
for the success of the emergent field of oxide heterostructures it is crucial to properly determine the relative work functions of the materials involved , since these work functions control the curvature of the valence and conduction bands ( vb and cb ) of the constituent materials , and ultimately the carrier concentration at the interfaces .
work functions of conventional metals and semiconductors have been studied for decades establishing the fundamental background for current electronics .
thus , for the next - generation electronic devices utilizing the complex properties of correlated - electron systems , such as high-@xmath0 cuprates and cmr manganites , determining the relative work functions of a variety of transition - metal oxides is equally important .
photoemission spectroscopy ( pes ) is an important technique in this context .
pes has provided fundamental information to uncover the properties of complex oxides . however , although pes techniques can be of considerable help for oxide heterostructures , only a limited number of experiments have been reported specifically addressing the work function of transition - metal oxides @xcite . by measuring the diffusion voltage @xmath10 ( or built - in potential ) of a junction between two materials of interest
, one can also extract the chemical potential differences between them .
this diffusion voltage is the potential barrier at the interface after the rearrangement of charge occurs . if there are no extra contributions to @xmath10 , such as those caused by interface polarities , impurities , or lattice reconstructions , then @xmath10 is equivalent to the work function difference between the two constituents .
if the work function of one of the materials is known , the work function of the other can be estimated .
this provides another procedure to study the band alignment of oxides .
the purpose of this section is to provide a rough estimation of the band diagram of perovskite transition - metal oxides ( cuprates and manganites ) using the experimental data currently available .
this is achieved by combining information from chemical potential shifts obtained using pes with diffusion voltage measurements on heterostructures . since the discovery of high-@xmath0 superconductivity in the cuprates , considerable experimental information has been accumulated about these compounds . in particular ,
the chemical - potential jump from the hole - doped to the electron - doped materials has been one of the important topics since it is closely related to the mott / hubbard ( or charge transfer ) gap . here
, let us first consider la@xmath4cuo@xmath11 ( lco ) and nd@xmath4cuo@xmath11 ( nco ) as the parent compounds of hole - doped and electron - doped cuprates , respectively .
it should be mentioned that these compounds have different structures : lco has the layered perovskite or @xmath12 ruddlesden - popper structure , while nco presents the so - called @xmath13-structure without apical oxygen . even though the electronic properties of the cuo@xmath4 planes are quite similar for lco and nco , the different overall structures suggest that small differences may exist in their chemical potentials , as discussed below . from the optical measurements
, it has been suggested that the chemical potential of lco is located 0.4 ev above the top of the valence band @xcite .
similar results were obtained with pes techniques @xcite .
polaronic effects are believed to be the cause of this somewhat `` mysterious '' shift between the top of the valence band and the actual location of the chemical potential in lco @xcite .
similarly , resonant - photoemission studies on nco revealed that the chemical potential of nco is located 0.7 ev above the valence band @xcite .
if the positions of the valence bands are identical between lco and nco , then their chemical potential jump becomes @xmath14 ev @xcite .
although an accurate determination of the individual chemical potentials of these materials remains to be done , estimations based on the shifts of the o 1@xmath15 and cu 2@xmath16 core levels indicate that the chemical potential jump is at most 0.5 ev @xcite .
then , the positions of the valence bands can be assumed to be the same , although with a @xmath17 ev uncertainty . to complete the band diagram , information about the location of the unoccupied conduction , or upper hubbard band , is needed . from optical spectroscopy , it was first suggested that the mott gaps of lco and nco are about 2 ev and 1.5 ev , respectively @xcite .
however , numerical studies of the hubbard model on finite - size clusters using appropriate hopping amplitudes for the cuprates revealed that the mott gap in cu oxides was in fact an indirect one @xcite : the top of the lower - hubbard band is at momenta @xmath18 while the bottom of the upper - hubbard band lies at @xmath19 and @xmath20 .
thus , the mott gap is indirect and smaller than the optical gap .
the indirect nature of this gap has been confirmed by a recent resonant x - ray study @xcite .
considering the ratio between the magnitudes of direct gap and indirect gap from the theoretical considerations and the optical gap obtained by the experiments , the separation between the vb and cb of lco and nco is estimated to be about 1.5 ev and 1 ev , respectively @xcite .
combining all this information together , the schematic band diagram for these materials can be constructed and it is shown in fig .
[ fig : nco_lco ] .
next , let us address another series of high-@xmath0 cuprates : ybco .
the diffusion voltage of the heterostructure between ybco ( oxygen contents unclear ) and 0.05wt% nb - doped srtio@xmath1 ( sto ) ( nb@xmath21-sto ) has been measured @xcite .
the chemical potential of ybco was found to be 1.5 ev below that of nb@xmath21-sto . to establish a connection with the single - layer cuprates ,
note that similar experiments on the diffusion voltage of the single - layer parent compound sm@xmath4cuo@xmath11 ( sco ) and 0.01wt% nb - doped srtio@xmath1 ( nb@xmath22-sto ) have been very recently performed @xcite .
sco is an isostructural material of nco , with the same @xmath13-structure and formal valences , therefore nco is expected to have a chemical potential very similar to that of sco .
the recent experiments of nakamura _ et al .
_ revealed that the chemical potential of sco is 1.3 ev below that of nb@xmath22-sto @xcite .
although there is a slight difference in the doping concentration of nb used for the two experiments ( 0.05wt% nb for ybco and 0.01wt% for sco ) , by reducing the nb concentration to 0.01wt% the diffusion voltage is expected to increase only by 0.1 ev or less because of band - gap narrowing effects @xcite .
thus , by combining the diffusion voltages on sco- and ybco - based heterostructures involving a nb - doped sto substrate , the chemical potential of sco is estimated to be @xmath23 ev above that of ybco .
furthermore , assuming that the behavior of sco and nco are identical , based on the similarity of their structures , then the chemical potential of lco can be estimated by considering the chemical potential jump between nco and lco .
figure [ fig : lco_sco_nbsto_ybco_lsmo ] summarizes the band diagram for the various high-@xmath0 cuprates studied here . from these results
, it is deduced that the parent compounds of hole - doped cuprates have fairly similar chemical potentials despite their different crystal structures .
the doping dependence of the chemical potential is important when using doped compounds for the oxide heterostructures .
the chemical potential shifts from their parent compounds of various high-@xmath0 cuprates have been intensively studied @xcite . for the benefit of the readers ,
the main results are summarized in fig .
[ fig : workfunctions ] ( a ) for hole - doped compounds , and fig .
[ fig : workfunctions ] ( b ) for electron - doped compounds @xcite . 1.0 cm there are still considerable discussions on the electronic properties of the underdoped region of la@xmath24sr@xmath25cuo@xmath11 ( lsco ) , involving concepts such as stripe formation , phase separation , and spin glass .
those may explain the flat chemical potential shifts of lsco at @xmath26 .
however , focusing on the overdoped region where the mixed - phase complexity is reduced , then the chemical potential shifts behave linearly with hole doping with a slope of about 0.2 ev per 0.1 carrier concentration , which is close to that of electron - doped nd@xmath24ce@xmath25cuo@xmath11 ( ncco ) .
this may indicate that the band structures of the cuo@xmath4 planes in those compounds are similar , as widely believed . on the other hand ,
the @xmath27 vs. carrier concentration slope for bi@xmath4sr@xmath4cacu@xmath4o@xmath28 ( bi2212 ) is larger than for the single - layer compounds , perhaps due to its bilayered nature .
as already discussed , both from the fundamental physics as well as the device engineering perspective , heterostructures involving high-@xmath0 cuprates and cmr manganites are important . here , we consider the cubic perovskite manganites first , and then turn to the double - layered manganites .
it should be mentioned that the @xmath29 measurement of the work function of cubic manganites has been performed by several groups using pes @xcite and the kelvin method @xcite , and results are approximately consistent with one another indicating an intrinsic work function @xmath30 ev for la@xmath2sr@xmath25mno@xmath1 with @xmath31 .
these experimental data are important in considering interfaces with the cuprates . however , as the individual work functions of many cuprates are not yet available but only their differences , then we have to consult other experiments in order to establish a connection between manganites and cuprates . here
, we consider pes experiments that measured the chemical potential shift in doped manganites @xcite , and those that provided the diffusion - voltage on la@xmath32sr@xmath33mno@xmath1 ( 20% hole - doped lsmo)/nb@xmath21-sto heterostructures @xcite . according to the experiment by matsuno _
et al_. @xcite , the chemical potential of la@xmath2sr@xmath25mno@xmath1 ( lsmo ) changes rather monotonically from the parent compound lamno@xmath1 to a value @xmath27=@xmath34 ev for la@xmath35sr@xmath36mno@xmath1 ( 60% hole - doped lsmo ) ( see fig . [
fig : workfunctions ] ( c ) ) . furthermore , diffusion voltage measurements on a la@xmath32sr@xmath33mno@xmath1 ( 20% hole - doped lsmo)/nb@xmath21-sto heterostructure , done by muramatsu _
@xcite , reported a value @xmath37 ev . combining these experiments , the bands of lsmo can be aligned relative to those of the cuprates obtained in the previous section .
figure [ fig : lco_sco_nbsto_ybco_lsmo ] summarizes the band alignment of cuprates and manganites , which is our main result of this section .
in particular , the chemical potential shifts between the various cuprates of potential relevance and the 20% doped lsmo are summarized as follows : ( i ) for sco @xmath38 ev , ( ii ) for lco @xmath39@xmath40 ev , and ( iii ) for ybco @xmath41 ev . in particular , it is natural to expect that electrons be transfered from lsmo to the valence band of hole - doped ybco at an interface between the two materials .
this has already been observed experimentally @xcite , thus it is reassuring that the simple procedure followed here is consistent with available experimental information @xcite .
consider now another class of mn oxides , the double - layered manganites with the so - called @xmath42 ruddlesden - popper structure .
direct measurements of work function using photoemission have actually been performed on la@xmath43sr@xmath44mn@xmath4o@xmath45 .
it has been reported that the work function of la@xmath43sr@xmath44mn@xmath4o@xmath45 increases with decreasing temperature across the curie temperature ( 125 k ) from @xmath46 3.5 ev at 180 k to @xmath46 3.56 ev at 60 k @xcite , while the simple double - exchange model predicts the opposite trend . this fact may indicate the importance of several ingredients neglected in the double - exchange model , such as electron - electron interaction , electron - lattice interaction , and orbital degeneracy .
the important point worth emphasizing here is that the work function of the double - layered manganite is more than 1 ev smaller than that of cubic lsmo ( whose work function is @xmath47 ev ) @xcite , which means that the chemical potential of the double - layer manganite is more than 1 ev @xmath48 than the cubic perovskite lsmo .
thus , at least naively , it should be possible to inject electrons to the undoped cuprates using the double - layer manganites .
however , there are issues that remain to be clarified for this conclusion to be valid , particularly the role of the surface condition : for double - layered manganites , the cleaved surface used in pes experiments is always ao@xmath49 ( a : a - site ion , such as la@xmath50 and sr@xmath51 ) , thus positively charged .
the low work function of bilayered manganites could be caused by this charged surface .
on the other hand , for cubic manganites the surface can be either positively charged ( ao termination ) or negatively charged ( mno@xmath4 termination ) . for this purpose ,
photoemission experiments on cubic manganites with controlled surface conditions are highly desirable .
summarizing , in this subsection we discussed the possible electron doping of cuprates from manganites at ideal interfaces . in the paragraphs
above , it was concluded that the parent compounds of electron - doped cuprates , such as nco and sco , are the best candidates for this purpose since they have the lowest conduction band ( the cb in lco is about 0.5 ev higher than that in nco , and the cb in ybco is expected to be even higher than those of nco and lco ) . to dope electrons into a cuprate parent compound , the minimal condition is that the chemical potential of manganites be higher than the bottom of the conduction band ( or the chemical potential if they are not the same ) of the undoped cuprate involved .
this condition seems to be satisfied for cubic manganites lsmo in a substantial doping range @xmath52 that includes fm and af states , when mixed with nco or sco @xcite .
possible improvements may be achieved by using double - layered manganites , which have about 1 ev higher chemical potential than cubic manganites .
the simple ideas described in the previous sections on the possibility of electron doping of cuprates at interfaces need to be checked in more theoretical detail to confirm their consistency . for this purpose ,
first a static mean - field study will be here discussed , followed in the next section by a more detailed analysis including dynamical effects .
the emphasis will be given to a manganite / cuprate interface , but the generation of superconductivity by electron doping can occur for any other combination where electrons are donated to the cuprates . the model used here is defined by the following hamiltonian on a three - dimensional cubic lattice : @xmath53 here @xmath54 is the electron creation operator at site @xmath55 with spin @xmath56 , @xmath57 is the electron density at site * i * , and @xmath58 is the nearest - neighbor hopping , which is @xmath59 on the @xmath60 plane and @xmath61 in the @xmath62 direction . @xmath63 is the electronic potential ( discussed in more detail later ) that will take into account effects related to the charge redistribution .
@xmath64 is the chemical potential , and @xmath65 ( @xmath66 ) are site potentials for the left ( right ) side of the system to regulate the transfer of charge .
@xmath67 contains the interaction terms . on the left side of the system ( the manganite )
the interaction is @xmath68 which is the standard `` double - exchange '' term .
@xmath69 are pauli matrices , and @xmath70 is a classical localized spin at site * i * ( @xmath71 ) representing the @xmath72 spins . on the right side of the system , the cuprate , the standard repulsive hubbard interaction ( @xmath73@xmath740 )
is supplemented by a nearest - neighbors attraction ( @xmath75@xmath760 ) that favors superconductivity in the @xmath77-wave channel : @xmath78 the number operator is @xmath79 , and @xmath80 indicates a pair of nearest - neighbor sites * i * and * j * on the @xmath60 plane .
it is well known that this @xmath59-@xmath73-@xmath75 model leads to a rich phase diagram , that includes superconductivity @xcite .
periodic ( open ) boundary conditions are used in the @xmath52 and @xmath81 ( @xmath62 ) directions for a lattice of size @xmath82 sites .
the focus of the results described below is on @xmath83=@xmath84=16 , but other sizes have been checked confirming that size effects are not strong in this study .
two comments are in order : ( 1 ) a fully realistic model for manganites should contain two @xmath85-orbitals and a jahn - teller electron - phonon coupling .
however , it is common practice in this context to use just one @xmath85-orbital for simplicity , since several phenomena are common to both one and two orbitals @xcite .
moreover , the focus here is not on the cmr regime but only on combining a standard homogeneous manganite state with a cuprate .
thus , the electron - phonon ( e - ph ) interaction , crucial to understand the cmr effect , is here neglected for simplicity .
in fact , lsmo does not have a large cmr effect , thus the e - ph coupling may not be strong in this material . ( 2 ) more severe is the approximation carried out on the cuprate side since a mean - field approximation with an explicitly attractive force is employed .
however , this approximation will be relaxed below and the hubbard model will be used without an explicit attraction , at the price of having to constraint the study to smaller systems than those that can be analyzed in the static mean - field approximation . to treat the many - body interactions described by @xmath86 , here a simple mean - field approximation is adopted .
for the first interaction term , the following replacement is introduced , @xmath87 .
\nonumber\ ] ] for the second term , a standard bcs model approximation is used : @xmath88,\nonumber } \end{array}\ ] ] where @xmath89 .
the effect of @xmath75 is restricted to generate in - plane superconducting correlations . to further simplify the analysis and to reduce computational efforts ,
two sublattices are assumed in the @xmath60 plane since it is known that this model has a tendency toward antiferromagnetism at half - filling : e.g. , @xmath90 and @xmath91 for @xmath92 ( a - sublattice ) and @xmath93 ( b - sublattice ) , respectively .
more details can be found in @xcite . to properly describe a stable junction made out of materials with different electronic densities , it is crucial to include long - range coulomb interactions between electrons and also with the background of positive charges . in this paper ,
the long - range coulomb interactions are treated within the hartree approximation by using the following poisson s equation @xcite : @xmath94,\ ] ] where @xmath95 ( @xmath96 , @xmath97 , and @xmath98 are the dielectric constant , electronic charge , and lattice spacing , respectively ) .
@xmath99 is the background positive charge density at site * i*. in the following , @xmath97 and @xmath98 are set to be 1 .
furthermore , @xmath99 is kept constant to a number @xmath100 ( @xmath101 ) on the left ( right ) side of the system .
the poisson s equation ( [ poisson ] ) is solved numerically using symmetric discretizations in the @xmath52 and @xmath81 directions , and a forward discretization in the @xmath62 direction , e.g. , @xmath102 ( * z * : unit vector in @xmath62 direction ) .
the use of the hartree approximation here is not a severe problem : the poisson equation is widely considered to be a reasonable starting point to account for charge transfer , and it is much used in the study of semiconducting systems . to mimic a ferromagnetic half - metallic ground state on the left side of the system , the localized spins @xmath103 are chosen to be ferromagnetically aligned , pointing into the @xmath62 direction . to mimic an a - type af state ,
these classical spins are chosen with the same orientation in the @xmath60 planes but opposite between adjacent planes .
on the cuprate side , the mean - field hamiltonian @xmath104 is first fourier transformed to momentum space in the @xmath60 plane , and the resulting bogoliubov - de gennes equation @xcite is solved numerically by diagonalizing a @xmath105 hamiltonian matrix for each momentum @xmath106 .
the @xmath77-wave order parameter is @xmath107 , where @xmath108 , and * x * and * y * are unit vectors in the @xmath52 and @xmath81 directions , respectively .
the description of the numerical results starts with a qualitative reproduction of the recent experiments @xcite , where transfer of charge from lcmo to ybco was found to destroy superconductivity at the interface . in fig .
[ fm_af_sc ] , the case when a fm metal forms an interface with a superconducting state ( as in fig . [ systems](a ) ) is studied .
the work functions and parameters are chosen such that a transfer of charge from the left to the right takes place , as shown in fig .
[ fm_af_sc](a ) where an accumulation of charge at the interface is found .
the band - bending picture , which appears to be true not only for doped band insulators but also for hubbard insulators @xcite , suggests that there will be a finite region where the density will be @xmath109=1 , with the chemical potential moving across the hubbard gap . in fig .
[ fm_af_sc](b ) the local magnetization is shown .
on the left , it is fm as expected in a one - orbital manganite model away from @xmath109=0 or 1 .
on most of the right , the magnetization is zero as in a regular superconductor .
however , in the interface region the magnetization develops a staggered character suggesting the presence of af order .
thus , as expected from the mean - field formalism the density @xmath12 is associated with antiferromagnetism .
figure [ fm_af_sc](c ) shows the suppression of the superconducting order parameter at the interface .
overall , these simple results reproduce properly the experiments where it was observed that transfer of charge from a manganite lcmo to ybco led to the suppression of superconductivity at the interface @xcite . after having crudely reproduced the main qualitative aspects of previous experiments , the same formalism can be used to formulate predictions for other systems .
of particular interest is the combination of an a - type af state ( as it occurs for instance in undoped or highly doped lsmo and also in bilayer compounds @xcite ) and an @xmath110 cuprate . if the transfer of charge occurs in the same directions as before , as already suggested by the discussion on experimental work functions
, then it would be expected that at the interface a density larger than @xmath109=1 would be produced ( see the sketch in fig .
[ systems](b ) ) .
if this doping is as large as 5 - 10% , then an electron doped superconductor could be created in a real system .
the actual calculations are conceptually simple and the results are shown in fig .
[ a - af_sc ] . in ( a )
, the density is presented : once again a robust region in parameter space is identified where the transfer of charge occurs from the af manganite to the af cuprate , leading in this case to @xmath109 larger than 1 at the interface . in ( b ) , the local magnetization is shown . in this case , the g - type af order is seen on most of the right ( cuprate ) region , but this magnetic order disappears at the interface due to the transfer of charge .
finally , in ( c ) the superconducting order parameter is shown to become nonzero at the interface .
thus , as anticipated from the introduction , an electron - doped superconductor is predicted to occur in the manganite / cuprate system described here , within a simple mean - field approximation , for a pair of oxides with the proper location of chemical potentials .
if the a - type af manganite state for the left side is replaced by a fm state , as before , then a similar transfer of charge occurs and within mean - field a sc state is also generated .
however , it is known that the proximity - effect influence of ferromagnetism is detrimental to singlet superconductivity , thus for experimental realizations of this scenario , a non - fm state appears more suitable .
_ note _ : if the transfer of charge would occur in the other direction , namely from the @xmath109=1 af to a doped fm material , then the cuprate interface would have less charge and a narrow layer with the properties of a @xmath111 cuprate could be expected .
thus , the simple motive of this effort works both ways at the mean - field level .
however , the analysis of work functions from the experimental viewpoint , already discussed , suggests that in order to generate interfacial superconductivity involving a manganite , electron doping is the most likely outcome since the chemical potential of lsmo is above that of lco , nco , sco , and ybco .
the generation of a superconductor via transfer of charge from another compound is certainly not restricted to occur only when lsmo is involved , but it should happen under far more general circumstances .
for instance , the case of a band - insulator ( bi ) forming an interface with an af was also studied .
figure [ bi_af](a ) shows the density profile , indicating that the parameters of the calculation are such that the transfer of charge this time occurs from the af to the bi , inducing a region in the cuprate with hole doping , and concomitant superconductivity ( as shown in ( b ) ) .
certainly , electron - doped sc can be induced as well by adjusting the work functions .
if instead of a bi , a standard metal is used ( modeled in our study by merely using hubbard @xmath73=0 in a tight - binding model ) , then for appropriate work functions also a charge transfer is to be expected ( see ( c ) and ( d ) ) . thus , the most important aspect of the problem is to identify materials with the proper relative location of work functions such that the charge transfer occurs in the proper direction , and also such that there is a good matching between lattice spacings to avoid generating extra complications for the transfer of charge to occur .
to place our findings on a firmer footing , we describe here the results obtained from cluster dynamical mean field theory ( cdmft ) @xcite generalized to a layered geometry for the case of a ferromagnetic mott insulator interface .
the previous section showed that there are many similarities in the use of fm or af manganites in the heterostructure , since their main role is the donation of carriers to the cuprate . since technically the case of fm is simpler , in this section we focus on a fm manganite .
cdmft is a powerful technique which maps a full many - body problem onto that of a cluster embedded in a self - consistent medium .
the model hamiltonian considered here is identical to eqs . 1 and 2 , with the addition of the local coulomb interaction represented by the hubbard @xmath73 on the mott insulating side of the interface ( and without the explicitly attractive nearest - neighbors attraction ) .
long - range coulomb interactions are treated again using the poisson equation as in eq .
cdmft can be generalized to treat the effects of a layered geometry in the same manner as dmft was adapted to treat a film geometry @xcite .
each layer is mapped onto an independent cluster impurity embedded in a medium .
the solution for the self - energy of each layer is then used via a self - consistency condition to obtain the local green s function of the entire lattice . hopping between the layers
is taken into account only at the level of the self - consistency condition . to make the problem computationally feasible
, we fix the number of layers to @xmath112 and @xmath113 for the ferromagnet and the mott insulator , respectively .
further , each layer is mapped to a 4-site cluster embedded in a bath of 8 sites .
sc correlations are included in the bath around the mi layers to allow for the sc instability and a paramagnetic solution is imposed .
individual layers are oriented in the @xmath60-plane and stacked in the @xmath62-direction .
the algorithm starts by making an initial guess for the local chemical potential and the bath parameters across the layered structure . solving the cluster impurity model in each layer using the lanczos method
we obtain the cluster green s function , cluster density , and self - energy .
the density in each layer is used to obtain the new local potential @xmath114 using the poisson equation ( eq . 4 ) .
the self - energy in each layer is used to obtain a local green s function for the lattice using the equation below : @xmath115 where @xmath116^{-1}. } \end{array}\ ] ] here , @xmath117 denotes the fourier transform of the hopping matrix ( both intralayer and interlayer ) , @xmath118 and @xmath119 denote the superlattice momenta within each layer , and @xmath62 denotes the layer index . @xmath120
includes all the terms that couple to the density , including the chemical potential , @xmath114 , site potential , and the work function on either side of the interface . the green s function @xmath121 is combined with the dyson equation to give a new weiss field that is then used to obtain a new set of bath parameters using a conjugate gradient minimization .
convergence is achieved when the bath parameters for each layer and the density profile across the layers do not change with further iterations . in fig .
[ interface](a ) the density profile is presented .
the bulk density is at @xmath109=0.7 on the left side ( mn oxide ) and @xmath109=1.0 on the right ( cu oxide , @xmath122=8 ) .
as expected from the previous discussions , electron doping of the mott insulator occurs at the interface due to a transfer of charge from the manganite to the cuprate .
figure [ interface](b ) provides the magnetization at each layer .
note that the effects of penetration of the fm moment onto the mott insulator side are not considered in this calculation . in a real setup ,
it is likely that a finite polarization will be found on the right hand side of the figure .
figure [ interface](c ) shows the @xmath77-wave sc order parameter , which in this heterostructure becomes non - zero in the cuprate s interfacial region and decays rapidly into the bulk of the mott insulator .
thus , an electron - doped @xmath77-wave superconductor can be realized at the interface of a manganite and a cuprate , due to the expected transfer of charge between them , even when dynamical effects in the mean - field approximation are taken into account .
in this manuscript , the possible charge transfer from a manganite to an undoped cuprate was discussed in the context of oxide heterostructures .
this issue is nontrivial since @xmath98 @xmath123 the existence of a robust gap in the undoped cu oxides would have suggested the lack of available states for manganite electrons to pour into the antiferromagnetic cuprates near the interfaces .
the recently discovered @xmath124 nature of the cu - oxides gaps effectively reduces the magnitude of the hubbard gap . taking this into consideration
, our analysis suggests that the fermi level of manganites , in a robust range of hole doping , lies above the chemical potential of several cuprates and even above the bottom of the upper hubbard band of sco and nco .
this opens the possibility of electron doping of high-@xmath0 s at interfaces with manganites ( and other oxides ) .
under idealized conditions , the doping discussed here may induce a superconducting state ( if the manganite is not ferromagnetic ) , an exotic metallic state containing polarized carriers in an antiferromagnetic background ( if the manganite is ferromagnetic ) , or still a superconducting state , likely with a triplet component , if the spin polarization of the carriers coming from the manganite is only partial
. an important component of our effort is the introduction of a systematic procedure to analyze photoemission and diffusion voltage experiments to predict the direction of flow of charge at interfaces .
a figure with the relative fermi levels and gap locations of several oxides was presented .
if the superconducting state is ever realized in clean interfaces , a fundamental issue to address is the unveiling of the phase diagram of cuprates in the absence of quenched disorder .
this is true for both electron and hole doping .
recent phenomenological calculations @xcite involving noninteracting carriers in interaction with the af and sc order parameters ( an extension of the traditional landau - ginzburg approach ) suggest that these phases should be separated by either a region of local coexistence or a first - order transition . a glassy state , as in the widely studied phase diagram of lsco , was reproduced @xmath125 incorporating quenched disorder . for a sketch of these diagrams
see fig .
[ phase_sc ] .
the generation of superconductivity at interfaces may reveal the true phase diagram of clean cuprates .
it is clear that our calculations can only be considered as suggestive , and its main purpose is to induce further work in this area .
important simplifications employed in the study , including the neglect of lattice reconstructions , vacancies , and polarity effects , need to be addressed .
@xmath126 @xmath127 calculations are crucial to clarify these issues . and , of course , the experimental realization of the interfaces discussed here would provide a definitive answer to the proposal of electron doping of cuprates at the interfaces .
we thank i. bozovic , h. y. hwang , m. kawasaki , t. kopp , h. kumigashira , s. pennycook , a. sawa , y. tokura , and m. varela for valuable discussions .
this work was supported in part by the nsf grant dmr-0443144 and by the division of materials sciences and engineering , office of basic energy sciences , u.s .
department of energy , under contract de - ac05 - 00or22725 with oak ridge national laboratory , managed and operated by ut - battelle , llc .
support by the ldrd program at ornl is also acknowledged . for recent work in this area see n. pavlenko , i. elfimov , t. kopp , and g. a. sawatzky , cond - mat/0605589 , and references therein . for related literature
see : v. koerting , q. yuan , p.j .
hirschfeld , t. kopp , and j. mannhart , phys .
b * 71 * , 104510 ( 2005 ) ; n. pavlenko and t. kopp , phys . rev .
b * 72 * , 174516 ( 2005 ) .
m. varela , a.r .
lupini , v. pea , z. sefrioui , i. arslan , n.d .
browning , j. santamaria , s.j .
pennycook , cond - mat/0508564 ; see also todd holden , h .- u .
habermeier , g. cristiani , a. golnik , a. boris , a. pimenov , j. humlicek , o. i. lebedev , g. van tendeloo , b. keimer , and c. bernhard , phys . rev .
b * 69 * , 064505 ( 2004 ) ; a. hoffmann , s. g. e. te velthuis , z. sefrioui , j. santamaria , m. r. fitzsimmons , s. park , and m. varela , phys .
rev . b*72 * , 140407(r ) ( 2005 ) ; v. pea , c. visani , j. garcia - barriocanal , d. arias , z. sefrioui , c. leon , j. santamaria , and c. a. almasan , phys .
b * 73 * , 104513 ( 2006 ) .
h. hilgenkamp and j. mannhart , rev .
* 74 * , 485 ( 2002 ) ; and references therein .
see also b. nikolic , j. k. freericks , and p. miller , phys .
b * 65 * , 064529 ( 2002 ) ; u. schwingenschlogl and c. schuster , cond - mat/0702098 .
z. sefrioui , d. arias , v. pea , j. e. villegas , m. varela , p. prieto , c. len , j. l. martinez , and j. santamaria , phys .
b * 67 * , 214511 ( 2003 ) . see also v. pea , z. sefrioui , d. arias , c. len , j. santamaria , j. l. martinez , s. g. e. te velthuis , and a. hoffmann , phys .
lett . * 94 * , 057002 ( 2005 ) ; j. chakhalian , j. w. freeland , g. srajer , j. strempfer , g. khaliullin , j. c. cezar , t. charlton , r. dalgliesh , c. bernhard , g. cristiani , h - u .
habermeier , and b. keimer , nature physics * 2 * , 244 ( 2006 ) .
the @xmath128 doping of an undoped cuprate when in combination with sto has already been reported in s. oh , m. warusawithana , and j. n. eckstein , phys .
b * 70 * , 064509 ( 2004 ) . see also j. n. eckstein and i. bozovic , annu .
sci . * 25 * , 679 ( 1995 ) ; g. yu .
logvenov , a. sawa , c. w. schneider and j. mannhart , app .
83 * , 3528 ( 2003 ) ; and references therein .
note that the chemical potential difference between lco and nco is still controversial .
other estimations arrive to a value @xmath129 ev for this quantity .
see p. g. steeneken , l. h. tjeng , g. a. sawatzky , a. tanaka , o. tjernberg , g. ghiringhelli , n. b. brookes , a. a. nugroho , and a. a. menovsky , phys .
lett . * 90 * , 247005 ( 2003 ) . the work function of ybco has been experimentally measured . according to t. hirano , m. ueda , k. matsui , t. fujii , k. sakuta , and t. kobayashi , jpn .
* 31 * , l1345 ( 1992 ) , the ybco work function is 6.1 ev .
since the work function of lsmo is @xmath30 ev @xcite , their difference is 1.3 ev , which is not too different ( given the uncertainties in these experimental results ) from the numbers deduced from fig .
[ fig : lco_sco_nbsto_ybco_lsmo ] . are there other transition - metal oxides that can replace manganites in this context ?
consider first the case of ruthenates .
in addition to the @xmath130 oxides , clearly the @xmath131 transition - metal oxides are another potentially important class of materials for our purposes . among them , srruo@xmath1 ( sro ) is known to be highly conductive and , therefore , a potential candidate for electrode in transition - metal based devices . however , through permittivity measurements , the work function of ( 001 ) sro is estimated to be 5.2 ev ( see x. fang and t. kobayashi , appl .
a * 69 * , s587 ( 1999 ) ) .
this sro work function is rather large compared with the value @xmath132 ev of the double - layered la@xmath43sr@xmath44mn@xmath4o@xmath45 and @xmath47 of the cubic perovskite lsmo @xcite .
this may indicate the following trend : the larger the atomic number of the transition - metal is , the larger the work function becomes .
in other words , if one wants to inject electrons to some material , one may need to use transition - metal oxides with @xmath133 atomic number than manganites ( since reducing the atomic number also reduces the electro - negativity ) from this point of view , electron - rich titanates ( via the addition of nb ) can provide another route for electron doping of cuprate parent compounds , as very recently experimentally reported @xcite .
our study is by no means restricted to manganites as electron donors .
e. dagotto , j. riera , y. c. chen , a. moreo , a. nazarenko , f. alcaraz , and f. ortolani , , 3548 ( 1994 ) ; and references therein .
see also r. micnas , j. ranninger , and s. robaszkiewicz , rev .
phys . * 62 * , 113 ( 1990 ) ; i. martin , g. ortiz , a. v. balatsky , and a. r. bishop , int .
j. of mod .
* 14 * , 3567 ( 2000 ) . | the electron doping of undoped high-@xmath0 cuprates via the transfer of charge from manganites ( or other oxides ) using heterostructure geometries is here theoretically discussed .
this possibility is mainly addressed via a detailed analysis of photoemission and diffusion voltage experiments , which locate the fermi level of manganites above the bottom of the upper hubbard band of some cuprate parent compounds . a diagram with the relative location of fermi levels and gaps for several oxides
is presented .
the procedure discussed here is generic , allowing for the qualitative prediction of the charge flow direction at several oxide interfaces .
the addition of electrons to antiferromagnetic cu oxides may lead to a superconducting state at the interface with minimal quenched disorder .
model calculations using static and dynamical mean - field theory , supplemented by a poisson equation formalism to address charge redistribution at the interface , support this view .
the magnetic state of the manganites could be antiferromagnetic or ferromagnetic .
the former is better to induce superconductivity than the latter , since the spin - polarized charge transfer will be detrimental to singlet superconductivity .
it is concluded that in spite of the robust hubbard gaps , the electron doping of undoped cuprates at interfaces appears possible , and its realization may open an exciting area of research in oxide heterostructures . |
the disorder induced metal - insulator transition , a genuine quantum phase transition is one of the most studied phenomena of condensed matter physics since the seminal paper published over five decades ago . @xcite according to the original problem , the hamiltonian @xmath7 describes the behavior of non - interacting spinless electrons in disorder . the first term in eq .
( [ eq : am ] ) represents an onsite disordered potential , where the energies , @xmath8 , are independent , uncorrelated random variables , drawn from a distribution function , @xmath9 , whose form is usually chosen to be uniform over an energy range that is symmetric around @xmath10 , but other forms , e.g. gaussian or binary distributions could be used , as well . the second term in eq .
( [ eq : am ] ) is the kinetic energy describing the hopping of the particles over a regular lattice , but restricted to nearest neighbors only .
the energy scale associated to the hopping process , @xmath11 , can be taken as the unit of energy ( @xmath12 ) .
the sites form a regular , usually simple cubic lattice .
the embedding dimension , @xmath13 , of the system is a very important parameter , since phase transition occurs for @xmath14 only .
@xcite besides diagonal disorder resembling substitutional disorder the other main cause of irregularity in condensed systems is structural disorder .
for the investigation of topological and structural disorder percolation is one of the most important and widely used models .
percolation in general has a wide applicability in many fields of physics .
@xcite in the bernoulli site - percolation problem every site is filled with probability @xmath15 and is empty with probability @xmath16 independently .
the main goal of classical percolation is to tell for a given @xmath15 whether an infinite cluster of filled sites may exist in the thermodynamical limit or not .
it turns out , that there is such a critical probability , @xmath17 , below which , @xmath18 , there is no infinite cluster but above which , @xmath19 , there is . in one dimension
@xcite @xmath20 , in two dimensions @xcite @xmath21 , in three dimensions @xcite @xmath22 . in the @xmath19 case
the existence of an infinite cluster ensures that the system can be treated as a conductor , since classical particles can travel through the whole system . on the other hand if @xmath18 , the system consists of a set of disjoint , finite clusters , and as a consequence , it behaves as an insulator , since no particle can escape from its initial finite cluster . for the electric conduction properties of a sample the electrons are responsible
whose behavior is described very well by quantum mechanics , therefore we shall investigate spinless non interacting electrons on a percolated lattice , this is called the quantum percolation model . omitting spin and interaction is necessary , because even with these simplifications the problem seems to be hard to solve .
the corresponding hamiltonian is @xmath23 where @xmath24 is the set of filled sites , @xmath25 is a constant on site energy , whose value can be safely set to zero without loss of generality .
note that the pure site
percolation problem is equivalent to a binary anderson model @xcite with constant @xmath26 and @xmath27 but taking the limit @xmath28 : @xmath29 this hamiltonian could describe an alloy of a perfect metal consisting of atoms @xmath24 and a perfect insulator consisting of atoms @xmath30 only .
all @xmath24 sites are equivalent , and the @xmath30 sites can not be reached due to their infinite on - site energy , therefore @xmath30 sites behave as if they were empty .
this suggests , that quantum percolation behaves similar to the anderson model . in our present work
we shall show many similarities .
the most important similarity with the anderson problem is the existence of a metal insulator transition for the quantum percolation model too , however , here @xmath15 , or strictly speaking @xmath31 , plays the role of disorder : for @xmath18 every state is localized onto finite , connected islands , thus the sample is an insulator .
increasing @xmath15 beyond @xmath17 , however , a classical particle can travel through the sample , the electron wave functions are localized due to strong interference effects caused by disorder , the sample still remains an insulator .
for @xmath15 values slightly below @xmath32 states are perturbed bloch - states , the sample is a metal . in between
there exists a mobility edge , @xmath33 , an energy
dependent quantum critical point , below which electronic eigenstates are anderson - localized giving rise to an insulator , and above which they are extended forming a metal . along the mobility edge , @xmath33 ,
the states are supposed to be multifractals . in sec .
[ sec : fss_qperc ] we argue , that the anderson model and the quantum percolation model belong to the same universality class . the organization of the paper is the following .
in the next section , sec . [
sec : theory ] we look at the peculiar properties of the density of states in quantum percolation and provide an overview about multifractality together with an introduction about the finite - size scaling analysis of the corresponding generalized dimensions .
[ subsec : fss_anderson ] we give a short overview of the technique of the latter analysis in the case of the 3d anderson transition , in sec .
[ sec : fss_qperc ] we provide with the methods applied in the present work , and in sect .
[ sec : gmfe_qperc ] we present the results of our analysis for the multifractal analysis .
finally sec .
[ sec : summ ] is left for a summary .
electronic conduction is only possible on an infinite cluster , so @xmath34 is expected , therefore the infinite cluster should be investigated , so only the @xmath19 regime is interesting for us . since numerically we can deal with a finite lattice only , we restricted our work on the largest finite cluster found by a hoshen - kopelman algorithm @xcite . in a finite size sample
the hamiltonian , eq .
( [ eq : qperc_qperchami ] ) is a huge sparse matrix . to obtain the spectrum and eigenfunctions we used the jacobi - davidson method encoded in the primme package @xcite with ilu preconditioning , using the ilupack package @xcite .
at first let us take a glance at the density of states ( dos ) because for the quantum percolation problem it deserves a special attention .
the dos of the giant cluster has itself an unusual form .
the evolution of this function with @xmath15 is depicted in fig .
[ fig : qperc_dos ] . with increasing disorder , in the present case
this means decreasing @xmath15 , more and more sharp peaks appear in the spectrum .
these peaks correspond to special , so - called molecular states , which are localized to a few sites @xcite .
these states are non - zero on a few sites only and exactly zero on every other one due to exact destructive interference .
therefore they are not localized in the sense of anderson localization , there is no exponential decay in the wave function envelope .
typical few - site structures and corresponding energies are given on the right side of fig .
[ fig : qperc_dos ] .
since the value @xmath35 appears for most clusters as an eigenvalue , the highest peak of the dos is at the middle of the band , and there is also a pseudo gap around it .
c c c figure1a ( 0,70)(a ) & figure1b ( 0,70)(b ) & [ 2]*[3.5 cm ] figure1e ( 0,103)(e ) + figure1c ( 0,70)(c ) & figure1d ( 0,70)(d ) & + considering other few - site clusters there is no reason for the eigenvalues to avoid any part of the band , therefore peaks in the dos corresponding to molecular states should appear densely in the thermodynamic limit .
the energy of a molecular state is a strict value , thus the peaks in the dos appear as a series of dirac - deltas .
as we can see , the spectrum consists of two parts : a dense point spectrum due to molecular states , and a continuous one due to all other states .
@xcite this statement has been rigorously proven recently in the case of a 2d square lattice , and for tree graphs corresponding to an effective infinite dimension , therefore it is conjectured to be true in any dimension .
@xcite since molecular states are strongly localized , they can not contribute to conduction .
therefore we restrict our investigation to the continuous part of the spectrum only . with the numerical method described above
we are able to compute one single eigenstate of the hamiltonian having an eigenenergy close to a given value of @xmath36 . in fig .
[ fig : qperc_dos ] it is shown , that in a finite system molecular states appear frequently at few special energies only , e.g. @xmath37 , therefore for our purpose we have chosen energy windows avoiding the peaks in the dos . the cubic lattice is a bipartite lattice and the hamiltonian ( [ eq : qperc_qperchami ] ) couples nearest neighbors only , therefore from one sublattice , @xmath38 , it is possible to hop to the other sublattice , @xmath39 , only . the hamiltonian anticommutes with an operator @xmath40 , which is @xmath32 on sublattice @xmath38 , and @xmath41 on sublattice @xmath39 , thus @xmath40 acts as a chirality transformation .
@xcite therefore the quantum percolation model is symmetric not only on average for the exchange of eigenenergies , @xmath42 , but for every single disorder realization . in the low ( high )
energy range the states have antibonding ( bonding ) character . in the middle of the band , around @xmath35 , chessboard - like
chiral states appear .
these chiral states exactly at @xmath35 are eigenfunctions of @xmath40 , as well , therefore they are protected against off - diagonal disorder . in order to understand the sub gap appearing around the middle of the band , @xmath35
, we invoke the arguments of ref . .
the square of the hamiltonian , @xmath43 , connects the sites of the same sublattice only , see fig .
[ fig : qperc_renorm_hami ] , thus one can , , renormalize @xmath43 acting on one of the sublattices @xcite .
\(1 ) @xmath38 ; ( 2 ) [ right of=1 ] @xmath39 ; ( 3 ) [ right of=2 ] @xmath38 ; ( 4 ) [ below of=1 ] @xmath39 ; ( 5 ) [ below of=2 ] @xmath38 ; ( 6 ) [ below of=3 ] @xmath39 ; ( 7 ) [ below of=4 ] @xmath38 ; ( 8) [ below of=5 ] @xmath39 ; ( 9 ) [ below of=6 ] @xmath38 ; \(1 ) edge [ bend right ] ( 3 ) edge [ ] ( 5 ) edge [ bend left ] ( 7 ) ( 3 ) edge [ ] ( 5 ) edge [ bend right ] ( 9 ) ( 5 ) edge [ ] ( 7 ) edge [ ] ( 9 ) ( 7 ) edge [ bend left ] ( 9 ) ; the vicinity of @xmath35 belongs to the low - energy regime of the spectrum of @xmath43 , therefore here antibonding states should appear , which are more or less visible in the wave functions themselves , too .
but the hopping elements to the diagonal - lying second neighbors in fig . [ fig : qperc_renorm_hami ] introduce triangles . triangles and the antibonding nature together lead to frustration . based on the frustration of the states around zero energy naumis _ et .
@xcite showed in two dimensions , that the width of the pseudogap around zero energy , @xmath44 , is connected to the peak at @xmath35 : @xmath45 , where @xmath46 stands for the weight of the zero energy states in the spectra .
they also showed , that the width of the pseudogap tends to zero in the non - disordered limit , @xmath47 .
the extension of these arguments to three dimensions should be valid , since the most important ingredient of their calculation is the coordination number of the lattice , and not the dimensionality itself explicitly .
the states close to @xmath35 belong to the edge of the spectrum of @xmath43 , which is a disordered hamiltonian .
therefore the pseudogap might be qualitatively interpreted as the lifshitz tail of @xmath43 , leading to localized states close @xmath35 . in recent high - precision calculations
@xcite the so - called multifractal exponents ( mfe ) have been used to describe the anderson metal insulator transition ( amit ) .
the renormalization flow of the amit as mentioned in the introduction has three fixed points : a metallic , an insulating and a critical one . in the metallic fixed point
every state is extended with probability one , thus with increasing system size , the effective size of the states also grows proportional to the volume .
so the fractal dimension of the states , that will be defined more precisely later , is just the embedding dimension @xmath48-independently , @xmath49 . in the insulating fixed point
every state is exponentially localized , their effective size does not change with growing system size , thus for @xmath50 @xmath51 , for @xmath52 @xmath53 . at criticality
the system does not change during renormalization , thus it must be statistically the same on all length scales showing scale independence , which means self similarity .
therefore wave functions are multifractals , in other words generalized fractals @xcite , see fig .
[ fig : fss_eigvec_anderson_qperc ] .
c c c metal / extended & critical / multifractal & insulator / localized + + figure3a ( 0,100)(a ) & figure3b ( 0,100)(b ) & figure3c ( 0,100)(c ) + figure3d ( 0,100)(d ) & figure3e ( 0,100)(e ) & figure3f ( 0,100)(f ) + in our case there is a @xmath13-dimensional hypercubic lattice with linear size @xmath54 , and a normalized wave function whose support is this lattice , @xmath55 , defining a probability distribution .
let us divide this lattice into smaller hypercubes ( boxes ) with linear size @xmath56 , and introduce the ratio @xmath57 .
then coarse graining @xmath58 , in other words summing all its values in the @xmath59th box we obtain : @xmath60 where @xmath61 is the weight associated to the @xmath59th box termed as box probability .
let us define the @xmath48th moment of the mass , frequently called generalized inverse participation ratio ( gipr ) , and its derivative as @xmath62 where @xmath63 is the finite system mass exponent . @xmath63 and its derivative read as : @xmath64
taking the @xmath65 limit , which is equivalent to taking the @xmath66 limit , the mass exponent and its derivative are @xmath67 @xmath68 can be written in the form @xmath69 where @xmath1 is the generalized fractal dimension . in this expression @xmath70 is the anomalous scaling exponent : @xmath71 the quantities @xmath68 , @xmath2 , @xmath1 , and @xmath70 are often referred as multifractal exponents ( mfes ) , while the finite system version of these exponents , @xmath72 , are called generalized multifractal exponents ( gmfes ) .
@xmath1 is directly related to the so - called _ rnyi - entropy _ , @xmath73 ,
which in the limit @xmath74 yields the well - known _ shannon - entropy _ , i.e. @xmath75 .
this is the reason why @xmath76 is also referred as information dimension : @xmath77 while another frequently used dimension is the correlation dimension , @xmath78 .
the latter dimension appeared often in recent studies of the physical relevance of multifractal eigenstates @xcite there is another way to characterize the multifractal nature of the wave functions .
for that purpose the box probability , @xmath79 can be transformed into another variable , @xmath80 assuming the fractal scaling @xmath81 let us denote the probability density function of the number of boxes having a value @xmath38 with @xmath82 .
the scaling of @xmath82 is described through the singularity spectrum @xmath5 , which is the fractal dimension of the number of boxes having a value @xmath38 : @xmath83 function @xmath5 is nothing else but the legendre - transform of @xmath68 : @xmath84 according to recent results a symmetry relation exists for @xmath2 and @xmath70 given in the form @xcite : @xmath85 this relation first obtained for some random matrix ensemble numerically and using the supersymmetric non - linear sigma model analytically @xcite was later confirmed for several two dimensional @xcite and three dimensional systems @xcite .
however , deviations have been detected in other cases .
@xcite the robustness of this relation has been investigated also for many - body localization .
@xcite finite size scaling techniques are very well described by rodriguez _ et .
@xcite for the anderson model .
we are going to use their notation , therefore we denote disorder by @xmath86 . in this subsection
we extend the formalism of ref . .
from the eigenfunction the @xmath87 and @xmath88 values can be computed for every state at different @xmath48 values . at fixed disorder ,
@xmath86 , system size , @xmath54 , and box size , @xmath56 , every gmfe is computable from these two quantities the following way @xcite : @xmath89 where @xmath90 stands for averaging : @xmath91 and @xmath92 denote the _ ensemble _ and _ typical _ averaging .
every gmfe approach the value of the corresponding mfe at the critical point only in the limit @xmath66 .
close to the critical point due to standard finite size scaling arguments we can suppose , that @xmath87 and @xmath88 shows scaling behavior determined only by the ratio of two length scales , @xmath54 and @xmath56 , and the localization / correlation length , @xmath93 , in the insulating / metallic phase : @xmath94 according to ( [ eq : fss_fsgmfe_sub1])([eq : fss_fsgmfe_sub4 ] ) for all gmfes the scaling - law holds independently from the type of averaging @xcite : @xmath95 equations ( [ eq : fss_tauwll])([eq : fss_deltawll ] ) can be summarized in one equation : @xmath96 @xmath97 on the left and @xmath98 on the right hand side can be changed to @xmath99 and @xmath100 : @xmath101 at fixed @xmath102 , @xmath103 in eq .
( [ eq : fss_scalinglaw_llambda ] ) can be considered as the constant term of @xmath104 , therefore @xmath105 where the constant @xmath102 has been dropped .
@xmath104 can be expanded with one relevant , @xmath106 , and one irrelevant operator , @xmath107 , the following way using @xmath108 : @xmath109 all the disorder - dependent quantities in the above formula can be expanded in taylor - series : @xmath110 the number of parameters is @xmath111 . for fixed @xmath56 the scaling law given in eq .
( [ eq : fss_scalinglaw_ll ] ) has to be considered .
the expansion of @xmath113 in ( [ eq : fss_scalinglaw_ll ] ) is @xmath114 choosing @xmath112 , and considering that in most cases @xmath115 and @xmath116 are constant , i.e. @xmath117 , the last term can be merged with the relevant part .
equation ( [ eq : fss_scalinglaw_ll ] ) has the following form for fixed @xmath112 : @xmath118 the taylor - expansions of the above functions are @xmath119 the number of parameters is @xmath120 .
we can see , that the number of parameters grows as @xmath121 for fixed @xmath112 , instead of @xmath122 as for fixed @xmath102 .
this makes the fitting procedure definitely much more difficult .
before turning to the analysis of our simulations on the 3d quantum percolation model , we briefly review the details of the finite size scaling using gmfes but first based on the 3d anderson model .
the aim of this section is twofold .
first of all we present the advantages and disadvantages of the various methods used and their applicability for our purposes .
second we show the precision of these techniques for the case of a well - studied case , the anderson transition .
our first goal was to check our numerical algorithm on the well - known anderson problem . based on ref .
we formulate two cases : at first fixing @xmath102 then fixing @xmath56 .
since the metal - insulator transition occurs at the band center @xcite ( @xmath35 ) at disorder @xmath123 , most works study the vicinity of this point . to have the best comparison
, we analyzed this regime also , therefore about @xmath124 disorder values were taken for the range @xmath125 .
system sizes were taken from the range @xmath126 , the number of samples were @xmath127 at least .
we considered only one wave function per realization , the one with energy closest to zero in order to avoid correlations between wave functions of the same system @xcite . from the wave function
the @xmath87 and @xmath88 multifractal moments were calculated in the range @xmath128 at fixed @xmath129 . in eqs .
( [ eq : fss_tauwll])([eq : fss_deltawll ] ) only two scaling functions are present , @xmath130 and @xmath131 , therefore we investigated @xmath132 and @xmath133 only using ensemble and typical averaging ( see sec .
[ sec : mfss ] ) . in order to fit the scaling law ( [ eq : fss_anderson_scalinglaw ] ) we used minuit . to find the best fit to the data obtained numerically the order of expansion of @xmath134 , @xmath135 and @xmath115 must be decided by choosing the values of @xmath136 and @xmath137 . since the relevant operator is more important than the irrelevant one we always used @xmath138 and @xmath139 . to choose the order of the expansion we used basically three criteria .
the first criterion we took into account was how close the ratio @xmath140 approached one .
@xmath141 is the sum of the squared differences between the data points and the best fit weighted by the inverse variance of the data points , and @xmath142 is the number of degrees of freedom , namely the number of data points minus the number of fit parameters .
a ratio @xmath143 means , that the deviations from the best fit are in the order of the standard deviation .
the second criterion was , that the fit has to be stable against changing the expansion orders , i.e. adding a few new expansion terms . from the fits that fulfilled the first two criteria we chose the simplest model , with the lowest expansion orders .
sometimes we also took into account the error bars , and we chose the model with the lowest error bar for the most important quantities ( @xmath144 , _ etc _ ... ) , if similar models fulfilled the first two criteria .
the error bars of the best fit parameters were obtained by a monte - carlo simulation .
the data points are results of averaging , so due to central limit theorem they have a gaussian distribution .
therefore we generated gaussian random numbers with parameters corresponding the mean and standard deviation of the raw data points and then found the best fit . repeating this procedure @xmath145 times provided us the distribution of the fit parameters .
we chose @xmath146 confidence level to obtain the error bars .
we performed fss for @xmath147 and @xmath148 .
the results were very similar to the ones obtained by rodriguez _
et al . _
@xcite . in the @xmath48-range we investigated the results
were @xmath48-independent for @xmath147 and @xmath148 within @xmath146 confidence interval .
the numerical values of @xmath149 , @xmath150 and @xmath151 have been obtained in excellent agreement with the results of ref . .
hence we concluded , that our method has been confirmed .
the disadvantage of this method is , that the constant term of @xmath104 does not equal to the corresponding mfe , since @xmath102 is fixed instead of tending to zero .
it would be possible to perform multifractal finite size scaling ( mfss ) at different @xmath102-s , and then obtain the mfes for @xmath152 .
the main goal of the present work is to investigate the quantum percolation problem , where a fraction of lattice points is missing . in this case performing the coarse graining technique defined above immediate difficulties arise .
it is not clear how the @xmath56-sized boxes have to be made , or how the boxes containing different number of filled sites should be compared .
one way to resolve this problem is to choose @xmath112 , meaning that a box contains only one site .
eventhough this choice eventually opens the possibility to extend the mfss method for irregular lattices or even for graphs and networks in the future , there is also a huge cost to be paid : the smoothing effect of the coarse graining is lost , and only the more complicated method of fixed-@xmath56 technique described in sec .
[ fss_fixed_l ] remains .
there is always some numerical noise on the data , which becomes even more relevant for the smallest wave function components . in case of
negative @xmath48 these uncertain small values are dominating the sums in @xmath87 and @xmath88 ( see .
( [ eq : sqrq ] ) ) .
coarse graining clearly suppresses this effect , because for @xmath153 in an @xmath154 sized box positive and negative errors can cancel each other .
another effect is , that in a box large and small wave function amplitudes appear together with high probability . this way the relative error of a @xmath61 box probability
is reduced with coarse graining , in other words coarse graining has a nice smoothing effect . at fixed @xmath112
this effect is missing , thus for @xmath155 the numerically obtained @xmath147 and @xmath148 ( see e.g. eqs .
( [ eq : fss_fsgmfe_sub1])([eq : fss_fsgmfe_sub4 ] ) ) values are very noisy .
this makes every attempt to get results for negative @xmath48 very hard if not impossible .
the other problem is , that the scaling law becomes more complicated , the leading number of fit parameters are growing as @xmath121 for fixed @xmath112 , instead of @xmath122 as for fixed @xmath102 . performing the mfss another problem appeared with eq .
( [ eq : fss_anderson_gammal ] ) . during the fit the irrelevant exponent , @xmath151 , converged to very small @xmath156 or very large @xmath157 values . in the first case the irrelevant term can be merged with the relevant one , since @xmath115 is in most cases constant . in the second case @xmath158
suppresses the irrelevant part .
this caused really large errors in the @xmath159 , and made the whole irrelevant part meaningless . to find out whether this is just a numerical problem or there is also some systematic physical reason behind this behavior we modeled the above problem :
first a dataset was made by evaluating the expression ( [ eq : fss_anderson_gammal ] ) at system sizes and disorder we used before , with some expansion parameter values similar to the ones provided by previous mfss procedures .
of course fitting eq .
( [ eq : fss_anderson_gammal ] ) to this dataset gave a perfect fit . now adding some small random noise to the initial dataset started to shift the resulting fit parameters a little .
by increasing the noise to the order of the standard deviation of the original dataset for the anderson model the fit showed the expected phenomenon : the irrelevant exponent , @xmath151 , converged to either large or small values .
this shows , that this is just a numerical artifact .
there is a shift on the @xmath160 curves for different system sizes , see fig .
[ fig : qperc_fss_alphaqdq_raw ] .
this shift comes mainly from the @xmath161 term in eq .
( [ eq : fss_anderson_gammal ] ) , and if noise is present it is numerically hard to determine the effect of the @xmath158 irrelevant part .
all in all , however , in a finite system irrelevant operators are always present , considering an irrelevant term will only increase the error of the fit parameters .
therefore it seems to be useful to drop the irrelevant part , and keep the relevant one only . this way the fitting function reads as @xmath162 we performed mfss in the range @xmath163 with this formula at fixed @xmath112 for the anderson model . similarly to the case of fixed @xmath102 at fixed energy , @xmath36 , and @xmath48 one has to decide the order of the taylor - expansion of the @xmath113 scaling function . to do this we used similar criteria as before .
the only difference was , that unfortunately the fits were not so stable against changing the expansion orders , @xmath164 and @xmath165 , as the ones for fixed @xmath102 , because at fixed @xmath112 we had to fit much more parameters to the same amount of data .
the value of the critical point must be @xmath48-independent , which contrary to the case of fixed @xmath102 we had to keep also as a criterion .
we had to compare fits at different @xmath48 values and choose the lowest expansion orders that led to a @xmath48-independent critical point , and still had @xmath140 ratio close to one . in some cases we also had to leave out the smallest system size(s ) ,
i.e. choose @xmath166 or @xmath167 instead of @xmath124 to fulfill the criteria above .
the results were acceptable only approximately in the range @xmath168 . if @xmath169 fit parameters started to shift , sometimes out of the confidence band of those obtained for smaller @xmath48 values , and error bars were growing extremely large .
similar effects of growing error bars for @xmath169 has been seen earlier , on a moderate level at fixed @xmath129 , where the help of the smoothing effect of coarse graining is present .
the reason behind this is , that increasing @xmath48 increases the numerical and statistical errors through the @xmath170 expression . as mentioned above
, increasing error on the data makes it really difficult to get acceptable results from the mfss . as a result , in the range @xmath168 the critical point , @xmath149 , and the critical exponent , @xmath150 , were found to be consistent with our results at fixed @xmath129 and based on the @xmath1 and @xmath2 exponents also with high precision result of rodriguez _ et .
we observe the expected symmetry ( [ eq : multifractals_deltaalphasymmety ] ) for @xmath70 and @xmath2 , our resulting mfes fulfill these conditions in the range @xmath168 . summarizing the results it is possible to perform an mfss at fixed @xmath112 and achieve good agreement with previous high precision results @xcite .
there are certainly numerical difficulties , however , that lead us to resort to the limited range of @xmath168 only , but with further averaging the widening of this @xmath48-range seems to be possible . the main goal in the present study was to find the mobility edge and the critical exponent of the 3d quantum percolation model , and investigate the multifractal properties of the critical wave functions . since the hamiltonian eq .
( [ eq : qperc_qperchami ] ) is symmetric for @xmath171 exchange , the @xmath172 interval is investigated only .
we used the same numerics as in sec [ sec : fss_anderson_l ] . to avoid the frequent molecular states
( see sec .
[ sec : qperc_dos ] ) , and cover the most interesting regions of the band we chose the following energies : @xmath173 . for averaging we considered only one wave function per realization with the eigenvalue closest to the chosen energy @xmath36 to avoid correlations .
we only used an eigenfunction if its energy was in a @xmath174 wide interval around @xmath36 , except for @xmath175 and @xmath176 , where @xmath177 and @xmath178 were used .
our @xmath179 energy intevals are so small , that it completely excludes the effect of molcecular states .
we ran a test after the finite size scaling was performed : molecular states have strict energy value , therefore at fixed system size , @xmath54 , disorder , @xmath15 , and energy , @xmath36 , we left out from our raw dataset all the wave functions with the same energy value ( at most 2% of the original raw dataset ) .
note , that these states are not necessarily molecular states , they can be regular ones , too , having the same energy within numerical precision .
we redid our whole finite size scaling procedure ( as described below ) , but this additional refinement had no effect on the results .
this test ensures , that we filtered out the molecular states very effectively , and if they were present in our raw dataset , their effect would be negligible . at every energy we searched for the critical point @xmath180 . from the approximately @xmath181 wide neighborhood around @xmath180 we picked about @xmath124 values of @xmath15 . for higher @xmath180 values at fixed system size , @xmath54 , there are more sites in the giant cluster , thus the hamiltonian matrix is larger , and it takes more time to find the closest eigenvalue to the given energy . on the other hand @xmath87 and @xmath88 are calculated from more data , thus they are more precise .
considering these arguments we investigated system sizes and number of samples listed in tab .
[ tab : qperc_fss_systemsize ] .
altogether @xmath182 wave functions were calculated .
.system sizes and number of samples of the simulation for the 3d quantum percolation model .
[ cols="^,^,^ " , ]
mfss provided us the points of the @xmath183 and @xmath184 surface at the investigated energies and @xmath48 values . by inversion of the mobility edge curve ,
@xmath33 one can derive the mfes as a function of @xmath180 and of @xmath48 , see fig .
[ fig : qperc_dq_alphaq_shift ] . since @xmath185 , at small @xmath48 values , i.e. @xmath186 , the results for @xmath1 are @xmath180-independent , but for larger values of @xmath48 the @xmath1 starts to shift down with decreasing @xmath180 , which shows up in the lower right corner of fig . [
fig : qperc_dq_alphaq_shift](a ) . in the lower regime of fig .
[ fig : qperc_dq_alphaq_shift](c ) this shift is visibly significant .
the same phenomenon can be detected for @xmath2 .
this suggests , that @xmath1 and @xmath2 seem not to behave as universal quantities .
c c figure8a ( 0,70)(a ) & figure8b ( 0,70)(b ) + figure8c ( 0,70)(c ) ( 52,13.8 ) & figure8e ( 0,70)(d ) ( 52,13.8 ) + at relatively larger values of @xmath180 , @xmath1 and @xmath2 fulfill the symmetry relation ( [ eq : multifractals_deltaalphasymmety ] ) , see fig .
[ fig : qperc_deltaalpha_symm_e ] ( a ) , ( b ) , ( e ) and ( f ) . however , at the bottom of the mobility edge , where @xmath180 is smaller , meaning that the lattice is more diluted or more irregular , deviations from the symmetry law seem to be prominent .
the @xmath1 and @xmath2 values remain the same at small @xmath48 , i.e. when @xmath186 , but drop down as @xmath48 increases . resulting in a conclusion , that the symmetry relation , eq , ( [ eq : multifractals_deltaalphasymmety ] ) , is violated in this regime ,
see for example fig .
[ fig : qperc_deltaalpha_symm_e ] ( c ) and ( d ) .
c c figure9a ( 0,70)(a ) & figure9b ( 0,70)(b ) + figure9c ( 0,70)(c ) & figure9d ( 0,70)(d ) + figure9e ( 0,70)(e ) & figure9f ( 0,70)(f ) + the non - universality of @xmath1 and @xmath2 would automatically imply the non - universality of @xmath68 , as well . on the other hand with a legendre - transform for @xmath68
, @xmath5 can be obtained , describing the scaling of the probability distribution of the wave function amplitudes .
this distribution should be universal , therefore @xmath5 should be universal , too . using eq .
( [ eq : multifractal_falpha_tauq ] ) and ( [ eq : multifractal_tauq_dq_deltaq ] ) immediately follows : @xmath187 from the @xmath2 and @xmath1 exponents presented in fig .
[ fig : qperc_dq_alphaq_shift](a ) and ( b ) we computed the @xmath5 curve , that is depicted in fig . [ fig : qperc_falpha ] . the values from different regimes of the mobility edge seem to form a unique curve , but this is mostly due to the scale on the axis .
the upper inset of fig .
[ fig : qperc_falpha ] shows significant differences between data points at different energies .
the approximate shape of the curve is a parabola , however , a quartic curve fits the data points slightly better .
figure10a ( 50,14 ) obtained from @xmath1 and @xmath2 computed at different energies , meaning different @xmath180 .
pentagons are the results for the anderson - model , solid line is a @xmath188th order polynomial , dashed line is a @xmath189nd order polynomial .
insets are magnified parts of the curve.,title="fig : " ] ( 50,34 ) obtained from @xmath1 and @xmath2 computed at different energies , meaning different @xmath180 .
pentagons are the results for the anderson - model , solid line is a @xmath188th order polynomial , dashed line is a @xmath189nd order polynomial .
insets are magnified parts of the curve.,title="fig : " ] according to eq .
( [ eq : gmfes_f_alpha_d ] ) @xmath190 corresponds to the fixed point of the @xmath5 function , @xmath191 . for different values of @xmath180 the exponent @xmath192
is not unique , leading to a linear regime of the @xmath5 function , see the lower inset of fig .
[ fig : qperc_falpha ] .
this makes the whole legendre - transformation difficult , since it needs strict convexity .
conversely an @xmath5 that is not strictly convex would lead to ill - defined @xmath68 , @xmath1 and @xmath2 , like in our case , which contradicts universality again .
a possible resolution of this contradiction could be , that our result for the mfes is just simply not complete , perhaps a @xmath15-dependent phenomenon has not been taken into account affecting the results . since the problem appeared at the bottom of the mobility edge ,
closest to the classical percolation threshold , one possible candidate for such phenomenon is the existence of an additional length scale , namely the correlation length of the classical percolation . in order to test it
we added this length scale to the fitting function leading to a 3-variable function with number of fit parameters @xmath193 , but we could not fit so many parameters to our dataset .
there is only a small difference between the values of the mfes for the quantum percolation model and for the anderson model , see fig .
[ fig : qperc_dq_alphaq_shift ] , and the symmetry relation ( [ eq : multifractals_deltaalphasymmety ] ) is almost valid within the error bar at the bottom of the mobility edge , too , see fig .
[ fig : qperc_deltaalpha_symm_e ] .
therefore another explanation would be , that somehow we underestimated the error bars of the mfes . in the @xmath194 limit ,
our exponents seem to be close to their value for the anderson model , that together with our former claim in sec .
[ sec : fss_qperc ] about their matching universality class corroborate this possibility further .
we believe , that there is a unique and universal @xmath1 , @xmath2 and @xmath5 curve for the quantum percolation method , and it is identical with the one for the anderson model , that fulfill the symmetry relation ( [ eq : multifractals_deltaalphasymmety ] ) . as a conclusion
the present coherent set of data with a coherent technology in deriving critical exponents fulfill our expectations for larger values of @xmath195 but unfortunately unexpected deviations occur for lower values , i.e. @xmath196 .
in the present work we have numerically investigated the quantum percolation model in 3d .
we developed the mfss method by rodriguez _ et .
al _ @xcite in order to use it for irregular lattices , or even for graphs in the future .
first we tested our method on the well - known anderson - model , however , certain numerical issues forced us to restrict our analysis to the interval @xmath168 , we found @xmath48-independent results in a good agreement with the previous high precision values of ref .
then we used our method to the quantum percolation model , where we found @xmath48-independent results again .
we numerically determined the mobility edge of the system , confirming previous calculations .
we also gave an explanation for the behavior of the mobility edge near @xmath35 and at high energy .
for the critical exponent we got energy - independent values within @xmath146 confidence level .
the average of these values is the same as the one for the critical exponent for the anderson model , implying that these models belong to the same universality class .
we also determined the mfes @xmath1 and @xmath2 along the mobility edge , and for larger values of @xmath180 we found no significant difference from the anderson model confirming the statement of the same universality class further . in this regime the symmetry relation ( [ eq : multifractals_deltaalphasymmety ] ) is fulfilled . on the other hand in the case of lower @xmath180 regime
the exponents started to deviate violating universality and ( [ eq : multifractals_deltaalphasymmety ] ) , probably caused by some unexpected @xmath15-dependent phenomenon .
this behavior deserves further attention .
the authors are indebted to dr .
a. stathopoulos for his help setting up the numerical method .
financial support from otka under grant no .
k108676 , the alexander von humboldt foundation are gratefully acknowledged .
99 p. w. anderson , _ phys .
rev . _ * 109 * , 1492 ( 1958 ) .
f. evers , a. d. mirlin , _ rev .
* 80 * , 1355 ( 2008 ) and references therein .
d. stauffert and j. g. zabolitzky , _ j. phys .
_ * 19 * 3705 ( 1986 ) .
k. christensen and n. r. moloney , _ complexity and criticality _ , imperial college press , ( 2005 ) .
m. e. j. newman and r. m. ziff , _ phys .
letters _ * 85 * , 4104 ( 2000 ) .
g. schubert , h. fehske , _ lec . not .
phys . _ * 762 * , 135 ( 2009 ) .
s. kirkpatrick , t. p. eggarter , _ phys .
b _ * 6 * , 3598 ( 1972 ) .
a. kusy , a. w. stadler , g. haldas , r. sikora , _ physica a _ * 241 * , 403 ( 1997 ) . c. m. soukoulis , q. li , g. s. grest , _ phys .
b _ * 45 * , 7724 ( 1992 ) .
j. hoshen and r. kopelman , _ phys .
b _ * 14 * , 3438 ( 1976 ) .
a. stathopoulos and j. r. mccombs _ acm transaction on mathematical software _ * 37 * , 2 , 21:121:30 ( 2010 ) o. schenk , m. bollhfer and r. a. rmer , _ siam review _ * 50 * , 91 ( 2008 ) .
bordenave , a. sen , b. virg , arxiv:1308.3755 g. g. naumis , ch . wang and r. a. barrio , _ phys .
b _ * 65 * , 134203 ( 2002 )
. a. rodriguez , l. j. vasquez , k. slevin and r. a. rmer , _ phys .
b _ * 84 * , 134209 ( 2011 )
. m. janssen , _ fluctuations and localization in mesoscopics electron systems _
( world scientific lecture notes in physics - vol .
64 , singapore , 2001 ) ; m. janssen , _ phys .
_ * 295 * , 1 ( 1998 ) .
e. cuevas and v. e. kravtsov , _ phys .
b _ * 76 * , 235119 ( 2007 ) ; a. d. mirlin , y. v. fyodorov , a. mildenberger , and f. evers , _ phys . rev . lett . _ * 97 * , 046803 ( 2006 ) .
a. mildenberger and f. evers , _ phys
b _ * 75 * , 041303(r ) ( 2007 ) .
f. evers , a. mildenberger , and a. d. mirlin , _ phys .
b _ * 245 * , 284 ( 2008 ) ; f. evers , a. mildenberger , and a. d. mirlin , _ phys . rev .
* 101 * , 116803 ( 2008 ) l. j. vasquez , a. rodriguez , and r. a. rmer , _ phys .
b _ * 78 * , 195106 ( 2008 ) ; a. rodriguez , l. j. vasquez , and r. a. rmer , _ phys . rev . b _ * 78 * , 195107 ( 2008 ) ; a. rodriguez , l. j. vasquez , and r. a. rmer , _ phys .
* 102 * , 106406 ( 2009 ) . i. travenec , int .
b , 22 , 5217 ( 2008 ) .
a. r. subramaniam , _ et al .
lett . _ * 96 * , 126802 ( 2006 ) . s. faez , a. strybulevych , j. h. page , a. lagendijk , and b. a. van tiggelen _ phys .
lett . _ * 103 * , 155703 ( 2009 ) . c. monthus and th .
garel , _ j. stat .
mech . _ * 2011 * , p05005 ( 2011 ) ; see also c. monthus , b. berche , and ch .
chatelain , _ j. stat .
mech . _ * 2009 * , p12002 ( 2009 ) .
g. schubert , h. fehske : quantum and semi - classical percolation and breakdown in disordered solids , _ lecture notes in physics _ * 762 * , ( 2009 ) , pp 1 - 28 .
l. j. root , j. d. bauer , j. l. skinner , _ phys . rev .
b _ * 37 * , 5518 ( 1988 ) .
koslowski , w. von niessen , _ phys .
b _ * 44 * , 9926 ( 1991 ) .
r. berkovits , y. avishai , _ phys .
b _ * 53 * , r16125(r ) ( 1996 ) . a. kaneko , t. ohtsuki , _ j. phys
* 68 * , 1488 ( 1999 ) . | the phase diagram of the metal - insulator transition in a three dimensional quantum percolation problem is investigated numerically based on the multifractal analysis of the eigenstates .
the large scale numerical simulation has been performed on systems with linear sizes up to @xmath0 .
the multifractal dimensions , exponents @xmath1 and @xmath2 , have been determined in the range of @xmath3 .
our results confirm that this problem belongs to the same universality class as the three dimensional anderson model , the critical exponent of the localization length was found to be @xmath4 .
however , the mulifractal function , @xmath5 , and the exponents @xmath1 and @xmath2 produced anomalous variations along the phase boundary , @xmath6 . |
the topic of parity - time ( @xmath9 ) symmetry and its relevance for physical applications on the one hand , as well as its mathematical structure on the other , have drawn considerable attention from both the physics and the mathematics community .
originally the proposal of bender and his collaborators @xcite towards the study of such systems was made as an alternative to the postulate of hermiticity in quantum mechanics . in view of the formal similarity of the schrdinger equation with maxwell s equations in the paraxial approximation
, it was realized that such @xmath9 invariant systems can in fact be experimentally realized in optics @xcite .
subsequently , these efforts motivated experiments in several other areas including @xmath9 invariant electronic circuits @xcite , mechanical circuits @xcite , and whispering - gallery microcavities @xcite .
concurrently , the notion of supersymmetry ( susy ) originally espoused in high - energy physics has also been realized in optics @xcite .
the key idea is that from a given potential one can obtain a susy partner potential with both potentials possessing the same spectrum ( with exception of possibly one eigenvalue ) @xcite .
an interplay of susy with @xmath9 symmetry is expected to be quite rich and is indeed useful in achieving transparent as well as one - way reflectionless complex optical potentials @xcite . in a previous paper @xcite
we explored the interplay between @xmath9 symmetry , susy and nonlinearity . in ref .
@xcite we derived the exact solutions of the general nonlinear schrdinger equation ( nlse ) with arbitrary nonlinearity in 1 + 1 dimensions when in an external potential given by a shape invariant @xcite supersymmetric and @xmath9 symmetric complex potential .
in particular , we considered the nonlinear partial differential equation @xmath10 for arbitrary nonlinearity parameter @xmath3 , with @xmath11 and the partner potentials arise from the superpotential @xmath12 giving rise to [ vpvm ] @xmath13 for @xmath14 , the _ complex _ potential @xmath15 has the same spectrum , apart from the ground state , as the _ real _ potential @xmath16 and we used this fact in our numerical study of the stability of the bound state solutions of the nlse in the presence of @xmath15 ( see ref .
@xcite ) . to complete our study of this system of nonlinear schrdinger equations in @xmath9 symmetric susy external potentials
, we will study the stability properties of the bound state solutions of nlse in the presence of the external real susy partner potential @xmath16 , and compare the stability regime of these solutions , which depends on the parameters @xmath17 , to the stability regime of the related _ solitary _ wave solutions to the nlse in the absence of the external potential , which depend only on the parameter @xmath3 .
because the nlse in the presence of @xmath16 is a hamiltonian dynamical system , we can use variational methods to study the stability of the solutions when they undergo certain small deformations .
we will compare the results of this type of analysis with a linear stability analysis based on the v k stability criterion @xcite . in our previous paper
@xcite we determined the exact solutions of the equation for @xmath14 and for @xmath15 , which was complex .
we studied numerically the stability properties of these solutions using linear stability analysis .
we found some unusual results for the stability which depended on the value of @xmath18 . in that paper , because of the complexity of the potential , the energy was not conserved and a hamiltonian formulation of the problem was not possible .
however for the partner potential @xmath16 , when @xmath14 , the potential is real .
we note that @xmath16 has the symmetry @xmath19 , so that we can obtain @xmath16 from two _ different _ superpotentials @xmath20 and @xmath21 that have the @xmath0 symmetric forms at arbitrary @xmath22 : [ ssw1w2 ] @xmath23 in particular , when @xmath14 we can determine the spectrum of bound states of the linear schrdinger equation with potential @xmath16 using the _ real _ superpotential @xmath24 and the real shape invariant sequence of partner potentials . when @xmath14 the first superpotential @xmath20 has the _ complex _ partner potential @xmath15 which we studied in @xcite , whereas @xmath25 is the usual _ real _ shape invariant susy potential .
both superpotentials yield the same real potential @xmath26 .
because this potential is real , one can use variational methods to study the stability of the exact solutions to the nlse in the potential @xmath16 .
we will consider both derrick s theorem @xcite as well as a time dependent variational approximation @xcite to study the stability of the exact solutions .
because of the similarity of the solutions to those of the nlse equation , we are able to use a variant of the v k stability criterion to study spectral stability of the solutions .
the results of the v
k analysis agree with the new regime of stability found from derrick s theorem and the time dependent variational approach .
the latter approach allows us to obtain the frequency of small oscillations of the perturbed solutions .
we are able to obtain exact analytic results because we can formulate the problem in terms of hamilton s action principle .
the euler - lagrange equations lead to the dynamical equations which have a conserved hamiltonian .
this is in sharp contrast with the stability analysis for the solutions in the presence of @xmath15 which had to be done numerically .
the latter system is dissipative in nature .
this paper is structured as follows . in sec .
[ s : ssmodel ] we review the non - hermitian susy model that we studied in ref .
@xcite . in sec .
[ s : hamprinc ] we consider hamilton s principle for the nlse in the real external potential @xmath16 .
section [ s : derrick ] describes the application of derrick s theorem for determining the domain of stability of the solutions .
here we determine analytically a new domain of stability for the solutions , when compared to the solutions in the absence of an external potential .
we find that for @xmath7 a new domain of stability exists .
[ s : collective ] contains a collective coordinate approach which allows one to study dynamically the blowup or collapse of the solution as well as small oscillations around the exact solution when slightly perturbed . by setting the frequency to zero
we obtain analytically a domain of stability that agrees with the result of derrick s theorem . in sec .
[ ns.s : linearstability ] we provide the details of a linear stability analysis based on the v - k stability criterion , which leads to identical conclusions as derrick s theorem .
section [ s : conclude ] contains a summary of our main results .
we are motivated by the @xmath27 symmetric susy superpotential @xmath28 which gives rise to the supersymmetric partner potentials given by eqs . .
in what follows we will specialize to the case @xmath14 . in that case , @xmath16 is the well known pschl - teller potential @xcite .
the relevant bound state eigenvalues assume an extremely simple form as @xmath29 ^ 2 \>.\ ] ] such bound state eigenvalues only exist when @xmath30 .
we notice that for the ground state ( n=0 ) to exist requires @xmath31 .
the existence of a first excited state ( n=1 ) requires @xmath32 .
we will find that the stability of the nlse solutions in this external potential will depend on the depth of the well , which will lead to a critical value of @xmath33 , above which the solutions are stable . in what follows we
will be concerned with the properties of the nlse in the presence of the external potential centered at @xmath34 , @xmath35 in particular we are interested in the bound state solutions of @xmath36 if we assume a solution of the form @xmath37 it is easy to show that an exact solution is given by : @xmath38 where @xmath39 and @xmath40 for the complex potential @xmath15 , the amplitude of the solution is given by @xcite @xmath41 so that there are two separate regimes where @xmath42 is real .
in contrast , for @xmath16 there is only one regime for an attractive @xmath16 where @xmath42 is real , namely @xmath43 the analysis of the stability of the solutions for @xmath15 in ref .
@xcite showed a very complicated pattern .
even for @xmath44 there is a regime of instability as a function of @xmath18 for the nodeless solution . for @xmath45
all solutions in that case found analytically and numerically were unstable . only for @xmath46 were the solutions stable .
in contrast to that analysis where each value of @xmath3 had to be investigated separately , in the case of the real potential @xmath16 , we are able to address the stability question for all @xmath3 using derrick s theorem as well as the v k stability criterion .
the mass @xmath47 of the bound state for the case @xmath16 is given by @xmath48^{1/\kappa } \
, \gamma[1/\kappa ] } { \gamma[1/2 + 1/\kappa ] } \>. \notag\end{aligned}\ ] ] if we turn off the external potential by setting @xmath49 , @xmath50 \to ( 1/\kappa + 1/2)$ ] and the mass @xmath47 of the bound state goes to the mass of the solitary wave solutions , @xmath51^{1/\kappa } \ , \gamma[1/\kappa ] } { \gamma[1/2 + 1/\kappa ] } \>.\ ] ]
let us first discuss hamilton s principle of least action for the usual nlse without a confining potential .
the nlse with arbitrary nonlinearity in 1 + 1 dimensions is given by @xmath52 the second term causes diffusion and the third term attraction and the competition allows for solitary wave blowup which depends on @xmath3 . here
@xmath53 can be scaled out of the equation by letting @xmath54 so that the linear equation for the rescaled equation is obtained in the limit @xmath55 . while the solitary waves are stable for @xmath56 , for @xmath57
there is a critical mass @xmath47 necessary for blowup to occur , where the width of solitary wave goes to zero . for @xmath58
, blowup occurs in a finite amount of time .
the classical action for the nlse is @xmath59 = \int l[\psi,\psi^{\ast } ] \dd{t}$ ] , where the lagrangian @xmath60 $ ] is given by @xmath61 & = \int \dd{x } \bigl \ { \ , i \ , \qty [ \psi^{\ast } ( \partial_t \psi ) - ( \partial_t \psi^{\ast } ) \psi ] / 2 \label{lagnlse } \\ & \qquad + ( \partial_x \psi^{\ast } ) ( \partial_x \psi ) - | \psi |^{2(\kappa+1 ) } / ( \kappa + 1 ) \ , \bigr \ } \>. \notag\end{aligned}\ ] ] the nlse follows from the hamilton s principle of least action , @xmath62 and @xmath63 , which leads to eq . with @xmath64 . multiplying this equation by @xmath65 and subtracting its complex conjugate
, it is easy to prove that the mass @xmath47 , defined by @xmath66 , is conserved .
we now want to add a _ real _ susy potential to the nlse .
we will consider the addition of @xmath16 given in eq . so that the equation of motion is now given by @xmath67 the action which leads to eq .
is given by @xmath59 = \int l[\psi,\psi^{\ast } ] \dd{t}$ ] where @xmath61 & = \int \dd{x } i \
, \qty [ \psi^{\ast } ( \partial_t \psi ) - ( \partial_t \psi^{\ast } ) \psi ] /2 - h[\psi,\psi^{\ast } ] \ > , \notag \\
h[\psi,\psi^{\ast } ] & = \int \dd{x } \bigl \ { \ , ( \partial_x \psi^{\ast } ) ( \partial_x \psi ) - | \psi |^{2(\kappa+1 ) } / ( \kappa + 1 ) \notag \\ & \qquad\qquad + \psi^{\ast } \ , v^{-}(x ) \ , \psi \ ,
\bigr \ } \>. \label{hdef}\end{aligned}\ ] ]
derrick s theorem @xcite states that for a hamiltonian dynamical system , for a solitary wave solution to be stable it must be stable to changes in scale transformation @xmath68 when we keep the mass of the solitary wave fixed .
that is the hamiltonian needs to be a minimum in @xmath69 space .
first let us look at the case of the nlse without an external potential : derrick s method is based on whether a scale transformation which keeps the mass @xmath47 invariant , raises or lowers the energy of a solitary wave . for the nlse with hamiltonian @xmath70
\\ & \equiv h_1 - h_2 \ > , \notag\end{aligned}\ ] ] where both @xmath71 and @xmath72 are positive definite .
a static solitary wave solution can be written as @xmath73 the exact solution has the property that it minimizes the hamiltonian subject to the constraint of fixed mass as a function of a stretching factor @xmath69 .
this can be seen by studying a variational approach as done in @xcite , or by directly studying the effect of a scale transformation that respects conservation of mass . in the latter approach , which generalizes the method used by derrick @xcite , we let @xmath68 , and consider the stretched wave function , @xmath74 so that @xmath75 is preserved by the transformation . defining @xmath76 as the value of @xmath77 for the stretched solution @xmath78
, one finds that @xmath79 is consistent with the equations of motion .
the stable solutions must then also satisfy : @xmath80{h_{\beta}}{\beta } \ge 0 \>.\ ] ] if we write @xmath77 in terms of the two positive definite pieces @xmath71 , @xmath72 , then @xmath81 we find @xmath82 so that @xmath83 .
this result is consistent with the equations of motion .
in fact for the nlse the exact solution has @xmath84 where @xmath85 .
one finds then using @xmath86 } { \gamma[1/2+r/2 ] } \>,\ ] ] that [ h1andh2eval ] @xmath87^{1/\kappa } \gamma[1/\kappa ] } { 2 \kappa^2 \ , \gamma[3/2 + 1/\kappa ] } \ > , \label{h1-val } \\ h_2 & = \frac { \sqrt{\pi } \ , \qty [ ( \kappa + 1)/\kappa^2 ] ^{1/\kappa } \gamma[1/\kappa ] } { \kappa^3 \ , \gamma[3/2 + 1/\kappa ] } \ > , \label{h2-val } \end{aligned}\ ] ] so that the exact solution is indeed a minimum of the hamiltonian with respect to scale transformations , with @xmath88 .
the second derivative is given by @xmath89{h_{\beta}}{\beta } = 2 \ , h_1 - \kappa(\kappa - 1 ) \
, \beta^{\kappa - 2 } \ , h_2 \>,\ ] ] which when evaluated at the stationary point yields @xmath90{h_{\beta}}{\beta } = 2 \ , ( 2 - \kappa ) \ , h_1 \ge 0 \>,\ ] ] for stability .
this result indicates that solutions are unstable to changes in the width , compatible with the conserved mass , when @xmath58 .
the case @xmath91 is a marginal case where it is known that blowup occurs at a critical mass ( see for example ref .
the result found above for the nlse has also been found by various other methods such as linear stability analysis and using strict inequalities .
numerical simulations ( see ref .
@xcite ) have been done for the critical case @xmath91 showing that blowup ( self - focusing ) occurs when the mass @xmath92 . for @xmath93 a variety of analytic and numerical methods have been used to study the nature of the blowup at finite time @xcite . in the case of the nonlinear schrdinger equation
, one can perform a linear stability analysis of the exact solutions .
namely one lets @xmath94 \ , e^{-i \omega t } \>,\ ] ] and linearizes the nlse to find an equation for @xmath95 to first order , @xmath96 and studies the eigenvalues of the differential operator @xmath97 .
if the spectrum of @xmath97 is imaginary , then the solutions are spectrally stable .
k showed @xcite that when the spectrum is purely imaginary @xmath98 .
also they showed that when @xmath99 , there is a real positive eigenvalue so that there is a linear instability .
for the nlse , there is a class of solutions with arbitrary nonlinearity parameter @xmath3 .
namely [ nlse.e:3 ] @xmath100 when we do not have an external potential , we know explicitly how the mass changes when we change @xmath101 at fixed @xmath3 .
that is @xmath102^{1/\kappa } \gamma[1/\kappa ] } { \kappa \ , \sqrt{-\omega } \ , \gamma[1/2 + 1/\kappa ] } \
> , \notag \end{aligned}\ ] ] where @xmath103 we find @xmath104 thus for @xmath58 the solitary waves are unstable .
this agrees with the result of derrick s theorem .
when we have an external potential , we will need to determine the solutions numerically as we change @xmath101 .
this will be accomplished in sec .
[ ns.s : linearstability ] .
so now let us look at our situation when we have in addition the real external potential : @xmath105 the exact solution to the nlse in the presence of @xmath16 is given by eq . .
this solution is similar in form to the usual solution to the nlse except this nodeless solution is pinned to the potential so that there is no translational invariance .
when @xmath106 this solution goes over to a particular solution of the nlse with width parameter @xmath107 . under the scale transformation @xmath108 , the stretched solution which preserves the mass @xmath47 is given by : [ dext.e:2 ] @xmath109 the stretched wave function @xmath110 is no longer an exact solution .
the stretched hamiltonian for the external potential case is now given by @xmath111 where @xmath112 } { 2 \kappa^2 \ , \gamma[3/2 + 1/\kappa ] } \
> , \end{aligned}\ ] ] and @xmath113 } { ( \kappa + 1 ) \ , \gamma[3/2 + 1/\kappa ] } \
> , \end{aligned}\ ] ] with @xmath114 now given by , and @xmath115 thus , we find @xmath116 \dd{x } \>.\end{aligned}\ ] ] using the identity , @xmath117 we obtain @xmath118 } { ( \kappa + 1 ) \ , \gamma[3/2 + 1/\kappa ] } \>.\ ] ] as in the case when @xmath119 , we again find @xmath120 for our exact solution .
so the stretched solution is again an extremum of @xmath76 with @xmath47 kept fixed . for the second derivative we have @xmath121{h_{\beta}}{\beta } } _ { \beta=1 } \!\!\ ! = 2 \ , h_1 - \kappa ( \kappa - 1 ) \eval { \pdv[2]{h_{3}}{\beta } } _ { \beta=1 } \>,\ ] ] where @xmath122{h_{3}}{\beta } } _ { \beta=1 } \!\!\ !
= ( 1/4 - b^2 ) \ , a^{2}(b,\kappa ) \ , \qty [ i_1 + i_2 + i_3 ] / \kappa \>,\ ] ] with @xmath123
we can again evaluate these integrals using the first identity eq . and
the identity : @xmath124{\sech^{2 + 2/\kappa } ( \lambda x ) } { \lambda } \label{dext.e:9 } \\ & \quad = ( 2/\kappa + 2)(2/\kappa + 3 ) \
, x^2 \sinh^2(\lambda x ) \ , \sech^{2/\kappa + 4}(\lambda x ) \notag \\ & \qquad\qquad - ( 2/\kappa + 2 ) \ , x^2 \ ,
\sech^{2/\kappa + 2}(\lambda x ) \>. \notag\end{aligned}\ ] ] we will also need the following hypergeometric function : @xmath125 using these results , eq
. gives @xmath126{h_{3}}{\beta } } _ { \beta=1 } \!\!\!\!\ !
& = ( 1/4 - b^2 ) \ ,
a^{2}(b,\kappa ) \ , f_3(\kappa ) \ > , \label{dext.e:11 } \\
f_3(\kappa ) & = - \qty [ \frac{4 \ , ug(\kappa)}{2 + 3\kappa } + \frac{4 \sqrt{\pi } \ , \kappa \ , \gamma[1 + 1/\kappa ] } { ( \kappa+1)(3\kappa+2 ) \ , \gamma[3/2 + 1/\kappa ] } ] . \notag\end{aligned}\ ] ] the critical value is determined from : @xmath126{h_{\beta}}{\beta } } _ { \beta=1 } \!\!\ ! & = a^{2}(b,\kappa ) \ , \bigl [ \ , 2 \ , f_1(\kappa ) \label{dext.e:12 } \\ & \hspace{-2em } - \kappa ( \kappa + 1 ) \ , a^{2\kappa}(b,\kappa ) \ ,
f_2(\kappa ) + ( 1/4 - b^2 ) \ , f_3(\kappa ) \ , \bigr ] \>. \notag\end{aligned}\ ] ] solving for the critical value of @xmath5 , we find @xmath127 the result of calculating the second derivative at @xmath107 and setting it equal to zero is that the domain of stability is now as follows : for @xmath56 and all @xmath18 in the range @xmath128 , the solution is stable , as it was for the solitary wave solutions of the nlse . here
@xmath106 corresponds to no external potential . when @xmath58 the solitary wave solutions of the nlse were _
unstable_. instead , in the presence of the confining potential , a new domain of stability occurs when @xmath58 as long as @xmath129 , where @xmath8 is given by eq . .
we see this in the result for @xmath130 shown in fig .
[ ns.f : dhdbeta ] . in fig .
[ ns.f : bvskappa ] , we show both @xmath131 and @xmath132 as a function of @xmath3 .
the region between @xmath106 and @xmath131 is unstable . as we will show in sec .
[ ns.s : linearstability ] , this analytic result for @xmath8 given by eq .
is confirmed by our linear stability analysis . just as there is a critical mass for instability in the nlse at @xmath91 , for @xmath133 we can interpret the critical value of @xmath18 in terms of a critical mass which depends on @xmath3 above which the solution is unstable . since from eq . , we have @xmath134^{1/\kappa } \ , \gamma[1/\kappa ] } { \gamma[1/2 + 1/\kappa ] } \>,\ ] ] we see that the mass @xmath47 _ decreases _ as we increase @xmath18 at fixed @xmath3 .
so as we go from the unstable case @xmath106 ( no potential ) and increase @xmath18 we decrease the mass until we reach an @xmath135 below which the solution is stable .
finally we reach the curve @xmath136 which corresponds to @xmath137 = 1/2 + 1/\kappa$ ] .
the different regimes are shown in fig .
[ ns.f : mcrit ] . in the lightly shaded regime
the solutions are unstable .
the maximum value of the mass is given by the case @xmath106 , when there is no longer a stabilizing potential .
the interval @xmath138 corresponds to the regime @xmath139 .
this is the stable regime denoted by the darker shaded area in fig .
[ ns.f : mcrit ] .
in order to follow the time evolution of a slightly perturbed solitary wave or bound solution to a hamiltonian dynamical system , without solving numerically the time dependent partial differential equations for @xmath140 , one can introduce time - dependent collective coordinates assuming that the general shape of the original solution is maintained apart from the height , width , and position , etc .
this will allow us to see whether these parameters just oscillate around the original values or whether the parameters grow or decrease in time .
when instabilities are seen in the variational results , it suggests that the exact solutions are also unstable . unlike derrick
s theorem when applied to the nlse , the collective coordinate method can be applied to the special case @xmath91 .
it also gives an approximate description of wave function blow - up or collapse in the unstable regime , and oscillation of the perturbed solution in the stable regime . in the next section
, we first apply this approach when there is no external potential .
let us remind ourselves of the collective coordinate variational approach to blow - up for the nlse @xcite with no external potential . using this method
, we found previously that when @xmath57 , there is a critical value of the mass required before blowup could take place .
derrick s theorem has nothing to say about the stability of the solitary wave solution for this case . to make the collective coordinate approach concrete , we assume self - similar solutions of the form : @xmath141 \\ & \qquad \times \exp [ i \qty ( v \ , y(t ) / 2 + \lambda(t ) \ , y^2(t ) - \omega t ) ] \>. \notag\end{aligned}\ ] ] here @xmath142 , @xmath143 , and @xmath144 are arbitrary real functions of time alone , and @xmath145 . for no external potential translation invariance
gives @xmath146 .
in particular at @xmath147 and @xmath148 , we will start with the exact solution of the form @xmath149 and assume that this solution just changes during the time evolution in amplitude and width . with this assumption one can derive the dynamical equations for @xmath143 and @xmath144 from hamilton s principle of least action with the lagrangian given in eq . .
noether s theorem yields three conservation laws : conservation of probability , conservation of momentum , and conservation of energy .
conservation of probability gives `` mass '' conservation : @xmath150 and allows one to rewrite @xmath143 in terms of the conserved mass @xmath47 , the width parameter @xmath144 , and a constant @xmath151 whose value depends on @xmath152 .
thus , @xmath153 we will therefore keep @xmath47 in our definition of @xmath143 since it will be a relevant parameter when @xmath91 . for @xmath154 , one obtains @xmath155}{\gamma[1/2 + 1/\kappa ] } \>.\ ] ] setting @xmath156 , in terms of the new collective coordinates @xmath157 $ ] , the lagrangian is given by @xmath158 = k[g,\lambda ] - h[g,\lambda ] \>,\ ] ] where [ var.e:6 ] @xmath159}{m } & = \frac{i}{2 m } \int \dd{x } \qty [ \psi^{\ast } ( \partial_t \psi ) - ( \partial_t \psi^{\ast } ) \psi ] \notag \\ & = \frac{1}{2 } \ , v^2 + \omega - \dot{\lambda } \ , g^2 \ , \frac{c_2}{c_1 } \ > , \label{var.e:6a } \\ \frac{h[g,\lambda]}{m } & = - \frac{1}{m } \int \dd{x } \qty [ ( \partial_x \psi^{\ast } ) ( \partial_x \psi ) + | \psi |^{2(\kappa+1 ) } / ( \kappa + 1 ) ] \notag \\ & \hspace{-3em } = \frac{v^2}{4 } + \frac{c_3}{c_1 } \ , \frac{1}{g^2 } + 4 \lambda^2 \ , \frac{c_2}{c_1 } \ , g^2 - \frac{1}{(\kappa + 1 ) } \ , \frac{c_4}{c_1 } \ , \qty ( \frac{m}{c_1 g } ) ^{\kappa } \ > , \label{var.e:6b}\end{aligned}\ ] ] where [ var.e:7 ] @xmath160}{2 \kappa \ , \gamma[3/2 + 1/\kappa ] } \ > ,
\notag \\
c_4 & = \int_{-\infty}^{\infty } \!\!\ !
f^{(2 \kappa + 2)}(z ) \label{var.e:7c } \\
& = \frac{\sqrt{\pi } \ , \gamma [ 1 + 1/\kappa ] } { 2 \kappa \ , \gamma[3/2 + 1/\kappa ] } = 2 \kappa \ , c_3 \>. \notag\end{aligned}\ ] ] collecting terms from and
, the lagrangian is given by @xmath161}{m } & = \frac{1}{4 } \
, v^2 + \omega - \dot{\lambda } \ , g^2 \ , \frac{c_2}{c_1 } - \frac{c_3}{c_1 } \ ,
\frac{1}{g^2 } \label{var.e:8 } \\ & \qquad - 4 \lambda^2 \ , \frac{c_2}{c_1 } \ , g^2 + \frac{1}{(\kappa + 1 ) } \ , \frac{c_4}{c_1 } \ , \qty ( \frac{m}{c_1 g } ) ^{\kappa } \>. \notag\end{aligned}\ ] ] from the euler - lagrange equations we obtain the second order differential equation for @xmath162 , @xmath163 and the relation @xmath164 . in solving these equations
, we will use for the mass @xmath47 , when we are not at the critical value @xmath91 , the expression for the mass for the solitary wave solution given by eq . .
if we do this , we can rewrite eq . as @xmath165 one notices that for @xmath166=1 $ ] , @xmath167 , as it must for an exact solution .
we see that to get @xmath168 when @xmath91 we need to have @xmath169 where @xmath170 is the value of the mass for the exact solution .
so initial conditions with a mass greater than this are necessary to see blow up at @xmath91 . by multiplying both sides of by @xmath171 and integrating with respect to time
we obtain a first integral of the second order differential equation , which up to a multiplicative factor is the same as setting the conserved hamiltonian divided by the mass @xmath47 to a constant @xmath172 .
this gives @xmath173 we notice that at the critical value of @xmath174 , the last two terms both go like @xmath175 .
self - focusing occurs when the width can go to zero .
since @xmath176 needs to be positive , this means that at @xmath91 , the mass has to be greater than @xmath177 for @xmath162 to be able to go to zero .
we find @xcite @xmath178 provided we use the exact solution ( which is a zero - energy solution ) for @xmath91 , namely @xmath179 .
this agrees well with numerical estimates of the critical mass @xcite and is slightly lower than the variational estimate obtained earlier by cooper _
_ @xcite using a post - gaussian trial wave functions instead of a trial wave function based on the exact solution . for @xmath180
, if we use the mass of the exact solitary wave solution , the energy conservation equation simplifies to @xmath181 in the supercritical case when @xmath182 , we have @xmath183 this `` mean - field '' result was obtained earlier in refs .
@xcite . to show the difference between the stability at @xmath184 and @xmath185 , we have solved eq . for the initial conditions @xmath186 , @xmath187 , with the results shown in fig .
[ f : nlse-3 - 5 ] . for small oscillations we can assume @xmath188
from which we obtain the equation , @xmath189 \>. \notag\end{gathered}\ ] ] setting @xmath190 in eq . , leads to the same criterion for the critical mass when @xmath91 .
the same equation gives the frequency of small oscillations when @xmath56 .
for @xmath191 , the predicted period of oscillation is @xmath192 in good agreement with fig .
[ f : nlse-3 - 5]a .
@xmath193 from eq . for ( a ) @xmath191 and ( b ) @xmath194 .
the latter case corresponds to blowup " . ]
potential fit for @xmath191 .
the solid line ( blue online ) is the exact derivative of the potential from eq .
, the dashed line ( red online ) is the fitted function @xmath195 of eq . .
] now we would like to see how this argument is modified when we add the external potential @xmath16 . in this case
the exact solution is `` pinned '' to the origin .
the lagrangian is again given by eq . with the addition of the potential term : @xmath196 / m & = \int_{-\infty}^{+\infty } \!\!\ !
\dd{x } \psi^{\ast}(x ,
t ) \ , v^{-}(x ) \ , \psi(x ,
t ) / m \\ & = - ( b^2 - 1/4)\ , { \mathcal{v}}[g,\kappa ] / c_1 \ > , \notag\end{aligned}\ ] ] where @xmath197 & = \int_{-\infty}^{+\infty } \!\!\ ! \dd{y }
\sech^{2/\kappa}(y ) \ , \sech^{2}(g y ) \
> , \\ & \xrightarrow{g \to 1 } \frac{\sqrt{\pi } \ , \gamma[1 + 1/\kappa]}{\gamma[3/2 + 1/\kappa ] } \>. \notag\end{aligned}\ ] ] the lagrangian now becomes : @xmath198}{m } = \frac{1}{4 } \
, v^2 + \omega - \dot{\lambda } \ , g^2 \ , \frac{c_2}{c_1 } - \frac{c_3}{c_1 } \
, \frac{1}{g^2 } - 4 \lambda^2 \ ,
\frac{c_2}{c_1 } \
, g^2 \notag \\ & \ > + \frac{1}{(\kappa + 1 ) } \ , \frac{c_4}{c_1 } \ , \qty ( \frac{m}{c_1 g } ) ^{\kappa } + \frac{(b^2 - 1/4)}{c_1 } \ , { \mathcal{v}}[g,\kappa ] \>. \label{var.e:17}\end{aligned}\ ] ] the euler - lagrange equations now give @xmath199}{g } \
> , \notag\end{aligned}\ ] ] where @xmath200}{g } & = - 2 \int_{-\infty}^{+\infty } \!\!\ !
\dd{y } y \ , \sech(gy ) \ , \sech^{3}(g y ) \ , \sech^{2/\kappa}(y ) \notag \\ & \xrightarrow{g \to 1 } - \frac{\kappa}{\kappa + 1 } \
, \frac{\sqrt{\pi } \ , \gamma[1 + 1/\kappa]}{\gamma[3/2 + 1/\kappa ] } \>. \label{var.e:19}\end{aligned}\ ] ] to solve this equation numerically we fit the numerical values of the integral in by a function of the form : @xmath201 using mathematica , one obtains an extremely accurate 4-parameter fit . for example , the result of this fit for @xmath191 is shown in fig .
[ f : vfit ] for @xmath202 , @xmath203 , @xmath204 , and @xmath205 .
different fit parameters are used for each value of @xmath3 .
plots of the solutions @xmath193 of eq . for different values of @xmath18 and for @xmath191 , @xmath206 , @xmath207 , and @xmath208 are shown in fig .
[ f : gt ] .
the solid ( blue online ) , dotted ( green online ) , and dashed ( red online ) lines are the solutions @xmath193 of eq . for @xmath191 , @xmath206 , @xmath207 , and @xmath208 , and ( a ) @xmath209 , for @xmath191 , ( b ) @xmath210 , for @xmath57 , ( c ) @xmath211 , for @xmath130 , and ( d ) @xmath212 , for @xmath194 .
] we can study analytically the stability of the solutions in this variational approximation by linearizing eq
. around the exact solution @xmath213 , @xmath214 to evaluate the effect of the external potential on the small oscillation equation we just need to know that : @xmath215 + \order { \epsilon^2 } \notag \>. \end{aligned}\ ] ] substitution of this expansion into gives @xmath216 \notag \\
\hspace{3em } - 4 \ , ( b^2 - 1/4 ) \ , \qty [ 2 \ , ug(\kappa ) - 3 \ , ug_2(\kappa ) ] / c_2 \notag \ > , \end{gathered}\ ] ] with @xmath217 , where @xmath218 is given by eq .
and where @xmath219 the collective coordinate method allows one to approximately calculate the small oscillation frequency as well as the time evolution of the system using eq . .
[ f : gt ] we assumed @xmath220 .
the relevant values of @xmath8 are @xmath221 for @xmath222 .
for @xmath184 we get oscillation for the entire range from @xmath106 to @xmath223 as seen in fig . [
f : gt]a . as predicted , for @xmath91 ,
once we get above @xmath224 , which is the case with no potential , then the solution is stable as seen in fig .
[ f : gt]b . for @xmath225 ,
once we get above @xmath226 , then the solution is stable as seen in fig .
[ f : gt]c .
for @xmath185 we get similar results to @xmath227 , as seen in fig . [
f : gt]d . in the stable regime ,
the oscillation periods are accurately predicted by eq . .
setting @xmath228 , determines the critical value of @xmath18 at a given @xmath3 below which the solutions are unstable for @xmath58 .
the expression for @xmath8 obtained this way is identical to the expression for @xmath8 obtained from derrick s theorem in eq . and shown in fig .
[ ns.f : bvskappa ] . in the domain of instability
one finds that if we look at initial conditions where @xmath229 and @xmath230 , @xmath231 , then for the minus sign one gets `` blow up '' ( @xmath168 ) , and for the plus sign we get collapse of the solution ( @xmath232 ) . in fig .
[ collapse ] , we give an example of collapse when @xmath233 and we are in the unstable regime .
a first integral of the second order differential equation resulting from the lagrange s equation for @xmath162 can be obtained by setting the conserved hamiltonian to a constant @xmath172 .
one then has @xmath234 \>. \label{econs}\end{aligned}\ ] ] from the energy conservation equation we can see immediately that at @xmath91 the exact solution we found does not blow up .
this is for two reasons : first , when the width parameter @xmath235 , then @xmath236 $ ] becomes a constant independent of @xmath162 and therefore the potential does not affect the small @xmath162 behavior of the differential equation ; secondly the mass of the exact solution depends now on @xmath18 and @xmath3 and it is lower than the critical mass needed for blowup .
that is , the mass of the bound solution is given by : @xmath237 \
, c_1[\kappa ] = \frac{\sqrt{\pi } \ , \qty ( b_{\text{max}}^2(\kappa ) - b^2 ) ^{1/\kappa } \ , \gamma[1/\kappa ] } { \gamma[1/2 + 1/\kappa ] } \notag \\ & \hspace{2em } \xrightarrow{\kappa \to 2 } \pi \ , \sqrt{1 - b^2 } \>. \label{mkappav } \end{aligned}\ ] ] the maximum value of this occurs when the external potential goes to zero at @xmath106 .
when @xmath31 , the mass of the exact solution is _ always _ less than @xmath170 , so that these solutions are always stable when @xmath91 . for the nlse with no external potential ,
when the stability depends on the mass of the initial wave function at @xmath91 , the critical value is that of the exact solitary wave solution .
see also fig .
[ ns.f : mcrit ] and the discussion thereof .
let us perform the linear stability analysis of the solitary wave solutions @xmath238 to the nonlinear schrdinger equation in the external potential .
we take a perturbed solitary wave solution in the form @xmath239 \ , e^{-i\omega t}$ ] and consider the linearized equation on @xmath240 $ ] , @xmath241 if the spectrum of @xmath242 has eigenvalues with positive real part , then the corresponding solitary wave is called linearly unstable ; otherwise , it is called spectrally stable .
in general , the spectral stability does not imply nonlinear stability , but for the nodeless solutions to the nonlinear schrdinger equation one can use the lyapunov - type approach to prove the orbital stability ; see e.g. ref .
@xcite .
the equation we are solving is @xmath243 with @xmath244 we are interested in the stability of the solitary wave solution @xmath245 to , with the amplitude @xmath246 satisfying @xmath247 \ , \phi_{\omega , b}(x)\>.\ ] ] for @xmath248 , one has the explicit expression @xmath249^{1/(2\kappa ) } \ , \sech^{1/\kappa}{x } \>,\ ] ] with @xmath217
. we will perform the spectral analysis of the linearization operator following the v
k approach @xcite .
we consider the perturbation of the solitary wave , @xmath250 \ , e^{-i\omega t}$ ] , with @xmath251 , and with @xmath252 and @xmath253 real .
the linearized equation on @xmath252 and @xmath253 is given by @xmath254 where the self - adjoint operators @xmath255 are given by [ lpmdefs ] @xmath256 the stationary equation satisfied by @xmath257 and its derivative with respect to @xmath101 give the relations @xmath258 we need to perform the spectral analysis of @xmath259 and @xmath260 .
we start with reviewing the v
k approach from @xcite for the case @xmath106 , when @xmath261 . for
a given value @xmath262 , let @xmath263 be the profile of a solitary wave for the case when @xmath264 ( when @xmath106 ) . by the v
k theory , the linearization at @xmath265 is such that @xmath266 has a simple eigenvalue @xmath267 as its smallest eigenvalue , corresponding to the eigenfunction @xmath268 , while @xmath269 has one simple negative eigenvalue on the subspace of even functions , and a simple eigenvalue at @xmath270 on the subspace of odd functions corresponding to the eigenfunction @xmath271 . for any nonzero eigenvalue @xmath272 of the linearization operator from ,
one has the relation @xmath273 with nonzero @xmath274 .
being in the range of @xmath275 , which is self - adjoint , @xmath274 is orthogonal to the null space of @xmath276 ; this allows us to arrive at @xmath277 hence @xmath278 .
thus , the linear instability could only be caused by a positive eigenvalue of @xmath279 .
from , one can see that one could have @xmath280 if the right - hand side of becomes positive for some @xmath274 orthogonal to the kernel of @xmath276 ; in other words , if the minimization problem @xmath281 gives a negative value of @xmath282 .
by @xcite , finding the minimum of under constraints @xmath283 and @xmath284 leads to the relation @xmath285 with @xmath286 lagrange multipliers ; pairing the above with @xmath274 shows that @xmath282 in and is the same . writing @xmath287 and taking into account that
@xmath284 , we see that we need to analyze the location of the first root of the v
k function @xmath288 which is defined for @xmath289 in the resolvent set of the operator @xmath290 restricted onto the subspace of even functions .
this domain includes the interval @xmath291 , where @xmath292 is the smallest negative eigenvalue of @xmath293 and @xmath294 is the next eigenvalue of @xmath293 , on the subspace of even functions . since clearly @xmath295 for @xmath296 , one has @xmath297 at some @xmath298 , @xmath299 ( hence stability ) if and only if @xmath300 , which leads to @xmath301 , and , using , we arrive at the v k stability condition @xmath302 an elementary computation based on shows that is satisfied ( for all @xmath303 ) if and only if @xmath304 . the left - hand side of becomes identically zero for @xmath57 and becomes positive for @xmath58 ( again , for all @xmath303 ) .
now let us consider @xmath305 with @xmath31 and @xmath248 .
as in the case of no potential , one has @xmath306 , with @xmath267 a simple eigenvalue corresponding to the eigenfunction @xmath307 . at @xmath308 ,
one has @xmath309 and @xmath310 we note that @xmath311 hence the smallest eigenvalue @xmath312 of @xmath313 ( assumed on the subspace of even functions ) is negative . at @xmath314 ,
one has @xmath315 hence for @xmath316 and @xmath317 , @xmath318 just as in the case @xmath106 which we considered above , the linear instability takes place when the minimization problem @xmath319 gives a negative value of @xmath282 .
as we already pointed out in the case @xmath106 , one has @xmath320 for @xmath304 ( equivalently , @xmath321 are linearly stable ) , and @xmath322 for @xmath91 .
due to , one then also has @xmath320 for @xmath323 , @xmath324 , @xmath325 and for @xmath91 , @xmath326 , @xmath327 .
thus , for these values of @xmath3 and @xmath18 , the solitary waves @xmath328 are spectrally stable . for @xmath6 , the story is different : while @xmath282 in is negative for @xmath106 corresponding to the linear instability of @xmath329 , @xmath282 could become positive if @xmath18 exceeds some critical value @xmath131 : @xmath330 numerically , we proceed as follows .
we pick @xmath6 and use the shooting method to construct a solitary wave @xmath257 and find the critical value @xmath331 above which @xmath332 becomes negative [ that is , when @xmath333 , a positive eigenvalue from the spectrum of @xmath334 collides with a negative eigenvalue , and they produce a pair of purely imaginary eigenvalues ; for @xmath335 , spectral stability takes place ] .
this gives us the critical values @xmath8
vs. @xmath3 in agreement with fig .
[ ns.f : bvskappa ] .
we find remarkably that derrick s theorem and the v
k spectral analysis of stability give identical results .
the same result for the stability regime was also obtained by setting the oscillation frequency for small oscillations around the exact solution to zero using the time dependent variational method . in distinction with the case without a potential , in the presence of the external potential @xmath16 the results of the stability analysis are much more interesting because of the additional @xmath18 dependence of the exact solution
. for @xmath93 it is possible to interpret the results of v - k and derrick s theorem in terms of a critical mass @xmath336 below which the solution is stable , or equally in terms of a critical depth @xmath18 for the confining potential above which the solution is stable .
in this paper we studied the stability of the exact solution of the nlse in a real pschl - teller potential which is the susy partner of a _ complex _ @xmath27 symmetric potential studied previously @xcite .
unlike the previous problem which required detailed numerical analysis for every value of the nonlinearity parameter @xmath3 , the real external potential problem here results in a hamiltonian dynamical system amenable to several variational approaches to the stability problem , such as derrick s theorem @xcite , v k theory @xcite , and a time dependent variational approach . using these methods we were able to show that for @xmath93 the pinned solution has a region of stability that was not available to the solitary wave solution of the nlse without an external potential .
the latter solutions are known to blow up in a finite time interval when perturbed appropriately .
the analytic result for the re - entry regime of stability found using derrick s theorem was corroborated by a numerical study of spectral stability based on the v k theory .
this result is different from the result found numerically for the stability of the solution for the complex susy partner external potential @xmath15 .
the analysis of the stability of the solutions for @xmath15 in @xcite showed a very complicated pattern .
even for @xmath44 there is a regime of instability as a function of @xmath18 for the nodeless solution . at @xmath45
all the solutions found for @xmath15 , were unstable due to oscillatory instabilities . only for @xmath46 were the solutions stable .
in contrast , for the @xmath16 potential we are able to address the stability question for all @xmath3 _ analytically _ and show that the effect of the external potential is to introduce a new domain of stability for all @xmath58 , when compared to the stability of the related solitary wave solutions in the absence of an external potential .
the stability properties of the solutions of the nlse in the presence of the partner potentials @xmath337 are quite different from one another due to the dissipative versus conservative nature of these potentials .
f.c . would like to thank the santa fe institute and the center for nonlinear studies at los alamos national laboratory for their hospitality .
is grateful to indian national science academy ( insa ) for awarding him insa senior scientist position at savitribai phule pune university , pune , india .
the research of a.c . was carried out at the institute for information transmission problems , russian academy of sciences at the expense of the russian foundation for sciences ( project 14 - 50 - 00150 ) .
b.m . and j.f.d .
would like to thank the santa fe institute for their hospitality .
b.m . acknowledges support from the national science foundation through its employee ir / d program .
the work of a.s . was supported by the u.s .
department of energy .
see special issues : h. geyer , d. heiss , and m. znojil , eds .
, j. phys . a : math
. gen . * 39 * , _ special issue dedicated to the physics of non - hermitian operators _ ( _ phhqp iv _ ) ( university of stellenbosch , south africa , 2005 ) ( 2006 ) ; a. fring , h. jones , and m. znojil , eds . , j. math .
. a : math theor . * 41 * , _ papers dedicated to the subject of the 6th international workshop on pseudo - hermitian hamiltonians in quantum physics _ ( _ phhqpvi _ ) ( city university london , uk , 2007 ) ( 2008 ) ; c.m .
bender , a. fring , u. gnther , and h. jones , eds .
, _ special issue : quantum physics with non - hermitian operators _ , j. math
. a : math theor .
* 41 * , no .
44 ( 2012 ) .
a. ruschhaupt , f. delgado , and j. g. muga , j. phys . a : math
* 38 * , l171 ( 2005 )
. k. g. makris , r. el - ganainy , d. n. christodoulides , and z. h. musslimani , phys .
lett . * 100 * , 103904 ( 2008 ) ; s. klaiman , u. gnther , and n. moiseyev , _ ibid_. * 101 * , 080402 ( 2008 ) ; o. bendix , r. fleischmann , t. kottos , and b. shapiro , _ ibid_. * 103 * , 030402 ( 2009 ) ; s. longhi , _ ibid_. * 103 * , 123601 ( 2009 ) ; phys . rev .
b * 80 * , 235102 ( 2009 ) ; phys . rev .
a * 81 * , 022102 ( 2010 ) .
a. guo , g. j. salamo , d. duchesne , r. morandotti , m. volatier - ravat , v. aimez , g. a. siviloglou , and d. n. christodoulides , phys .
* 103 * , 093902 ( 2009 ) ; c. e. rter , k. g. makris , r. el - ganainy , d. n. christodoulides , m. segev , and d. kip , nature phys . * 6 * , 192 ( 2010 ) ; a. regensburger , c. bersch , m .- a .
miri , g. onishchukov , d. n. christodoulides , and u. peschel , nature * 488 * , 167 ( 2012 ) .
m. ali miri , m. heinrich , r. el - ganainy , and d.n .
christodoulides , phys .
lett . * 110 * , 233902 ( 2013 ) ; m. heinrich , m. ali miri , s. sttzer , r. el - ganainy , s. nolte , a. szameit , and d.n .
christodoulides , nat .
comm . * 5 * , 3698 ( 2014 ) . | we discuss the stability properties of the solutions of the general nonlinear schrdinger equation ( nlse ) in 1 + 1 dimensions in an external potential derivable from a parity - time ( @xmath0 ) symmetric superpotential @xmath1 that we considered earlier [ kevrekidis _ et al .
_ phys .
rev .
e * 92 * , 042901 ( 2015 ) ] .
in particular we consider the nonlinear partial differential equation @xmath2 for arbitrary nonlinearity parameter @xmath3 .
we study the bound state solutions when @xmath4 , which can be derived from two different superpotentials @xmath1 , one of which is complex and @xmath0 symmetric . using derrick s theorem , as well as a time dependent variational approximation ,
we derive exact analytic results for the domain of stability of the trapped solution as a function of the depth @xmath5 of the external potential .
we compare the regime of stability found from these analytic approaches with a numerical linear stability analysis using a variant of the vakhitov - kolokolov ( v k ) stability criterion .
the numerical results of applying the v - k condition give the _ same _ answer for the domain of stability as the analytic result obtained from applying derrick s theorem .
our main result is that for @xmath6 a _ new _ regime of stability for the exact solutions appears as long as @xmath7 , where @xmath8 is a function of the nonlinearity parameter @xmath3 . in the absence of the potential the related solitary wave solutions of the nlse are _ unstable _ for @xmath6 . |
after the discovery of the higgs boson , the atlas @xcite and cms @xcite collaborations have continued searching for new exotic particles , such as the randall - sundrum spin-2 boson or new heavy scalar particles @xcite .
particularly , atlas collaboration is carrying out searches for new scalar resonances decaying into two photons .
though up to the moment the searches for new scalar resonances has been fruitless , it is expected that the experimental collaborations atlas and cms continue the seek for new particles at the tev energy scale .
these quests are supported by the improvements implemented in the cms and atlas detectors @xcite along with the fact that the experimental collaborations have been able to develop a reliable detection machinery of spin-0 resonances .
thus , the perspectives of searching for new physics phenomena at the tevs scale are hopeful .
complementarily , several theoretical studies are found in the literature where the predictive power of extended models is tested via production of new - heavy neutral scalars decaying into standard model ( sm ) particles @xcite .
the aforementioned experimental results offer theoretical arguments to explore new physics process related with extended higgs sectors or sm extensions at the tev energy scale @xcite . in this meaning , there exist many theoretical approaches that predict more content of scalar particles , such as the two - higgs doublet models ( 2hdms ) @xcite , three - higgs doublet model ( 3hdm ) @xcite , higgs - singlet extension model @xcite , little higgs models ( lhms ) @xcite , etc . among the variety of lhms , the case of the littlest higgs model ( lthm ) is very interesting since it does no present new degrees of freedom beyond the sm under tev scale .
additionally , the lthm has a reduced spectrum of new scalar particles @xcite .
as value added , this type of models offer a possible solution to hierarchy problem , which emerges when the higgs boson mass is affected by one - loop corrections .
also , the lhms represents another approach to the electroweak symmetry breaking pattern @xcite , based on the dimensional deconstruction @xcite , where the quadratic divergences cancel .
expressly , the quadratic divergence generated at the one - loop level because of the sm gauge bosons is canceled by the quadratic divergence introduced by the new gauge bosons at the same perturbative level . the one - loop quadratic divergence induced in the sm yukawa sector is eliminated by introducing new heavy fermions in such a way that the quadratic divergence coming from the sm top quark cancels . in lhms , in agreement with the breaking of the global symmetry at the energy scale of tevs , the new higgs field get mass becoming pseudo - goldstone bosons , where in addition a massless higgs arises .
the quadratic - divergent corrections to the higgs boson mass show up at loop level , which assures a light higgs boson . with regard to lthm @xcite ,
the new particles arising at the tev energy scale are grouped in a new set of four gauge bosons with the same quantum numbers as the sm gauge bosons , namely , @xmath10 , @xmath11 , and @xmath12 , an exotic quark with the same charge as the sm top quark , and a new heavy scalar triplet , which contains six physical states : double - charged scalars @xmath13 , single - charged scalars @xmath14 , a neutral scalar @xmath15 , and a neutral pseudo scalar @xmath16 .
all the details of the lthm model can be found in ref .
@xcite .
the aim of this work is to study in detail the one - loop level @xmath17 ( @xmath1 ) decays in the context of the lthm . in particular , our calculations are placed on the basis of the so - called linearized theory of lthm @xcite , which means that it is performed a first - order expansion of the @xmath18 field around its vacuum expectation value ( vev ) in powers of @xmath19 @xcite . regarding the @xmath20 , @xmath21 and @xmath22 vertexes , at the one - loop level , these present a completely different lorentz structure from what it has in similar interactions with a scalar particle @xcite .
therefore , it is interesting to deepen the analytical study of the transition amplitudes for these interactions .
although the parameter space of the lthm has been severely restricted by the higgs discovery channels and electroweak precision observables @xcite , our proposal could be of interest as far as of the search for new scalar particles refers .
our study takes in consideration joined results from experimental and phenomenological analyzes , where it is proposed a simulated scenario as realistic as possible @xcite .
the viable analysis region for the energy scale @xmath8 is sustained by a phenomenological study based on the experimental searches of the sm - higgs boson , where it is used the signal strength modifier along with electroweak precision data . even when the parameter space of the lthm is strongly constrained , there is still room to study a wide variety of processes at the tev energy scale . for consistency with electroweak precision data , a lower limit for the energy scale @xmath8
around 2 - 4 tev is set @xcite .
the present paper is organized as follows . in section [ mod - fra ] ,
a survey of the lthm is presented . in section [ ana - cal ] ,
the analytical calculations of the one - loop level @xmath0 decays are described . in section [ num - res ] ,
the numerical results are discussed .
finally , the conclusions are summarized in section [ conclu ] .
the lthm is established by a nonlinear sigma model with @xmath23 global symmetry together with the gauged group @xmath24\otimes [ su(2)_2\otimes u(1)_2]$ ] @xcite . at first
, the @xmath23 group is spontaneously broken to the @xmath25 group at the energy scale @xmath8 . at par ,
the @xmath24\otimes [ su(2)_2\otimes u(1)_2]$ ] group is broken to its subgroup @xmath26 , the latter being the sm electroweak gauge group . at the energy scale @xmath8 ,
the spontaneous global symmetry breaking of the @xmath23 group is generated by the vev of the @xmath18 field , identified as @xmath27 @xcite .
the explicit form of the @xmath18 field is given by @xmath28 with @xmath29 and @xmath30 being the goldstone boson matrix with the following structure @xmath31 here , the fields @xmath32 and @xmath33 represent a doublet and a triplet under the @xmath26 sm gauge group , respectively @xcite , having the following representation @xmath34 the global symmetry breaking produces 14 goldstone bosons which transform under the @xmath26 group as a real singlet @xmath35 , a real triplet @xmath36 , a complex doublet @xmath37 , and a complex triplet @xmath38 @xcite . by means of this mechanism both the real singlet and the real triplet
are absorbed by the longitudinal components of the gauge bosons at the scale @xmath8 , where the complex doublet and the complex triplet remain massless .
the complex doublet is precisely the sm higgs field . through the coleman - weinberg type potential when the global symmetry of the group @xmath25 breaks down the complex triplet gets mass of the order of the energy scale @xmath8 . on the other hand , the effective lagrangian invariant under the @xmath24\otimes [ su(2)_2\otimes u(1)_2]$ ] group is @xcite @xmath39 where @xmath40 contains the gauge bosons kinetic contributions
, @xmath41 represents the fermions kinetic contributions , @xmath42 includes the nonlinear sigma model contributions of the lthm , @xmath43 comprises the yukawa couplings of fermions and pseudo - goldstone bosons .
the last term corresponds to the coleman - weinberg potential . for the purposes of our work
, we only present a brief overview of the lthm . for a more detailed description of this model see ref .
@xcite .
the nonlinear sigma model sector is described by the following lagrangian @xmath44 where the covariant derivative is expressed as @xmath45.\ ] ] explicitly , the @xmath46 represent @xmath47 gauge fields , the @xmath48 are the @xmath49 gauge fields , @xmath50 are the generators of the @xmath47 gauge group , the @xmath51 denote @xmath49 gauge group generators , @xmath52 are the coupling constants of the @xmath47 group , and @xmath53 are the coupling constants of the @xmath49 group @xcite . after the spontaneous symmetry breaking ( ssb ) around @xmath27 , for the gauge bosons , mass eigenstates of order of @xmath8
are generated @xcite @xmath54 here , @xmath55 and @xmath56 , for @xmath57 ; moreover , @xmath58 , @xmath59 , @xmath60 , and @xmath61 . in the procedure of ssb , the @xmath18 field is expanded around its vev ( @xmath27 ) preserving dominant terms in the lagrangian of the nonlinear sigma model sector @xcite .
consequently , the masses of the new heavy gauge bosons are of order @xmath62 @xmath63 at this stage of the ssb the @xmath64 and @xmath65 fields still do not acquire mass . in the lthm , the @xmath66 parameter , @xmath67 , takes the value closes to one at the leading order @xcite , in order to have similar values for the new gauge bosons masses ; as it happens in the electroweak sector of the sm @xcite .
while the @xmath68 parameter is related to the mixing angles between gauge bosons of the @xmath47 and @xmath69 groups , it directly affects the mass of the heavy photon ( @xmath10 ) .
therefore , in desire of keep masses of the order of tevs , the @xmath68 parameter should be restricted to large mixing angles . at the fermi energy scale ,
the sm gauge bosons get masses by the ssb mechanism , where also mixings between sm and new heavy gauge bosons are induced . with regard to the scalar sector , in the lthm
the higgs potential is induced via one - loop radiative corrections at the leading order .
this potential comprises contributions coming from gauge boson and fermion loops .
when the @xmath18 field is expanded around its vev into the nonlinear sigma lagrangian the coleman - weinberg potential is achieved @xcite @xmath70 the @xmath71 s quantities are given as @xmath72 + 8 a^{\prime } \lambda_1 ^ 2 , \nonumber \\
\lambda_{h \phi h } & = & -\frac{a}{4 } \left [ g^2 \frac{(c^2-s^2)}{s^2c^2 } + g^{\prime 2 } \frac{(c^{\prime 2}-s^{\prime 2 } ) } { s^{\prime 2}c^{\prime 2 } } \right ] + 4 a^{\prime } \lambda_1 ^ 2 , \nonumber \\
\lambda_{h^4 } & = & \frac{a}{8 } \left [ \frac{g^2}{s^2c^2 } + \frac{g^{\prime 2}}{s^{\prime 2}c^{\prime 2 } } \right ] + 2 a^{\prime } \lambda_1 ^ 2 = \frac{1}{4 } \lambda_{\phi^2}.\end{aligned}\ ] ] the parameters : @xmath66 , @xmath73 ( @xmath68 , @xmath74 ) represent mixing angles related with the gauge coupling constants of the @xmath47 ( @xmath49 ) symmetry group .
the @xmath75 and @xmath76 parameters depict unknown ultraviolet ( uv ) physics at the cutoff scale @xmath77 .
their values depend on the uv completion details at the @xmath77 scale @xcite .
the @xmath78 quantity is a free parameter that receives evenly significant contributions coming from one - loop logarithmic and two - loop quadratically divergent parts @xcite .
the vev @xmath79 ( @xmath80 ) of the doublet ( of the triplet ) is obtained after minimizing the coleman - weinberg potential , where there is fulfilled the relations @xmath81 by diagonalizing the higgs mass matrix heavy scalar bosons get masses @xcite .
thereby , the gauge eigenstates of the higgs sector can be expressed depending on mass eigenstates in the following manner @xmath82 here , @xmath83 symbolizes the higgs boson , @xmath15 is a new neutral scalar , @xmath9 represents the neutral pseudoscalar , @xmath84 and @xmath85 are the charged and doubly charged scalars , respectively .
the @xmath86 and @xmath87 fields are the goldstone bosons that are eaten by the massless @xmath88 and @xmath89 bosons @xcite . at the leading order ,
the masses of the new scalar bosons are degenerate , being written as @xcite @xmath90 where @xmath91 .
the previous expression is only definite positive if @xmath92 . for the rest of particle content ,
the lthm predicts new fermions which couple to the higgs field in such a manner that the quadratic divergence of the top quark is annulled @xcite .
the new set of heavy fermions is settled as a vector - like pair ( @xmath93 ) with quantum numbers @xmath94 and @xmath95 , respectively .
thus , the yukawa sector has the following structure @xmath96 where @xmath97 ; @xmath98 and @xmath99 are antisymmetric tensors for @xmath100 and @xmath101 @xcite . here ,
@xmath102 and @xmath103 represent free parameters , where @xmath103 can be fixed in such a way that for given values of @xmath104 , the top quark mass should adjust to its experimental measurement @xcite . by expanding the @xmath18 field around @xmath27 keeping terms up to @xmath105 and then
diagonalizing the mass matrix , it can be found the mass states @xmath106 , @xmath107 , @xmath108 , and @xmath109 , which correspond to the sm top quark and the new top quark , respectively @xcite .
all the remaining contributions to the lthm lagrangian can be consulted in ref .
our analytical results were computed by using the set of new feynman rules presented in ref .
in order to analyze the one - loop level decays of the @xmath9 boson , the total decay width of the @xmath9 ( @xmath111 ) , which will include only sm final states , must be computed .
the most important contributions to @xmath111 are the tree - level decays of @xmath9 into @xmath112 .
other subdominant contributions turn out to be the final states : @xmath113 , which we will also consider .
the feynman diagrams that depict two- and three - body decays of @xmath9 boson at three level are shown in figs .
[ feyn2vv ] and [ feyn2abc ] , respectively
. decays at tree level . ] on this basis , the decay width of the @xmath114 process is calculated , resulting in @xmath115 where @xmath116 is the mass of the pseudoscalar boson , @xmath117 is the color factor equal to 3 for quarks and @xmath8 is the global symmetry breaking scale . the @xmath118 decay width can be written as follows @xmath119^{3/2}.\end{aligned}\ ] ] concerning the three - body decays , these processes are mediated by sm particles , new scalar bosons and the @xmath11 gauge boson ( see fig .
[ feyn2abc ] ) . the analytic expression for the decay width of the @xmath9 scalar boson decaying into three bodies
can be computed by using the standard formulation described in ref .
@xcite .
boson at tree level . ]
the amplitude for the @xmath120 decay is @xmath121 \epsilon^{*}_{\mu}(k_1)\epsilon^{*}_{\nu}(k_2),\end{aligned}\ ] ] where @xmath122 . due to degeneracy in the masses of new scalar bosons ,
@xmath123 was employed .
@xmath124\nonumber\\ & \times&\epsilon^{*}_{\mu}(k_1)\epsilon^{*}_{\nu}(k_2)\epsilon^{*}_{\alpha}(k_3).\end{aligned}\ ] ] owing to the fact that the main impact of new particles on the different decay modes of the @xmath9 boson would be manifesting at tree level , we will analyze this type of contributions exclusively at this level . for this purpose
, it is necessary to fix the @xmath66 parameter or at least to explore a neighborhood for it with the objective of maximizing the possible contributions of new physics .
particularly , our election seeks to avoid the decoupling in the @xmath125 interaction .
thereby , it is assumed that @xmath126 @xcite , which indicates a small mixing angle and promotes an enhanced contribution of the @xmath11 boson to the @xmath127 decay .
the numerical results tell us that new heavy particle corrections are less than 1% .
all the analytical formulae necessary to calculate the decay widths were computed by using the feyncalc package @xcite .
firstly , we will discuss relevant details of @xmath20 vertex in order to present the analytic results for the @xmath128 decay .
secondly , the analytical study of the @xmath21 vertexes , where ( @xmath1 ) , will be placed in the context of @xmath129 decays .
lastly , analytical results for the one - loop amplitude of the @xmath130 process are presented . motivated by the fact that the @xmath20 vertex is absent at tree - level in the lthm , it is interesting to analyze this coupling at one - loop level .
this can be done through the analysis of the @xmath5 process . in what follows ,
all the one - loop calculations will be carried out by making use of the unitary gauge . in the lthm
, this decay only receives contributions from quarks , however , the main one corresponds to sm top quark .
the dominant feynman diagrams that represent the @xmath5 decay are shown in fig .
[ diaww ] . after performing dimensional regularization for the one - loop amplitudes related with the feynman diagrams in fig .
[ diaww ] , we find that the total amplitude for the @xmath5 decay can be written as @xmath131 where @xmath132.\end{aligned}\ ] ] it should be stressed that the form factor @xmath133 is finite , being , @xmath134 and @xmath135 , the passarino - veltman scalar functions ( pv ) .
thus , the decay width of the @xmath5 process is @xmath136 process at one - loop level . ]
the following describes the analytical expressions for the amplitudes of the @xmath137 processes . in the lthm context
, the @xmath137 decays are only mediated by sm quarks and the new exotic top quark @xmath138 . even though the new top quark induces a contribution different from zero ,
this is suppressed at least by two orders of magnitude with respect to the sm top quark contribution , which is the dominant one regarding the remaining sm quarks . moreover ,
when one @xmath138 quark fluctuates into the @xmath139 decay , its amplitude is exactly zero since there is no @xmath140 vertex .
therefore , we do not present this in the feynman diagrams . in fig .
[ unchargedweaks ]
the feynman diagrams associated with the @xmath2 decays are shown .
( @xmath1 ) decays at one - loop level . ]
the one - loop amplitude for the @xmath139 decay can be written as follows @xmath141 the form factor @xmath142 is given in terms of pv , which can be seen below @xmath143 where @xmath144 is the three - point pv .
after performing algebraic operations , the decay width of the @xmath139 decay is expressed as @xmath145 in this way , the one - loop amplitude for the @xmath146 decay can be computed , being equal to @xmath147 the form factor @xmath148 has the following structure @xmath149 where @xmath150 .
thereby , the associated decay width can be expressed as follows @xmath151 the respective decay amplitude for the @xmath152 process turns out to be @xmath153 in this case , the form factor @xmath154 is given by @xmath155,\end{aligned}\ ] ] where @xmath156 and @xmath157 .
following a similar algebraic procedure as above , the decay width of the @xmath152 process can be obtained , being @xmath158 it should be emphasized that the form factors : @xmath159 and @xmath154 are free of ultraviolet ( uv ) divergences and its corresponding lorentz structures satisfy gauge invariance .
decay at one - loop level . ]
finally , we exhibit the analytical result for the one - loop amplitude of the @xmath4 decay .
also , for this process , we have only included the sm top quark contribution since the exotic top quark contribution is suppressed at least two orders of magnitude ( see fig . [ gluons ] ) .
said amplitude can be appreciated below @xmath160 the form factor @xmath161 is written as follows @xmath162 where @xmath163 . in this manner , the decay width of the @xmath4 process is given as @xmath164 notice that the form factor @xmath161 is also uv finite and the respective amplitude complies gauge invariance .
the appearance of the levi - civita tensor in all the one - loop decay amplitudes of the @xmath9 boson is a distinctive manifestation of the pseudoscalar nature of this particle @xcite . in all our analytical calculations
it has been assumed that @xmath165 @xcite . the numerical evaluation for all the @xmath0 processes
was carried out by using the looptools package @xcite .
this work represents a follow - up study to the one made by some of us concerning the two - body decays of the @xmath15 scalar boson at one - loop level @xcite , because now we are interested in investigating phenomenological details regarding the two - body decays of the pseudoscalar boson at one - loop level in the context of the lthm . in this sense ,
a scenario where @xmath116 is of the order of unities of tevs is settled down , where the one - loop level decays of @xmath9 into @xmath166 and @xmath167 will be useful to test the consistency of the current parameter space of the lthm @xcite . before proceeding , we remark that now the energy scale @xmath8 is the only free parameter which we can play with .
this symmetry - breaking energy scale is restricted by the experimental data to lower limits by around 2 - 4 tev @xcite . in fig .
[ widthc ] , it can be seen that the decay width of the @xmath127 process essentially does not depend on @xmath66 parameter variations .
this plot shows a variation of the @xmath66 parameter from 0.1 to 0.9 , for three distinct energy scales , i.e. @xmath168 tev , @xmath169 tev and @xmath170 tev . because of the @xmath127 decay is the only one that depends on the @xmath66 parameter , and this is not the dominant decay mode of the @xmath9 boson , we assume that the total decay width of the @xmath9 boson will not depend of the @xmath66 parameter .
however , we will adopt a specific value for the @xmath66 parameter such that the @xmath11 contribution to @xmath127 is enhanced .
thus , we fix @xmath171 , in order to avoid the decoupling in the @xmath125 and @xmath172 interactions , which also indicates presence of small mixing angles @xcite .
process as a function of the @xmath66 parameter for @xmath173 tev . ] in fig .
[ brf](a ) , the decay widths for the @xmath174 processes are displayed . from fig .
[ brf](a ) , it can be observed that the dominant contributions come from tree - level decays of @xmath9 , specifically , into @xmath175 , @xmath176 and @xmath177 , which are close to the value of 1 gev around @xmath170 tev .
the two - body decays of the @xmath9 boson at tree level have already been calculated @xcite , however , between the energies studied only a small range is consistent with current constraints that come from electroweak precision data @xcite . at the one - loop level , we found a subdominant decay mode which corresponds to the @xmath130 decay , whose associated decay width is of the order of @xmath6 gev near to @xmath168 tev .
the decay width of the @xmath128 process is two orders of magnitude less than @xmath178 , being of the order of @xmath7 gev close to @xmath168 tev .
in contrast , the numerical evaluation tell us that the decays to electroweak neutral bosons are very suppressed . in specific ,
@xmath179 gev , @xmath180 gev and @xmath181 gev for @xmath8 around 2 tev .
furthermore , we have computed their corresponding branching ratios as a function of the energy scale @xmath8 , as can be seen in fig .
[ brf](b ) .
the total decay width of the @xmath9 pseudoscalar boson contains the decay modes : @xmath182 . in fig .
[ brf](b ) , it is clearly appreciated that the dominant branching ratios correspond to tree - level decays of @xmath9 .
nevertheless , the one - loop decay of the @xmath130 process offers a branching ratio on the order of @xmath6 for @xmath168 tev .
the remaining one - loop branching ratios turn out to be @xmath183 , @xmath184 , @xmath185 and @xmath186 , for @xmath168 tev . according to all the one - loop electroweak decay modes
, the @xmath128 process stands out as the most interesting one in order to search for new scalar bosons of pseudoscalar nature .
however , in phenomenological terms , this is not the most important decay mode at one - loop level , since the @xmath9 decay to two gluons would be the most likely to be found . in this section
we present an approximate study for the production cross section of the exotic pseudoscalar @xmath9 in the context fo the lthm at lhc , decaying into different final states mentioned above .
although this analysis is not entirely accurate , it can provide valuable information as for the order of magnitude of the production cross section of the @xmath9 particle is concerned . to perform this , we employ the breit - wigner resonant cross section @xcite . in this approximation
, the production cross section via gluon fusion can be computed by means of the branching ratios @xmath187 and @xmath188 , where @xmath189 .
thus , our breit - wigner cross section is written as follows @xmath190 where @xmath191 is estimated just at the resonance of the pseudoscalar boson @xmath9 @xcite . in gluon fusion as a function of the @xmath8 energy scale .
[ phiprod ] ] we have numerically calculated the production cross section of @xmath9 as a function of the @xmath8 parameter from @xmath192 tev to @xmath193 tev ( see fig . [ phiprod ] ) . from fig .
[ phiprod ] , it is evident that the region that has the greater predictive importance corresponds to @xmath8 around 2 tev .
because the branching ratios for the @xmath129 processes are very suppressed , we will focus our analysis on the decays @xmath128 and @xmath130 .
for @xmath168 tev , the production cross section of @xmath9 with @xmath194 final states is @xmath195 fb , whereas that @xmath196 fb . the expected integrated luminosity of the lhc at the last stage of operation
is projected to be around 3000 fb@xmath197 @xcite .
therefore , by considering this experimental scenario it would be very difficult to observe some event related with the @xmath128 , nonetheless , there would still be an observation gap for the @xmath4 process , since almost one event could be observed at lhc .
if we compare our results with those obtained in the context of the @xmath15 production at the lhc @xcite , we find that the branching ratio for the @xmath198 decay is of the same order of magnitude that @xmath199 .
in contrast , @xmath200 , @xmath201 and @xmath202 ; the last discrepancy is mainly due to that the @xmath203 decay is induced at tree level while the @xmath204 decay appears at one - loop level .
although our study does not represent a detailed calculation of the production mechanism of a new exotic scalar particle via gluon fusion at the lhc , it could offer a most complete experimental guide in order to search for new exotic scalar particles with masses of the order of tevs @xcite .
alternatively , the production mechanism of new heavy scalars could be studied in the context of the compact linear collider ( clic ) bearing in mind that this collider would offer much cleaner collisions also to a high integrated luminosity @xcite .
the lthm is based on a nonlinear sigma model together with a @xmath23 global symmetry and a gauged subgroup @xmath24\otimes[su(2)_2\otimes u(1)_2]$ ] , where the presence of new particles with masses at energy scales of tev is proposed . among all the variety of new particles
predicted , a new neutral pseudoscalar particle , identified as @xmath9 , is the main subject of this work . in specific , to explore the predictability of the lthm the @xmath17 decays were studied .
even though the parameter space of the lthm is severely restricted by electroweak precision data , there is still opportunity to test this model at low energies between 2 - 4 tev .
we have settled down an analysis region from 2 tev to 4 tev for the energy scale @xmath8 , which indicates a mass interval for @xmath116 between 1.66 tev and 3.32 tev , respectively . on this energy range
, it has been found that the relevant branching ratios correspond to the @xmath130 and @xmath128 decays , being at most of the order of @xmath6 and @xmath7 , respectively .
a roughly estimate for the production cross section of the @xmath9 boson via gluon fusion was implemented . for @xmath168 tev ,
the numerical results tell us that in the last planned stage of operation of the lhc , events corresponding to the @xmath128 decay would be very difficult to detect , however , the possibility of detection for the @xmath130 process is more likely to be present .
this work has been partially supported by conacyt , sni - conacyt , and cic - umsnh .
+ k. fujii , h. hano , h. itoh , n. okada and t. yoshioka , phys .
d * 78 * , 015008 ( 2008 ) ; p. s. bhupal dev , r. n. mohapatra and y. zhang , jhep * 1602 * , 186 ( 2016 ) ; m. bauer , m. neubert and a. thamm , phys .
rev . lett . * 117 * , 181801 ( 2016 ) ; b. fuksa , d. w. kang , s. c. park and m .- s .
seo , phys .
b * 761 * , 344 ( 2016 ) ; p. s. bhupal dev , r. n. mohapatra and y. zhang , arxiv:1703.02471 .
i. p. ivanov , v. keus , and e. vdovin , j. phys .
* 45 * , 215201 ( 2012 ) ; i. p. ivanov and e. vdovin , eur .
phys . j. c * 73 * , 2309 ( 2013 ) ; v. keus , s. f. king , and s. moretti , jhep * 1401 * , 052 ( 2014 ) . n. arkani - hamed , a. g. cohen , and h. georgi , phys . lett .
b * 513 * , 232 ( 2001 ) ; n. arkani - hamed , a. g. cohen , t. gregoire , and j. g. wacker , jhep * 0208 * , 020 ( 2002 ) ; n. arkani - hamed , a. g. cohen , e. katz , a. e. nelson , t. gregoire and j. g. wacker , jhep * 0208 * , 021 ( 2002 ) ; i. low , w. skiba , and d. smith , phys .
d * 66 * , 072001 ( 2002 ) . | we study two - body decays of a new neutral pseudoscalar into gauge bosons within the context of the littlest higgs model .
concretely , the @xmath0 processes induced at the one - loop level , with @xmath1 , are considered .
since the branching ratios of the @xmath2 decays result very suppressed , only the @xmath3 processes are thoroughly studied .
the branching ratios for the @xmath4 and @xmath5 decays are of the order of @xmath6 and @xmath7 , respectively , for @xmath8 around 2 tev , which represents the global symmetry breaking scale of the theory .
the production cross section of the @xmath9 boson via gluon fusion at lhc is estimated . |
intermittency has been observed in a variety of real settings as well in a vast number of numerical models
. a great deal of effort has therefore gone into understanding these modes of behaviour in the context of deterministic dynamical systems theory .
these studies have demonstrated the existence of a number of different types of intermittency ( such as pomeau
manneville @xcite , crisis @xcite , on - off @xcite intermittencies ) , each with their own associated signatures and scalings .
many of these forms of intermittency have in turn been concretely shown to be present in experiments and numerical studies of dynamical systems in a variety of settings ( see @xcite and references therein ) .
an important potential domain of applicability of such behaviour arises in understanding the mechanisms underlying the intermediate time scale variability in the sun @xcite - the occurrence of the so called _ maunder _ or _ grand minima _ - during which solar activity ( as deduced from the sunspot numbers ) was greatly diminished @xcite .
this behaviour is also confirmed by evidence coming from the analysis of proxy data @xcite .
there is also some evidence for similar types of variability in solar - type stars @xcite .
the idea that some type of dynamical intermittency may under pin the grand minima type variability in the sunspot record ( the _ intermittency hypothesis _ @xcite ) goes back at least to the late 1970 s @xcite .
this idea has been the subject of intense study over the recent years and has involved the employment of various classes of dynamo models , including ordinary differential equations ( ode ) ( e.g.@xcite as well as partial differential equations ( pde ) models ( e.g. @xcite ) .
in addition to the phenomenological evidence for the presence of intermittent - type behaviours in dynamo models @xcite , concrete evidence has recently been found for the presence of particular types of intermittency in both ode dynamo models @xcite as well as a recently discovered generalisation of on - off intermittency , referred to as _ in - out _
intermittency @xcite , in pde models @xcite . here
we wish to report concrete evidence for the occurrence of two other types of intermittency , namely the crisis induced and pomeau manneville type - i intermittencies , in pde mean field dynamo models .
the organisation of the paper is as follows . in sec .
ii we briefly introduce the model studied here .
iii summarises our evidence demonstrating the presence of these types of intermittencies in this model and finally in sec .
iv we draw our conclusions .
ideally one would wish to employ the full 3-dimensional dynamo models with a minimum number of approximations and simplifying assumptions . despite a number of important recent attempts @xcite , the difficulty of dealing with small scale turbulence makes a detailed and extensive self consistent study of such fully turbulent regimes in stars still computationally impractical ( see e.g. @xcite . in view of this an alternative approach in studies of stellar dynamos
has been to employ mean field models @xcite .
we should mention that there is an ongoing debate regarding the nature and realistic value of such models @xcite .
nevertheless , 3-d turbulence simulations do seem to produce magnetic fields whose global properties ( such as field parity and time dependence ) are similar to those expected from corresponding mean field dynamo models @xcite . in this way
mean field dynamo models seem to reproduce certain features of the more complicated models and allow the study of certain global properties of magnetic fields in the sun and solar - type stars ( see for example @xcite ) . the standard mean field dynamo equation
is given by @xmath0 where @xmath1 and @xmath2 are the mean magnetic field and mean velocity respectively and the turbulent magnetic diffusivity @xmath3 and the coefficient @xmath4 arise from the correlation of small scale turbulent velocities and magnetic fields ( @xmath4 effect ) @xcite .
we consider the usual algebraic form of @xmath4quenching namely @xmath5 where @xmath6 and @xmath7 is the co - latitude .
we solve eq .
[ dynamo ] in an axisymmetric configuration and in the following , as is customary @xcite , we shall discuss the behaviour of the solutions by monitoring the total magnetic energy , @xmath8 , where @xmath9 the induction constant , and the integral is taken over the dynamo region .
we split @xmath10 into two parts , @xmath11 , where @xmath12 and @xmath13 are respectively the energies of the antisymmetric and symmetric parts of the field with respect to the equator .
the overall parity @xmath14 is given by @xmath15/e$ ] , so @xmath16 denotes an antisymmetric ( dipole - like ) pure parity solution and @xmath17 a symmetric ( quadrupole - like ) pure parity solution . for the numerical results reported in the following section ,
we used a modified version of the axisymmetric dynamo code of brandenburg _ et al . _
( 1989 ) @xcite employed recently in @xcite .
these models are constructed from a complete sphere of radius @xmath18 by removing an inner concentric sphere of radius @xmath19 and a conical section of semi - angle @xmath20 about the rotation axis , from both the north and south polar regions ( see @xcite for details of the model and the relevant parameters ) . to test the robustness of the code we verified that no qualitative changes were produced by employing a finer grid and different temporal step length ( we used a grid size of @xmath21 mesh points and a step length of @xmath22 in the results presented in this paper ) . for the following results we use @xmath23 , which give the magnitude of the differential rotation and @xmath24 .
the magnitude of the @xmath4-effect is given by the dynamo parameter @xmath25 .
in the next section we show in turn concrete evidence for the occurrence of crisis induced and pomeau manneville type - i intermittencies .
as far as their detailed underlying mechanisms and temporal signatures are concerned , crises come in three varieties @xcite . here
we shall be concerned with only one of these types , referred to as `` attractor merging crisis '' , whereby as a system parameter is varied , two or more chaotic attractors merge to form a single attractor .
there are both experimental and numerical evidence for this type of intermittency ( see for example @xcite and references therein ) . in particular , this type of behaviour has been discovered in a 6-dimensional truncation of mean field dynamo models @xcite .
[ crisis.series ] shows the plots of the energy and parity for the above model as a function of time , calculated with @xmath26 and @xmath27 which show a bimodal behaviour , switching intermittently between two different chaotic states . #
1#20.46#1 to determine the nature of this behaviour more precisely , we have plotted in fig .
[ return_map.eps ] the return maps for the pde models ( 1 ) , showing the attractors before and after the merging .
as can be seen the resulting merged attractor is , as expected , larger than the superposition of the two pre - existing attractors .
these results can be taken as indications for the presence of crisis induced intermittency in this model . to substantiate this further
, we recall that another important signature of this type of intermittency is the way @xmath28 , the average time between switches , scales with the system parameter , in this case , @xmath29 . according to grebogi _
@xcite , for a large class of dynamical systems this relation takes the form @xmath30 where the real constant @xmath31 is the critical exponent characteristic of the system under consideration and @xmath32 is the critical value of @xmath29 at which the two chaotic attractors merge . #
1#20.43#1 the model under study here is a pde system which is formally infinite dimensional .
such pde models are numerically costly to integrate over long enough intervals of time ( sometimes in excess of 5000 time units ) necessary in order to obtain the scaling of the type ( [ scaling ] ) .
furthermore , the demonstration of such scaling requires a precise determination of the critical value @xmath32 which is difficult since as one approaches this value @xmath28 diverges and the integration time becomes prohibitive . despite these difficulties ,
we have succeeded to obtain strong evidence for the presence of such a scaling as depicted in fig .
[ crisis.statistics ] , with the corresponding @xmath33 .
@xcite conjecture that there may be a general tendency for @xmath31 to be larger for higher dimensional attractors .
we do have a value of @xmath31 higher than the previous one found for a related six dimensional ode dynamo model @xcite but much lower than the value range suggested by grebogi _
therefore , the conjectured range may need modification for large high dimensional systems .
# 1#20.46#1 there is also evidence for an enlargement of the final attractor after merging , as shown by the larger amplitudes of variation in the parity , in the sense that the parity gets closer to @xmath34 after the merging , as depicted in fig.[return_map.eps ] .
this helped us to numerically arrive at a better estimate for the critical value @xmath32 .
these indicators , taken together , amount to strong evidence for the presence of crisis induced intermittency for this model .
this type of intermittency , which is brought about through a tangent bifurcation , results in the system switching back and forth between a `` ghost '' periodic orbit and sudden bursts of chaotic behaviour @xcite .
there are both experimental and numerical evidence for this type of intermittency ( see for example @xcite and references therein ) . in particular this type of behaviour has been discovered in a 12-dimensional truncation of mean field dynamo model @xcite . to demonstrate the presence of this type of intermittency in the above pde dynamo model , we have plotted in fig .
[ typei.series ] the energy and parity as a function of time for the parameter values @xmath35 and @xmath36 , which clearly demonstrates switches between nearly periodic behaviour and sudden bursts .
we note that interestingly the energy in this case shows strong modulation which could be of interest in accounting for the occurrence of grand type minima in sunspot activity .
# 1#20.46#1 another signature of this type of intermittency is provided by the specific characteristics of its corresponding power spectrum . by employing finite dimensional maps@xcite
, it has been shown that the corresponding spectra have a broad - band feature whose shape obeys approximately the inverse - power law @xmath37 for @xmath38 , where @xmath39 is the saturation frequency .
below this frequency there is a flat plateau induced by noise that causes arbitrarily long laminar phases to become finite . as further evidence for this type of intermittency in the model ( 1 ) ,
we have plotted in fig .
[ typei.statistics ] the power spectrum at @xmath40 , obtained by averaging over 16 different initial conditions corresponding to different initial parities .
as can be seen , the power spectrum shows both the flat plateau and the @xmath37 power law scaling .
taken together , these indicators amount to strong evidence for the presence of pomeau manneville type - i intermittency for this model .
we have obtained concrete evidence , in terms of phase space signatures , spectra and scalings to demonstrate the presence of crisis induced and the pomeau manneville type - i intermittencies in axisymmetric mean field pde dynamo models . despite the rather idealised nature of these models , this is of potential importance since it shows the occurrence of two more types of intermittency ( in addition to in out intermittency recently discovered @xcite ) in these models which may in turn be taken as an indication that more than one type of intermittency may occur in solar and stellar dynamos .
this suggests that any observational programme for identifying the mechanisms underlying grand minima type variability needs to take into account the possibility that multiple intermittency mechanisms may be operative in different stars of the similar type , or even in the same star over different epochs .
this would also be of importance in the interpretation of proxy data . in this way a more appropriate hypothesis regarding such variability would be that of _ multiple intermittency hypothesis_. + we would like to thank axel brandenburg for providing us with the original code and andrew tworkowski for helpful discussions .
ec is supported by grant bd/5708/95
praxis xxi , jnict .
rt benefited from pparc uk grant no .
l39094 .
999 y. pomeau , and p. manneville , _ commun .
phys . _ * 74 * , 189 ( 1980 ) . c. grebogi , e. ott , and j. a. yorke , _ phys .
lett _ * 48 * , 1507 ( 1982 ) ; c. grebogi , e. ott , f. romeiras , and j. a. yorke , _ phys . rev .
a. _ * 36 * , 5365 ( 1987 ) .
n. platt , e. spiegel , and c. tresser , _ phys .
lett . _ * 70 * , 279 ( 1993 ) ; e. spiegel , n. platt , and c. tresser , _ geophys .
astrophys .
fluid dyn . _ * 73 * , 146 ( 1993 ) .
karakotsou , and a. n. anagnostopoulos , _ physica d _ * 93 * , 157 ( 1986 ) ; w. l. ditto _ et al .
_ , _ phys .
rev . lett . _ * 63 * , 923 ( 1989 ) .
m. n. bussac , and c. meunier , _
j. de phys . _
* 43 * , 585 , ( 1982 ) ; r. richter , a. kittel , g. heinz , g. fltgen , j. peinke , and j. parisi , _ phys .
b _ * 49 * , 8738 ( 1994 ) .
b. so , and h. mori , _ prog .
_ , * 72 * , 1258 ( 1984 ) ; b. so , and h. mori , _ physica d _ * 21 * , 125 ( 1986 ) ; d. l. feng , j. zheng , w. huang , c. x. yu , and w. x. ding , _ phys . rev .
e _ * 54 * , 2839 ( 1996 ) .
j. a. eddy , _ science _ * 192 * , 1189 ( 1976 ) ; n. o. weiss , and s. m. tobias , in solar and heliospheric plasma physics , ed .
g. m. simnett , c. e. alissandrakis , and l. vlahos , springer , berlin , ( 1997 ) .
j. beer , s. m. tobias , and n. o. weiss , _ solar physics _ * 181 * , 237 ( 1998 ) .
m. stuiver , and t. f. braziunas , _ nature _ * 338 * , 405 ( 1989 ) .
_ , _ ap . j. _ * 438 * , 269 ( 1995 ) .
r. tavakol , and e. covas , 1999 , meeting `` stellar dynamos : non - linearity and chaotic flows '' , submitted to proceedings of astronomical society of the pacific .
preprint available at web address _ http://www.maths.qmw.ac.uk/@xmath41eoc _
r. tavakol , _ nature _ * 276 * , 802 ( 1978 ) .
b. ya .
zeldovich , a. a. ruzmaikin , and d. d. sokoloff , in _ magnetic fields in astrophysics _ , ( gordon and breach , new york , 1983 ) ; n. o. weiss , f. cattaneo , and c. a. jones , _ geophys .
astrophys .
fluid dyn .
_ * 30 * , 305 ( 1984 ) .
e. a. spiegel , in _ chaos in astrophysics _ , edited by j. r. butcher , j. perdang , and e. a. spiegel ( reidel , dordrecht , 1985 ) ; u. feudel , w. jansen , and j. kurths , _ int .
j. of bifurcation and chaos _ * 3 * , 131 ( 1993 ) .
s. schmalz , and m. stix , _
a&a _ * 245 * , 654 ( 1991 ) .
d. schmitt , m. schssler , and a. ferriz - mas , _
a&a _ * 311 * , l1 ( 1996 ) .
a. tworkowski , r. tavakol , a. brandenburg , j. m. brooke , d. moss , and i. tuominen , _ mnras _ * 296 * , 287 ( 1998 ) . s. m. tobias , _ a&a _ * 307 * , l21 , ( 1996 ) .
j. m. brooke , j. pelt , r. tavakol , and a. tworkowski , _ a&a _ * 332 * , 339 ( 1998 ) . j. m. brooke , _ europhysics letters _ * 37 * , 3 ( 1997 ) .
e. covas , and r. tavakol , _ phys .
e. _ * 55 * , 6641 ( 1997 ) .
e. covas , p. ashwin , and r. tavakol , _ phys .
e. _ * 56 * , 6451 ( 1997 ) .
p. ashwin , e. covas , and r. tavakol , _ nonlinearity _ * 9 * , 563 ( 1999 ) .
e. covas , r. tavakol , p. ashwin , a. tworkowski , and j. m. brooke , submitted to _ phys .
a _ ( 1999 ) .
preprint available at web address p. a. gilman , _ apj .
suppl . _ * 53 * , 243 ( 1983 ) .
. nordlund , a. brandenburg , r. l. jennings , m. rieutord , j. ruokolainen , r. f. stein , and i. tuominen , _ apj _ * 392 * , 647 ( 1992 ) .
a. brandenburg , r. l. jennings , .
nordlund , m. rieutord , r. f. stein , and i. tuominen , _ jfm _ * 306 * , 325 ( 1996 ) .
f. cattaneo , d. w. hughes , and n. o. weiss , _ mnras _ * 253 * , 479 ( 1991 ) .
d. moss , d. m. barker , a. brandenburg , and i. tuominen , _
a&a _ * 294 * , 155 ( 1995 ) .
s. i. vainshtein , f. cattaneo , _ ap .
j. _ * 393 * , 165 ( 1992 ) ; f. cattaneo , d. w. hughes , _ phys .
e. _ * 54 * , 4532 ( 1996 ) .
m. steenbeck , and f. krause , _ astron .
nachr . _ * 291 * , 49 ( 1969 ) ; p.
h. roberts , and m. stix , _
a&a _ * 18 * , 453 ( 1972 ) . a. brandenburg , i. tuominen , f. krause , r. meinel , and d. moss , _ a&a _ * 213 * , 411 ( 1989 ) ; a. brandenburg , d. moss , and i. tuominen , _ geophys .
astrophys .
fluid dyn .
_ * 40 * , 129 ( 1989 ) .
a. brandenburg , 1999 , in : _ helicity and dynamos _ , eds .
a. a. pevtsov , american geophysical union , florida .
a. brandenburg , 1999 , in : _ theory of black hole accretion discs _ , eds .
m. a. abramowicz , g. bjrnsson , and j. e. pringle , cambridge university press .
f. krause , and k .- h .
rdler , _ mean - field magnetohydrodynamics and dynamo theory _ , ( pergamon , oxford , 1980 ) .
e. covas , r. tavakol , a. tworkowski , a. brandenburg , j. m. brooke , and d. moss , 1999 , _
a&a _ * 345 * , 669 ( 1999 ) . j. e. hirsch , b. a. huberman , and d. j. scalapino , _ phys . rev . a _ , * 25 * , 519 ( 1982 ) | we find concrete evidence for the presence of crisis induced and pomeau manneville type - i intermittencies in an axisymmetric pde mean field dynamo model .
these findings are of potential importance for two different reasons .
firstly , as far as we are aware , this is the first time detailed evidence has been produced for the occurrence of these types of intermittency for such deterministic pde models . and secondly , despite the rather idealised nature of these models , the concrete evidence for the occurrence of more than one type of intermittency in such models makes it in principle possible that different types of intermittency may occur in different solar - type stars or even in the same star over different epochs . in this way a
_ multiple intermittency framework _ may turn out to be of importance in understanding the mechanisms responsible for grand - minima type behaviour in the sun and solar - type stars and in particular in the interpretation of the corresponding observational and proxy evidence . |
single quantum dots grown by molecular beam epitaxy are one of the most promising sources of non - classical light due to their stable and sharp emission lines and easy integration on a chip via the mature iii - v semiconductor fabrication technology . in particular , inas quantum dots emitting at wavelengths around 950 nm have proved to be pure sources of single indistinguishable photons @xcite and entangled photons @xcite and a powerful platform for spin initialization and manipulation , spin - photon and remote spin entanglement @xcite . to encode information in single photons and transmit it over long distances , sources of quantum light emitting at the so - called telecommunication wavelengths are most desirable .
advances in the development of superconducting detectors operating at cryogenic temperatures @xcite allow detection efficiencies exceeding 90% @xcite , making single - photon experiments and technologies eminently feasible .
the growing interest in the field of long wavelength quantum dots is demonstrated by recent achievements such as the demonstration of bright sources of indistinguishable photons @xcite , interference of photons emitted by dissimilar sources @xcite , entangled photon pair generation @xcite and exciton fine - structure splitting manipulation @xcite in the telecom wavelength band . however , the growth and fundamental characterisation of quantum dots emitting at telecom wavelengths is less mature compared to emitters at wavelengths @xmath1 1 @xmath2 m .
the longer emission wavelength can be achieved by growing quantum dots in a quantum well ( the so - called dot - in - a - well or dwell structures @xcite ) , a technique that partially relaxes the strain accumulated during the stranski - krastanow growth , resulting in larger quantum dot dimensions .
the ingaas quantum well provides local strain relief and also preserves the quantum dot composition and height during growth by reducing the out - diffusion of in during capping of the dot layer @xcite .
a larger physical confining potential implies a reduced energy separation between the quantum dot confined states and radiative electron - hole recombinations for inas / gaas quantum dots can reach wavelengths around 1.3 @xmath2 m both at room temperature , for example for quantum dot lasers @xcite , and at low temperature for single - quantum dot applications @xcite . besides the technological interest associated with lower transmission losses , telecom - wavelength quantum dots present interesting fundamental properties due to very different confinement of electron and hole wavefunctions and potentially larger oscillator strengths compared to shorter - wavelength quantum dots . in order to translate the more developed technology of 950 nm - band quantum dots towards longer wavelengths ,
the fundamental properties of the emitters need to be further understood . to this end , the application of external electric and magnetic fields as well as coulomb blockade is a means to characterise the electron and hole wavefunctions and the coulomb interactions between carriers . since quantum dots emitting around 1300 nm are physically larger and have a higher in - composition than shorter wavelength quantum dots , the different composition and morphology can result in a different wavefunction extensions and electron - hole overlap , impacting their fundamental response to applied fields .
in this direction , analysis of the emission properties of quantum dots emitting at wavelengths > 1.2 @xmath2 m in the presence of an external magnetic field @xcite and of 1300 nm quantum dots in the presence of external strain @xcite have been reported .
however , the full characterisation of the fundamental properties of quantum dots allowing direct comparison of emitters at 950 nm and at telecom wavelengths is still incomplete . here
, we report on magneto - optical studies of the emission properties of single telecom - wavelength quantum dots grown within a charge - tunable structure .
we extract the physical properties of the electron and hole wavefunctions , including the confinement energies , interaction energies , wavefunction lengths , and @xmath0-factors . for excitons ,
we measure the permanent dipole moment , polarizability , diamagnetic coefficient , and zeeman splitting .
the sample investigated was grown by molecular beam epitaxy .
the dwell layer was grown at 500@xmath3c by initial deposition of 1 nm in@xmath4ga@xmath5as , followed by a deposition of nominally 1.8 monolayers ( ml ) of inas to form the quantum dots , at a growth rate of 0.016 mls@xmath6 .
sample rotation was stopped during growth of the quantum dot layer to provide a variation in inas coverage across the wafer , resulting in a variation in quantum dot density across the wafer .
the quantum dot layer was subsequently capped by 6 nm in@xmath4ga@xmath5as and a further 4 nm gaas at 500@xmath3c , before the substrate temperature was raised to 580@xmath3c for growth of the remaining structure . a cross - section transmission electron microscopy ( tem ) image of a dwell layer grown under similar conditions
is shown in ref.@xcite .
analysis of tem images indicates the dots have a base width of 20 - 30 nm and a relatively large capped height of around 8 - 10 nm , preserved due to reduced out - diffusion of in during capping by the ingaas layer .
a sketch of the energy diagram of the field effect structure with a single layer of dwell quantum dots is shown in fig .
1a . by applying a voltage between the semi - transparent nicr schottky gate on the sample surface and the doped ( @xmath7 ) gaas layer ( ohmic contact ) , discrete charging of the dots with single electrons
can be achieved @xcite .
the sample is placed in a cryostat at @xmath8 4 k and , using a microscope in confocal geometry , a fiber - coupled non - resonant ( 830 nm - wavelength ) laser is used to excite the emitters .
a zirconia super solid - immersion lens is positioned on the surface of the sample to reduce the excitation spot ( and therefore be able to excite single emitters in the relatively high density sample ) and increase the collection efficiency of the photoluminescence signal @xcite .
the emission from single quantum dots is then coupled to a single - mode fiber and sent into a grating spectrometer equipped with an ingaas array detector for spectral characterisation .
an external magnetic field can then be applied to the sample either parallel or orthogonal to the growth axis .
indicates the fermi energy level .
( b , c ) photoluminescence spectra collected as a function of the applied gate voltage v under non - resonant excitation ( @xmath9 = 830 nm ) at t = 4 k. the external magnetic field b applied along the quantum dot growth axis @xmath10 is 0 t in ( b ) and 9 t in ( c ) .
the emission lines corresponding to the different charge states of a single quantum are labeled accordingly ( x@xmath11 = neutral exciton , x@xmath12 = negatively - charged exciton , etc . ) .
, height=680 ]
examples of photoluminescence spectra acquired as a function of applied voltage are shown in fig .
distinct emission lines are visible and can be attributed to single exciton ( @xmath13 ) and negatively - charged exciton ( @xmath14 and @xmath15 ) recombinations .
discrete jumps in the emission wavelength , visible when varying the applied voltage , are the signature of the coulomb blockade effect occuring when an extra electron is added to the quantum dot bound states @xcite .
we observe a partial co - existence of excitonic lines of different charged states of the same quantum dot around the transition voltages due to the comparable rates of electron tunneling into the quantum dot from the fermi sea and exciton recombination ( see fig .
1b ) @xcite . by applying a perturbative coulomb blockade model @xcite to single quantum dots
, we can extract important physical parameters including the electron - electron ( electron - hole ) interaction energy , the electron ( hole ) confinement energy , and the effective lengths of the electron ( hole ) wavefunctions .
these results are shown in fig .
we find that the coulomb interaction energies are much smaller than the confinement energies and they can , therefore , be treated as perturbations .
thus , we consider the trapped carriers to be in the so - called strong confinement regime . compared to typical 950 nm quantum dots @xcite , we derive wavefunction effective lengths and confinement energies about a factor two larger and carrier - carrier interaction energies about a factor two smaller .
this is compatible with a larger quantum dot physical size .
it is worth noting that engineering of larger wavefunction lengths could lead to larger oscillator strengths for the excitonic transitions , relevant for quantum electrodynamics experiments with single quantum dots .
the larger confinement energies have important consequences in the tunability of the transition energies of the quantum dot confined states .
while the transition energy of quantum dots emitting below 1 @xmath2 m can be tuned by about 1 mev in typical field - effect transistor devices @xcite , the telecom wavelength quantum dots under study show a tunability up to 7 mev .
this can be explained by the larger electron and hole confinement energies which enable larger electric fields to be applied before the charges tunnel out of the confining potential .
this larger wavelength tunability is important for potential applications such as the mutual tuning of the emission lines with respect to optical cavities for quantum optics experiments , or cancellation of the fine structure splitting to create sources of entangled photon pairs by applying an external electric field @xcite .
measurements of the fine structure splitting of quantum dots , from the same growth and within the same structure as the ones presented in this work , were reported in ref.@xcite .
further , such large confinement energies yield carriers more decoupled from the electron reservoirs and have potential for reduced impact of phonon - induced dephasing . when we apply an external bias v , exciton transition energies ( e@xmath16 ) experience a stark shift , following the relation e@xmath16 = e@xmath17 - @xmath18f + @xmath19f@xmath20 , where the electric field f = -(v@xmath21 - v@xmath17)/d is a function of the schottky barrier height v@xmath17 and the distance @xmath22 between the back gate and sample surface
, @xmath18 is the permanent dipole moment and @xmath19 the polarizability .
given the structure of the sample under study ( see fig .
1a ) , we use v@xmath21 = 0.62 v , and @xmath22 = 400 nm and fit the exciton lines with quadratic functions , as shown in fig .
3a , to extract the permanent dipole moment @xmath18 and the polarizability @xmath19 .
we observe permanent dipole moments with @xmath23 values ( where @xmath24 is the electron charge ) ranging from -0.5 to -3.0 nm , values similar to those reported for quantum dots emitting around 950 nm @xcite , indicating that the electron and hole wavefunctions are centered within the quantum dot . the observed polarizabilities ( ranging between -0.5 to -1.2 @xmath2ev/(kv / cm)@xmath20 ) are slightly smaller than the values reported for shorter wavelength quantum dots @xcite , which can be again explained by the stronger confinement of the carriers in larger telecom wavelength quantum dots .
the negative sign of the polarizability implies that the hole is confined near the base of the dot , while the electron wavefunction , given its lighter effective mass , is delocalized over the quantum dot .
the polarizability and the permanent dipole moment , as expected @xcite , are linearly related , as shown in fig . 3b . and @xmath25 , respectively ) , as well as the electron and hole confinement energies ( @xmath26 and @xmath27 , respectively ) and the electron and hole wavefunction extension ( @xmath28 and @xmath29 , respectively ) .
the wavelength on the @xmath30-axis corresponds to the emission wavelength for the @xmath13 line in the middle of the emission tuning range .
, height=718 ] , measured from several single quantum dots , plotted as a function of polarizability @xmath19 .
the solid red line is a linear fit with slope 3.2 @xmath31 0.2 nm/(mev/(kv / cm)@xmath20 ) and the error bars are obtained from the errors in the fits.,title="fig:",height=340 ] , measured from several single quantum dots , plotted as a function of polarizability @xmath19 .
the solid red line is a linear fit with slope 3.2 @xmath31 0.2 nm/(mev/(kv / cm)@xmath20 ) and the error bars are obtained from the errors in the fits.,title="fig:",height=340 ]
to further investigate the properties of the bound - state wavefunctions of the telecom - wavelength quantum dots under study , we apply an external magnetic field * b * in the faraday ( * b * parallel to the quantum dot growth direction ) and in the voigt ( * b * orthogonal to the quantum dot growth direction ) configurations .
examples of photoluminescence spectra collected at a field of 9 t in the faraday geometry , when varying the electric field applied to the charge tunable structure are shown in fig .
a clear splitting of each excitonic transition is visible ; this will be analysed and discussed in more detail in the following sections .
we first consider the faraday geometry , where we apply magnetic fields up to 9 t parallel to the quantum dot growth axis and collect photoluminescence spectra as a function of the voltage applied to the charge tunable structure .
examples of the spectra collected for the negatively- charged exciton are shown in fig .
4a : the energy of the emission lines experiences the so - called diamagnetic shift in the presence of an external magnetic field , following the expression e=@xmath32b@xmath20 , where @xmath32 is the diamagnetic coefficient .
the values of @xmath32 obtained from individual quantum dots are shown in table i. the diamagnetic coefficient is related to the exciton binding and confinement energies , and therefore to the microscopic properties of each specific quantum dot .
for the telecom - wavelength quantum dots under study , we generally observe a modest ( about 10@xmath33 ) increase of @xmath32 between the neutral exciton and the charged exciton ( see table i ) . in the weak confinement regime , due to
electron - electron interaction the addition of a second electron to a neutral exciton can reduce @xmath32 by up to a factor 2 @xcite .
the fact that we do not observe a reduction , but rather a modest increase , in @xmath32 further validates that the carriers are in the strong confinement regime in the quantum dots under study , as reported also in ref .
we observe significant differences between the electron and hole wavefunction extents , as expected due to the difference in confinement potentials and effective masses . by applying the external magnetic field orthogonal to the quantum dot growth axis ( voigt configuration ) ,
the rotational symmetry of the wavefunctions is broken .
this results in the mixing of the originally bright and dark neutral exciton states , with the latter becoming visible in the photoluminescence spectra @xcite .
if we consider quantum dots in the 950 nm emission range , the difference between the diamagnetic coefficients measured in faraday and voigt configuration reaches values up to about a factor 3 @xcite .
interestingly , for the telecom wavelength quantum dots under study , the diamagnetic coefficient is one order of magnitude smaller in the voigt configuration compared to the results obtained in the faraday configuration . as the diamagnetic coefficient is a measure of the effect of confinement , this striking result confirms the unique morphology of the dwell quantum dots compared to typical self - assembled quantum dots emitting near 950 nm . as @xmath32 is an order of magnitude larger for applied fields in - plane versus out - of - plane ,
we conclude that the confinement in the growth direction is less for dwell quantum dots as expected .
examples of the spectra collected for the negatively - charged exciton in the faraday configuration are shown in fig .
4a : the emission line in the presence of the magnetic field is split by the so - called zeeman splitting ( see fig .
4b ) with a magnitude @xmath34e given by @xmath34e=@xmath35b , where @xmath36 is the bohr magneton and @xmath0 is the land factor .
the zeeman splittings that we measure range between about 10 and 40 @xmath2ev / t , considerably smaller than for 950 nm quantum dots ( that were reported to be 120 @xmath31 30 @xmath2ev / t @xcite ) .
this is consistent with theoretical calculations showing that an increase in the quantum dot size implies a reduction of the @xmath0-factor @xcite . by applying the external magnetic field orthogonal to the quantum dot growth axis ( voigt configuration ) ,
the rotational symmetry of the wavefunctions is broken .
as shown in fig .
4c , the single quantum dot negatively - charged exciton line splits into 4 separate contributions ( e@xmath37 , e@xmath38 , e@xmath39 , e@xmath40 ) , as a result of the zeeman splitting of the bright and dark states . the s - shell electron and hole @xmath0-factors can be determined from the exciton energies : by fitting the energies of each transition , one can determine @xmath41= e@xmath42-e@xmath43=e@xmath44-e@xmath45 and @xmath46= e@xmath42-e@xmath44=e@xmath43-e@xmath45 @xcite .
the values that we extract from these measurements are plotted in table i. we attribute the variations in the values of the @xmath0-factors for different quantum dots to be due to the dependence of @xmath0 on the quantum dot shape @xcite . the magnitudes of the @xmath0-factors measured for the negatively - charged states are consistently higher than the ones measured for the neutral exciton , in accordance with previous reports @xcite .
the electron @xmath0-factors are two to four times larger than hole @xmath0-factors since the electron wavefunctions are less confined than the hole ones and are therefore more sensitive to the external magnetic field . from the measurements shown in fig .
4c , one can also see that the four transitions are linearly polarised and , as expected , two of the four transitions disappear when polarisation is resolved into horizontal and vertical components .
in summary , we have fully characterized the exciton and carrier properties of single quantum dots emitting at telecom wavelengths ( near 1.3 @xmath2 m ) under applied electric and magnetic fields . via the coulomb - blockade model
, we extract the electron and hole wavefunction lengths as well as multi - particle coulomb interaction and confinement energies .
the results are consistent with a strong - confinement picture .
the confinement energies of these quantum dots are found to be a factor of two larger than 950 nm quantum dots . due to the larger confinement energies
, the excitonic transitions can be tuned over a larger range than 950 nm band quantum dots in comparable devices .
additionally , the deeper confinement holds promise for better decoupled spins from fermi or phonon reservoirs . with applied external magnetic fields we extract the zeeman and diamagnetic coefficients as well as electron and hole @xmath0-factors .
compared to 950 nm quantum dots , the zeeman splittings are significantly smaller and the diamagnetic coefficient shows a drastic shift when changing from the faraday to the voigt configuration due to the unique dwell morphology .
these results give insights into the fundamental properties of telecom wavelngth quantum dots .
further investigations into the impact of the dwell morphology on the strain in the dot ( and its effect e.g. on the heavy - hole / light - hole mixing in the valence band and the electron and hole spin coupling to nuclear spins ) are needed to validate the promising potential of telecom - wavelength quantum dots for future quantum information applications .
@xmath47(nm ) & excitonic state & @xmath48 & @xmath32(@xmath2ev / t@xmath20 ) + 1274 & x@xmath49 & [email protected] & [email protected] + & x@xmath12 & [email protected] & [email protected] + 1281 & x@xmath49 & [email protected] & [email protected] + & x@xmath12 & [email protected] & [email protected] + 1282 & x@xmath49 & [email protected] & [email protected] + & x@xmath12 & [email protected] & [email protected] + |p4cm|p4cm|p1.92cm|p1.92cm|p4cm| + @xmath47(nm ) & excitonic state & g@xmath50 & g@xmath51 & @xmath32(@xmath2ev / t@xmath20 ) + 1290 & x@xmath49 & [email protected]&[email protected] & [email protected] + & x@xmath12 & [email protected]&[email protected] & [email protected] + 1292 & x@xmath12 & [email protected]&[email protected] & [email protected] + 1300 & x@xmath49 & [email protected]&[email protected] & [email protected] + & x@xmath12 & [email protected]&[email protected] & [email protected] + 1310 & x@xmath12 & [email protected]&[email protected] & [email protected] + 1317 & x@xmath12 & [email protected]&[email protected] & [email protected] + | we report on the optical properties of single inas / gaas quantum dots emitting near the telecommunication o - band , probed via coulomb blockade and non - resonant photoluminescence spectroscopy , in the presence of external electric and magnetic fields .
we extract the physical properties of the electron and hole wavefunctions , including the confinement energies , interaction energies , wavefunction lengths , and @xmath0-factors . for excitons ,
we measure the permanent dipole moment , polarizability , diamagnetic coefficient , and zeeman splitting .
the carriers are determined to be in the strong confinement regime .
large range electric field tunability , up to 7 mev , is demonstrated for excitons .
we observe a large reduction , up to one order of magnitude , in the diamagnetic coefficient when rotating the magnetic field from faraday to voigt geometry due to the unique dot morphology .
the complete spectroscopic characterization of the fundamental properties of long - wavelength dot - in - a - well structures provides insight for the applicability of quantum technologies based on quantum dots emitting at telecom wavelengths . |
one major task in the theory of dynamics is to establish some kind of dynamic irreducibility of a system . of principal interest
are the systems that display some kind of persistent irreducibility .
the two main examples of this concept are robust transitivity and stable ergodicity .
blenders were introduced by c. bonatti and l. daz in @xcite , to produce a large class of examples of non - hyperbolic robustly transitive diffeomorphisms .
there , they showed that these objects appeared in a neighborhood of the time - one map of any transitive anosov flow .
these systems are partially hyperbolic .
to establish robust transitivity , they showed that the strong invariant manifolds entered a small ball ( the _ blender _ ) , where things got mixed .
this blending was then distributed all over the manifold by means of the strong invariant manifolds .
this phenomenon is robust , whence they get robust transitivity . in @xcite , c. bonatti and l. daz showed that blenders appear near co - index one heterodimensional cycles .
this provides a local source of robust transitivity . in ( * ? ? ?
* theorem c ) , the authors showed that , surprisingly , blenders also provide a local source of stable ergodicity .
this arouses some interest in the appearance of blenders in the conservative setting .
indeed , the presence of blenders near pairs of periodic points of co - index one in the conservative setting , allows the authors to prove a special case of a longstanding conjecture by c. pugh and m. shub , namely that stable ergodicity is @xmath1-dense among partially hyperbolic diffeomorphisms with 2-dimensional center bundle @xcite .
the aim of this paper is to obtain conservative diffeomorphisms admitting blenders near conservative diffeomorphisms with a pair of hyperbolic periodic points with co - index one .
this result is crucial in the proof of the pugh - shub conjecture @xcite .
let us remark that , in fact , what we need for the proof of the pugh - shub conjecture is a very special case of theorem [ main.theorem ] .
we think that in view of the new importance of blenders for the conservative setting , it is interesting to state the result in its full generality .
the main result of this paper is the following .
[ main.theorem ] let @xmath2 be a @xmath0 diffeomorphism preserving a smooth measure @xmath3 such that @xmath2 has two hyperbolic periodic points @xmath4 of index @xmath5 and @xmath6 of index @xmath7 .
then there are @xmath0 diffeomorphisms arbitrarily @xmath1-close to @xmath2 which preserve @xmath3 and admit a @xmath8-blender associated to the analytic continuation of @xmath4 .
the proof of theorem [ main.theorem ] closely follows the scheme in @xcite .
all preliminary concepts are in section [ section.preliminaries ] .
let @xmath2 be a @xmath0 diffeomorphism preserving a smooth measure @xmath3 such that @xmath2 has two hyperbolic periodic points @xmath4 of index @xmath5 and @xmath6 of index @xmath7 .
a first step is to prove : [ proposition.heterodimensional.cycle ] let @xmath2 be a @xmath0 conservative diffeomorphism such that @xmath2 has two hyperbolic periodic points @xmath4 of index @xmath5 and @xmath6 of index @xmath7 , then @xmath1-close to @xmath2 there is a @xmath0 conservative diffeomorphism such that the analytic continuations of @xmath4 and @xmath6 form a co - index one heterodimensional cycle .
the proof of this involves a combination of recurrence results by c. bonatti and s. crovisier , the connecting lemma , and a recent result by a. vila concerning approximation of conservative @xmath1 diffeomorphisms by smooth conservative diffeomorphisms .
we prove this proposition in section [ section.real.central.eigenvalues ] .
the goal of of the rest of the paper is to reduce this heterodimensional cycle to a standard form , in which perturbations are easily made .
note that in a co - index one heterodimensional cycle associated to the points @xmath4 and @xmath6 with periods @xmath9 and @xmath10 , both @xmath11 and @xmath12 have @xmath13 contracting eigenvalues and @xmath7 expanding eigenvalues .
there is a remaining ( central ) eigenvalue which is expanding for @xmath11 and contracting for @xmath14 .
this center eigenvalue , however , could be complex or have multiplicity bigger than one
. this could complicate getting a simplified model of the cycle .
next step is to show , like in @xcite , the following : [ teo.real.central.eigenvalues ] let @xmath2 be a @xmath0 conservative diffeomorphism having a co - index one cycle associated to the periodic points @xmath4 and @xmath6
. then @xmath2 can be @xmath1-approximated by @xmath0 conservative diffeomorphisms having co - index one cycles with real central eigenvalues associated to the periodic points @xmath15 and @xmath16 which are homoclinically related to the analytic continuation of @xmath4 and @xmath6 .
the fact that @xmath15 and @xmath16 are homoclinically related to the analytic continuation of @xmath4 and @xmath6 is important due to the fact that if @xmath17 is a @xmath8-blender associated to @xmath4 and @xmath15 is homoclinically related to @xmath4 then @xmath17 is also a @xmath8-blender associated to @xmath15 . see remark [ remark.blender.homoclinic ] .
the proof of theorem [ teo.real.central.eigenvalues ] is in section [ section.real.central.eigenvalues ] .
the proof of theorem [ main.theorem ] shall now follow after proving theorems [ teo.strong.homoclinic.intersection ] and [ teo.blenders ] below .
[ teo.strong.homoclinic.intersection ] let @xmath2 be a @xmath0 conservative diffeomorphism having a co - index one cycle with real central eigenvalues
. then @xmath2 can be @xmath1-approximated by @xmath0 conservative diffeomorphisms having strong homoclinic intersections associated to to a hyperbolic periodic point with expanding real central eigenvalue .
theorem [ teo.strong.homoclinic.intersection ] is the more delicate part .
next theorem is now standard after c. bonatti , l. daz and m. viana s work @xcite : [ teo.blenders ] let @xmath2 be a @xmath0 diffeomorphism preserving @xmath3 with a strong homoclinic intersection associated to hyperbolic periodic point @xmath4 with expanding real center eigenvalue
. then @xmath2 can be @xmath1-approximated by a @xmath0 diffeomorphism preserving @xmath3 and having a @xmath8-blender associated to @xmath4 . in the creation of conservative blenders
, we closely follow the scheme in @xcite .
it will be proved that after a @xmath1-perturbation , we obtain a @xmath0 conservative diffeomorphism such that the co - index one cycle with real central eigenvalue has local coordinates where the dynamics of the cycle is affine and partially hyperbolic , with one dimensional central direction .
this is called a _ simple cycle _ , see definition [ definition.simple.cycle ] .
let us note that , unlike in @xcite , we can not use sternberg s theorem to linearize , due to the obvious resonance .
we use instead the pasting lemmas by a. arbieto and c. matheus @xcite .
the construction of simple cycles @xmath1-close co - index one cycles with real center eigenvalues is proved in section [ section.simple.cycle ] .
we shall afterwards produce a continuous family of perturbations @xmath18 of @xmath2 shifting the unstable manifold of @xmath6 in @xmath19 so that it does not intersect @xmath20 , thus breaking the cycle .
these perturbations preserve the bundles @xmath21 , @xmath22 and @xmath23 . in this form ,
they induce maps of the interval on @xmath22 . by carefully choosing a small parameter @xmath24 , and possibly touching " the center expansion / contraction
, we may obtain that there is an @xmath25 plane that is periodic by two different itineraries of @xmath26 .
now , the dynamics @xmath26 on this periodic plane is hyperbolic , using a markovian property , we get two periodic points that are homoclinically related within this plane . using the @xmath27-lemma we obtain a periodic point with a strong homoclinic intersection .
this is done in section [ section.strong.homoclinic.intersection ] . in this way
we can apply the well - known techniques of bonatti , daz and viana @xcite and of bonatti , daz which give blenders near points with strong homoclinic intersections @xcite . note that these perturbations can be trivially made so that the resulting diffeomorphism be @xmath0 and conservative .
from now on , we shall consider a smooth measure @xmath3 on a smooth manifold @xmath28 , and a @xmath0 diffeomorphism @xmath29 preserving @xmath3 .
we shall say that @xmath2 is conservative if @xmath2 preserves @xmath3 .
given a hyperbolic periodic point @xmath4 of @xmath2 , the index of @xmath4 is number of expanding eigenvalues of @xmath11 , counted with multiplicity , where @xmath9 is the period of @xmath4 .
[ definition.heterodimensional.cycle ] a diffeomorphism @xmath2 has a heterodimensional cycle associated to two hyperbolic periodic points @xmath4 and @xmath6 of @xmath2 if their indices are different , and the stable manifold @xmath20 of @xmath4 meets the unstable manifold @xmath30 of @xmath6 , and the unstable manifold @xmath31 of @xmath4 meets the stable manifold @xmath32 of @xmath6 .
when the indices of @xmath4 and @xmath6 differ in one , we say @xmath4 and @xmath6 are a co - index one heterodimensional cycle , or co - index one cycle .
we shall say that @xmath2 is partially hyperbolic on an @xmath2-invariant set @xmath33 , or @xmath33 is partially hyperbolic for @xmath2 if there is a @xmath34-invariant splitting @xmath35 such that for all @xmath36 and all unit vectors @xmath37 , @xmath38 we have : @xmath39 for some suitable riemannian metric on @xmath28 .
we require that both @xmath40 and @xmath41 be non - trivial .
our goal is to produce a @xmath1-perturbation admitting a blender .
here is the definition of blender we shall be using : [ definition.blender ] let @xmath4 be a partially hyperbolic periodic point for @xmath2 such that @xmath34 is expanding on @xmath22 and @xmath42 .
a small open set @xmath17 , near @xmath4 but not necessarily containing @xmath4 , is a @xmath8-blender associated to @xmath4 if : 1 .
every @xmath5-strip well placed in @xmath17 transversely intersects @xmath20 .
2 . this property is @xmath1-robust .
moreover , the open set associated to the periodic point contains a uniformly sized ball .
a @xmath5-strip is any @xmath5-disk containing a @xmath7-disk @xmath43 , so that @xmath43 is centered at a point in @xmath17 , the radius of @xmath43 is much bigger than the radius of @xmath17 , and @xmath43 is almost tangent to @xmath41 , i.e. the vectors tangent to @xmath43 are @xmath1-close to @xmath41 .
a @xmath5-strip is well placed in @xmath17 if it is almost tangent to @xmath44 .
see fgure [ figure.blender ] .
naturally , it makes sense to talk about robustness of these properties and concepts , since there is an analytic continuation of the periodic point @xmath4 and of the bundles @xmath40 , @xmath22 and @xmath41 .
we can define @xmath45-blenders in a similar way . for @xmath45-blenders
we will consider a partially hyperbolic point such that @xmath22 is one - dimensional and @xmath34 is contracting on @xmath22 .
@xmath8-blender associated to @xmath4,width=226 ] @xmath8-blender associated to @xmath4,width=226 ] this is the definition of blender we shall be using in this work , and , in particular what we obtain in theorem [ main.theorem ] .
we warn the reader that there are other definitions of blenders . in @xcite ,
chapter 6.2 .
there is a complete presentation on the different ways of defining these objects .
our definition corresponds to definition 6.11 of @xcite ( the operational viewpoint ) . in some works of c. bonatti and l. daz ,
see for instance @xcite and @xcite , what we call @xmath17 is known as the _ characteristic region of the blender _ ; and what they call _ blender _ is in fact a hyperbolic set which is the maximal invariant set of a small neighborhood of @xmath6 ( definition 6.9 .
of @xcite ) . however , let us remark that , under the hypothesis of theorem [ main.theorem ] , we obtain a @xmath1-perturbation preserving @xmath3 and admitting blenders also in the sense of definition 6.9 of @xcite , and of @xcite .
the existence of blenders in the sense of @xcite implies the existence of blenders in the sense of definition [ definition.blender ] .
[ definition.blender.homoclinic ] let @xmath15 be a partially hyperbolic periodic point for @xmath2 such that @xmath34 is expanding on @xmath22 , with @xmath42 .
a small open set @xmath46 is called @xmath8-blender associated to @xmath15 if @xmath47 , where @xmath4 is a partially hyperbolic periodic point homoclinically related to @xmath4 and @xmath17 is a @xmath8-blender near @xmath4
. we shall also denote @xmath48 the @xmath8-blender associated to @xmath15 .
[ remark.blender.homoclinic ] let us note that if @xmath48 is a @xmath8-blender associated to a hyperbolic periodic point @xmath15 and @xmath17 is a @xmath8-blender near @xmath4 , where @xmath4 is homoclinically related to @xmath15 , then it follows from the @xmath27-lemma that @xmath49 transversely intersects every @xmath5-strip well placed in @xmath17 , and that this property is robust .
the @xmath27-lemma also implies that a @xmath8-blender associated to @xmath15 is also a @xmath8-blender associated to @xmath50 if @xmath15 and @xmath50 are homoclinically related .
if @xmath2 is a partially hyperbolic diffeomorphism in @xmath28 , then we can define @xmath8-blender associated to a periodic point @xmath15 directly as in definition [ definition.blender ] , without requiring that @xmath48 be near @xmath15 .
indeed , this requirement is only needed to guarantee the existence of the @xmath5-strips well placed in @xmath48 . in the case
that @xmath2 is partially hyperbolic , the @xmath41 and @xmath22 bundles are globally defined , and so definition [ definition.blender ] makes sense even without asking that @xmath48 be near @xmath15 .
let us consider two hyperbolic periodic points @xmath4 of index @xmath5 and @xmath6 of index @xmath7 , with periods @xmath9 and @xmath10 respectively .
let us denote by @xmath51 , @xmath52 the eigenvalues of @xmath11 and @xmath12 respectively , with @xmath53 , ordered in such a way that : @xmath54 s are contracting and the @xmath55 s are expanding .
moreover , @xmath56 is expanding for @xmath4 and contracting for @xmath6 .
note that @xmath56 could be equal to @xmath57 , and could be a complex eigenvalue . a co - index one heterodimensional cycle associated to two hyperbolic periodic points @xmath4 and @xmath6 as described above has real central eigenvalues if @xmath58 and @xmath59 .
note that when a cycle has real central eigenvalues , then @xmath60 and @xmath61 have multiplicity one . in the remainder of this subsection
, we shall work with heterodimensional cycles with real central eigenvalues . in this case , both orbits of @xmath4 and @xmath6 admit a partially hyperbolic splitting @xmath62 , with @xmath42 , @xmath63 and @xmath64
. we shall denote @xmath65 the strong stable manifold of @xmath4 , this means , the invariant manifold tangent to @xmath66 that has dimension @xmath13 .
analogously we define @xmath67 , the strong unstable manifold of @xmath4 .
note that the unstable manifold of @xmath4 could be of dimension @xmath5 , in which case it would contain @xmath67 .
a partially hyperbolic periodic point @xmath4 has a strong homoclinic intersection if @xmath68 the first goal is to obtain a @xmath1-perturbation of @xmath2 admitting a co - index one cycle with some local coordinates such that the dynamics in a neighborhood of the points are affine : [ definition.simple.cycle ] a co - index one cycle associated to the periodic points @xmath4 of index @xmath5 and @xmath6 of index @xmath7 is called a simple cycle if it has real central eigenvalues and : 1 .
@xmath4 and @xmath6 admit neighborhoods @xmath19 and @xmath69 on which the expressions of @xmath70 and @xmath71 are linear and partially hyperbolic , with central dimension 1 .
we denote the coordinates of a point @xmath72 or @xmath73 according to wether it belongs to @xmath19 or @xmath69 .
2 . there is a quasi - transverse heteroclinic point @xmath74 such that @xmath75 for some @xmath76 , and a neighborhood @xmath77 of @xmath78 satisfying @xmath79 for which @xmath80 is an affine map preserving the partially hyperbolic splitting , which is contracting in the @xmath13-direction , expanding in the @xmath7-direction , and an isometry in the central direction .
there is a point @xmath81 , with @xmath82 such that @xmath83 with @xmath84 for some @xmath85 , and a neighborhood @xmath86 of @xmath87 , such that @xmath88 and @xmath89 is an affine map preserving the partially hyperbolic splitting , and is contracting in the @xmath13-direction , expanding in the @xmath7-direction , and an isometry in the central direction .
4 . there is a segment @xmath90_p^c$ ] contained in @xmath91 , such that the form of @xmath92 in @xmath69 is @xmath93^c_q$ ] .
we call the affine maps @xmath94 and @xmath95 the transitions of the simple heterodimensional cycle . in this subsection we state the preliminary results that shall be used in this work .
the following , theorem 1.3 of @xcite , allows us to approximate conservative diffeomorphisms by transitive diffeomorphisms preserving @xmath3 : [ teo.bonatti.crovisier ] there exists a residual set of the set of diffeomorphisms preserving @xmath3 such that all diffeomorphisms in this set is transitive .
moreover , @xmath28 is the unique homoclinic class .
we shall use this theorem in combination with the connecting lemma below : [ connecting.lemma ] let @xmath96 be hyperbolic periodic points of a @xmath0 transitive diffeomorphism @xmath2 preserving a smooth measure @xmath3 .
then , there exists a @xmath1-perturbation @xmath97 preserving @xmath3 such that @xmath98 .
the recent remarkable result by a. vila allows us to approximate @xmath1 conservative diffeomorphisms by @xmath99 conservative diffeomorphisms : [ a. vila @xcite][teo.avila ] @xmath99 diffeomorphisms are dense in the set of @xmath1 diffeomorphisms preserving @xmath3 . this following conservative version of franks lemma is proposition 7.4 of @xcite : [ franks.lemma ]
let @xmath2 be a @xmath0 diffeomorphism preserving a smooth measure @xmath3 , @xmath100 be a finite set .
assume that @xmath46 is a conservative @xmath101-perturbation of @xmath34 along @xmath100 .
then for every neighborhood @xmath102 of @xmath100 there is a @xmath1-perturbation @xmath103 preserving @xmath3 , coinciding with @xmath2 on @xmath100 and out of @xmath102 , such that @xmath104 is equal to @xmath46 on @xmath100 . the fundamental tools in order to adapt the construction of c. bonatti and l. daz @xcite to the conservative case
are the pasting lemmas of a. arbieto and c. matheus @xcite .
we shall need two such lemmas , one for vector fields and the other for diffeomorphisms .
the following is theorem 3.1 . of @xcite , and states that we can paste " two sufficiently @xmath1-close @xmath0 vector fields , so that one gets the value of the first one on one set and the value of the second one on a disjoint set : [ pasting.lemma.para.campos ] given @xmath105 and @xmath106 there exists @xmath107 such that if @xmath108 are two @xmath0 vector fields preserving a smooth measure @xmath3 that are @xmath109 @xmath1-close on a neighborhood @xmath110 of a compact set @xmath111 , then there exist an @xmath3-preserving vector field @xmath112 @xmath101
@xmath1-close to @xmath113 and two neighborhoods @xmath102 and @xmath114 of @xmath111 such that @xmath115 satisfying @xmath116 and @xmath117 . we shall also need the pasting lemma for diffeomorphisms , which states that we can produce a @xmath0 diffeomorphism by pasting " a conservative diffeomorphism @xmath2 with its derivative on a neighborhood of a point : [ pasting.lemma.para.difeos]if @xmath2 is a @xmath118 diffeomorphism preserving a smooth measure @xmath3 and @xmath119 is a point in @xmath28 , then for any @xmath120 there exists a @xmath0 diffeomorphism @xmath121 preserving @xmath3 , @xmath101-@xmath1-close to @xmath2 and two neighborhoods @xmath102 and @xmath110 of @xmath119 such that @xmath122 and @xmath123 and @xmath124 ( in local charts ) .
let @xmath2 be a @xmath0 diffeomorphism preserving a smooth measure @xmath3 , with two hyperbolic periodic points @xmath4 of index @xmath5 and @xmath6 of index @xmath7 .
our first step is to show that @xmath2 can be @xmath1-approximated by a @xmath0 diffeomorphism preserving @xmath3 such that the analytic continuations of @xmath4 and @xmath6 form a co - index one heterodimensional cycle .
namely : [ lemma.heterodimensional.cycle ] let @xmath2 be a @xmath0 diffeomorphism preserving a smooth measure @xmath3 such that @xmath2 has two hyperbolic periodic points @xmath4 and @xmath6 , then @xmath2 is @xmath1-approximated by @xmath0 diffeomorphisms preserving @xmath3 such that the continuations of @xmath4 and @xmath6 either form a heterodimensional cycle in case they have different indices , or else they are homoclinically related .
let us consider the case where @xmath4 has unstable index @xmath7 and @xmath6 has unstable @xmath125 .
the case @xmath126 is simpler , and follows analogously .
we shall apply the connecting lemma ( theorem [ connecting.lemma ] ) to produce the heterodimensional cycle .
however , this lemma has transitivity as a hypothesis .
so , let us consider a conservative diffeomorphism @xmath127 @xmath1-close to @xmath2 so that @xmath127 is transitive
. this @xmath127 exists due to theorem [ teo.bonatti.crovisier ] by c. bonatti and s. crovisier .
note that @xmath127 is a priori only @xmath1 and has two hyperbolic periodic points @xmath128 and @xmath129 which are the analytic continuations of @xmath4 and @xmath6 .
now , due to its transitivity , the connecting lemma applies , and we can find a @xmath1-close conservative diffeomorphism @xmath130 such that @xmath131 intersects @xmath132 , where @xmath133 and @xmath134 are the analytic continuations of @xmath4 and @xmath6 .
we can even ask that this intersection be transverse ( it will be @xmath135-dimensional ) . note that since it is transverse , this intersection persists under @xmath1-perturbations .
we will also ask , by using theorem [ teo.bonatti.crovisier ] again , that @xmath130 be transitive . then
, we apply the connecting lemma again , and obtain a new @xmath1 conservative diffeomorphism @xmath136 @xmath1-close to @xmath130 so that @xmath137 and @xmath138 , we can even ask that this last intersection be quasi - transverse , i.e. @xmath139 does not contain a non - trivial vector . but @xmath136 could be not @xmath0 a priori .
theorem [ teo.avila ] of a. vila yields a @xmath99 conservative diffeomorphism @xmath140 @xmath1-close to @xmath136 .
since the stable and unstable manifolds of @xmath141 and @xmath142 vary continuously , we obtain that @xmath143 , and @xmath144 is close to @xmath145 near the point @xmath146 .
there is a @xmath99 conservative diffeomorphism @xmath147 , @xmath1-close to @xmath140 , so that @xmath148 intersects @xmath149 .
this last perturbation can be made in fact @xmath99 .
this gives the desired heterodimensional cycle .
theorem [ teo.real.central.eigenvalues ] follows from lemma [ lemma.real.central.eigenvalues ] below and lemma [ lemma.heterodimensional.cycle ] .
the idea of lemma [ lemma.real.central.eigenvalues ] is as in lemma 4.2 . of @xcite .
see also its generalization to dimension @xmath150 in lemmas 1.9 and 4.16 of @xcite .
the only difference is that we shall apply the pasting lemmas and the conservative franks lemma instead of the corresponding results .
[ lemma.real.central.eigenvalues ] let @xmath2 be a conservative @xmath0 diffeomorphism with a hyperbolic periodic point @xmath4
. then @xmath2 is @xmath1-approximated by a @xmath0 conservative diffeomorphism with a hyperbolic periodic point @xmath15 having the same index as @xmath4 such that the inequalities in equation ( [ equation.real.central.eigenvalues ] ) are strict .
let us assume @xmath4 is a hyperbolic fixed point for @xmath2 and let us denote the eigenvalues of @xmath152 as in equation ( [ equation.real.central.eigenvalues ] ) .
we may assume , by using the conservative franks lemma ( proposition [ franks.lemma ] ) that the multiplicity of all the eigenvalues , complex or real , is one ; and that any pair of complex eigenvalues with the same modulus are conjugated , and have rational argument .
we shall also assume that @xmath153 , and that @xmath60 and @xmath154 are complex conjugated expanding eigenvalues .
we shall prove that there is @xmath15 such that @xmath155 , whence @xmath15 will have index @xmath5 , just like @xmath4 .
in fact , this is all we need , since the same argument applies to show there is @xmath16 of index @xmath7 with real central ( contracting ) eigenvalue of multiplicity one .
lemma [ lemma.real.central.eigenvalues ] follows from an inductive argument , since the fact that the eigenvalue is the central one is not used in the argument . using the connecting lemma ( theorem [ connecting.lemma ] ) and genericity arguments we obtain a transverse homoclinic intersection @xmath156 . using the pasting lemma for diffeomorphisms ( theorem [ pasting.lemma.para.difeos ] ) we can linearize @xmath2 in a small neighborhood @xmath102 of @xmath4 , so that @xmath2 remains the same outside a neighborhood @xmath157 . by considering sufficiently large iterates
, we may assume that @xmath158 , and @xmath159 are such that the tangent spaces to @xmath20 at @xmath146 and to @xmath31 at @xmath119 are close enough to @xmath66 and @xmath160 .
the tangent space to @xmath20 at @xmath119 and to @xmath31 at @xmath146 are @xmath66 and @xmath160 due to the linearization .
let @xmath167 and @xmath168 . applying the conservative franks lemma we obtain a perturbation such that @xmath169 a new perturbation allows us to fix " the eigenspace associated to @xmath60 and @xmath154 , and obtain normal bases on which the derivative @xmath170 has the form : @xmath171 where @xmath172 is a @xmath173 matrix acting on a subspace @xmath174 , where @xmath175 is the 2-dimensional eigenspace associated to the eigenvalues @xmath60 and @xmath154 .
for simplicity we shall assume that @xmath4 and @xmath6 are fixed points , and that the co - index cycle is a simple cycle .
we shall call @xmath239 and @xmath240 .
we shall assume that @xmath241 and @xmath242 , since no greater complication appears in the cases @xmath243 and @xmath244 .
since the cycle is simple , there are coordinates @xmath245 and @xmath246 in suitable neighborhoods of @xmath247 and @xmath248 on which the expression of @xmath2 is : @xmath249 where @xmath250 , @xmath251 are contractions , and @xmath252 , @xmath253 are expansions .
we recall that there are points @xmath78 in the quasi - transverse intersection of @xmath254 , and @xmath255 in the transverse intersection of @xmath190 such that on suitable neighborhoods @xmath77 and @xmath86 the transitions @xmath256 and @xmath257 have the form : @xmath258 and @xmath259 where @xmath260 , @xmath261 are contractions and @xmath262 , @xmath263 are expansions .
we shall produce a continuous family of perturbations @xmath18 of @xmath2 shifting the unstable manifold of @xmath6 in @xmath19 so that it does not intersect @xmath65 , see figure [ figure.perturbation ] .
these perturbations preserve the bundles @xmath21 , @xmath22 and @xmath23 . in this form ,
they induce maps of the interval on @xmath22 . by eventually changing the original @xmath264 and @xmath241 ( so that @xmath26 continues to be conservative ) , and carefully choosing a small parameter @xmath24
, we may obtain that the @xmath25 plane containing the point @xmath255 is periodic by two different ( large ) itineraries .
now , the dynamics @xmath26 on this periodic plane is hyperbolic , then using a markovian property , we get two periodic points that are homoclinically related within this plane .
the @xmath27-lemma gives now a periodic point with a strong homoclinic intersection .
proceeding as in section [ section.simple.cycle ] we take a divergence - free vector field @xmath113 supported in a small neighborhood of @xmath265 , so that the composition @xmath26 of @xmath2 with the time-@xmath266 map of @xmath113 form a @xmath1-family of @xmath0 conservative diffeomorphisms admitting transitions @xmath227 of the form ( [ equation.transition2 ] ) and @xmath267 of the form : @xmath268 where @xmath230 is as in formula ( [ equation.transition1 ] ) since @xmath269 and @xmath270 , the formulas ( [ equation.linear.expresion.f ] ) hold for all small @xmath24 . note that if the composition @xmath271 makes sense for some point and takes a small neighborhood of @xmath255 into @xmath19 then , due to the formulas above , its center coordinate takes the form : @xmath272\qquad\mbox{where}\quad\delta y = y^--y^+\ ] ] conversely , we have the following : [ lemma.periodic.points.m.n ] if @xmath273 are such that @xmath274 with sufficiently large @xmath275 , then there is a point @xmath276 in the @xmath25 plane through @xmath255 that is @xmath26-periodic , with period @xmath277 .
its center eigenvalue is @xmath278 .
let @xmath24 be a small parameter , and let @xmath284 be sufficiently large .
suppose that @xmath285 .
then it is easy to see that the @xmath286 image of the @xmath287-disc @xmath288^s\times\{y^+\}\times[-1,1]^u$ ] contains a cylinder of the form @xmath289^u,\ ] ] for some sufficiently small @xmath290 , where @xmath291 denotes the @xmath13-disc of radius @xmath109 centered at @xmath119 .
analogously , we obtain that the @xmath286 pre - image of the @xmath287-disc @xmath288^s\times\{y^+\}\times[-1,1]^u$ ] contains a cylinder of the form @xmath292^s\times \{y^+\}\times b^u_\delta(z),\ ] ] for some @xmath293 and some suitable small @xmath109 , which can be taken equal to the previous one .
@xmath294 denotes the @xmath7-disc of radius @xmath109 centered at @xmath295 .
this implies the existence of a periodic point @xmath283 of period @xmath277 .
the fact that the transitions are isometries on the bentral direction implies that @xmath296 .
if @xmath279 are such that @xmath280 , then the previous argument gives us a periodic point @xmath297 which , by construction , is different from @xmath283 . due to linearity ,
the unstable manifolds of @xmath283 , @xmath297 are , respectively , the @xmath7-discs @xmath298^u$ ] and @xmath299^u$ ] . also , @xmath300^s\times(y^+,z_{m , n})$ ] and @xmath301^s\times(y^+,z_{m',n'})$ ] .
the @xmath7-discs transversely intersect the @xmath7-discs in the @xmath287-plane .
therefore , @xmath283 and @xmath297 are homoclinically related in the @xmath287-plane , that is @xmath302 the @xmath27-lemma implies the existence of a point @xmath303 in the intersection of the strong stable and strong unstable manifolds of @xmath283 .
the expression of @xmath26 in @xmath19 implies that this intersection is quasi - transverse .
this is proved in lemma 3.11 of @xcite .
the proof of theorem [ teo.strong.homoclinic.intersection ] now ends by taking @xmath312 so that @xmath313 . with the techniques used in section [ section.simple.cycle ] we can produce a @xmath1 perturbation so that to obtain a @xmath0 conservative diffeomorphism where the linear expressions ( [ equation.linear.expresion.f ] ) are such that the center coordinate expansions are , respectively @xmath314 and @xmath315 .
theorem [ teo.blenders ] has been proved in @xcite and in theorem 2.1 of @xcite , see also section 4.1 of @xcite .
note that the perturbations can be trivially made so that the resulting diffeomorphism be @xmath0 and conservative .
rrrrrr a. arbieto , c. matheus , a pasting lemma and some applications for conservative systems , _ erg .
th . & dyn .
* 27 * , 5 ( 2007 ) 1399 - 1417 .
arnaud , cration de connexions en topologie @xmath316 , _ erg .
th . & dyn . systems _ ,
* 21 * , 2 ( 2001),339 - 381 .
a. vila , on the regularization of conservative maps , _ preprint _
( 2008 ) arxiv:0810.1533 c. bonatti , s. crovisier , rcurrence et gnricit , _ inv . math .
_ , * 158 * , 1 ( 2004 ) , 33 - 104 . c. bonatti , l. daz , persistent nonhyperbolic transitive diffeomorphisms .
( 2 ) * 143 * ( 1996 ) , no . 2 , 357 - 396 . c. bonatti , l. daz , robust heterodimensional cycles and @xmath1-generic dynamics , _ journal of the inst . of math .
jussieu _ * 7 * ( 3 ) ( 2008 ) 469 - 525 .
c. bonatti , l. daz , e. pujals , a @xmath1-generic dichotomy for diffeomorphisms : weak forms of hyperbolicity or infinitely many sinks or sources , _ ann . math .
_ * 158 * ( 2003 ) 355 - 418 . c. bonatti , l. daz , e. pujals , j. rocha , robustly transitive sets and heterodimensional cycles , _ astrisque _ * 286 * ( 2003 ) 187 - 222 . c. bonatti , l. daz , m. viana , discontinuity of the hausdorff dimension of hyperbolic sets , _ c. r. acad .
paris sr .
_ * 320 * ( 1995 ) , no . 6 , 713 - 718 . c. bonatti , l. daz , m. viana , _ dynamics beyond uniform hyperbolicity . a global geometric and probabilistic perspective . _
encyclopaedia of mathematical sciences , 102 . mathematical physics , iii .
springer - verlag , berlin , 2005 .
l. daz , e. pujals , r. ures , partial hyperbolicity and robust transitivity , _ acta math _ ,
* 183 * ( 1999 ) 1 - 43 .
f. rodriguez hertz , m. rodriguez hertz , a. tahzibi , r. ures , a criterion for ergodicity of non - uniformly hyperbolic diffeomorphisms , _ era - ms _ * 14 * , ( 2007 ) 74 - 81 .
f. rodriguez hertz , m. rodriguez hertz , a. tahzibi , r. ures , new criteria for ergodicity and non - uniform hyperbolicity , _ preprint _
we may assume now that @xmath2 is a @xmath0 conservative diffeomorphism with a co - index cycle having real central eigenvalues .
our main result will be established if we prove theorems [ teo.strong.homoclinic.intersection ] and [ teo.blenders ] . to do this
, we shall perturb that in order to obtain a simplified model of the cycle , that is , a simple cycle .
the creation of simple cycles follows the same arguments as in ( * ? ? ?
* proposition 3.5 ) and ( * ? ? ?
* lemma 3.2 ) .
the only difference is that we shall use the pasting lemmas to linearize in the conservative setting .
the goal of this section is to produce a simple cycle .
let us suppose that @xmath4 and @xmath6 are hyperbolic fixed points of indices @xmath5 and @xmath7 respectively .
we may also assume that @xmath32 and @xmath31 have a non - trivial transverse intersection , and @xmath20 and @xmath30 have a point of quasi transverse intersection .
on @xmath178 we have a partially hyperbolic splitting @xmath62 , where @xmath64 , @xmath42 and @xmath63 with @xmath179 . using the pasting lemma for diffeomorphisms , we obtain two neighborhoods @xmath19 and @xmath69 on which we can linearize @xmath2 .
we call this new @xmath0 conservative diffeomorphism @xmath121 .
@xmath121 equals @xmath152 on @xmath19 and @xmath180 on @xmath69 , the strong stable and unstable manifolds are , respectively , the @xmath13- and @xmath7-planes parallel to @xmath66 and @xmath160 , or @xmath181 and @xmath182 .
the center lines parallel to @xmath183 and @xmath184 are also invariant under @xmath121 .
we can choose @xmath121 so that there is a point of transverse intersection @xmath113 of @xmath31 and @xmath32 , and a point of quasi - transverse intersection @xmath185 of @xmath20 and @xmath30 .
there is a sufficiently large iterate @xmath186 so that @xmath187 and @xmath188 .
take @xmath189 .
now , generically @xmath190 is transverse to the strong unstable in @xmath19 and to the strong stable foliation in @xmath69 .
so , take a curve @xmath191 and iterates @xmath186 so large that @xmath192 is the graphic of a map @xmath193 , and @xmath194 is the graphic of a map @xmath195 .
@xmath196 and @xmath197 are small segments contained , respectively in @xmath198 and @xmath199 .
note that @xmath192 approaches @xmath198 exponentially faster than it approaches @xmath4 ; analogously @xmath200 approaches @xmath199 exponentially faster than it approaches @xmath6 .
hence we can choose @xmath186 so large that @xmath201 and @xmath202 and are @xmath1-close to zero .
let us define @xmath0 vector fields @xmath203 and @xmath204 in suitable neighborhoods of @xmath205 and @xmath206 .
we define @xmath203 as a vector field that is constant the hyperplanes parallel to @xmath207 , such that @xmath208 .
that is , @xmath203 assigns to each point its center coordinate .
since @xmath203 is constant along the hyperplanes parallel to @xmath207 it is divergence free and it is very close to the null vector field .
then we can apply the pasting lemma for flows ( theorem [ pasting.lemma.para.campos ] ) and paste @xmath100 with the null vector field obtaining a @xmath0-vector field @xmath209 that is @xmath1-close to the null vector field . by composing our diffeomorphism with the time - one map of @xmath210 we have a perturbation of @xmath121 ( which we continue to call @xmath121 ) such that @xmath211 is contained in @xmath198 .
analogously , we obtain a @xmath1-perturbation , so that @xmath212 . in this way
, we have obtained so far the points @xmath78 , @xmath213 , @xmath87 and @xmath214 .
let us apply the pasting lemma for diffeomorphisms again , so that we obtain a new perturbation for which the transitions @xmath215 and @xmath216 are affine maps , where @xmath102 and @xmath114 are small neighborhoods of @xmath78 and @xmath87 respectively .
we loose no generality in assuming that the images of the hyperplanes @xmath21 , @xmath23 , @xmath217 , @xmath218 and @xmath219 are in general position . by taking @xmath76 sufficiently large
, one obtains that the image of the center - unstable foliation becomes very close to the @xmath219 in @xmath19 . a small perturbation using the pasting lemma for vector fields like in the previous paragraph gives us an invariant center - unstable foliation .
indeed , there exists a matrix @xmath220 with @xmath221 , close to the identity , taking the image of the center - unstable foliation in @xmath86 into the center - unstable foliation of @xmath222 . but now , there exists a vector field @xmath223 such that the time - one map of @xmath223 is @xmath220 .
we use the pasting lemma for vector fields to paste @xmath223 in a neighborhood of @xmath214 in @xmath224 with the identity outside of @xmath224 .
composing @xmath121 with the time one map of this vector field , we get a @xmath0 diffeomorphisms @xmath1-close to @xmath121 such that @xmath225 leaves the center - unstable foliation invariant .
we replace @xmath224 by @xmath226 , and @xmath227 by this new affine transition .
let us continue to call @xmath121 this new @xmath0 conservative diffeomorphism , and @xmath227 the new transition . in order to get the invariance of the strong stable foliation , note that by the previous construction , the center unstable foliation is preserved by backward iterations .
the backward iterations of the strong stable foliation approach the strong stable foliation in @xmath19 . if necessary
, we may replace @xmath87 by a large backward iterate , and @xmath114 by a corresponding iterate in @xmath19 .
there is a matrix @xmath46 with @xmath228 close to the identity , that preserves the center - unstable foliation and is such that @xmath229 preserves the strong stable and center - stable foliations .
proceeding as in the previous paragraph , we obtain a @xmath1-close diffeomorphism preserving these foliations .
we can repeat now the same argument inside the center - unstable foliation in order to get the invariance of the strong unstable and the center foliations . in this way we obtain an affine partially hyperbolic transition @xmath227 preserving @xmath21 , @xmath23 and @xmath22 .
analogously we obtain @xmath230 .
we only need to show that we can perturb in order to obtain that the transitions @xmath230 and @xmath227 are isometries on the center foliations .
now we can replace @xmath230 by @xmath231 with large @xmath232 on a suitable small neighborhood of @xmath233 .
there are infinitely many @xmath232 such that the center eigenvalues of @xmath234 are in a bounded away from zero finite interval . considering @xmath235 and @xmath236 sufficiently large , and changing @xmath78 by a point @xmath113 with coordinates of the same form , we obtain a @xmath1-perturbation in a small neighborhood of the segment of orbit @xmath113 , @xmath237 , where @xmath238 , such that the action of @xmath234 in the central direction is an isometry .
the perturbation is produced using the pasting lemma for vector fields as in the previous paragraphs . in analogous way we obtain a transition @xmath227 acting as an isometry in the central direction .
let @xmath2 be a @xmath0 conservative diffeomorphism having a co - index one cycle with real central eigenvalues associated to the points @xmath4 and @xmath6 .
then @xmath2 can be @xmath1-approximated by @xmath0 conservative diffeomorphisms having simple cycles associated to @xmath4 and @xmath6 . | in this work we prove that each @xmath0 conservative diffeomorphism with a pair of hyperbolic periodic points of co - index one can be @xmath1-approximated by @xmath0 conservative diffeomorphisms having a blender . |
in this work we are concerned with identifiability in a particular kind of blind source separation ( bss ) motivated by different applications in digital communication ( see e.g.,@xcite ) , but also in cancer genetics ( see e.g.,@xcite ) . a prominent example is the separation of a mixture of audio or speech signals , which has been picked up by several microphones , simultaneously ( see e.g. , @xcite ) . in this case
the different speech signals correspond to the sources and the recordings of the microphones to the mixture of signals with unknown mixture weights . from this mixture
the individual signals have to be separated .
more generally , in bss problems one observes @xmath0 mixtures of @xmath1 sources and aims to recover the original sources from the available observations . in this paper
we focus on the linear case ( for the non - linear case see e.g. , @xcite ) , where the blindness refers to the fact that neither the sources nor the mixing weights are known .
we also treat the case of unknown number of sources . a minimal requirement underlying any recovery algorithm for sources and mixture weights ( in a deterministic or noisy setting ) to be valid is identifiability , i.e. , the unique decomposition of the mixture into sources and mixing weights . without any additional information on the source signals @xmath2 identifiability can not hold , of course , as from a linear system @xmath3 the matrices @xmath4 and @xmath2 are not uniquely determined , in general . in particular
this applies to single linear mixtures ( @xmath5 ) .
however , if we assume that the values of the sources are attained in a known finite set of real numbers ( finite alphabet ) , then bss identifiability holds under certain conditions , which is necessary for any recovery algorithm to be valid .
therefore , the aim of this work is to give a comprehensive discussion of such conditions .
we are able to give ( under weak assumptions ) a complete combinatorial characterization of identifiability . from this
we derive sufficient conditions , which are easy to verify . moreover , these conditions yield an explicit construction to recover sources and weights in the noiseless case from the deterministic mixture . more specifically , using the notation @xmath6 for a vector @xmath7 and a matrix @xmath8 with @xmath9 rows and @xmath10 columns and @xmath11 for the corresponding row and column vectors , respectively , we assume from now on that the observed signal is linked to the sources via @xmath12 where @xmath13 is the deterministic mixture , @xmath14 are the unknown mixing weights , @xmath15 are the unknown source signals , and @xmath16 is a known alphabet , i.e. , the set of possible values the sources can attain .
the row vectors @xmath17 with @xmath18 denote the single source signals ( of length @xmath19 ) and the row vectors @xmath20 with @xmath21 denote the mixing vectors ( of length @xmath1 ) of the different mixtures . in an observational model
we often have @xmath22 where @xmath23 is some random noise term with zero mean .
all identifiability results for ( [ model ] ) transfer to this situation as well , of course .
however , we stress that the primarily aim of this paper is not to provide a method which reconstructs the mixing weigths @xmath4 and the sources @xmath2 from the ( possibly ) noisy observations @xmath24 , but rather to clarify the scope of scenarios under which this is possible . to this end
, we will analyze necessary and sufficient conditions under which the decomposition in ( [ model ] ) is unique . for the moment
, we assume that the number of sources @xmath1 is known . however , we point out that most of our results remain true even when @xmath1 is unknown ( see section [ sec : um ] ) , a case which has never been treated before to best of our knowledge . as far as we are aware of , identifiability of model ( [ model ] ) has not yet been considered in the literature in this general form , although various special cases and variations of this bss problem have been addressed .
the particular case of a binary alphabet , i.e. , when @xmath25 and @xmath26 , has been considered in @xcite .
diamantaras and papadimitriou @xcite and rostami et al .
@xcite assume the alphabet to be equally spaced , i.e. , @xmath27 .
we consider arbitrary finite alphabets @xmath28 .
moreover , several authors ( e.g.,@xcite ) assume a specific distribution on the alphabet , e.g. , uniform .
our results show that it is already sufficient to observe some specific combinations of alphabet values ( which are minimal conditions in a sense ) , hence we do not need to assume such a specific distribution .
diamantaras @xcite works with a general finite alphabet as well but considers a mixture of two sources , @xmath29 .
@xcite gave necessary and sufficient identifiability criteria for sparse signals , i.e. , signals having many zero entries , in contrast to our work .
although they consider underdetermined mixtures , their results require at least two sensors ( @xmath30 ) and do not hold for single linear mixture ( @xmath31 ) . bofill and zibulevsky @xcite
suggest for @xmath32 a method for estimating the mixing weights for sparse signals as well , although without giving any explicit identifiability criteria .
diamantaras @xcite considers a general finite alphabet but assumes the mixing weights to be complex .
thus , he works with a 2-dimensional signal . combing the results from @xcite and @xcite yields a sufficient identifiability criterion for finite alphabet sources with complex mixing weights .
however , this result does not hold when the mixing weights are real as considered here .
there are further variations of the bss problem .
some of them are associated with independent component analysis ( ica ) ( see e.g. , @xcite ) , which is based on the stochastic independence of the different sources ( assumed to be random ) .
ica can be a powerful tool for ( over)determined models ( @xmath33 ) @xcite and there are approaches for underdetermined multiple linear mixture models ( @xmath34 ) as well @xcite . however , ica is not applicable for single linear mixtures ( @xmath31 ) , as the error terms of the single sources sum up to a single error term such that stochastic independence of the sources becomes irrelevant .
also conceptually related is blind deconvolution ( see e.g. , @xcite ) , however , the convolution model makes analysis and identifiability severely different @xcite .
another related problem is non - negative matrix factorization ( see e.g. , @xcite ) , where one assumes ( [ model ] ) , but instead of @xmath35 , both @xmath2 and @xmath4 are non - negative .
indeed , the identifiability conditions derived in @xcite are quite related in nature to ours in section [ subsec : sc ] , where their simpliciality condition on @xmath4 corresponds to our condition ( [ emptyi ] ) and their separability condition on @xmath2 corresponds to our assumption a[assusuff ] .
however , whereas their assumptions necessarily imply @xmath36 , ours yield identifiabiliy for single linear mixtures ( @xmath5 ) , explicitly exploring the finite alphabet .
for the non - blind scenario , i.e. , when @xmath4 in model ( [ model ] ) is known , @xcite considers identifiability in a probabilistic framework . to the best of our knowledge ,
a comprehensive characterization and unifying treatment of identifiability of the mixing weights and the sources for model ( [ model ] ) has been elusive .
this issue is , however , fundamental for identifying the scope of possible scenarios where recovery algorithms for @xmath2 and @xmath4 in ( [ model ] ) are applicable ( see , e.g. , @xcite ) . in this sense
our work provides an almost `` minimal '' set of conditions under which any recovery algorithm for the bss problem only can be expected to be valid ; in the noiseless case as well as for the case with random error .
we will start our analysis by making two simplifications , which lead to a better interpretation of the corresponding identifiability conditions .
first , we will assume that @xmath5 , i.e. , that we observe a single linear mixture in ( [ model ] ) .
clearly , when @xmath0 increases the identification problem becomes easier , as more mixtures of the same sources are observed .
thus , the case @xmath5 corresponds to the most difficult scenario and therefore we treat this case in detail . generalizations to arbitrary @xmath0 will then follow easily from this case and are given in section [ sec : mlm ] .
second , we start with considering probability mixing weights , i.e. , @xmath37 and @xmath38 for all @xmath39 .
this is because the corresponding identifiability conditions have an easier interpretation when the mixing weights are positive .
when we allow for negative mixing weights , the identifiability issue becomes slightly more difficult and the corresponding conditions become more complicated .
generalizations to negative mixing weights are given in section [ sec : abw ] .
the paper is organized as follows . after introducing a rigorous formulation of the problem and model ( section [ sec : probstat ] )
, we will give a necessary and sufficient identifiability criterion for the mixing weights and the sources ( section [ subsec : nsc ] ) . to this end
, we will first characterize the identifiability issue as a purely combinatorial problem . in sections [ subsec : sc ]
we will then generalize a result from diamantaras and chassioti @xcite in order to derive a simple sufficient identifiability criterion .
this characterizes those source signal combinations which make the variation of the mixture rich enough in order to become identifiable .
this condition also provides an explicit construction ( see the algorithm in figure [ fig : algodetmu ] ) for recovery of the weights @xmath4 and the sources @xmath2 from the mixture @xmath13 in ( [ model ] ) . in section
[ sec : sp ] we will shortly discuss how likely it is for the identification criterion of section [ subsec : sc ] to be satisfied when the underlying sources @xmath40 are discrete markov processes . to this end
, we will bound the probability of identification from below by a phase - type distribution . using a stopping time argument
, we will show that the mixture becomes identifiable exponentially fast , which reveals identifiability as less a critical issue in many practical situations as one might expect . although , we assume the number of source components @xmath1 to be known , in section [ sec : um ] we show that most of our results remain true even when @xmath1 is unknown . in section [ sec :
sim ] we simulate source signals from a two - state markov chain to illustrate how far our derived sufficient identifiability conditions ( section [ subsec : sc ] ) are from being necessary in this setting . to this end ,
our results from section [ subsec : nsc ] are fundamental as they give an explicit way to decide whether a given mixture is identifiable or not .
our simulation results reveal the simple sufficient condition in section [ subsec : sc ] as quite sharp as we find that the number of observations needed for this to hold is quite close to the actual number of observations required for identifiabiliy .
this establishes the exponential bound in section [ sec : sp ] as a useful tool to estimate the required number of observations guaranteeing identifiability with high probability .
we conclude in section [ sec : conc ] .
as mentioned above , for ease of presentation , we start with analyzing identifiability in ( [ model ] ) for single linear mixtures , i.e. , @xmath5 ( for arbitrary @xmath0 see section [ sec : mlm ] ) and probability mixing weights ( for arbitrary mixing weights see section [ sec : abw ] ) . to this end , let @xmath41 denote the set of positive ( probability ) mixing vectors . note that @xmath42 for @xmath43 is always necessary to ensure identifiability of different sources @xmath44 ( if @xmath45 interchanging of @xmath46 results in the same mixture @xmath47 ) . for a given finite alphabet @xmath48 , with @xmath49 , and
given @xmath50 let @xmath51 and @xmath52 .
then the observed values are given by @xmath53 , i.e. , @xmath54 let @xmath55 ; @xmath56 as in ( [ modeld ] ) .
then we denote the vector @xmath57 and the matrix @xmath2 as ( jointly ) identifiable from the observational vector @xmath47 when there exists exactly one @xmath58 such that @xmath59 . in other words
, identifiability means that @xmath60 in ( [ modeld ] ) implies that @xmath61 and @xmath62 . for simplicity
, we refer to @xmath57 and @xmath2 being identifiable from @xmath47 just by saying that @xmath63 is identifiable .
the aim of this paper is to study under which conditions @xmath64 is identifiable from @xmath47 .
even though we assumes @xmath1 to be known , most of our results remain true when @xmath1 is unknown , i.e. , @xmath55 with @xmath65 ( see section [ sec : um ] ) . to illustrate the problem and notation , let us start with a simple example of model ( [ modeld ] ) , where @xmath66 and the alphabet is binary with @xmath67 .
this means that we consider mixing vectors of the form @xmath68 with @xmath69 , @xmath70 and two different sources @xmath71 , @xmath72 with @xmath73 for @xmath74 and @xmath75 .
the question we would like to answer is , under which conditions on @xmath57 and @xmath2 is @xmath63 uniquely determined via @xmath55 . for a given observation @xmath76 the underlying source vector @xmath77 equals one of the four different values @xmath78 and hence , @xmath79 clearly , if any two of the four values in the set on the r.h.s . of ( [ exampleeq2 ] )
coincide , then two different source values in ( [ exampleeq1 ] ) lead to the same mixture value for @xmath76 and hence the sources are not identifiable , i.e. , they can not be distinguished .
consequently , a necessary condition for identifiability is that all values in the r.h.s . of ( [ exampleeq2 ] )
are different , which is equivalent to @xmath80 in other words , it is necessary that the alphabet values in @xmath81 are well separated via the mixing weights @xmath82 . a generalization of this argument to arbitrary alphabets and number of sources
is done later in ( [ emptyi ] ) .
further , we may assume w.l.o.g .
@xmath83 , i.e. , we denote that source as @xmath84 which comes with the smaller weight .
( [ exampleeq3 ] ) alone , however , is necessary but not sufficient for identifiability .
for instance , if @xmath85 for all @xmath75 then @xmath86 and hence , @xmath57 is not identifiable from @xmath47 .
thus , a certain variability of the two sources @xmath87 and @xmath88 is necessary to guarantee identifiability of @xmath57 . in this simple example
, it is easy to check that a necessary and sufficient variability of @xmath87 and @xmath88 is that @xmath2 either takes the value @xmath89 ( i.e. , @xmath90 for some @xmath91 ) or @xmath92 ( i.e. , @xmath93 for some @xmath91 ) as by ( [ exampleeq3 ] ) and @xmath70 it follows that @xmath94 . in other words , it is necessary that the mixing weight @xmath95 ( or @xmath96 respectively ) is seen somewhere in the mixture @xmath47 on its own , without the influence of the other mixing weight
. a generalization of these assumptions and this argument to general systems ( [ model ] ) is done later in theorem [ theoalgo ] and theorem [ theoalgoabw ] , respectively .
in model ( [ modeld ] ) every observation @xmath76 , for @xmath97 , is given by a linear combination of @xmath98 ( unknown ) with one of the finitely many vectors in @xmath81 .
so in order to identify @xmath99 , we have to determine the corresponding vector in @xmath81 .
note that multiple observed values leave this identification problem invariant , i.e. , do not contribute further to identifiability .
hence , w.l.o.g . we assume all observations @xmath100 to be pairwise different .
note , that this implies @xmath101 .
of course , when for a given mixing vector @xmath98 there exist @xmath102 with @xmath103 and one observes this value , it is not possible to identify the underlying sources @xmath40 uniquely .
consequently , a necessary condition for identifiability is that those values are not observed , i.e. , @xmath104 for all @xmath97 . for arbitrary sources this
is comprised in the condition of a positive _ alphabet separation boundary _
, i.e. , @xmath105 let @xmath106 be the collection of injective maps from @xmath107 to @xmath108 , i.e. , for @xmath109 the vector @xmath110 corresponds to a selection of elements from @xmath111 .
[ theoid ] assume model ( [ modeld ] ) with @xmath112 @xmath113 .
let @xmath114 be an arbitrary invertible @xmath115 matrix with elements in @xmath116 .
assume that @xmath117 and [ asstheoid ] there exists @xmath109 such that @xmath118 .
then @xmath63 is identifiable if and only if [ condit ] there exists exactly one @xmath119 such that for @xmath120 @xmath121 i.e. , @xmath122 is a valid mixing weight and can reproduce all observations . for a proof see appendix [ app : a ] .
theorem [ theoid ] is fundamental for the following as a[condit ] provides a necessary and sufficient condition for identifiability of @xmath64 , given a[asstheoid ] holds . in section [ subsec : sc ]
, it will serve to derive a simple sufficient identifiability condition which is easy to check . in the following
, we will discuss condition a[asstheoid ] and a[condit ] more detailed .
first , we consider assumption a[asstheoid ] . without further restrictions on the matrix @xmath123 , a[asstheoid ]
simplifies to @xmath124 ( in particular implying @xmath125 ) , which is , indeed , an almost minimal condition . by simple linear algebra , it is easy to check that @xmath126 implies that for any @xmath98 exists an @xmath127 such that @xmath128 , i.e. , @xmath64 is not identifiable .
when we allow for arbitrary mixing weights ( see section [ sec : abw ] ) , i.e. , not necessarily summing up to one , then by the same argument @xmath124 becomes even a necessary condition .
intuitively , a[asstheoid ] ensures that the sources @xmath129 differ sufficiently such that one can identify @xmath57 from their mixture .
for instance , if @xmath130 , it follows that @xmath131 , irrespective of @xmath57 .
note that if @xmath132 different values @xmath133 are observed , i.e. , @xmath134 ( recall that w.l.o.g . in this section
@xmath19 equals the number of pairwise different @xmath76 s ) , then it must hold true that @xmath135 .
thus , a[asstheoid ] follows trivially for any invertible @xmath115 matrix @xmath123 with elements in @xmath116 . however , a[asstheoid ] is a much weaker assumption than @xmath134 .
second , we comment on assumption a[condit ] . given a[asstheoid ]
, assumption a[condit ] reveals @xmath57 as identifiable as soon as we can assign a collection of observations to rows in @xmath123 in a unique way .
when @xmath57 is determined , identifiability of @xmath2 follows from @xmath136 . for a[condit ] to check , we need to assume a fixed , given @xmath123 which depends on @xmath137 in a[asstheoid ] .
however , @xmath137 does not need to be known , in order to be able to verify condition a[condit ] , i.e. , one does not need to know which observations correspond to rows in @xmath123 . in practice , the choice of @xmath123 will depend on the specific application and different choices of @xmath123 lead to different identifiability conditions a[condit ] . in section [ subsec : sc ]
we will show that for some specific choices of @xmath123 a[condit ] always holds , i.e. , a[asstheoid ] already implies identifiabiliy .
the following example shows that this is not true in general , i.e. , not for any choice of @xmath123 .
[ gsi ] with the notation of ( [ modeld ] ) and theorem [ theoid ] , let @xmath138 , @xmath139 , @xmath140 , and @xmath141 i.e. , @xmath137 in a[asstheoid ] is the identity map and @xmath142 for @xmath143 we find that @xmath144 which is a valid mixing weight .
hence , @xmath64 is not identifiable .
as mentioned before , in section [ subsec : sc ] we will show that some specific choices of @xmath123 already lead to uniqueness of the selection @xmath145 in a[condit ] , and thus ensure identifiability .
the following remark illustrates how specific choices of rows in @xmath123 fix some subdomain of @xmath145 in a[condit ] .
[ remark1 ] if @xmath134 with @xmath146 , then @xmath147 , the smallest observed value , corresponds to the situation when all sources @xmath148 take the smallest value of the alphabet ( denoted with @xmath149 ) , i.e. , if @xmath150 , then for all @xmath145 satisfying a[condit ] @xmath151 .
the second smallest observed value , @xmath152 , corresponds to the situation when all sources @xmath153 take the smallest value @xmath149 , but the source @xmath154 with the smallest weight @xmath95 takes the second smallest value @xmath155 , i.e. , if @xmath156 , then for all @xmath145 satisfying a[condit ] @xmath157 .
analogous holds for the largest observed value and the second largest observed value .
we have seen in theorem [ theoid ] that the problem of identifying @xmath57 and @xmath2 from the observations @xmath100 reduces to identify the corresponding vectors @xmath158 for sufficiently many observations @xmath76 .
namely , one has to find the corresponding observations @xmath76 for @xmath1 linear independent rows of an invertible @xmath115 matrix @xmath123 with elements in @xmath116 .
remark [ remark1 ] points out that some observations @xmath76 can always be uniquely assigned to source vectors @xmath158 and thus , limit the possible maps @xmath145 in a[condit ] .
the next theorem shows that there is even more structure in the observations @xmath100 and that certain variations in the sources , i.e. , certain choices of @xmath123 in a[asstheoid ] , already ensure identifiability .
moreover , the proof of the following theorem gives an explicit construction of the unique @xmath64 from @xmath47 .
[ theoalgo ] assume model ( [ modeld ] ) with @xmath112 @xmath113 . furthermore , assume that @xmath136 and [ assusuff ] there exists @xmath109 such that @xmath159 with @xmath160 .
then @xmath63 is identifiable .
before we give a proof ( which is based on an explicit algorithm to compute @xmath57 from @xmath100 ) we will discuss relationships and differences between the previous results and their assumptions a[asstheoid ] - a[assusuff ] .
assumption a[assusuff ] of theorem [ theoalgo ] has a simple interpretation .
it means that each of the mixing weights @xmath161 appears somewhere in the mixture on its own ( without the influence of any other mixing weight @xmath162 ) via the mixture value @xmath163 .
for instance , if the alphabet is of the form @xmath164 a[assusuff ] simplifies to the condition that the mixing weights appear somewhere in the mixture , i.e. , @xmath165 for all @xmath166 .
intuitively , a[assusuff ] means that for each @xmath166 there exists one mixture observation @xmath76 such that only @xmath167 is active ( taking the value @xmath155 ) and all other sources @xmath168 with @xmath169 are silent ( taking the value @xmath155 ) .
it is easy to check that the choice of alphabet values @xmath149 and @xmath155 ( the smallest and second smallest alphabet value ) in theorem [ theoalgo ] can be replaced by @xmath170 and @xmath171 ( the largest and second largest alphabet value ) .
obviously , assumption a[assusuff ] arises from assumption a[asstheoid ] for a specific choice of @xmath123 in theorem [ theoid ] , namely with @xmath172 consequently , if the matrix in ( [ famatrix ] ) is invertible , a[assusuff ] implies a[asstheoid ] .
the following lemma shows that this holds under mild conditions on @xmath116 and @xmath1 .
[ aa ] for model ( [ modeld ] ) let a[assusuff ] be as in theorem [ theoalgo ] and a[asstheoid ] as in theorem [ theoid ] .
+ if @xmath173 , then a[assusuff ] implies a[asstheoid ] . for a proof see appendix [ app : b ]
figure [ arelation ] summarizes all relations between a[asstheoid ] - a[assusuff ] in a diagram .
\(m ) [ matrix of math nodes , row sep=1em , column sep=1em , minimum width=1em ] & & & & & a3 + & & & & & + & & & & & + & & & & & + a1 & & & ( a2 & & ) + ; ( m-1 - 6 ) edge [ double ] node[sloped , midway , above ] ( m-5 - 1 ) ( m-1 - 6 ) edge [ double ] node[sloped , midway , above ] ( m-5 - 6 ) ( m-5 - 1 ) edge [ double ] node[sloped , midway , below ] ( m-5 - 4 ) ; now we turn to the proof of theorem [ theoalgo ] , which is proven by explicit recovery of @xmath64 .
this generalizes an algorithm of diamantaras and chassiot @xcite for the binary alphabet @xmath174 to a general finite alphabet .
assumption a[assusuff ] implies that @xmath175 and hence , @xmath176 thus , it suffices to determine the map @xmath137 and the values @xmath177 , respectively , in order to determine @xmath57 .
when @xmath57 is determined , identifiability of @xmath2 follows from @xmath136 .
recall remark [ remark1 ] and note that @xmath178 which determines @xmath95 as in ( [ remarkomega ] ) .
the following two lemmas show that successively all the other @xmath179 ( and hence @xmath180 ) for @xmath181 can be determined as well , which finishes the proof .
let @xmath182 be the @xmath183-matrix , where the @xmath184th column of @xmath185 is the number @xmath186 written in the positional notation based on the number @xmath187 , identifying @xmath188 with @xmath149 , @xmath189 with @xmath155 , and so on , i.e. , @xmath190 and @xmath191 the @xmath132 dimensional vector of all possible values that @xmath47 can take .
[ lemmad ] from @xmath192 one can determine @xmath193 uniquely .
[ lemmacr ] it holds that @xmath194 . for proofs
see appendix [ app : bb ] and [ app : c ] , respectively .
@xmath100 @xmath195 @xmath196 @xmath197 @xmath198 determine @xmath199 with lemma [ lemmad ] ( using @xmath200 ) .
@xmath201 @xmath202 @xmath203 @xmath204 determine @xmath193 with lemma [ lemmad ] ( using @xmath192 ) .
@xmath205 @xmath206 + @xmath1 and @xmath207 the proof of theorem [ theoalgo ] gives an explicit recovery construction for @xmath64 which is summarized in figure [ fig : algodetmu ] . for noisy data one may use algorithm [ fig : algodetmu ] to proceed similar as diamantaras and chassiot @xcite who suggest a clustering approach for estimating @xmath57 and @xmath2 from noisy observations of @xmath47 .
however , as the purpose of this paper is not to propose a practical method for recovery from noisy data , but rather to analyze the scope of scenarios under which this is possible in principle , we are not going to follow this approach here further .
in this subsection we will shortly discuss how likely it is for the identifiability condition of theorem [ theoalgo ] to be satisfied when @xmath208 is a stochastic process .
therefore , let @xmath209 be as in theorem [ theoalgo ] , and define the hitting times @xmath210 for @xmath211 , and the stopping time @xmath212 then it follows from theorem [ theoalgo ] that @xmath213 note that this bound only depends on the distributions of the hitting times @xmath214 , which are often explicitly known or good estimates exist
. a prominent class of examples for modeling the distribution of the source signals are markov processes including iid sequences ( see e.g. , @xcite ) .
[ theoexp ] assume that the source signals @xmath208 in ( [ modeld ] ) constitute an irreducible markov process on the finite state space @xmath81 , with transition matrix @xmath215 , where we identify the first @xmath1 states of @xmath81 with @xmath216 from theorem [ theoalgo ] .
let @xmath217 be such that @xmath218 and @xmath219 then @xmath220 and , if @xmath136 , @xmath221 for a proof see appendix [ app : e ] .
bernoulli model[exb ] let us consider ( [ modeld ] ) for the simple case where we have two sources @xmath84 and @xmath222 that can take two different values , i.e. , @xmath223 .
for instance , the source signals could come from a binary antipodal alphabet ( @xmath224 ) as they appear in many digital modulated schemes .
if we assume that @xmath154 and @xmath225 are independent and identically distributed ( i.i.d . ) for all @xmath226 with @xmath227 and @xmath228 for @xmath74 and @xmath75 , then @xmath229 constitutes an irreducible markov process on the state space @xmath230 with transition matrix @xmath231 hence , @xmath232 , @xmath233 and @xmath234 .
thus , theorem [ theoexp ] yields @xmath235 in this simple setting we can even calculate the probability of identifiabiliy exactly .
note that @xmath63 is identifiable if and only if @xmath236 or @xmath237 is observed as @xmath238 and @xmath239 therefore , @xmath240 example [ exb ] shows that the bound in theorem [ theoexp ] does not need to be sharp in general but captures the exponential decay ( in @xmath19 ) well .
this is mainly because in theorem [ theoexp ] the probability of @xmath64 being identifiable is bounded using the sufficient ( and not necessary ) identifiabiliy condition a[assusuff ] from theorem [ theoalgo ] . in section [ sec : sim ] the gap between this bound and the true probability @xmath241 in ( [ piexpb ] ) is further explored in a simulation study .
after analyzing the most difficult scenario of a single linear mixture with @xmath5 in ( [ model ] ) , generalizations to arbitrary number of mixtures @xmath0 now follow easily . to this end , for a vector @xmath242 let @xmath243 denote the @xmath244-norm and define the set of @xmath0-mixtures as @xmath245 again , note that @xmath246 for @xmath43 is necessary to ensure identifiability of different sources . then for @xmath247 and @xmath248 the observed values are given by @xmath3 , i.e. , @xmath249 identifiability means to decompose the matrix @xmath250 uniquely into matrices @xmath251 and @xmath35 for given finite alphabet @xmath252 with @xmath253 and given @xmath254 .
analog to before we define the alphabet separation boundary of a mixture matrix @xmath248 as @xmath255 clearly , @xmath256 is a necessary condition on @xmath4 for @xmath257 to be identifiable .
theorem [ theoid ] , theorem [ theoalgo ] , and theorem [ theoexp ] assume that @xmath31 .
it is straight forward to check that theorem [ theoid ] holds unchanged when @xmath36 with @xmath258 replaced by @xmath259 .
the same is true for theorem [ theoalgo ] , where in the proof the minimum of a set of observations @xmath260 must be replaces by the minimum defined in terms of the ordering of @xmath261 .
thus , clearly theorem [ theoexp ] also holds unchanged when @xmath36 .
so far , we assumed the mixing weights to be positive and to sum up to one .
however , in some applications this assumption is not satisfied ( e.g. , in digital communications @xcite ) and in the following we discuss such generalizations .
let @xmath262 be an arbitrary subset of mixing weights @xmath263 .
note that w.l.o.g .
@xmath264 in order to assign the mixing weight to a source .
it is easy to check that theorem [ theoid ] holds unchanged with @xmath258 replaced by @xmath265 .
note , however , that if @xmath266 condition a[condit ] becomes more restrictive , i.e. , a mixture @xmath64 which is identifiable with respect to @xmath258 might not be identifiable with respect to @xmath265 .
analogously , theorem [ theoalgo ] can be generalized for @xmath266 , where now the corresponding identifiabiliy assumption a[assusuff ] becomes more restrictive .
the following theorem considers the most general case of arbitrary mixing weights in @xmath267 .
[ theoalgoabw ] assume model ( [ modeld ] ) @xmath55 with @xmath268 .
furthermore , assume that @xmath136 and there exists @xmath269 such that @xmath270 with @xmath271 for @xmath272 defined as @xmath273 for @xmath274 and @xmath275 .
then @xmath63 is identifiable .
the proof of theorem [ theoalgoabw ] is given in appendix [ sec : appthealgoabw ] .
recall that for positive mixing weights the identifiability condition a[assusuff ] in theorem [ theoalgo ] had a very simple interpretation , namely that each of the sinlge mixing weights @xmath161 appears somewhere in the mixture @xmath47 on its own , without the influence of any of the other mixing weights @xmath276 for @xmath169 .
the interpretation of ( [ assusuffabw ] ) is somewhat more difficult , but similar . in the case of probability mixing weights @xmath98 as in theorem [ theoalgo ] both , the sum and the absolute sum of the mixing weights were fixed via @xmath277 and this determined the scaling in which the mixing weights appear in the mixture @xmath47 .
now for general mixing weights @xmath278 as in theorem [ theoalgoabw ] both , the sum and the absolute sum ( or equivalently the sum of the negative mixing weights and the sum of the positive mixing weights ) are unknown and thus , additional conditions to determined these unknown scaling parameters are needed .
these correspond to @xmath279 and @xmath280 .
they ensure that the smallest possible mixture value ( which corresponds to @xmath279 ) and the largest possible mixture value ( which corresponds to @xmath280 ) are observed and thus determine the scaling parameters .
now analog to @xmath281 in a[assusuff ] of theorem [ theoalgo ] , @xmath281 and @xmath282 in ( [ assusuffabw ] ) of theorem [ theoalgoabw ] ensure that @xmath161 appears somewhere in the mixture @xmath47 on its own and can thus be determined . however , as the sign of @xmath161 is now unknown , too , we get the additional unambiguity that a mixture value can be increased either by increasing a source which corresponds to a positive mixing weight or by decreasing a source which corresponds to a negative mixing weight . from theorem [ theoalgoabw ]
it follows directly that theorem [ theoexp ] holds with @xmath1 replaced by @xmath283 , when we allow for arbitrary mixing weights in @xmath267 .
so far , we assumed that the number of sources @xmath1 is fixed and known .
now we consider the case where @xmath1 is unknown , i.e. @xmath55 with @xmath284 while it is not clear how to generalize theorem [ theoid ] for @xmath63 as in ( [ ddmum ] ) , condition a[assusuff ] in theorem [ theoalgo ] is still sufficient for identifiability when @xmath1 is unknown . to see this , note that the proof of theorem [ theoalgo ] ( in particular lemma [ lemmad ] ) does not require the number of sources @xmath1 to be known , where @xmath1 is determined via @xmath285 and thus , we obtain the following theorem .
[ cor : idm ] assume model ( [ modeld ] ) with @xmath112 @xmath286 .
furthermore , assume that @xmath136 and a[assusuff ] holds .
then @xmath63 ( and thus @xmath1 ) is identifiable .
analogously , theorem [ theoexp ] and theorem [ theoalgoabw ] do not require @xmath1 to be known .
finally , we explore in a simulation study how far assumption a[assusuff ] from theorem [ theoalgo ] is from being necessary when the sources come from an irreducible markov process ; which corresponds to exploring the tightness of the bound in theorem [ theoexp ] . to this end ,
theorem [ theoid ] is fundamental as it enables us to explicitly examine identifiabilty of @xmath64 .
we consider an example with a binary alphabet @xmath287 and a mixture of @xmath288 sources .
the matrix @xmath123 in theorem [ theoid ] was chosen randomly over the set of invertible @xmath115 matrices with elements in @xmath116 .
simulation runs were always @xmath289 .
first , we assume the mixing weights @xmath290 .
note that @xmath291 for all @xmath292 , i.e. , @xmath136 . assuming the sources to be i.i.d . for all @xmath293 and @xmath75
, with @xmath294 we find that @xmath63 is already identifiable on average for @xmath295 observations .
for the sufficient identifiability condition from theorem [ theoalgo ] to hold we find an average value of @xmath296 observations .
figure [ fig : id_hist ] shows the corresponding histograms and cumulative distribution functions .
the results indicate that the number of observations needed for the sufficient identifiability condition a[assusuff ] from theorem [ theoalgo ] to be satisfied is not considerably higher then the actual number of observations until @xmath63 is identifiable and thus the bound in theorem [ theoexp ] is quite sharp in this example .
is identifiable ( left ) and until the sufficient identifiability condition from theorem [ theoalgo ] is fulfilled ( right ) , bottom row : corresponding empirical cumulative distribution function , with @xmath288 , @xmath287 , @xmath290 and @xmath167 i.i.d . on @xmath116 . ]
we consider a more general markov model for generating the sources , i.e. , we assume the sources to be independent markov processes on the state space @xmath287 with transition matrix @xmath297 in figure [ fig : id_markov ] we display the average numbers of observations until @xmath64 is identifiable and until the sufficient identifiability condition a[assusuff ] from theorem [ theoalgo ] is fulfilled , respectively , for each @xmath298 . note that @xmath299 corresponds to i.i.d .
observations , with @xmath300 . is identifiable ( left ) and until the sufficient identifiability condition from theorem [ theoalgo ] is fulfilled ( right ) in dependence of @xmath301 and @xmath302 from ( [ transm ] ) . ] from figure [ fig : id_markov ] we draw that identifiability is achieved faster when @xmath301 and @xmath302 are close to @xmath303 , which corresponds to the i.i.d .
bernoulli model from section [ subsec : simb ] .
this is explained by condition a[condit ] in theorem [ theoid ] , where a richer variation in the sources , i.e. , many different observations @xmath304 , reduces the set of possible valid mixing weights and thus favor identifiability .
the sufficient identifiability condition a[assusuff ] from theorem [ theoalgo ] , however , requires repeated occurrence of the smallest alphabet values @xmath149 .
consequently , small @xmath301 and large @xmath302 discriminate against those variations .
now , we consider multiple linear mixtures , i.e. , @xmath36 . therefore , for each run , we draw @xmath0 mixing weights , each of length @xmath288 , independently from the uniform distribution on @xmath258 ( implying @xmath136 ) . for the sources ,
we consider a bernoulli model as in section [ subsec : simb ] .
we find that @xmath63 is identifiable on average after @xmath305 for @xmath306 observations , revealing that identifiability ( condition a[condit ] in theorem [ theoid ] ) depends much more on the variability of the sources than on the specific mixing weights .
this is confirmed in figure [ fig : id_hist_mlm ] , which shows the corresponding histograms and cumulative distribution functions .
the histograms and cumulative distribution functions for @xmath5 and @xmath307 differ only slightly and for @xmath31 they look almost the same as in figure [ fig : id_hist ] , although in figure [ fig : id_hist ] @xmath57 is fixed , whereas in figure [ fig : id_hist_mlm ] it is random . for the sufficient identifiability condition from theorem [ theoalgo ] to hold we find an average value of @xmath308 observations .
note that this condition depends on the sources only and , thus , is the same for all @xmath0 .
is identifiable ( left ) and until the sufficient identifiability condition a[assusuff ] from theorem [ theoalgo ] is fulfilled ( right ) , bottom row : corresponding empirical cumulative distribution function , with @xmath288 , @xmath287 , @xmath57 uniformly distributed on @xmath258 and @xmath167 i.i.d . on @xmath116 . ]
is identifiable ( left ) and until the sufficient identifiability condition from theorem [ theoalgo ] is fulfilled ( right ) , bottom row : corresponding empirical cumulative distribution function , with @xmath288 , @xmath287 , @xmath57 uniformly distributed on @xmath309^{{\ensuremath { n } } } : { { \ensuremath { a}}}_1 < \ldots < { { \ensuremath { a}}}_{{\ensuremath { n}}}\}$ ] and @xmath167 i.i.d . on @xmath116 . ] finally , we consider arbitrary mixing weights in @xmath267 . therefore , for each run , we draw mixing weights ( @xmath5 , @xmath288 ) independently from the uniform distribution on @xmath309^{{\ensuremath { n } } } : { { \ensuremath { a}}}_1 < \ldots < { { \ensuremath { a}}}_{{\ensuremath { n}}}\}$ ] and for the sources we consider a bernoulli model as in section [ subsec : simb ] . we find that @xmath63 is identifiable on average after @xmath310 observations . confirming , that identifiability , i.e. , condition a[condit ] in theorem [ theoid ] , is achieved slower , when we allow for larger sets of possible mixing weights . for the sufficient identifiability condition from theorem [ theoalgoabw ] to hold we find an average value of @xmath311 observations .
figure [ fig : id_hist_aw ] shows the corresponding histograms and cumulative distribution functions . in summary ,
our simulations show that the number of observations needed for our simple sufficient identifiability condition a[assusuff ] to hold is relatively close to the actual number of observations until @xmath64 is identifiable and thus , serves as a good benchmark criterion for identifiability
. this can be used as a simple proxy for validating the applicability of any recovery procedure in practice .
in this paper we have established identifiability criteria for single linear mixtures of finite alphabet sources as well as its matrix analogue . we gave not only sufficient but also necessary criteria for identifiability .
our work reveals the identification problem as a combinatorial problem utilizing the one to one correspondence between the mixture values and the mixing weights .
we generalized the method of diamantaras and chassioti @xcite to an arbitrary finite alphabet in order to derive a simple sufficient identifiability criterion .
the proof uses the specific hierarchical structure of possible mixture values leading to successive identification of the weights .
thus , our results characterize and extend the range of settings under which recovery algorithms ( for statistical data ) are applicable .
notably , we showed that our identifiability conditions extend to unknown number of sources @xmath1 .
this lays the foundation to design algorithms to recover the number of active sources from a mixture and sketches a road map to pursue this in future research .
finally , we showed that the probability of identifiability converges exponentially fast to @xmath189 when the underlying sources come from a discrete markov process .
this provides a useful and simple tool to pre - determine the required number of observations in order to guarantee identifiability at a given probability .
the derived sufficient identifiability conditions were briefly investigated in a simulation study and the required sample size for their validity was found to be quite close to the minimal sample size for identifiability . this work is intended to give a solid theoretical background for a model that is used in a variety of applications in digital communications , but also in bioinformatics .
+ for @xmath312 we define @xmath313 .
`` @xmath314 '' + by assumption a[asstheoid ] @xmath315 , i.e. , @xmath316 and , consequently , @xmath317 which , by assumption a[condit ] , is not fulfilled for any other @xmath119 .
thus , @xmath57 is uniquely determined . moreover , as @xmath136 , @xmath2 is uniquely determined as well .
+ `` @xmath318 '' + assume a[condit ] does not hold , i.e. , there exists @xmath319 such that @xmath320 fulfills @xmath321 as we assume all observations to be pairwise different , @xmath322 and @xmath122 with the corresponding @xmath323 lead to the same observations @xmath100 . therefore , @xmath63 is not identifiable .
obviously , a[assusuff ] arises from a[asstheoid ] when we choose the matrix @xmath123 in theorem [ theoid ] as in ( [ famatrix ] ) .
hence , a[assusuff ] implies a[asstheoid ] if the matrix in ( [ famatrix ] ) is invertible .
( [ famatrix ] ) can be written as @xmath324 and consequently , the matrix in ( [ famatrix ] ) has zero determinant if and only if @xmath325 is an eigenvalue of @xmath326 i.e. , @xmath327 or @xmath328 . as @xmath329 the assertion follows .
for @xmath330 the assertion is obvious .
so let @xmath331 .
then for @xmath332 we have that @xmath333 where @xmath334 are the entries of the matrix @xmath185 in ( [ bmatrix ] ) . by definition @xmath335 .
therefore , @xmath336 and for @xmath337 @xmath338 if @xmath339 , then obviously @xmath340 .
if @xmath341 , then by definition of @xmath185 there exists an @xmath342 such that @xmath343 , and therefore , @xmath344 consequently , @xmath345 .
let @xmath214 be as in ( [ tr ] ) and let @xmath346 be the initial distribution of @xmath347 .
define the stopped process @xmath348 for @xmath349 , which is a markov process as well ( see e.g. , ( * ? ? ?
* proposition 4.11.1 . ) ) .
it is obvious that for the markov process @xmath350 the state @xmath209 is absorbing and all other states are transient .
moreover , when we reorder the states in @xmath81 such that @xmath209 is the first state , the transition matrix of @xmath350 is given by @xmath351 the distribution of @xmath214 is a discrete phase type distribution ( see e.g. , ( * ? ? ?
* section 2.2 . ) ) , i.e. , @xmath352 as @xmath218 @xmath353 with @xmath354 for @xmath355 .
consequently , all row sums of @xmath356 are smaller than @xmath189 , i.e. , @xmath357 and hence @xmath358 .
next , we show by induction that @xmath359 for all @xmath360 . for @xmath361
this holds by definition .
so assume that @xmath362 for all @xmath363 and define @xmath364 , i.e. , @xmath365 if @xmath366 , then @xmath367 as @xmath368 and @xmath369 . if @xmath370 , then @xmath371 and @xmath372 , with @xmath373 .
therefore , @xmath374 with ( [ tq ] ) and ( [ poi ] ) it follows that @xmath375 and as @xmath376 the assertion follows .
assume that @xmath377 .
otherwise , we can multiply all observations by @xmath378 , such that the new alphabet becomes @xmath379 , which then fulfills @xmath377 . further , note that @xmath136 implies that @xmath380 for all @xmath381 .
let @xmath382 be the set of the pairwise different observations .
( [ assusuffabw ] ) implies that there exist @xmath383 such that @xmath384 and @xmath385 for @xmath386 .
first , note that @xmath387 and thus @xmath388 if @xmath389 , all weights are positive and , as @xmath390 is identified and thus w.l.o.g equal to one , theorem [ theoalgo ] applies .
thus , assume that @xmath391 and define @xmath392 and @xmath393 i.e. , @xmath394 .
second , note that analog to ( [ remarkc1 ] ) @xmath395 and thus @xmath396 hence , if @xmath397 we find that @xmath398 and if @xmath399 that @xmath400 thus , we have identified the first weight , namely @xmath401 now assume that we have identified @xmath402 different weights , @xmath403 .
if @xmath404 , all the remaining weights are positive and theorem [ theoalgo ] applies .
thus assume that @xmath405 and define @xmath406 , with @xmath407 and @xmath408 note that analog to lemma [ lemmacr ] @xmath409 and thus @xmath410 hence , if @xmath411 we find that @xmath412 and if @xmath413 that @xmath414 thus , we have identified the @xmath415-th weight as @xmath416 by induction , we can identify all weights and thus , by @xmath417 the assertion follows .
the authors acknowledge support of dfg crc 803 , 755 , rtg 2088 , and for 916 .
helpful comments of an editor , two referees , c. holmes , p. rigollet , and h. sieling are gratefully acknowledged .
s. l. carter , k. cibulskis , e. helman , a. mckenna , h. shen , t. zack , p. w. laird , r. c. onofrio , w. winckler , b. a. weir _
et al . _ , `` absolute quantification of somatic dna alterations in human cancer , '' _ nature biotechnology _ , vol . 30 , no . 5 , pp . 413421 , 2012 .
b. liu , c. d. morrison , c. s. johnson , d. l. trump , m. qin , j. c. conroy , j. wang , and s. liu , `` computational methods for detecting copy number variations in cancer genome using next generation sequencing : principles and challenges , '' _ oncotarget _ , vol . 4 , no .
11 , p. 1868 , 2013 . a. assa - el - bey , k. abed - meraim , and y. grenier , `` underdetermined blind audio source separation using modal decomposition , '' _ eurasip journal on audio , speech , and music processing _ , vol .
2007 , no . 1 ,
pp . 1414 , 2007 .
m. castella , `` inversion of polynomial systems and separation of nonlinear mixtures of finite - alphabet sources , '' vol .
56 , no . 8 , pp . 39053917 .
[ online ] .
available : http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=4567638 g. ha , a. roth , j. khattra , j. ho , d. yap , l. m. prentice , n. melnyk , a. mcpherson , a. bashashati , e. laks _ et al .
_ , `` titan : inference of copy number architectures in clonal cell populations from tumor whole - genome sequence data , '' _ genome research _ , vol .
24 , no . 11 , pp . 18811893 , 2014 .
c. yau , o. papaspiliopoulos , g. o. roberts , and c. holmes , `` bayesian non - parametric hidden markov models with applications in genomics , '' _ journal of the royal statistical society : series b ( statistical methodology ) _ , vol .
73 , no . 1 , pp . 3757 , 2011 .
s. talwar , m. viberg , and a. paulraj , `` blind separation of synchronous co - channel digital signals using an antenna array .
_ i. algorithms _
, '' _ ieee transactions on signal processing _ , vol .
44 , no . 5 , pp . 11841197 , 1996 .
k. diamantaras and e. chassioti , `` blind separation of n binary sources from one observation : a deterministic approach , '' _ international workshop on independent component analysis and blind signal separation _ , pp . 9398 , 2000 .
f. gu , h. zhang , n. li , and w. lu , `` blind separation of multiple sequences from a single linear mixture using finite alphabet , '' _ international conference on wireless communications and signal processing _ , pp . 15 , 2010 .
m. rostami , m. babaie - zadeh , s. samadi , and c. jutten , `` blind source separation of discrete finite alphabet sources using a single mixture , '' _ ieee statistical signal processing workshop ( ssp ) _ , pp . 709712 , 2011 .
y. li , s .-
amari , a. cichocki , d. w. ho , and s. xie , `` underdetermined blind source separation based on sparse representation , '' _ ieee transactions on signal processing _
54 , no . 2 ,
423437 , 2006 .
lee , m. s. lewicki , m. girolami , and t. j. sejnowski , `` blind source separation of more sources than mixtures using overcomplete representations , '' _ ieee signal processing letters _ , vol . 6 , no . 4 , pp .
8790 , 1999 .
k. diamantaras and t. papadimitriou , `` blind deconvolution of multi - input single - output systems using the distribution of point distances , '' _ journal of signal processing systems _ , vol .
65 , no . 3 , pp .
525534 , 2011 .
s. arora , r. ge , r. kannan , and a. moitra , `` computing a nonnegative matrix factorization - provably , '' in _ proceedings of the forty - fourth annual acm symposium on theory of computing _ , 2012 , pp .
145162 .
a. aissa - el - bey , d. pastor , s. m. a. sbai , and y. fadlallah , `` sparsity - based recovery of finite alphabet solutions to underdetermined linear systems , '' _ information theory , ieee transactions on _ , vol .
61 , no . 4 ,
pp . 20082018 , 2015 . | we give under weak assumptions a complete combinatorial characterization of identifiability for linear mixtures of finite alphabet sources , with unknown mixing weights and unknown source signals , but known alphabet .
this is based on a detailed treatment of the case of a single linear mixture .
notably , our identifiability analysis applies also to the case of unknown number of sources .
we provide sufficient and necessary conditions for identifiabilty and give a simple sufficient criterion together with an explicit construction to determine the weights and the source signals for deterministic data by taking advantage of the hierarchical structure within the possible mixture values .
we show that the probability of identifiability is related to the distribution of a hitting time and converges exponentially fast to one when the underlying sources come from a discrete markov process .
finally , we explore our theoretical results in a simulation study .
our work extends and clarifies the scope of scenarios for which blind source separation becomes meaningful .
blind source separation ; bss ; finite alphabet signals ; single mixture ; instantaneous mixtures ; markov processes , stopping time |
since the first options on stocks appeared in an organized exchange ( the cboe ) in 1973 there has been a continuous growth in worldwide option markets .
this growth is not only apparent in the increasing volumes of options traded in organized and otc markets but also in the never ending production of new products providing alternatives to satisfy the necessities of investors , banks and other financial institutions @xcite .
these necessities may include risk transference , speculative leveraging , portfolio diversification , etc .
the purpose of this article is to present a new family of options specially designed to satisfy the necessities of speculators .
nowadays we have at our disposal a wide spectrum of products that can virtually serve at whatever market view that a speculator may have .
the main inconvenient with these ` standard ' products is that a change in the speculator s market view may require a change in his / her position forcing unwanted transactions costs .
from that perspective , it would be very convenient for aggressive intra - day speculators to have a product that could be adjusted to their changing expectations as many times as desired and without costs associated to these adjustments .
this is exactly the spirit of mirror options .
basically a mirror option is a european or american option where the holder has a well - defined payoff function ( for example @xmath0 if we are talking about a call ) and the right to make an undetermined number of changes in the real path of the underlying during the life of the option .
we will call these changes ` mirrorings ' and we will say that the holder has the right ( not the obligation ) to ` mirror ' the underlying an undetermined number of times . in the next section we will state exactly what we mean by mirroring but now it will be enough to think that the holder has the right to change the real path of the underlying by a kind of reflected path in the sense that , after the mirroring , downwards movements in the real path become upwards movements in the reflected path and vice - versa . the dates for these mirrorings are not fixed and can be chosen freely by the holder . when the option is exercised , the payoff is calculated _ not _ with the real underlying but with the virtual underlying resultant from the mirroring process performed by the holder up to this time . the freedom to perform mirrorings whenever during the contract allows the holder to adapt the option to his / her particular view of the market and , what is more important , to do it without suffering the associated costs that this policy would imply trading traditional products like futures , calls , puts , etc . taking this into account
we think that mirror options offer a new universe of possibilities that is not available using nowadays standard and otc products .
the organization of the article is as follows : in section [ defining ] we define the concept of mirroring and immediately after that we state what we understand by a mirror option in mathematical terms . in section [ valuating ]
we proceed to valuate mirror options within the simplest market hypothesis in order to obtain a general valuation formulae . there
we obtain pricing expressions for european mirror options in the case where the payoff is a monotonic function .
we also present as examples the cases of the mirror call , mirror put and mirror forward . in section [ hedging ]
we show explicitly a very important feature of the hedging strategy for mirror options and moved by the conclusions of this section we generalize , in section [ finitem ] , our valuation formulae to the case where there is a fixed number of allowed mirrorings .
finally in section [ conclusions ] we discuss some interesting characteristics of our results and point out possible alternatives of further work .
in order to give a definition of mirror options we have to clearly state the meaning of _ mirroring _ and _ mirror path_. by mirroring an underlying @xmath1 at a certain time @xmath2 we understand constructing a new path @xmath3 ( mirror path ) defined simply by @xmath4 we can see how a mirror path looks like in the example of fig.([mirrorfig1 ] ) . in fig.([mirrorfig2 ] ) we have plotted the same example in logarithmic scale where we can clearly see the reason for the name ` mirror ' path . in eq.([single ] ) we have given the definition of a mirror path when a single mirroring time is allowed .
if we allow for a second mirroring at a time @xmath5 with @xmath6 we can define a new path by just applying eq.([single ] ) over the former single mirrored path .
note that in the case @xmath7 we recover the original path without mirrorings . in general
if we allow for several consecutive mirrorings at times @xmath8 it is easy to see that @xmath9 where @xmath10 is the starting time of the option contract . note that after an even number of mirrorings @xmath11 equals the value of the underlying @xmath12 affected by a leveraging factor depending on the historical values taken by the underlying at the mirroring times selected by the holder .
after an odd number of mirrorings we have also a path dependent factor but in this case multiplying the inverse value of the underlying . in fig.([mirrorfig3 ] ) we can see the effects of two consecutive mirrorings over the sample path of fig.([mirrorfig1 ] ) .
= 8.0 cm = 8.0 cm = 8.0 cm once it is clear what we mean by mirrorings we can define a mirror option as following : [ def1]given a european or american option with payoff @xmath13 we define its mirror counterpart as the one with payoff @xmath14 given by @xmath15 the holder of the mirror option has the right ( not the obligation ) at each time between @xmath10 ( start of the contract ) and @xmath16 ( maturity of the contract ) to perform mirrorings .
the value @xmath11 of the mirror underlying at time @xmath17 is given by eq.([several ] ) where @xmath18 is the whole set of times where the holder has chosen to perform mirrorings .
besides the right to perform mirrorings the holder has all the rights associated to the standard option counterpart ( e.g. early exercise in the american case ) .
please note that we understand the holder is not allowed to mirror the past history .
only future and therefore non - predictable history is affected by the holder s mirrorings .
note also that the definition is sufficiently general to accommodate mirror versions of plain - vanilla and exotic options including the path - dependent case .
however , we do not intend here to give a static definition .
we want instead to make clear which is the spirit of the definition and therefore allow the reader to adapt it to other possibilities .
just to give an example we can consider mirror exchanges or rainbows @xcite where we have a payoff that depends on several underlyings and where the holder is allowed to perform independent mirrorings on each of them .
from the definition it is now clear that depending on the holder ability to foresee the tendency of the underlying the leveraging factor resulting from judiciously chosen mirrorings can generate a much more convenient payoff than the one obtained with the standard option counterpart .
for example a holder of a mirror call will try to perform mirrorings in order to maximize the rise of @xmath3 and vice - versa for a mirror put . in the case of options with more general payoffs
the holder ability must be directed towards the maximization of the payoff whatever it could be . in some sense mirror options are reminiscent of chooser options @xcite with the difference that here the holder ` chooses ' all the time during the contract .
for the valuation of mirror options we will assume the simplest market model we can start with .
namely a one factor model consisting of a single underlying @xmath1 , paying the continuous dividend rate @xmath19 , together with a riskless bonus @xmath20 the pair of assets satisfy @xmath21 where @xmath22 is a wiener process and where the drift @xmath23 , volatility @xmath24 and riskless interest rate @xmath25 are taken constant .
we also assume that there are no transactions costs
. therefore we will work with a complete and efficient model without transaction costs .
this model can be complicated considerably and can be adapted to the particular underlying or sets of underlyings of the corresponding option .
evidently correlations enter into the game when several underlyings are involved .
moreover , the constancy of the drift and volatility is no longer acceptable when the option is a compounded one and therefore the underlying is another option @xcite . even for a single equity underlying one can consider multifactor models with stochastic volatility and term structure @xcite , inefficient market models @xcite , jump - diffusion models @xcite , models with transaction costs @xcite etc .
however , in this article we want to present a general valuation formulae obtained with the simplest hypothesis leaving refinements and the treatment of particular cases to further developments . using eq.([several ] ) after @xmath26 mirrorings
we can apply it lemma @xcite to eq.([model ] ) to obtain @xmath27 with @xmath28 from this fact one can be tempted to make the following incorrect argument : the volatility of @xmath3 is unaffected by the mirroring operations made by the holder and by definition [ def1 ] the payoff of the mirror option is given by the same function that corresponds to the standard counterpart but computed using @xmath3 instead of @xmath1 .
moreover , since in the valuation using a replicating self - financing portfolio the drift of the underlying drops out from the calculations @xcite we can naively conclude that the value of the mirror option is the same as the value of its standard counterpart . as the reader has surely noted
the reason why the above argument is wrong is that @xmath3 _ is not the value of any tradable _
@xcite_. _ despite of that , we will see that the value of the mirror option is closely related to the value of its standard counterpart , at least for a family of payoffs where the valuation can be done in a closed form ( monotonic payoffs as we will see ) . before starting the valuation
we need some preliminary results . for @xmath29 and @xmath30
we define @xmath31 where @xmath13 is the payoff function . using the chapman - kolmogorov relation satisfied by gaussians
we immediately obtain @xmath32 for @xmath33 and @xmath34 we also define @xmath35 then performing a change of variables and using eq.([tower ] ) we obtain the relation @xmath36 now we state the following simple result [ lemma1]_the payoff _ @xmath13 _ is a monotonic increasing ( decreasing ) function if and only if _
@xmath37 _
_ @xmath38 _ is a monotonic increasing ( decreasing ) function . _
* proof . *
suppose that @xmath37 @xmath39 is a monotonic increasing ( decreasing ) function , then in particular we have @xmath40 and since the exponential is a monotonic increasing function we immediately obtain that the payoff is a monotonic increasing ( decreasing ) function .
conversely , if the payoff _ _
@xmath13 is a monotonic increasing ( decreasing ) function then @xmath41 is also a monotonic increasing ( decreasing ) function and finally since @xmath42 is the convolution of a positively defined function ( a gaussian in this case ) with @xmath41 we immediately obtain that @xmath38 is a monotonic increasing ( decreasing ) function @xmath37 .
@xmath43 to perform the valuation we take time as discrete points and at the end we will take the continuous limit .
we define the time to maturity @xmath44 as @xmath45 with @xmath46 and @xmath47 arbitrary and constant . since @xmath47 is arbitrary taking the limits @xmath48 , @xmath49 with @xmath50 fixed we obtain the value of the option for arbitrary time @xmath51 .
let us suppose that at time @xmath52 the holder has performed a certain number of mirrorings generating a virtual underlying value @xmath53 given by eq.([several ] ) .
that value can be very different from the real underlying value @xmath54 ( note also that the value @xmath53 is not affected by a mirroring at time @xmath55 ) . to begin with let us introduce some notation .
we define @xmath56 hence using eq.([several ] ) we obtain the relation @xmath57 where @xmath58 takes values @xmath59 and is equal to @xmath60 with @xmath61 the number of mirroring decisions made by the holder up to time @xmath17 * inclusive*. from eq.([several ] ) we can see that @xmath62 also depends on the history of mirrorings up to time @xmath17 inclusive . now we need to prove the following result [ lemma2 ] @xmath63 * proof . * from eq.([several ] ) , when passing from time @xmath17 to time @xmath64 , we have the following four possibilities : @xmath65 and @xmath66 where @xmath67 and @xmath68 are factors that depend on the values taken by the underlying at mirroring times * previous to * @xmath17 ( see eq.([several ] ) ) . hence taking logarithms in eqs.([fi1 ] ) and
( [ fi-1 ] ) we obtain @xmath69 and using eq.([fibeta ] ) we immediately obtain @xmath70 @xmath43 with this result and eq.([relat ] ) we are ready to start
. we will denote the value of the mirror option at time to maturity @xmath44 by @xmath71 .
it is obvious that @xmath72 also has a certain dependence on the history of the underlying up to time @xmath73 .
for example at maturity time we have @xmath74 where @xmath75 and @xmath76 are clearly history dependent . in general to avoid raveling more the notation we will assume such dependence implicitly . however , it is important to realize that because of eq.([atmat ] ) mirror options enter into the category of asian options with the peculiarity that the path dependence is dictated by the holder with his / her mirroring decisions .
hereafter we will analyze the particular case where the payoff function @xmath13 is monotonic . denoting by @xmath77 ( @xmath78 ) the value of the mirror option
whenever the payoff is monotonic increasing ( decreasing ) we define @xmath79 now will demonstrate by induction the main result of this work , that is : [ theorem1]defining @xmath80 for the european case we have @xmath81 * proof . * using eqs.([atmat ] ) and ( [ redef ] ) the result is trivially true for @xmath82 . supposing it valid for @xmath45 we will demonstrate it for @xmath83 and therefore for arbitrary @xmath84 by induction . according to the standard procedure @xcite for our market model ( [ model ] ) ,
the value of a european claim at @xmath83 is given by the @xmath85discounted mean value of the claim at @xmath45 using the _ risk neutral measure _
( @xmath86 ) .
let us for a moment remember here the case of an american option .
there the value of the option was given by the maximum between that discounted value and the value obtained by early exercise . here , like in the american case , we have to choose between two values , namely @xmath87 where @xmath88 means the maximum taken between the values of the expression between braces evaluated with and without mirroring at time to maturity @xmath44 .
note that the risk neutral measure used in eq.([fund ] ) is just the one arising from no - arbitrage arguments based on the standard @xmath89-hedging strategy for our model ( [ model ] ) .
using the inductive hypothesis given by eq.([result ] ) along with eq.([fibeta ] ) we have @xmath90 now using eqs.([deffb ] ) and ( [ relat ] ) we can write @xmath91 and from lemma [ lemma2 ] we have @xmath92 noting that @xmath93 is invariant under mirrorings at time @xmath17 we have @xmath94 finally , using the payoff monotonicity hypothesis and lemma [ lemma1 ] we obtain @xmath95 @xmath43 now we can trivially take the continuous limit @xmath48 , @xmath49 with @xmath45 fixed .
in this limit and using eq.([redef ] ) we obtain the value of the mirror option for general monotonic payoff functions as @xmath96 this is a remarkable result , remembering that the value @xmath97 for the standard european option is given by @xmath98 we immediately note that at the beginning of the contract ( @xmath99 ) if we have a monotonic increasing payoff ( like in a call for example ) we have that the mirror option has the same ( for @xmath100 ) or bigger ( for @xmath101 ) value than its standard counterpart .
analogously , if we have a monotonic decreasing payoff ( like in a put for example ) we have that the mirror option has the same ( for @xmath101 ) or bigger ( for @xmath102 ) value than its standard counterpart .
thus we have arrived to the conclusion that , at the beginning of the contract , the deviation of the value of the mirror option from that of its standard counterpart depends critically on the interplay between dividends , interest rates and volatility .
note also that here dividends appear in a non trivial way , that is the rule to incorporate continuous dividends given by @xmath103 is not valid for mirror options .
another point that should be clear is that even though we have derived our expressions for the monotonic payoff case , we can easily carry on the demonstration of theorem [ theorem1 ] in the limit case where @xmath104 _ for arbitrary payoff functions_. in this ` zero - measure ' case we obtain that eq.([finalresult ] ) remains valid and therefore at the beginning of the contract we have @xmath105 where we define the critical volatility @xmath106 as @xmath107 with this definition we can write eq.([finalresult ] ) in the more suggestive way @xmath108 where @xmath109 another point to remark is that now we can clearly see the asian characteristics of mirror options . from the payoff definition of mirror options
we already knew that they were asian options , but now it is clear also that we have a dependence on @xmath11 for _ any _ time @xmath17 during the contract .
moreover , from eq.([several ] ) , we can interpret @xmath11 as a kind of weighted geometric average in a set of ` mirrored ' values up to time @xmath110 this set is , by definition , dictated by the holder with his / her mirrorings decisions at each time up to @xmath17 . to illustrate these results we will consider some examples , but before that let us clarify some potentially confusing language .
it is standard in financial language to talk about a _
long _ option position with payoff @xmath13 and about a _
short _ option position with payoff @xmath111 . here
we will continue with this nomenclature but bearing in mind that unlike what happens with standard contracts a _ short _ mirror option _
_ position _ is not _ equivalent to a _ sold _ mirror option position . in the last case we are taking about the position of the _ issuer _ of the long mirror option that is _
not _ equivalent to any position of a _ holder _ because it is only the later who takes the mirroring decisions .
because of this fact holding both a _ short _ and a @xmath112 mirror option position has in general some non - vanishing value .
for the monotonic payoff case this value is @xmath113 where the upper ( lower ) sign corresponds to the case where the payoff @xmath13 for the long position is monotonic increasing ( decreasing ) .
note that in this case the holder has the right to perform a certain set of mirrorings for his / her long position and a _ different _ set of mirrorings for his / her short position and therefore as time goes by he / she will have in general a mirror path @xmath114 for the long position and a another mirror path @xmath115 for the short position . at the beginning of the contract ( @xmath116 ) we have @xmath117 and therefore using lemma [ lemma1 ] we can clearly see that the value given by eq.([longshort ] ) at this time is always _
positive definite_. moreover the unique possibility for this value to be zero is the limit case where @xmath118 which is the case where all mirror options take the same value as their standard counterparts . a european long ( short ) call has the payoff @xmath119where the upper ( lower ) sign corresponds to the long ( short ) position .
modeled with eq.([model ] ) the long ( short ) call has the value @xmath120 given by the black - scholes formula @xcite @xmath121 with @xmath122 hence from eq.([finalresult ] ) and since in this case the payoff is a monotonic increasing ( decreasing ) function of @xmath1 we immediately obtain the value of the corresponding european mirror call as @xmath123 with @xmath124 note that necessarily one of the two mirror positions ( short or long ) has the same value as its standard counterpart at the beginning of the contract .
a european long ( short ) put has the payoff @xmath125 where the upper ( lower ) sign corresponds to the long ( short ) position .
modeled with eq.([model ] ) the long ( short ) put has the value @xmath120 given by the black - scholes formula @xcite @xmath126 with @xmath122 hence from eq.([finalresult ] ) and since in this case the payoff is a monotonic decreasing ( increasing ) function of @xmath1 we immediately obtain the value of the corresponding european mirror put as @xmath127 with @xmath128 note again that necessarily one of the two mirror positions ( short or long ) has the same value as its standard counterpart at the beginning of the contract a long ( short ) forward contract has the payoff @xmath129 where the upper ( lower ) sign corresponds to the long ( short ) position . using simple no - arbitrage arguments @xcite we obtain its value @xmath120 given by @xmath130 with @xmath122 hence from eq.([finalresult ] ) and since in this case the payoff is a monotonic increasing ( decreasing ) function of @xmath1 we immediately obtain the value of the corresponding mirror forward contract as @xmath131 with @xmath132 like what happens with the standard forward contract here we can also choose the strike @xmath133 ( the forward price ) in order to make the price of the forward contract zero .
note also that from eq.([forward ] ) the forward prices for the long and short positions are different .
however , as we have seen , one of them is necessarily equal to the standard value .
we can use this simple result to exemplify the positive definite value arising from holding both a long and a short position at the beginning of the contract . according to eq.([longshort ] ) we have @xmath134 where @xmath135 ( @xmath136 ) is the value of the mirror underlying for the long ( short ) position . at time @xmath10
( the beginning of the contract ) we have @xmath137 so for this time we have @xmath138 which is clearly positive definite as we have already shown in general
. from the holder s perspective the payoff at maturity will be @xmath139 with the break - even given by @xmath140
in this section our aim is to show a very important feature appearing in the hedging of mirror options that is highly relevant when transaction costs and illiquidity of the market are taken into account .
this in turn will provide us with another way to envisage mirror options as a kind of ` swaps ' of transaction costs between the holder and the issuer . to calculate the value of mirror options in section [ valuating ] we have used the neutral risk probability arising in a @xmath89-hedging strategy .
let us now calculate explicitly such @xmath89 from result ( [ finalresult ] ) .
we define @xmath141 hence using eqs.([finalresult ] ) and ( [ standard ] ) we obtain @xmath142 with @xmath143 and @xmath144 the most important point to note in eq.([delta ] ) is _ not _ the difference between @xmath145 and @xmath146 , that was already discussed in the previous sections , but the sign flip given by eq.([mirrorsign ] ) . in other words
we have obtained that , from the hedger perspective , a mirroring made by the holder implies a change in the hedging position in the underlying from a given @xmath147 to the opposite one .
this result is in fact very intuitive ; since mirror options were created as instruments allowing the holder costless changes in his / her position it is natural to have the cost of these changes transferred to the issuer ( zero in our model ) .
this is , again , very reasonable because issuers are in general large investment banks or financial institutions that are exposed to much lower transaction costs than individual speculators and therefore can absorb such changes in position much more easily . moreover , such financial institutions do not hedge single instruments individually because the impact of transactions costs is highly non - linear @xcite .
as it is well known , it is the _ complete portfolio _ of options and other products what is hedged as a whole .
hedging each product individually is not realistic because such policy do not profit from cancellations between different positions and reductions of transaction costs due to economies of scale .
therefore the real cost of hedging mirror options is minimized with respect to the total premium received when large quantities of mirror options are issued in the marker hedging them as a whole along with the rest of products in the portfolio . however , in situations of large illiquidity ( like those appearing in crashes or rallies ) collective behavior of mirror option holders can lead to large changes in delta that can not be easily canceled out . because of that is not unrealistic to conceive that in a real market of mirror options a risk premium depending in some way on the number of allowed mirrorings can appear . in order to admit such correction we need to know the value of mirror options when only a finite number of mirrorings is allowed .
this is the subject of the next section .
in this section we will obtain the value of a mirror option for the case where the holder is allowed to perform @xmath26 mirrorings at most .
again we will be able to obtain closed expressions for the monotonic payoff case or , at most , for arbitrary payoffs in the critical point @xmath148 . with the same notational conventions we used in section [ valuating ] we define the value of the mirror option with at most @xmath26 allowed mirrorings as @xmath149 in this case we have [ theorem2 ] defining @xmath80 for the european case , where the holder is allowed to perform @xmath26 mirrorings at most , we have @xmath150 where with @xmath151
we mean that at time @xmath73 the holder has at least one mirroring to perform and with @xmath152 we mean that at time @xmath73 the holder has exhausted all his / her allowed mirrorings .
* proof . * using eqs.([atmat ] ) and ( [ redef ] ) the result is trivially true for @xmath82 .
again , supposing it valid for @xmath50 we will demonstrate it for @xmath153 and therefore for arbitrary @xmath84 by induction . using the same arguments of theorem [ theorem1 ] we obtain @xmath154 where @xmath155 and @xmath156 using the inductive hypothesis given by eq.([result2 ] ) along with eqs.([deffb ] ) , ( [ relat ] ) , ( [ fibeta ] ) and lemma [ lemma2 ] we obtain @xmath157 where it is clear that when considering @xmath158 we have @xmath159 note that so far we have not used the payoff monotonicity hypothesis . for @xmath160 using the inductive hypothesis given by eq.([result2 ] ) along with eq.([fibeta ] ) we have @xmath161 where @xmath162 hence using eqs.([deffb ] ) , ( [ relat ] ) and lemma [ lemma2 ] we obtain @xmath163 now we have to analyze two cases .
if @xmath164 , remembering that @xmath165 and using lemma [ lemma1 ] we obtain @xmath166 and finally if @xmath167 we obtain @xmath168 @xmath43 note that as expected we have @xmath169 moreover , from eq.([result2 ] ) we can see ( at least for the monotonic payoff case ) that the value of the mirror option _ before the holder has exhausted his / her last mirroring _ do not depend on the allowed number of mirrorings @xmath170 immediately after the last allowed mirroring is executed the value of the mirror option always falls to a lower or equal value .
the discontinuity arises whenever @xmath171 where the upper ( lower ) sign corresponds to the monotonic increasing ( decreasing ) payoff case .
for example we will have a discontinuous fall immediately after the last mirroring in a long mirror call if @xmath26 is even ( odd ) and @xmath172 ( @xmath173 ) .
note also that the expression given by eq.([result2 ] ) in the trivial case @xmath174 reduces to the standard result ( [ standard ] ) .
so far the reader may find that our results are counterintuitive since at the beginning of the contract two mirror options differing only in the number of allowed mirrorings have exactly the same value ( if at least 1 mirroring is allowed in both ) .
however in our simplified model our results are correct and must be taken as a base for further developments incorporating the corresponding corrections ( see the conclusions for a brief discussion ) .
in this work we have created a new family of options specially crafted to satisfy the necessities of those speculators who are convinced of their abilities to call the market and do not want to assume the transaction costs of daily trading . moreover , we were able to obtain general formulae for the value of these options in the case where the payoff is a monotonic function ( which is the case in calls , puts , futures , spreads , asset or nothing options @xcite , etc . ) . moreover , for the general payoff case we have identified a critical volatility @xmath175 for which the value of _ all mirror options , _ at the beginning of the contract , equals that of their standard counterparts .
we conclude our work with two sets of general remarks .
the first set regards general aspects of the pricing of mirror options within the hypothesis of our model .
the second set considers the impact other models may have on our results . from simple financial arguments we know that holding a long call and a long put with the same strike is equivalent to holding a long straddle .
note however that holding a long mirror call and a long mirror put with the same strike is not equivalent to holding a long mirror straddle .
this is because in the last case the holder has the right to perform a set of mirrorings for the call and another different set of mirrorings for the put .
this is evidently not equivalent to the situation of holding a single mirror straddle .
this fact is rather general in the sense that care must be taken when considering the valuation of synthetic options made of mirror options .
since one long mirror call plus one long mirror put is not one long mirror straddle , which is the price of a long mirror straddle then ? or more generally : which is the price of a mirror option with non - monotonic payoff ?
the answer can be obtained following the steps of valuation of sections [ valuating ] and [ finitem ] with the difference that since the straddle has not monotonic payoff we can not use this property and because of that only a numerical method ( the binomial tree for example ) can be applied in principle . in general , the difficulty of valuating mirror options with not monotonic payoffs is equivalent to the difficulty of valuating american options where free boundary problems arise @xcite . here
the free boundary is given , for each time @xmath176 by the value @xmath11 for which mirroring is equivalent to no mirroring ; in a complete analogous way to the free boundary arising in the american option case given by the equivalence of early exercise or not . in this work
we have valuated european mirror options , but what about american mirror options ?
again , the valuation can be obtained using the steps of sections [ valuating ] and [ finitem ] but now at each time stage we have to take the maximum over the possibility of mirroring or not and the possibility of early exercise or not .
this , also , will in general force us to use numerical methods like for example a binomial tree approach .
an important fact according to our results is that no matter what the market conditions are , all mirror options with monotonic increasing payoffs or all mirror options with monotonic decreasing payoffs will have the same initial value as their standard counterparts .
this is a really surprising result since in this case probably nobody would buy the equal value standard options !
moreover when we restrict the holder to make a maximum number of mirrorings @xmath26 , we have obtained that at the beginning of the contract the value of any mirror option with monotonic payoff and @xmath177 does not depend on @xmath170 again this seems to be counterintuitive since we expect holders to prefer mirror options with more allowed mirrorings than with lesser ones .
even though these results are correct in our simplified market model we expect corrections to appear in more involved models
. in particular inefficient market models@xcite , where we have correlations between assets returns at different times , may lead to significant deviations from our results depending on the degree of inefficiency in the model .
this can be expected on the intuitive basis that partial knowledge of future returns can be used by the holder to improve his performance with adequate mirrorings . if inefficiencies can bring about corrections to our formulae the incompleteness of the market @xcite is not less important ( at least equally important as it is in the valuation of standard options ) .
incompleteness in the market can be generated by several reasons , for example it can appear in mixed jump - diffusion models @xcite , stochastic volatility and term structure models @xcite , etc .
in any case both from the point of view of future implementations of mirror options as real otc products or from the theoretical point of view it is also interesting to consider the modifications these models can produce in our valuation formulae .
finally we can not overview deviations appearing in our results in the presence of transaction costs @xcite .
in particular we have to remember that even though we have derived our formulae in the absence of such costs it is just in the real world _ with transaction costs _ where mirror options are attractive ! therefore we have to consider that absence just as a simplification to the valuation procedure , in the same way as with standard options . here
it is worth to note the following : when analyzing the hedging of mirror options in section [ hedging ] we have realized that each mirroring made by the holder implies a change of sign in the @xmath89 of the hedger .
this is in fact intuitive since mirror options can be thought as a way to transfer transactions costs from the holder to the issuer .
however , since the issuer deals in general with large portfolios where cancellations between different positions occur , we are in fact effectively transferring only a fraction of those costs . in general such fraction will be lower the bigger the portfolio of mirror and standard options the issuer has .
however as we warned in section [ hedging ] in situations of large illiquidity ( like those appearing in crashes or rallies ) collective behavior of holders of mirror option can lead to large changes in delta that can not be easily canceled out . because of that we have considered in section [ finitem ] the valuation of mirror options when only a finite number @xmath26 of mirrorings is allowed .
then we can take the value of a real mirror option to be the one of section [ finitem ] ( or the numerical result if the payoff is not monotonic ) plus a correction depending on @xmath26 and other factors incorporating the inefficiency , incompleteness and transactions costs of the market .
however , as it well known , when considering transaction costs it is _ the hedger s whole portfolio of options _
( mirror and standard options ) what should be valued @xcite . what it is not well known , to our knowledge , is how to ascribe a fraction of such value to each component of the portfolio .
this is necessary if we want to obtain a realistic correction to our results due to transaction costs . considering the above remarks we can fairly say that a lot of work can still be done in order to have practical implementations of mirror options
however , because of their attractive characteristics we believe that innovative financial institutions can make real otc mirror options available in the near future .
j.m . would like to thank j. masdemont , a. gisbert , j. masoliver and j. perell for their useful comments .
the author also acknowledges a fellowship from generalitat de catalunya , grant 1998fi-00614 .
aurell e. , baviera r. , hammarlid o. , serva m. , vulpiani a. , _ a general methodology to price and hedge derivatives in incomplete markets _ , international journal of theoretical and applied finance * 3 * , 1 - 24 ( 2000 ) dewynne j. n. , whalley a. e. , and wilmott , _ path - dependent options and transaction costs , _ mathematical models in finance , edited by s. d. howison , f. p , kelly and p. wilmott , chapman & hall , london ( 1995 ) | in this work we present a new family of options ( mirror options ) specially crafted to satisfy the necessities of aggressive speculators .
the main ideas behind mirror options are : 1 ) a product that can be adjusted by the holder to agree with his / her market view at any time during its life .
2 ) the holder s right to make an arbitrary number of those adjustments without penalizing costs . after defining mirror options as ` super - versions ' of standard options
we derive general formulae for their value in the case where the payoff is a monotonic function of the underlying ( which is the case in calls , puts , futures , spreads etc . ) .
we briefly discuss also their valuation for general payoffs and the american case .
finally we analyze the situation where the number of allowed adjustments is restricted and we point out directions for further developments . 155 mm -15pt |
we are grateful to v. kartvelishvili for useful discussions .
this work was supported by the epsrc and pparc .
999 m. a. shifman , a. i. vainshtein and v. i. zakharov , nucl .
* b147 * ( 1979 ) 385 , 448 . l. j. reinders , h. rubinstein and s. yazaki , phys .
reports * 127 * ( 1985 ) 1 .
h. shiomi and t. hatsuda , nucl .
* a594 * ( 1995 ) 294 .
t. meissner , phys . rev .
* c52 * ( 1995 ) 3386 .
l. j. reinders , h. rubinstein and s. yazaki , nucl .
* b213 * ( 1983 ) 109 . b. l. ioffe , nucl .
* b188 * ( 1981 ) 317 ; * b191 * 591(e ) .
b. l. ioffe and a. v. smilga , nucl .
phys . * b232 * ( 1984 ) 109 . j. f. donoghue , e. golowich and b. r. holstein , _ dynamics of the standard model _ ( cambridge university press , cambridge , 1992 )
. s. h. lee , s. cho , t. d. cohen and d. k. griegel , phys .
* b348 * ( 1995 ) 263 . v. m. belaev and ya .
i. kogan , phys .
lett . * 136b * ( 1983 ) 273 . b. l. ioffe , university of bern preprint butp-94/25 , hep - ph/9501319 ( 1995 ) . b. l. ioffe , itep preprint 62 - 95 , hep - ph/9511401 ( 1995 ) .
y. koike , phys .
* d48 * ( 1993 ) 2313 .
d. b. leinweber , university of washington preprint doe / er/40427 - 17-n95 , nucl - th/9510051 .
v. a. novikov , m. a. shifman , a. i. vainshtein , m. b. voloshin and v. i. zakharov , nucl .
* b237 * ( 1984 ) 525 .
v. g. kartvelishvili and m. v. margvelashvili , z. phys .
* c55 * ( 1992 ) 83 ; v. kartvelishvili , phys .
* b287 * ( 1992 ) 159 .
e. braaten , s. narison and a.pich , nucl .
* b373 * ( 1992 ) 581 .
s. narison , phys .
* b361 * ( 1995 ) 121 .
d. v. bugg , a. a. carter and j. r. carter , phys .
lett . * 44b * ( 1973 ) 278 . v. stoks , r. timmermans and j. j. de swart , phys .
* c47 * ( 1993 ) 512 ; + r. a. arndt , r. l. workman and m. pavan , phys .
* c49 * ( 1994 ) 2729 ; + f. bradamante , a. bressan , m. lamanna and a. martin , phys .
* b343 * ( 1995 ) 431 ; + t. e. o. ericson _
et al . _ ,
* 75 * ( 1995 ) 1046 ; + d. v. bugg and r. machleidt , phys .
* c52 * ( 1995 ) 1203 .
. dependence on the square of the borel mass of the @xmath0 coupling constant determined from the ratio of sum rules ( [ ratio ] ) .
the values of the parameters used are given in the text .
the solid line shows the value of @xmath2 corrected for the mixed continuum term @xmath71 , the dashed line the uncorrected l.h.s . of ( [ ratio ] ) . | we evaluate the @xmath0 coupling constant using a qcd sum rule based on the pion - to - vacuum matrix element of the correlator of two interpolating nucleon fields .
the part of the correlator with dirac structure @xmath1 is used , keeping all terms up to dimension 5 in the ope and including continuum contributions on the phenomenological side .
the ratio of this sum rule to the nucleon sum rule involving condensates of odd dimension yields stable results with values of @xmath2 in the range @xmath3 .
the sources of uncertainty are discussed .
understanding hadron interactions from first principles is one of the main goals of quantum chromodynamics ( qcd ) .
since the solution of qcd for hadron interactions at low energies is still far off it is useful to consider tackle problems of hadron dynamics with approaches that lie as close as possible to qcd .
one of them , the method of qcd sum rules@xcite , has proved to be a very powerful tool to extract information about hadron properties .
we present here a sum rule analysis of the @xmath0 coupling constant @xmath4 , which is one of the most important parameters in hadron physics .
this quantity has previously been examined within the framework of qcd sum rules in refs .
@xcite .
reinders , rubinstein and yazaki @xcite worked with two different sum rules for @xmath2 , one obtained from a correlator of three interpolating fields , and one based on the pion - to - vacuum matrix element of the correlator of two interpolating nucleon fields , @xmath5 : @xmath6 however those authors included only the leading term of operator product expansion ( ope ) and neglected continuum contributions .
shiomi and hatsuda@xcite developed the sum rule based on the part of this two - point correlation function with dirac structure @xmath7 .
they calculated @xmath4 in the soft - pion limit , including condensates up to dimension 7 in the ope , radiative corrections and a perturbative estimate of continuum contributions .
the method based on the two - point correlator ( [ twopt ] ) has a significant advantage in that it can be used for low values of the momentum transfer to the nucleon .
in contrast the ope of the three - point correlator is valid only for large spacelike meson momenta and so a determination of the coupling constant requires an extrapolation to zero momentum ( see , for example@xcite ) .
this procedure is dangerous because of the higher - order terms that have been omitted from the ope .
the contributions of these terms give rise to corrections that are proportional to large inverse powers of the meson momentum @xmath8 , making the extrapolation of a truncated ope unreliable .
estimates of the coupling constant from the coefficient of @xmath9 determined at large @xmath10 , as in refs.@xcite , can not distinguish the meson pole term from the contributions of higher - mass states in the same meson channel .
in the soft - pion limit the ope for the @xmath7 part of the two - point correlator for @xmath2 has exactly the same form as that for the nucleon sum rule@xcite involving condensates of odd dimension , up a factor of @xmath11 @xcite . if continuum corrections are neglected , the ratio of these two sum rules has the form of the goldberger - treiman relation with @xmath12@xcite .
shiomi and hatsuda@xcite showed that this continues to hold for the higher - dimension terms in the ope , provided that the continuum thresholds are taken to be the same for both sum rules . using different thresholds in the two sum rules they were able to get around this problem with the implied value of @xmath13 .
however , using the usual soft - pion theorem@xcite , the correlator ( 1 ) can be expressed as @xmath14|0\rangle = { i\over 2f_\pi}\{\gamma_5\tau^a,\langle 0|t(\eta(x ) , \eta^\dagger(0))|0\rangle\}\ ] ] where @xmath15 is the axial charge and we have made use of the transformation properties of the interpolating field under axial rotations@xcite , @xmath16=-{1\over 2}\gamma_5\tau^a\eta$ ] .
the anticommutator with @xmath7 picks out the part of the two - point correlator proportional to the unit dirac matrix .
the phenomenological side of the resulting sum rule is thus @xmath17 times the corresponding expression for the odd - condensate nucleon sum rule .
this matches exactly with the structure found for the ope side in refs.@xcite .
the soft - pion limit for the @xmath7 piece of the correlator ( 1 ) thus yields a sum rule for @xmath18 .
the value for @xmath2 determined from such a rum rule follows from the odd - condensate sum rule for the nucleon mass and the goldberger - treiman relation ( or an approximation to it taking @xmath12 ) .
the sum rule can be thought of as just a chiral rotation of the odd - condensate nucleon sum rule and _ not _ an independent determination of @xmath2 . by considering terms beyond the soft - pion limit ,
we obtain here a value for @xmath2 that is not simply a consequence of chiral symmetry .
the sum rule we use is thus analogous to that for the @xmath19 coupling@xcite and the approach can be applied to calculations of other meson - baryon couplings .
moreover , a potentially important piece of the phenomenological side of the sum rule for @xmath2 has been omitted in both calculations .
this term corresponds to transitions of where a ground - state nucleon absorbs the pion and is excited into the continuum . since it is not suppressed by the borel transformation such a term should be included in a consistent sum - rule analysis , as pointed out long ago@xcite and stressed recently by ioffe@xcite . in the soft - pion limit ,
such terms generate contact interactions where the pion couples directly to the nucleon field , @xmath20 , and which are essential if the correct soft - pion limit is to be obtained .
although the need for these terms is particularly clear if pseudovector @xmath0 coupling is used ( see for example@xcite ) , they should also be included for pseudoscalar coupling .
their omission in refs.@xcite can explain why the correct goldberger - treiman relation was not found there .
indeed , as the authors of@xcite point out , a quick estimate of these unsuppressed @xmath21 contributions suggests that they could be as large as 25% : enough to remove the discrepancy with the goldberger - treiman relation .
here we start from the two - point correlator ( [ twopt ] ) just discussed , but instead of the piece with with dirac structure @xmath7 considered in refs.@xcite we work with the structure @xmath1 , where @xmath8 is the pion momentum .
we work here to leading order in a chiral expansion , neglecting higher - order terms in the pion momentum or current quark mass .
we consider the two - point correlation function @xmath22 where we use the ioffe interpolating field@xcite for the proton , @xmath23\gamma_{5 } \gamma^{\mu}d^{c}(x),\ ] ] where @xmath24 are the colour indices and @xmath25 is the charge conjugation matrix .
( the corresponding neutron field @xmath26 is obtained by interchanging @xmath27 and @xmath28 quark fields . )
more general choices of interpolating field are possible , as discussed in detail by leinweber@xcite . for the odd - condensate nucleon sum rule , which we make use of in our determination of @xmath2 , it turns out that the ioffe field is close to the optimal one as determined in ref.@xcite and so we do not consider more general fields here . in the deeply euclidean region ( @xmath29 large and negative ) the ope of the product of two interpolating fields takes the following general form @xmath30 where @xmath31 are the wilson coefficients and @xmath32 are local operators constructed out of quark and gluon fields ( all renormalised at some scale @xmath33 ) . using this ope in the correlator ( [ correl ] ) , we find that only operators of odd dimension contribute .
the leading term in this expansion involves operators with dimension 3 and has the form @xmath34 where terms that do not contribute to the dirac structure of interest , @xmath1 , have been suppressed .
the matrix element here is just the usual one for pion decay : @xmath35 where @xmath36 mev is the pion decay constant . at dimension 5
the only relevant contribution arises from the second - order term in the covariant expansion of the nonlocal operator @xmath37 .
it has the form @xmath38 up to corrections of higher order in the current mass , this can easily be re - expressed in terms of a mixed quark - gluon condensate @xmath39 some further manipulation allows one to rewrite this in the form @xmath40 where @xmath41 .
( we use the convention @xmath42 . )
the second term in this expression is of higher order in the chiral expansion ( see ref.@xcite for details ) and so we neglect it here . the first matrix element in ( [ cond5 ] )
was extracted by novikov _
et al._@xcite from two qcd sum rules for the pion .
they expressed it in the form @xmath43 and obtained @xmath44 gev@xmath45 .. however we have checked that our coefficient of @xmath46 in the expansion of @xmath47 ( which should be independent of convention ) agrees in both sign and magnitude with that of the corresponding term in the expansion of @xmath48 given in refs.@xcite .
] a crucial contribution in both of their sum rules is the four - quark condensate , @xmath49 .
novikov _
et al._@xcite used the factorisation approximation for this quantity but direct determinations of it from other sum rules suggest significantly larger values@xcite , at least 23 times bigger than those obtained from factorisation .
these give correspondingly larger values for @xmath50 , a point we shall come back to in the discussion of our results below . as an estimate of the importance of higher dimension condensates
, we have also calculated the contribution of what we hope is the most important dimension-7 operator in the ope .
this is a mixed quark - gluon condensate , which we evaluate in the factorised approximation .
keeping only this contribution explicitly , the dimension-7 piece of the correlator is @xmath51 where @xmath52 is the gluon condensate in vacuum . on the phenomenological side ,
the term of interest in the correlator ( [ correl ] ) is the one with a double pole at the nucleon mass , since this contains the @xmath0 coupling constant .
however the nucleon interpolating field does not just create ground - state nucleons ; there are also continuum contributions which can not be ignored .
the continuum - to - continuum pieces are modelled in the usual manner , in terms of the spectral density associated with the imaginary part of the ope expression for the correlator .
this continuum is assumed to start at some threshold @xmath53 .
after borel transformation , it can be taken over to the ope side of the sum rule where it modifies the coefficients of the terms involving @xmath54 .
in addition one must include nucleon - to - continuum terms since borel transformation does not suppress these with respect to the double - pole term@xcite .
to first order in @xmath8 , the correlator has the form @xmath55+\cdots , \label{phenom}\ ] ] where the continuum - continuum terms ( and terms with other dirac structures ) have not been written out . here
@xmath56 is the strength with which the interpolating field couples to the nucleon : @xmath57 note that the strength of the @xmath1 piece of the double - pole term is the same for both pseudoscalar and pseudovector @xmath0 coupling . equating the ope and phenomenological expressions for the correlator ( [ correl ] ) and borel transforming@xcite
, we get the sum rule @xmath58 where @xmath59 is the borel mass and @xmath60 with @xmath61 .
the second term on the r.h.s . of this sum rule , involving the undetermined constant @xmath62 , is the borel transform of the nucleon pole term of the nucleon - to - continuum piece in ( [ phenom ] ) .
it contains the same exponential as the nucleon double - pole term and so can not be ignored .
the second nucleon - to - continuum term in ( [ phenom ] ) leads to a term that is suppressed by an exponential involving the masses of states in the continuum .
it is thus typically a factor of 34 smaller than the term included in ( [ gpinsr ] ) . provided that the first of these mixed terms is a reasonably small correction to the sum rule
, it should be safe to neglect the second , as discussed by ioffe@xcite .
we now turn to the numerical analysis of this sum rule .
first , one should get rid of the unknown constant @xmath62 . multiplying the sum rule by @xmath63
, we see that the r.h.s .
becomes a linear function of @xmath64 . by acting on this form of the sum rule with @xmath65@xcite ( or equivalently by fitting a straight line to the l.h.s . and extrapolating to @xmath66@xcite )
we in principle can determine the value of @xmath2 .
however we are unable to find a region of borel mass in which the l.h.s .
is approximately a linear function of @xmath64 , and hence there is no region of stability for the extracted @xmath2 .
this lack of stability is similar to the situation for the nucleon sum rules , where two sum rules can be derived@xcite ( involving either odd or even dimension operators ) but neither shows good stability .
nonetheless the ratio of these leads to a more stable expression for the nucleon mass .
we have therefore taken the ratio of our sum rule ( [ gpinsr ] ) to the nucleon sum rules .
we obtain the most stable results from the ratio to the odd - dimension sum rule , @xmath67 and so we present here only the results for that case . taking such
a ratio also has the advantage of eliminating the experimentally undetermined strength @xmath56 from the sum rules .
note that we have allowed for a different continuum threshold @xmath68 in the nucleon sum rule and have defined @xmath69 .
the ratio of the sum rules ( [ gpinsr ] ) and ( [ noddsr ] ) can be written in the form @xmath70 and the method discussed above used to eliminate the unknown mixed nucleon - to - continuum term , @xmath71 ( @xmath72 ) . in fig .
1 we show results for @xmath2 as a function of @xmath64 , for typical values of the condensates and thresholds : @xmath73 , @xmath74 gev@xmath75 , @xmath76 gev@xmath45 , @xmath77 gev and @xmath78 gev .
stable values of @xmath79 are found over a region @xmath80 gev@xmath45 .
corrections due to the @xmath71 term are small , at most 5% . the second such term in ( [ phenom ] ) is expected to be smaller by a factor of 34 , and so we are justified in neglecting it . the threshold @xmath53 has been adjusted to give stability for @xmath64 around 1 gev@xmath45 , since one may hope that in this region the borel transformed sum rule is not too sensitive to the approximations that have been made on both the ope and phenomenological sides of the sum rule .
the existence of a window of stability provides a check on the consistency of this assumption .
we also demand that the thresholds @xmath68 and @xmath53 should lie significantly above this window so that the continuum is not too heavily weighted in the borel transform .
we find that the window of stability moves rapidly upwards as @xmath53 is increased for fixed @xmath68 .
for the typical parameter values above , only the region 2.05 gev@xmath81 gev@xmath45 satisfies these requirements . the value of @xmath2 varies by at most @xmath82 over this region . as a further check on our results
, we have examined whether the individual sum rules ( [ gpinsr ] ) and ( [ noddsr ] ) satisfy the criteria suggested by leinweber@xcite .
we find that the highest dimension condensates contribute less that 10% of the ope to both sum rules for @xmath83 gev@xmath45 .
the continuum forms about 40% of the phenomenological side of the differentiated version of the @xmath2 sum rule ( [ gpinsr ] ) for @xmath64 up to 1.4 gev@xmath45 , the point at which the continuum reaches 50% of the odd - condensate sum rule ( [ noddsr ] ) .
the region @xmath84 gev@xmath45 thus provides a window within which our results are both stable with respect to the borel mass and not too sensitive to our approximations .
we have examined the dependence of our results on the other input parameters .
variation of the threshold in the nucleon sum rule @xmath68 from 2.2 to 2.8 gev@xmath45 , readjusting @xmath53 to maintain stability , changes @xmath2 by @xmath82 . to estimate the sensitivity of our sum rule to the contributions of dimension-7 condensates and to uncertainties in the gluon condensate , we have varied the dimension-7 term in ( [ gpinsr ] ) between zero and twice its standard value .
our results for @xmath2 change by @xmath85 over this range .
one of the most important input parameters in our sum rule is the matrix element @xmath50 , defined by ( [ delta2 ] ) .
as already mentioned , this parameter was extracted by novikov _
et al._@xcite from an analysis of two sum rules for the pion .
their results depend crucially on the four - quark condensate , @xmath49 , for which they made the factorisation approximation and took a value of about @xmath86 gev@xmath87 . with this input ,
both of their sum rules yield consistent results for @xmath50 in the region @xmath88 gev@xmath45 .
however , sum rules analyses of @xmath89 decay and @xmath90 annihilation into hadrons lead to significantly larger values of the four - quark condensate ( see@xcite and references therein ) , in the range @xmath91 gev@xmath87 .
using these in the sum rules of ref.@xcite leads to values for @xmath50 ranging from 0.28 to 0.45 , although the two sum rules do not then give consistent results . as a conservative estimate of the uncertainty in @xmath50
we have considered the range 0.20 to 0.45 gev@xmath45 .
the corresponding variation in @xmath2 is @xmath92 , when the other parameters are held at their values above and @xmath53 is changed to keep the window of stability around 1 gev@xmath45 .
a second significant source of uncertainty is the quark condensate @xmath93 which appears in the odd - dimension sum rule for the nucleon .
standard " values for this lie in the range @xmath94 and @xmath95 .
the values of the nucleon mass determined from sum rules@xcite are strongly correlated with this condensate .
there is also a weaker correlation with the chosen value of the threshold @xmath68 .
since we are dividing by @xmath96 in the ratio ( [ ratio ] ) , our results are rather sensitive to the value of this condensate .
one would like to use values of @xmath93 and @xmath68 that give the nucleon mass correctly , but the ratio of the odd and even dimension nucleon sum rules does not yield completely stable results for @xmath96 .
the best we can do is to rule out values of @xmath97 below @xmath98 since they can not reproduce the nucleon mass within the region of borel mass and threshold that we consider . varying the quark condensate between @xmath99 and @xmath95
, we find that @xmath2 changes by @xmath100 .
our final result for the @xmath0 coupling constant is thus @xmath101 , where the uncertainty is dominated by @xmath50 and @xmath102 .
this is to be compared with values deduced from @xmath103 and @xmath0 scattering . for many years
the accepted value was @xmath104@xcite but this coupling has been the subject of some debate in recent years .
more recent analyses lead to values in the range 12.713.6@xcite .
our result is obviously consistent with any of these .
the rather large uncertainty in it could be reduced if the quark condensate could be determined more precisely .
in addition , the sum rules of novikov _ et al._@xcite should be re - examined using larger values of the four - quark condensate to try to pin down @xmath50 more exactly . in summary
, we have calculated the @xmath0 coupling constant using a qcd sum rule based on the pion - to - vacuum matrix element of a two - point correlator of interpolating nucleon fields .
this approach avoids the need for extrapolation from large spacelike meson momenta .
we have included nucleon - to - continuum terms omitted from previous analyses .
our sum rule is based on the part of the correlator with dirac structure @xmath1 and includes all terms up to dimension 5 in the ope .
stable results are obtained from the ratio of this sum rule to one for the nucleon mass and the unsuppressed nucleon - to - continuum contributions are found to be small .
contributions from higher - dimension operators and omitted continuum contributions are estimated to be small .
this demonstrates the practicability of this type of sum rule for calculation of other meson - baryon couplings , whose values are at present not well determined . |
what is the intrinsic connection between correlated random variables ? how much interaction is necessary to create correlation ?
many fruitful efforts have been made to quantify correlation between two random variables .
each quantity is justified by the operational questions that it answers .
covariance dictates the mean squared error in linear estimation .
shannon s mutual information is the descriptive savings from side information in lossless source coding and the additional growth rate of wealth due to side information in investing .
gcs and krner s common information @xcite is the number of common random bits that can be extracted from correlated random variables .
it is less than mutual information .
wyner s common information @xcite is the number of common random bits needed to generate correlated random variables and is greater than mutual information .
this work provides a fresh look at two of these quantities mutual information and wyner s common information ( herein simply `` common information '' ) .
both are extreme points of the channel simulation problem , introduced as follows : an observer ( _ encoder _ ) of an i.i.d .
source @xmath0 describes the sequence to a distant random number generator ( _ decoder _ ) that produces @xmath1 ( see figure [ figure channel simulation ] ) .
what is the minimum rate of description needed to achieve a joint distribution that is statistically indistinguishable ( as measured by total variation ) from the distribution induced by putting the source through a memoryless channel ?
[ ] [ ] [ 0.8]@xmath2 [ ] [ ] [ 0.8]@xmath3 [ ] [ ] [ 0.8]@xmath4 bits [ ] [ ] [ 0.8]@xmath5 [ ] [ ] [ 0.8]@xmath6 [ ] [ ] [ 1.0]channel simulator : @xmath7 and @xmath8 .
the first processor , @xmath9 , observes @xmath10 and the second processor , @xmath8 , generates @xmath11 after receiving a message at rate @xmath12 from @xmath9 .
the minimum rate needed is the common entropy of @xmath10 and @xmath11.,title="fig:",width=288 ] channel simulation is a form of random number generation . the variables @xmath2 come from an external source and @xmath6 are generated to be correlated with @xmath2 .
the channel simulation is successful if the total variation between the resulting distribution of @xmath13 and the i.i.d .
distribution that would result from passing @xmath2 through a memoryless channel is small .
this is a strong requirement .
it s stricter than the requirement that @xmath13 be jointly typical as in the coordinated action work of cover and permuter @xcite .
this total variation requirement means that any hypothesis test that a statistician comes up with to determine whether @xmath2 was passed through a real memoryless channel or the channel simulator will be virtually useless .
wyner s result implies that in order to generate @xmath2 and @xmath6 separately as an i.i.d .
source pair they must share bits at a rate of at least the common information @xmath14 of the joint distribution . in the channel simulation problem
these shared bits come in the form of the description of @xmath2 .
however , the `` reverse shannon theorem '' of bennett and shor @xcite suggests that a description rate of the mutual information @xmath15 of the joint distribution is all that is needed to successfully simulate a channel . how can we resolve this apparent contradiction
the work of bennett and shor assumes that common random bits , or _
common randomness _
, independent of the source @xmath2 are available to the encoder and decoder . in that setting ,
the common randomness provides a second connection between the source @xmath2 and output @xmath6 , in addition to the description of @xmath2 . remarkably , even though it is independent from the source @xmath2 , the common randomness assists in generating correlated random numbers and allows for description rates smaller than the common information @xmath14 . in this work ,
we characterize the tradeoff between the rate of available common randomness and the required description rate for simulating a discrete memoryless channel for a fixed input distribution , as in figure [ figure channel simulation with common randomness ] .
indeed , the tradeoff region of section [ section main result ] confirms the two extreme cases .
if the encoder and decoder are provided with enough common randomness , sending @xmath15 bits per symbol suffices . on the other hand , in the absence of common randomness one must spend @xmath14 bits per symbol .
[ ] [ ] [ 0.8]@xmath2 [ ] [ ] [ 0.8]@xmath3 [ ] [ ] [ 0.8]@xmath16 bits [ ] [ ] [ 0.8]@xmath5 [ ] [ ] [ 0.8]@xmath6 [ ] [ ] [ 1.0]channel simulator : @xmath7 [ ] [ ] [ 0.8]@xmath17 bits and @xmath8 .
the first processor , @xmath9 , observes @xmath10 and common randomness independent of @xmath10 at rate @xmath18 .
the second processor , @xmath8 , generates @xmath11 based on the common randomness and a message at rate @xmath19 from @xmath9.,title="fig:",width=288 ] this result has implications in cooperative game theory , reminiscent of the framework investigated in @xcite .
suppose a team shares the same payoff in a repeated game setting .
an opponent tries to anticipate and exploit patterns in the team s combined actions , but a secure line of communication is available to help them coordinate .
of course , each player could communicate his randomized actions to the other players , but this is an excessive use of communication .
a memoryless channel is a useful way to coordinate their random actions .
thus , common information is found in section [ section game theory ] to be the significant quantity in this situation .
we represent random variables as capital letters , @xmath10 , and their alphabets are written in script , @xmath20 .
sequences , @xmath21 are indicated with a superscript @xmath2 .
distribution functions , @xmath22 , are usually abbreviated as @xmath23 when there is no confusion .
accented variables , @xmath24 , indicate different variables for each accent , but their alphabets are all the same , @xmath20 .
similarly , distribution functions written with an accent or different letter , such as @xmath23 versus @xmath25 , represent different distributions .
markov chains , satisfying @xmath26 , are represented with dashes , @xmath27 .
( wyner s ) common information : @xmath28 conditional common information : @xmath29 total variation distance : @xmath30 a source @xmath2 is distributed i.i.d . according to @xmath31 . a description of the source at rate @xmath19 is represented by @xmath32 .
a random variable @xmath33 , uniformly distributed on @xmath34 and independent of @xmath2 , represents the common random bits at rate @xmath18 known at both the encoder and decoder .
the decoder generates a channel output @xmath6 based only on @xmath35 and @xmath33 .
the channel being simulated has a the conditional distribution @xmath7 , thus the _ desired joint distribution _ is @xmath36 .
a @xmath37 _ channel simulation code _ consists of a randomized encoding function , @xmath38 and a randomized decoding function , @xmath39 the description @xmath35 equals @xmath40 , and the channel output @xmath6 equals @xmath41 . since randomized functions are specified by conditional probability distributions , it is equivalent to say that a @xmath37 channel simulation code consists of a conditional probability mass function @xmath42 with the properties that @xmath43 , @xmath44 , and @xmath45 .
the _ induced joint distribution _ of a @xmath37 channel simulation code is the joint distribution on the quadruple @xmath46 . in other words , it is the probability mass function , @xmath47 where @xmath48 by construction .
a sequence of @xmath37 channel simulation codes for @xmath49 is said to _ achieve _
@xmath7 if the induced joint distributions have marginal distributions @xmath50 that satisfy @xmath51 a rate pair @xmath52 is said to be _ achievable _ if there exists a sequence of @xmath37 channel simulation codes that achieves @xmath7 .
the _ simulation rate region _ is the closure of achievable rate pairs @xmath53 .
[ theorem main result ] for an i.i.d .
source with distribution @xmath31 and a desired memoryless channel with conditional distribution @xmath7 , the simulation rate region is the set , @xmath54 where @xmath55
two extreme points of the simulation rate region @xmath56 fall directly from its definition . if @xmath57 , the second inequality in ( [ definition s ] ) dominates .
thus , the minimum rate @xmath19 is the common information @xmath14 .
this coincides with the intuition provided by wyner s result in @xcite . at the other extreme , using the data processing inequality on the first inequality of ( [ definition s ] ) yields @xmath58 no matter how much common randomness is available , and this is achieved when @xmath59 .
does nt necessary have to be as large as @xmath60 for @xmath61 to be in the simulation rate region .
] source coding results and the coordinated action work of cover and permuter in @xcite illustrate that with a description rate of @xmath15 we can create a codebook of output sequences in such a way that we ll likely be able to find a jointly typical output sequence for each input sequence from the source .
consequently , we can then randomize the codebook using common randomness to actually simulate the channel , as bennett and shor proved in @xcite . for a bernoulli - half source @xmath10 ,
let us demonstrate the simulation rate region for the binary erasure channel .
@xmath11 is an erasure with probability @xmath62 and is equal to @xmath10 otherwise .
the distributions in @xmath63 that produce the boundary of the simulation rate region are formed by cascading two binary erasure channels as shown in figure [ figure erasure test channel ] , where @xmath64 , \nonumber \\ p_1 & = & 1 - \frac{1-p_e}{1-p_2}. \nonumber\end{aligned}\ ] ] the mutual information terms in ( [ definition s ] ) become @xmath65 where @xmath66 is the binary entropy function .
figure [ figure bec 075 tradeoff ] shows the boundary of the simulation rate region for erasure probability @xmath67 .
the required description rate @xmath19 varies from @xmath68 bits to @xmath69 bits as the rate of common randomness runs between @xmath70 and @xmath71 bits .
[ ] [ ] [ 1.0]@xmath10 [ ] [ ] [ 1.0]@xmath72 [ ] [ ] [ 1.0]@xmath11 [ ] [ ] [ 0.8]@xmath70 [ ] [ ] [ 0.8]@xmath73 [ ] [ ] [ 0.8]@xmath74 [ ] [ ] [ 0.8]@xmath74 [ ] [ ] [ 0.8]@xmath75 [ ] [ ] [ 0.8]@xmath75 [ ] [ ] [ 0.8]@xmath70 [ ] [ ] [ 0.8]@xmath76 [ ] [ ] [ 0.8]@xmath73 that give the boundary of the simulation rate region for the binary erasure channel with a bernoulli - half input are formed by cascading two erasure channels.,title="fig:",width=240 ] [ ] [ ] [ 0.9]bec simulation rate region , @xmath67 [ ] [ ] [ 0.8]@xmath15 [ ] [ ] [ 0.8]@xmath14 and a bernoulli - half input , where @xmath19 is the description rate and @xmath18 is the rate of common randomness . without common randomness ,
a description rate of @xmath14 is required to simulate the channel . with unlimited common randomness , a description rate of @xmath15 suffices.,title="fig:",width=288 ]
let @xmath53 be an achievable rate pair .
then for each @xmath77 there exists a @xmath37 channel simulation code with an induced joint distribution @xmath78 such that @xmath79 let the random variable @xmath80 be uniformly distributed over the set @xmath81 .
the variable @xmath80 will serve as a random time index .
the joint distribution of the sequences @xmath13 is close in total variation to an i.i.d .
distribution , so we can extend lemma 2.7 of @xcite to obtain two bounds : @xmath82 where @xmath83 notice that @xmath84 .
define an epsilon rate region , @xmath85 where @xmath86 [ lemma epsilon rate region ]
@xmath87 we use familiar information theoretic inequalities , and the fact that @xmath2 and @xmath33 are independent , to bound @xmath19 and the sum rate @xmath88 .
@xmath89 we then lower bound the r.h.s . of ( [ equation ixij ] ) and ( [ equation ixyij ] ) using similar steps . here
we proceed from ( [ equation ixyij ] ) .
@xmath90 the second inequality comes from ( [ equation entropy bound ] ) , and the last inequality comes from ( [ equation information bound ] ) . the joint distribution of the pair @xmath91 can be shown to satisfy the total variation constraint in ( [ definition d epsilon ] ) . finally , we acknowledge the markovity of the triple @xmath92 to complete the proof of the lemma .
( the cardinality bound of @xmath72 in ( [ definition d epsilon ] ) is shown to be satisfiable via a generalized caratheodory theorem . )
the epsilon rate regions decrease to the simulation rate region as epsilon decreases to zero .
[ lemma lower semi continuity ] @xmath93
one key tool for the achievability proof is summarized in lemma [ lemma quantized input ] .
this lemma is implied by the resolvability work of han and verd in @xcite , but the concept was first introduced by wyner in theorem 6.3 of @xcite .
[ lemma quantized input ] for any discrete distribution @xmath94 and each @xmath95 , let @xmath96 be a `` codebook '' of sequences each independently drawn according to @xmath97 . for a fixed codebook ,
define the distribution @xmath98 then if @xmath99 , @xmath100 where the expectation is with respect to the randomly constructed codebooks @xmath101 .
assume that @xmath52 is in the interior of @xmath56 .
then there exists a distribution @xmath102 such that @xmath103 and @xmath104 . for each @xmath95 , let @xmath105 be uniformly distributed on @xmath106 .
we apply lemma [ lemma quantized input ] twice , once with @xmath107 and again with @xmath108 , to assert that there exists a sequence of `` codebooks '' @xmath109 , @xmath110 with the properties @xmath111 where @xmath112 and @xmath113 are marginal distributions derived from the joint distribution @xmath114 in an indirect way , we ve constructed a sequence of joint distributions @xmath115 from which we can derive channel simulation codes that achieve @xmath7 .
the markovity of @xmath116 implies the markov property @xmath117 .
let @xmath118 considering ( [ equation close joint distribution ] ) and ( [ equation close to independent ] ) with the properties of total variation and @xmath116 in mind , it can be shown that @xmath119 is a sequence of channel simulation codes that achieves @xmath7 .
this channel simulation scheme requires randomization at both the encoder and decoder .
in essence , a codebook of independently drawn @xmath120 sequences is overpopulated so that the encoder can choose one randomly from many that are jointly typical with @xmath2 .
the decoder then randomly generates @xmath6 conditioned on @xmath120 .
our framework finds motivation in a game theoretic setting .
consider a zero - sum repeated game between two teams .
team a consists of two players who on the @xmath121th iteration take actions @xmath122 and @xmath123 .
the opponents on team b take combined action @xmath124 .
all action spaces @xmath125 , and @xmath126 are finite .
the payoff for team a at each iteration is a time - invariant finite function @xmath127 and is the loss for team b. each team wishes to maximize its time - averaged expected payoff .
assume that team a plays conservatively , attempting to maximize the expected payoff for the worst - case actions of team b. then the payoff at the @xmath121th iteration is @xmath128.\end{aligned}\ ] ] clearly , ( [ loss function ] ) could be maximized by finding an optimal mixed strategy @xmath129 that maximizes @xmath130 $ ] and choosing independent actions each iteration .
this would correspond to the minimax strategy .
however , now we introduce a new constraint : the players on team a have a limited secure channel of communication .
player 1 , who chooses the actions @xmath2 , communicates at rate @xmath12 to player 2 , who chooses @xmath6 .
let @xmath72 be the message passed from player 1 to player 2 .
we say a rate @xmath12 is achievable for payoff @xmath131 if there exists a sequence of random variable triples @xmath132 that each form markov chains for all @xmath133 . ]
@xmath134 and such that @xmath135 and @xmath136 & > & \theta.\end{aligned}\ ] ] let @xmath137 be the infimum of achievable rates for payoff @xmath131 .
we claim that @xmath137 is the least average common information of all combinations of strategies that achieve average payoff @xmath131 .
define , @xmath138 \right ] \geq \theta.\end{aligned}\ ] ] [ theorem game theory ] @xmath139 _ converse sketch : _ + the important elements of the converse are the inequalities @xmath140 for all @xmath141 , where @xmath80 is uniformly distributed on @xmath81 .
now identify the tuple @xmath142 as the auxiliary random variable @xmath143 .
_ achievability comment : _ + the random variable @xmath143 serves as a time sharing variable to combine strategies of high and low correlation .
the author would like to thank his advisor , tom cover , for encouraging the study of coordination via communication , and young - han kim for his suggestions and encouragement .
this work is supported by the national science foundation through grants ccf-0515303 and ccf-0635318 .
p. gcs and j. krner , `` common information is far less than mutual information , '' _ problems of control and info .
149 - 162 , 1973 .
a. wyner , `` the common information of two dependent random variables , '' _ ieee trans .
info . theory _ ,
it-21 , no .
2 , march 1975 .
t. cover and h. permuter , `` capacity of coordinated actions , '' isit 2007 , nice , france . c. h. bennett and p. w. shor , `` entanglement - assisted capacity of a quantum channel and the reverse shannon theorem , '' _ ieee trans .
info . theory _ ,
v. anantharam and v. borkar , `` common randomness and distributed control : a counterexample , '' _ systems and control letters _
7 - 8 , july 2007 .
i. csiszr and j. krner , _ information theory : coding theorems for discrete memoryless systems .
_ new york : academic , 1981 .
t. s. han and s. verd , `` approximation theory of output statistics , '' _ ieee trans .
info . theory _ ,
3 , may 1993 . | two familiar notions of correlation are rediscovered as extreme operating points for simulating a discrete memoryless channel , in which a channel output is generated based only on a description of the channel input .
wyner s `` common information '' coincides with the minimum description rate needed .
however , when common randomness independent of the input is available , the necessary description rate reduces to shannon s mutual information .
this work characterizes the optimal tradeoff between the amount of common randomness used and the required rate of description . |
the concept of an eccentric , `` precessing '' disk was a crucial ingredient of the tidal - thermal instability model of superoutbursts proposed by osaki ( 1996 , 2005 and references therein ) .
results of recent analysis of superoutbursts of z cha ( smak 2007 , 2008a ) definitely show , however , that superoutbursts are due to a major enhancement in the mass transfer rate .
it is worth to recall that an alternative model for superoutbursts , involving such enhanced mass transfer resulting from variable irradiation of the secondary component , was considered earlier by osaki himself ( osaki 1985 ; see also mineshige 1988 ) .
further work in this direction appears now highly desirable .
eccentric , `` precessing '' disk continues to be an essential ingredient of the commonly accepted model for superhumps ( see also osaki 1996 , 2005 and references therein ) .
observational evidence for eccentric , `` precessing '' disks in dwarf novae during their superoutbursts was presented during the last three decades by numerous authors . under those circumstances
the question posed in the title of the present paper can be considered rather provocative .
even more provocative , however , will be our negative answer to that question .
the organization of the present paper will be as follows . in section 2
we will clarify problems of terminology .
the main part of the paper will then be devoted to a detailed critical review of the observational evidence for eccentric , `` precessing '' disks resulting from the analysis of radial velocities ( section 3 ) , of the observed ( disk ? ) eclipses ( section 4 ) , and of the spot of eclipses ( section 5 ) .
results will be discussed in section 6 .
three periodicities are observed during superoutbursts : the orbital period @xmath2 , the superhump period @xmath3 , and the corresponding beat period @xmath4 @xmath5 accordingly , three sets of phases are to be considered : the orbital phase @xmath6 ( or simply @xmath7 ) , the superhump phase @xmath8 , and the beat phase @xmath9 it is commonly assumed that the beat period @xmath4 represents the apsidal motion of an eccentric disk ( or an outer eccentric ring ) , usually referred to
incorrectly as `` precession '' .
consequently , the beat period is called the `` precession '' period while the beat phase the `` precession '' phase .
the orientation of an eccentric disk / ring can be defined by the position angle @xmath10 of the periastron with respect to the line of sight .
this position angle is directly related to the beat phase @xmath11 .
in general , they could differ by a constant phase shift . in practice , however , using various kinds of observational evidence , nearly all authors ( e.g. vogt 1981 , hessman et al .
1992 ) assume or conclude that the beat phase @xmath12 corresponds to the situation when the periastron is facing the observer , i.e. that @xmath13
using textbook formulae for the two body problem we find that the radial velocity of a point on an elliptical ring is given by @xmath14~,\ ] ] where @xmath15 is describing the orbital motion , @xmath16 is the position angle of the point considered with respect to the periastron , @xmath10 the position angle of the periastron with respect to the line of sight ( both counted counterclockwise ) , and @xmath17^{1/2}~,\ ] ] is an equivalent of the standard expression for the rotational velocity of the disk / ring . here
@xmath18 is the major semi - axis of the ring and @xmath19 its eccentricity . for the radial velocity of an element of the ring seen in projection on the white dwarf and the central parts of the disk ( to be discussed in sections 3.2 and 3.3 ) we obtain @xmath20 for the maximum and minimum radial velocity , corresponding to the red and blue peaks of the emission line ( to be used in section 3.4 ) we get @xmath21 the first evidence for an eccentric disk in a dwarf nova during the superoutburst came from spectroscopic observations of z cha made by vogt ( 1981 ) during its 1978 march / april superoutburst .
vogt measured radial velocities from the central absorption components of the double emission lines and found ( see his fig.7 ) that those measured on 1978 march 28/29 were very high ( his @xmath22 km / s ) while those measured on 1978 march 27/28 ( i.e. on the previous night ) very low ( @xmath23 km / s ) . to intepret those results vogt proposed the following model ( see fig.9 of his paper ) : the central absorption components are produced in the outer eccentric ring , specifically in the region which is seen in projection on the central parts of the disk . the difference between @xmath24 and @xmath25 results from the apsidal motion of the ring : the value of the beat period @xmath26d implies that the orientation of the eccentric disk on two consecutive nights differs by about 180@xmath27 .
this interpretation , however , can be easily shown to be incorrect .
to do so we calculate the beat phases corresponding to the times of vogt s spectroscopic observations . for the night of march 27/28 we obtain @xmath28 , while for the night of march 28/29 we get @xmath29 . from eq.6
it immediately follows that radial velocities measured on march 27/28 should be systematically _ positive _ , while those on march 28/29 systematically _
negative_. this , however , is just opposite to the observed effect . in other words ,
the orientations of the ring during those two nights were roughly opposite to the two orientations shown in vogt s fig.9 . to complete our discussion let us return to the problem of the origin of central absorption components .
vogt assumed arbitrarily that they are formed in the outer eccentric ring . in reality , however , the double emission lines , including their central absorption components , are produced in the atmosphere of the entire disk .
furthermore , in the particular case of z cha , an additional , variable contribution due to the absorption in the overflowing parts of the stream ( smak 2007 ) is likely to be present .
such a possibility is suggested by the fact that , according to vogt , the absorption components are strongest near the orbital phase @xmath30 , i.e. when the overflowing stream is directly projected against the central parts of the disk . to strengthen his interpretation of z cha vogt presented additional evidence based on absorption - line @xmath31-velocities of wz sge measured by several authors during its 1978 superoutburst .
his fig.8 did indeed show sinusoidal variations of the @xmath31-velocities with the beat phase .
however , to calculate @xmath11 vogt adopted an incorrect value of the beat period : @xmath32d . the corrected version of his @xmath33 diagram , with @xmath11 calculated with @xmath34d ( patterson et al . 1981 ) , shows only large scatter with _ no _ obvious dependence of @xmath31-velocities on @xmath11 .
the eccentricity of the disk can be determined , using eq.7 , from vogt s radial velocities measured from the peaks of the double emission lines .
for this purpose we use radial velocities e@xmath35 and e@xmath36 listed in his table 6 ( excluding only four uncertain values of e@xmath35 with @xmath6 between 0.10 and 0.40 which were near - blends with the absoprtion component ) .
first , we correct them for the orbital motion @xmath37 using two , rather different sets of parameters determined from observations at quiescence , namely : ( 1 ) @xmath38 km / s , @xmath39 , @xmath40 km / s ( vogt 1981 ) , and ( 2 ) @xmath41 km / s , @xmath42 , @xmath43 km / s ( marsch et al .
1987 ) .
results , obtained with those two sets of parameters , are : ( 1 ) @xmath44 km / s , @xmath45 , @xmath46 , and ( 2 ) @xmath47 km / s , @xmath48 , @xmath49 . as can be seen the eccentricity , if any , is close to zero .
worth noting is that in both cases the resulting disk radius turns out , as expected , to be close to the mean radius of the roche lobe @xmath50 .
the evidence discussed below for oy car ( section 4.2 ) and iy uma ( section 4.3 ) was based on the dependence of various eclipse parameters on the beat phase , interpreted in terms of the apsidal motion of an eccentric disk .
such an interpretation , however , assumes implicitly that the observed eclipses are _ pure disk _ eclipses .
this is obviously not true .
we now have evidence , based on the detailed analysis of eclipses in z cha ( smak 2007 ) , showing that , depending on the beat phase , they represent a combination of eclipses of several different sources : ( 1 ) the disk , ( 2 ) the `` standard '' spot ( around @xmath0 ) , or ( 3 ) the `` peculiar '' spot , involving stream overflow ( around @xmath51 and 0.75 ) , and ( 4 ) the superhump source ( around @xmath52 ) .
their relative contributions depend strongly on the beat phase .
consequently , the resulting dependence of the observed eclipse parameters on @xmath11 is quite complicated . under those circumstances the only practical solution is to decompose the observed light curves into their separate components ( as it was done in the case of z cha ; smak 2007 ) .
it can be hoped that the existing light curves of oy car and iy uma will eventually be analyzed in such a way . for the time being we can only mention that the shapes of the _ pure disk _ eclipse light curves obtained for
z cha ( smak 2007 ) do not show any dependence on the beat phase .
krzemiski and vogt ( 1985 ) collected light curves of oy car during its january 1980 superoutburst and determined several parameters describing the shape of the observed eclipses : ( 1 ) deviations of the eclipse amplitude @xmath53 from the amplitude @xmath54 vs. epoch @xmath55 relation , ( 2 ) deviations of the eclipse width @xmath56 from the width @xmath57 vs. epoch @xmath55 relation , ( 3 ) the eclipse asymmetry @xmath58 , measured at half - depth , and ( 4 ) the values of ( @xmath59 ) .
they found that all those parameters show roughly sinusoidal variations with the beat phase ( see their figs.7 - 11 ) and interpreted them as being due to the apsidal motion of an eccentric disk . in what follows
we will show that this interpretation was incorrect .
first of all , as noted by krzemiski and vogt themselves , variations predicted by their eccentric disk model ( see their fig.12 ) for @xmath56 and @xmath58 do not agree with their observed variations .
specifically , their model predicts _
double sinusoidal _
variations for @xmath56 , while in the case of @xmath58 the model predicted variations are shifted in @xmath11 by @xmath60 . to interpret the observed variations of @xmath53 krzemiski and vogt assumed that the outer parts of the disk , on the other side of the white dwarf , remain uneclipsed . using system parameters of oy car ( wood et al .
1989 ) we find , however , that this assumption would require the disk radius to be larger than @xmath61 , which is already larger than the upper limit to the disk radius @xmath62 , obtained from the condition that it must be smaller than the distance from l@xmath63 to the white dwarf .
this implies that the far side of disk is fully eclipsed .
the geometry of eclipses of the hypothetical eccentric disk in oy car is shown schematically in fig.1 .
it can immediately be seen that , regardless of the surface brightness distribution , the eclipse amplitude should show _
double sinusoidal _ variations with @xmath11 , which is inconsistent with observations .
= 1.5 cm the observed variations of @xmath53 with @xmath11 can , in fact , be qualitatively explained by the variable contribution from the hot spot . in particular , around @xmath0 , when the `` standard '' hot spot is present , its
eclipses make the full eclipse amplitude larger .
patterson et al . ( 2000 ) presented results of their extensive photometry of iy uma covering the declining part of its 2000 january superoutburst . from the analysis of the observed light curves they found that all eclipse parameters depend strongly on the beat phase .
in particular , around @xmath0 the eclipse amplitude is largest and all characteristic phases are systematically more positive ( see their figs.7 and 9 ) .
patterson et al . interpreted those effects as being to due the apsidal motion of a strongly eccentric ( @xmath64 ) disk .
there are several arguments against such an interpretation .
first of all , we can recall our general comments from section 4.1 .
secondly , we may note that the observed dependence of eclipse parameters on @xmath11 ( similar to the case of oy car ; section 4.2 ) is most likely due to the variable contribution from the hot spot which on account of the orbital inclination ( @xmath65 ) being much higher than in the case of oy car ( @xmath66 ) is expected to be relatively larger . around @xmath0 , when the `` standard '' hot spot is present , the contribution from its eclipses makes the observed eclipses deeper and shifted toward later phases . in partcular ,
the large values of @xmath67 observed around @xmath0 ( fig.9 in patterson et al .
2000 ) correspond probably to the egress of the `` standard '' hot spot .
the presence of the hot spot in a dwarf nova during its superoutburst was detected for the first time by schoembs ( 1986 ) .
his light curves of oy car obtained during the final decline from its 1980 november / december superoutburst showed clearly the characteristic eclipses of the hot spot .
six years later hessman et al .
( 1992 ) published results of their photometry of oy car , made also during the final decline from another of its superoutbursts , their light curves showing also characteristic spot eclipses .
unfortunately , the significance of these discoveries was not fully appreciated at that time , the presence of those eclipses being interpreted in terms of the _ reappearance _ of the quiescent hot spot .
the hot spot eclipses during superoutburst maximum were first detected by patterson et al .
( 2002 ) in wz sge during its 2001 superoutburst .
regretfully , this interpretation was challenged by osaki and meyer ( 2003 ) who argued that those eclipses were not due to the hot spot but rather due to _ `` the superhump light source itself''_. situation was definitely clarified only very recently by the results of our analysis of eclipse light curves of z cha and oy car ( smak 2007,2008b ) observed during maxima of their superoutbursts .
they were decomposed into their disk and spot components , the resulting hot spot light curves showing not only the characteristic shape of the eclipse but also the characteristic maximum around the orbital phase @xmath68 . according to the commonly accepted `` standard '' model
, the hot spot is produced by the collision of the stream with the outer parts of the disk .
in such a case , the distance @xmath69 of the spot from the white dwarf , to be referred to as `` spot distance '' , determined from spot eclipses , is identical with the radius of the disk @xmath70 .
it is worth to emphasize that in the case of an eccentric disk this @xmath70 is the _ local _ radius of the disk at a specific position angle corresponding to the point of collision .
furthermore , when such an eccentric disk rotates due to the apsidal motion , the dependence of @xmath70 on the position angle produces its periodic variations with the beat phase . in the analysis of eclipses of the `` standard '' hot spot we use the phases of mid - ingress @xmath71 and mid - egress @xmath72 which are related to the mass ratio @xmath73 ( defining the shape of the stream trajectory ) , the orbital inclination , the spot distance @xmath69 , and its ellongation parameter @xmath74 ( cf .
smak 1996,2007 ) .
when the mass ratio and inclination are known , the two phases @xmath71 and @xmath72 can then be used to obtain two indpendent determinations of the spot distance : @xmath75 it is obvious that in the case of a `` standard '' hot spot they must be identical @xmath76 the observed luminosity of the hot spot depends on the `` impact parameter '' @xmath77 ( i.e. on the square of the relative velocity of collision between the stream and the disk ) .
the value of this impact parameter increases with decreasing radial distance @xmath69 .
therefore in the case of variable disk radius the luminosity @xmath78 of the `` standard '' spot should vary as @xmath79 the last two relations provide two crucial tests for the applicability of the concept of a `` standard '' hot spot and in particular for assuming that the radial distance of the spot ( @xmath80 or @xmath81 ) is identical with the radius of the disk @xmath70 .
= 9.0 cm recent analysis of hot spot eclipses in z cha during its superoutbursts ( smak 2007 ) provides good illustration of different situations to be encountered in this area .
the `` standard '' hot spot eclipses are observed only at beat phases @xmath82 . in particular , the spot distances @xmath80 and @xmath81 determined from those phases are practically identical . on the other hand ,
the spot eclipses observed at intermediate beat phases ( around @xmath83 and around @xmath84 ) are rather peculiar .
first of all , the resulting spot distances are generally smaller , with @xmath80 being systematically larger than @xmath81 .
this could be seen already from fig.6 of smak ( 2007 ) and is shown here in the lower part of fig.2 .
secondly , the luminosities of the spot observed at intermediate beat phases are systematically lower ( smak 2007 , fig.5 ) .
those peculiarities were interpreted ( smak 2007 ) as being due to a substantial stream overflow .
supporting this interpretation is a correlation between the duration of egress @xmath85 and the spot distance @xmath81 , mentioned already in section 5.3 of that paper , and shown here in the upper part of fig.2 .
the lesson learned from z cha can be summarized in the form of the following warning : the spot distances @xmath86 , obtained from the analysis of peculiar spot eclipses , observed at intermediate beat phases , _ do not provide any reliable information about the radius of the disk_. this warning becomes even stronger when the observed spot luminosities imply @xmath87 . as already mentioned above , schoembs ( 1986 ) and hessman et al .
( 1992 ) presented extensive photometric coverage of the advanced stages of the 1980 november / december and 1987 february superoutbursts of oy car . during the final decline their light curves showed characteristic
eclipses of the hot spot .
hessman et al . (
1992 ) analyzed those light curves in order to determine the location of the hot spot .
specifically , using the phases of mid - ingress , they obtained spot distances @xmath80 and found that they show strong dependence on the beat phase @xmath11 , interpreted by them as being due to the `` precession '' of a strongly eccentric ( @xmath88 ) disk .
this interpretation , however , can easily be challenged . to begin with
, we must note that the variations of the spot luminosity @xmath78 and the spot distance @xmath80 with @xmath11 , shown in their fig.8 are strikingly similar to the case of z cha ( cf .
section 5.2 ; see also figs.5 and 6 in smak 2007 ) .
this immediately suggests that like in the case of z cha the `` standard '' hot spot was present in oy car only around @xmath0 and that spot distances @xmath80 obtained from intermediate beat phases should not be treated as a reliable measure of the disk radius @xmath70 .
= 7.5 cm to clarify the situation definitely we analyze the available light curves of seven spot eclipses ( eclipses 7 - 1 , 7 - 3 , 9 - 1 , 9 - 2 , 10 - 1 , and 10 - 2 from fig.3b of schoembs 1986 , and the eclipse from fig.1-bottom of hessman et al .
for each of them we determine the phases of mid - ingress @xmath71 and mid - egress @xmath72 and the duriation of egress @xmath85 .
then , using geometrical system parameters ( wood et al . 1989 ) and adopting two different values of the spot ellongation parameter @xmath89 and 0.04 , we obtain the spot distances @xmath80 and @xmath81 .
results , presented in fig.3 , show that the situation is indeed nearly identical with that in the case of z cha ( see fig.2 ) .
in addition , we have @xmath90 ( fig.8 in hessman et al . 1992 ) . taking all this into account
we must conclude that at intermediate beat phases away from @xmath0 we are dealing with a `` peculiar '' rather than `` standard '' spot and that consequently the spot distances @xmath80 obtained at those phases do not provide any reliable information about the disk radius @xmath70 .
patterson et al .
( 2002 ) published results of a worldwide photometric campaign covering the 2001 superoutburst of wz sge .
one of their most important results was the detection for the first time of hot spot eclipses during the main part of superoutburst maximum .
they analyzed those eclipses in a standard way finding that all hot spot parameters varied significantly with the `` precession '' ( i.e. beat ) phase .
this was interpreted by them as being due to the apsidal motion of a strongly eccentric ( @xmath91 ) disk .
this interpretation , however , can easily be challenged . in our analysis
we use the values of @xmath92 and eclipse widths ( fig.21-left of patterson et al .
2002 ) to recover the phases of mid - ingress ( @xmath71 ) and mid - egress ( @xmath72 ) . then , using system parameters from steeghs et al .
( 2007 ) we determine the spot distances @xmath93 and @xmath94 .
situation is somewhat complicated here by the fact that the zero - point of the orbital phases , defined by mid - eclipse of the spot , does not correspond to the true zero - phase .
the values of the corresponding phase shift , obtained by different authors from the analysis of spot eclipses ( e.g. smak 1993 ) or from radial velocity curves ( e.g. spruit and rutten 1998 , steeghs et al .
2001 ) , range from @xmath95 to @xmath96 .
in our analysis we adopt : @xmath97 .
for the spot ellongation parameter we also adopt two values : @xmath98 .
= 3.5 cm = 9.0 cm fig.4 shows an example of the resulting spot distances @xmath93 and @xmath94 plotted _ versus _ the beat phase and fig.5 presents a comparison of those two parameters obtained using four combinations of @xmath74 and @xmath99 . as can be seen , the situation is very similar to the case of z cha ( fig.2 ) and oy car ( fig.3 ) . in particular ,
the condition @xmath100 is fulfilled only by the largest values of those parameters , corresponding to @xmath0 , while at other beat phases we have @xmath101 .
in addition , the spot visibility and its luminosity depend strongly on @xmath11 ( fig.21-top in patterson et al .
2002 ) , this behavior being strikingly similar to the case of z cha . taking all this into account we conclude that away from @xmath0 we are dealing with a `` peculiar '' spot and that consequently the values of @xmath93 and @xmath94 obtained from those phases are not representative for the radius of the disk @xmath70 .
results presented above in sections 3 - 5 can be summarized in the form of the following conclusions : * the spectroscopic evidence for eccentric disks in z cha and wz sge , presented by vogt ( 1981 ) , based on radial velocities measured from absorption components , is shown to be an artifact , resulting from miscalculated beat phases . * the photometric evidence for eccentric disks in oy car ( krzemiski and vogt 1985 ) and iy uma ( patterson et al . 2000 ) , based on the observed dependence of various eclipse parameters on the beat phase , involved an implicit assumption that the observed eclipses are _ pure disk _ eclipses .
this assumption was incorrect .
in particular , the observed variations of eclipse parameters are very likely due to the variable contributions from the hot spot and from the superhump source , both of them depending strongly on the beat phase . * the photometric evidence for eccentric disks in oy car ( hessman et al 1992 ) and wz sge ( patterson et al .
2002 ) , resulting from the analysis of hot spot eclipses , was based on an implicit assumption that the spot distances are always identical with the radius of the disk .
this assumption , however , is correct only in the case of eclipses of the `` standard '' hot spot which are observed around @xmath0 . at other beat phases we are dealing with `` peculiar '' spots ( involving the stream overflow ) ,
their eclipses giving spot distances @xmath101 which are generally smaller than the radius of the disk .
to summarize : there is no reliable evidence for eccentric disks in dwarf novae during their superoutbursts .
this conclusion is strengthened by our new determination of disk eccentricity in z cha , based on vogt s radial velocities measured from emission components , which gave : @xmath1 . | the evidence presented earlier by several authors for the substantial disk eccentricity in dwarf novae during their superoutbursts is shown to result either from errors , or from arbitrary , incorrect assumptions .
\(1 ) the evidence for z cha and wz sge ( vogt 1981 ) , based on radial velocities measured from absorption components , was an artifact , resulting from miscalculated beat phases .
\(2 ) the evidence for oy car ( krzemiski and vogt 1985 ) and iy uma ( patterson et al . 2000 ) , based on the observed dependence of eclipse parameters on the beat phase , involved an implicit assumption that the observed eclipses are _ pure disk _ eclipses , which is not true .
in particular , the observed variations of eclipse parameters are likely due to the contributions from the hot spot and from the superhump source , which depend strongly on the beat phase .
\(3 ) the evidence for oy car ( hessman et al 1992 ) and wz sge ( patterson et al . 2002 ) , resulting from the analysis of hot spot eclipses , was based on the assumption that the spot distances are identical with the radius of the disk , which is not always correct . in particular , in the case of eclipses of `` peculiar '' spots ( involving the stream overflow ) , observed at beat phases away from @xmath0 , the resulting spot distances are smaller that the radius of the disk .
new determination of disk eccentricity in z cha , using vogt s radial velocities measured from emission components , gives @xmath1 .
* key words : * _ accretion , accretion disks binaries : cataclysmic variables , stars : dwarf novae , stars : individual : oy car , z cha , wz sge , iy uma _ |
the exploration of cp violation and determination of the cabibbo - kobayashi - maskawa ( ckm ) matrix elements motivate extensive interests of b meson decay . from another point of view
, b decays provide a good place to study the fruitful dynamics of qcd . up to now
, we have not a truly successful method to calculate the non - perturbative qcd and the mechanism of quark confinement is still unknown .
the study of exclusive b decays is usually difficult because of the complicate qcd dynamics .
however , the experiments from the belle and babar collaborations have accumulated and will continue to accumulate a large amount of data of b decays .
the large theoretical uncertainties can not compete with the more precise experimental data .
we come to one stage that experiment goes ahead of theory .
the theorists in b physics have to meet great challenge from the experiment .
the problem of exclusive b decays lies in a very large number of degrees of freedom .
the experiment observes the hadron states such as b meson and pion , kaon etc . in the qcd lagrangian ,
only quark and gluon degrees of freedom appear .
we do nt know accurately how the hadron are formed by quarks and gluons . from the energy scale standpoint , the b decays
usually contain many scales : the weak interaction scale @xmath0 , the @xmath1 quark mass @xmath2 , the qcd scale @xmath3 and possible intermediate scales due to the soft spectator quark in b meson .
the momenta of quarks or gluons are not restricted .
they can be highly virtual , very soft or highly energetic but collinear to the fast moving pion .
the fact that we have to treat all the degrees of freedom in one process if we think qcd is the correct theory of strong interaction leads to great theoretical complications .
one method to treat the multi - scales problem is factorization .
the factorization is a key ingredient of perturbative qcd ( pqcd ) @xcite .
its basic idea is to separate the short - distant dynamics from the long - distance physics .
it has been widely used in the hard qcd processes where the large momentum transfer @xmath4 is involved .
another method is the effective field theory .
it is a useful toll to study the process with several separate scales .
the heavy quark effective theory ( hqet ) @xcite is a low energy effective theory .
it allows model - independent predictions in some cases of the heavy meson system , such as @xmath5 form factor at zero recoil .
the developments of the two methods are nearly independent although some ideas in them are related . as we will show that these two lines of thought converge in the study of exclusive b decays , .
the factorization had been introduced in exclusive b decays for a long time .
the old form which we call the naive factorization approach is to divide a hadronic matrix element into the multiplication of a form factor and hadron decay constant .
this idea influences b physics for more than 30 years .
much efforts were done to interpret and generalize it .
now , the idea of factorization has been developed as a central idea of b physics . in this talk
, we will discuss the theoretical struggle of studying the exclusive b decays along the line of factorization .
the success and limitations of each theoretical approach will be analyzed .
we are focus on the conceptual developments from the naive factorization approach to the qcd factorization approach .
we will show how the effective field theory enters into b physics and modifies our view .
the first thing to do in b decays is to integrate out the heavy degrees of freedom of w , z bosons and top quark in the standard model .
the method is to construct an effective theory where the above heavy particles do not appear .
the theoretical technic is mature now .
it uses the operator product expansion ( ope ) and renormalization group equation ( rge ) . for non - leptonic
b decays , the relevant effective weak hamiltonian is h_eff=_i v^i_ckmc_i()q_i . where @xmath6 is the fermi constant and @xmath7 are current - current operators .
the scale @xmath8 is chosen of order of @xmath2 .
the amplitude of @xmath9 decay isa(bm_1m_2)=_i v^i_ckm + c_i()m_1m_2|q_i|b ( ) . where @xmath10 are hadronic matrix elements . the remained work is to calculate the hadronic matrix element .
the introduction of factorization to simplify the the hadronic matrix element may be firstly given in @xcite up to knowledge of the author .
i can not trace out this history but refer to @xcite as our start of discussion .
bauer , stech and wirbel consider the non - leptonic two meson decays where the final mesons are energetic .
they made assumptions that only the asymptotic part of the hadron field is effective and the current are proportional the hadron field .
all the initial state interaction and final state interactions are neglected .
based on the above assumptions , one hadron and its associated current are separated out .
the hadronic matrix element is factorized into a multiplication of decay constant and the form factor which represented by the matrix element of the other current .
take @xmath11 decay as an example , ^+^-|(|u b)_v - a(|d u)_v - a| |b^0 = + ^-|(|d u)_v - a|0 ^+|(|u b)_v - a||b^0 .
the idea of the above factorization is simple but it has a deep influence .
the application of the above naive factorization approach into the non - leptonic two body b decays is successful in early days of b physics when the experimental data are rare . for a long time
, this approach is nearly the only method to give a theoretical prediction of exclusive non - leptonic b decays although the accuracy is at the qualitative level for many processes .
the factorization approach plays a similar role as the feynman s parton model in dis , we can call this naive approach as the parton model in b physics .
one might expect that the factorization is a limit case of a more general theory .
the understanding of the factorization from field theory of qcd is a long way .
the first step comes from bjorken s intuitive space - time picture @xcite .
it is bjorken who proposed the famous scaling in dis which lead to the rise of qcd . for @xmath11 decay ,
the quark level decay is @xmath12 . in order to form the final energetic hadron
, the quark pair @xmath13 has to choose a nearly collinear configuration . because the pions move fast
, the formation time of @xmath14 will be long because of the relativistic time - dilation .
the dilation ratio is @xmath15 .
that means the hadronization occurs @xmath16 fm away from the remained system . before the hadronization , the @xmath17 quark pair produced from the pointlike , color - singlet weak interaction is a a small color dipole .
the small color dipole has little interaction with the other quarks .
the above consideration is usually called color transparency argument " . from the above argument , one may guess that the factorization approach is the leading order contribution of heavy quark limit where @xmath18 and the non - factorizable corrections come from the interactions of small color dipole with the remained quarks at short distance . what is color transparency ?
it is a concept outside of b physics . according to the discussions in @xcite ,
the color transparency is a phenomenon of pqcd @xcite .
it says that a small color - singlet object can pass freely through nucleon target as if the target is transparent .
the large target acts as a filter which removes the large transverse separation component of the hadron .
the @xmath19 decay provides a similar environment .
@xmath20 transition is at long distance .
the energetic @xmath14 selects the @xmath13 quark pair at small transverse separations .
the long distance processes caused by emitting or absorbing soft gluons have destructive effects and cancel . in @xcite , politzer and wise apply the pqcd method into the exclusive processes in @xmath19 decay .
they proposed a factorization formula that the hadronic matrix element can be written as the product of a matrix element of @xmath21 and @xmath22 mesons in hqet and a convolution by a hard scattering amplitude t and the pion distribution amplitude @xmath23 as [ eq : pw]d^+^-|(|c b)_v - a(|d u)_v - a| |b^0 = d^+|(|c_v b_v)_v - a||b^0 + _ 0 ^ 1 dx t(x , m_c / m_b , ) _ ( x , ) .s where @xmath24 are effective fields for heavy quarks in hqet .
they point out the above factorization formulae is the leading order result in @xmath25 .
one loop calculate is done and the result show @xmath26 correction to leading contribution is small .
however , they do nt give proof of the factorization .
the success of hqet motivates theorists to use effective field theory into wider range of application .
dugan and grinstein had a new idea to establish a foundation for factorization on the effective field theory .
they use the effective field theory to replace the intuitive color transparency argument " . in @xcite ,
dugan and grinstein proposed a large energy effective theory ( leet ) to describe the interaction of the energetic collinear quark with soft gluons .
they consider one kinematic case that the energy of the collinear quark is much lager than the momentum of soft gluon , i.e. , @xmath27 .
the central idea in leet is that the energy of the collinear quark is unchanged which is analogous to the velocity superselection rule in hqet .
the leet is very similar to the hqet .
the leet lagrangian is [ eq : leet]l_leet=^ind . where @xmath28 is a light - like vector which the direction is along the motion of the collinear quark and the collinear field @xmath29 satisfies @xmath30 .
if we choose the light - cone gauge @xmath31 , the soft gluons decouple from the collinear and factorization is a trivial result . from this point of view
, the color transparency is explained by that only the longitudinal gluons couple to the collinear quark and thus decouple .
although the proof of factorization in this way is too simple to be correct , the leet is very impressive .
it provide a new view of factorization . in pqcd ,
the diagrammatic analysis @xcite is the most familiar method to prove factorization .
the leet provide an operator description and the result is automatical to all orders .
the proposal of an effective lagrangian permits us to use gauge symmetry at the lagrangian level .
the proof of factorization can be easily done in a gauge invariant way , i.e. , we need not have to choose the light - cone gauge @xmath31 .
as we will show , the leet is one part of the soft - collinear effective theory . the leet lagrangian given in eq .
( [ eq : leet ] ) is just the leading order result .
the biggest problem of the leet is that it can not reproduce the long - distance physics of qcd .
the reason is simple because it misses the collinear gluon degrees of freedom . without collinear gluon ,
the collinear quark pair @xmath13 can not form the energetic bound state @xmath14 .
the neglect of collinear gluon is pointed out in @xcite .
aglietti and corb point out one problem of leet and modify the leet by including the transverse degrees of freedom @xcite .
the improved leet lagrangian is given by [ eq : leet2]l=^(ind + ) .
where @xmath32 is the energy of the collinear quark . in this modified version
, the collinear gluon is still missing . the effective theory which includes
the collinear gluon is more complicated than eq .
( [ eq : leet2 ] ) because the energy @xmath32 is not conserved . except the missing of the collinear gluon degrees of freedom , both the leet and its modified version
have the problem that there is no systematic power counting to support them .
they have to wait for the next step development .
the application of the pqcd method in @xcite into the exclusive b decays had been explored by many theorists .
there are several different perturbative approaches appeared in the literatures . due to the scope of this review
, we focus our discussions on the recently proposed qcd factorization approach in @xcite .
beneke , buchalla , neubert and sachrajda want to establish a rigorous framework for the exclusive non - leptonic b decays .
the basic idea of the qcd factorization approach is that in the heavy quark limit the naive factorization is the lowest order approximation and the corrections to the naive factorization can be formulated as a factorization form up to corrections of order @xmath25 .
the heavy quark limit @xmath33 is the kinematic foundation of the qcd factorization approach . for @xmath34 decay ,
the factorization formula is [ eq : qfa]|q_i||b^0=f^b(0)_0 ^ 1 dx t^i_i(x ) _ ( x ) + + _ 0 ^ 1 ddxdy t^ii_i(,x , y ) _
b()_(x)_(y ) . where @xmath35 is a @xmath36 form factor at @xmath37 , @xmath38
are light - cone distribution amplitudes of the pion and b meson .
the @xmath39 are perturbatively calculable hard scattering kernels . compared to eq .
( [ eq : pw ] ) , the difference lies in the second term due to the hard spectator correction .
@xcite can be considered as a systematic introduction of the pqcd method into b physics for the first time . in @xcite ,
a power counting is used to argue the validity of the qcd factorization approach .
this power counting largely uses the endpoint behavior of the distribution amplitudes of mesons .
annihilation diagrams and higher fock states of the mesons are proved to be power suppressed .
the hard spectator interaction in @xmath19 decay is suppressed if one assumes @xmath40 quark is heavy .
the validity of factorization can not be based on the intuitive arguments only .
it should be proved to all orders that all the soft and collinear divergences cancel or can be absorbed into the universal non - perturbative functions and the hard scattering kernels are infrared insensitive . for @xmath41 decays ,
the factorization is proved to hold in @xmath26 order .
but it is not sufficient to guarantee the validity of factorization because all the soft and collinear divergences cancel is not general .
the general case is that the infrared divergences may not cancel but they can be separated out and absorbed into the definition of the non - perturbative functions . up to now
, the factorization beyond @xmath26 order in @xmath41 decays has not been truly proved . in @xcite ,
the factorization proof for @xmath42 decays at two - loop order is given .
two - loop order is equivalent to @xmath43 order because the hard spectator interaction is power suppressed for the heavy - light final states .
the authors consider 62 non - factorizable " diagrams at two loop order .
maybe their two loop order proof of factorization is the most detailed analysis in the literatures up to my knowledge .
the eikonal approximation and the ward identity are used implicitly . at two - loop order ,
the infrared divergences in the soft - soft , soft - collinear and collinear - collinear momentum regions cancel .
the hard - collinear and hard - soft contributions contain non - cancelling infrared divergences .
they can be factorizaed out and absorbed into the definition of distribution amplitude and form factor respectively .
the proof of factorization in the diagrammatic analysis is intuitive , but it is impossible to go to all orders .
a factorization proof of all orders is necessary . from the leet , we know that the effective field theory can simplify the analysis of the infrared physics and the factorization in it is automatic to all orders .
one natural idea is : can we construct an effective field theory for the soft and collinear particles which reproduce all the infrared physics of qcd and can simplify the factorization proof ?
bauer el al .
aim at developing a soft - collinear effective theory by generalizing the idea of leet .
they start from the study of summing sudakov double logarithms in inclusive @xmath44 decay . for @xmath44 decay near the endpoint of the photon spectrum
, it contains energetic light particle .
the sudakov double logarithms will appear in one loop order due to the non - cancellation of the soft and collinear divergences .
the large double logarithms make the perturbation expansion ill - behaved and need to be resumed to all orders .
the sudakov resummation is usually considered as an important but difficult part in the conventional pqcd method . by matching the full theory to a new effective theory ,
the large logarithms cancel and the sudakov doule logarithms are summed by using the renormalization group equations @xcite .
one important thing is that sudakov resummation in the effective theory is simpler than that in the full theory .
the analysis in @xcite also shows another important thing that the effective field theory can be used in the case where the wilson s short - distance operator product expansion ( ope ) is not applicable .
the idea is matching the full theory onto the effective theory where the effective operators provide the long - distance information of qcd . to this externt
, the effective field theory develops the idea of ope .
the formalism of the soft - collinear effective theory is provide in @xcite . at this stage ,
only the ultrasoft gluons are included .
the lowest order effective lagrangian is written by l_eff=|_n_-,p + _
n_-,p .where @xmath45 , @xmath46 , @xmath47 and @xmath48 are two light - like vectors . because the formulation is given in a hybrid position - momentum space . a momentum label operator @xmath49 has to be introduced because the large momentum is not conserved .
beneke , chapovsky , diehl and feldmann developed a position space formalism in order to avoid the complicate momentum label operator in @xcite . in @xcite ,
i propose a soft - collinear effective theory which includes the soft gluons in the position space .
the final scet languagian is[eq : scet]l_scet=| . where the covariant derivative is defined by @xmath50 .
from the above effective lagrangian , it is easy to obtain the ( ultra)soft and collinear wilson lines .
the scet has rich symmetry structures .
the lorentz and gauge invariance are interesting in scet .
there is a new symmetry , scale symmetry .
this symmetry provide an interpretation of the bjorken s scaling and the scaling law in high energy scattering .
the application of scet to prove the factorization in @xmath42 decays is given in @xcite .
i give a factorization proof in dis in @xcite and scet can also be applied into multi - body b decays , such as @xmath51 decays @xcite .
although scet provides a new theoretical framework , the practical calculations of exclusive b decays still rely on some factorization formulae , such as the qcd factorization approach . by use of this opportunity ,
i want to express my personal opinions on the qcd factorization approach .
i will ask some conceptual questions about it . i. is
the factorization in @xmath41 decays proved ?
as we have discussed above , the factorization in @xmath41 decays is given only at @xmath26 order .
this is not sufficient for the validity of factorization .
we can say that the qcd factorization approach has not been proved as a `` theory '' . even for @xmath42 decays ,
the factorization are based on some kinematical assumptions , such as @xmath52 or the ratio @xmath53 is fixed .
the real world is : @xmath54 , @xmath55 .
although one can assume a limit case for the theoretical purpose , the relation between the ideal world and the real world needs to be explored .
\ii . is the qcd factorization approach correct in the @xmath56 limit ? this question is related to the first question .
this time we will not concern the technical complications about the factorization proof but the general argument of factorization . the argument is bjorken s color transparency " .
one strange thing for me is why the color transparency can be applied for one pion but not for another pion in @xmath41 decays .
the two pions are both energetic and have the same momentum in the rest frame of b meson , but they are treated unequally . the factorization formula in eq .
( [ eq : qfa ] ) contains two terms , one is related to the naive factorization approach , the other likes the hard scattering approach .
the two different terms make the factorization formula un - natural .
one associated problem is : the renormalization scale @xmath8 in first term is at order of @xmath2 , while in the next term @xmath57 .
there are two large scales in @xmath58 decays .
so it is a multi - scales problem .
my oppinion is that we should be more serious about his problem .
is @xmath59 form factor baisc ?
this is a very controversial topic . in the qcd factorization approach
, it is assumed as a basic function . but this function is different from other fundamental non - perturbative qcd fucntions such as the isgur - wise function , pion distribution amplitudes etc . @xmath59
form factor is not and can not be represented as a dimensionless function . from the soft - collinear effective theory which is scale invariant , the basic non - perturbative functions should be dimensionaless except the dependence on the renormalization @xmath8 . the @xmath59 form factor does not satisfy this criterion .
in fact , @xmath59 form factor contains very complicate qcd dynamics which includes scales of @xmath60 and scales between @xmath61 and @xmath3 . from this point of view
, the qcd factorization seems more like a phenomenological approach " rather than a theory " .
we have reviewed the theoretical developments from the naive factorization approach to the soft - collinear effective theory . in order to understand the factorization in exclusive b decays , new ideas and new theories or approaches
are produced .
the soft - collinear effective theory is a rigorous theory .
it provides a theoretical foundation of the factorization theorem in pqcd and a unified framework to study the inclusive and exclusive hard qcd processes .
the soft - collinear effective theory is the second theoretical contribution of b physics to qcd .
the first is hqet .
the qcd factorization approach can not solve the full complications of exclusive b decays .
the application of the soft - collinear effective in b decays is still limited because the momentum of quarks in hadrons depend on the non - perturbative dynamics for exclusive processes .
the final solution of b decays must rely on the solution on the quark confinement and non - perturbative problems .
if these fundamental problems of the strong interaction were solved by the theorists in b physics , it will not be a miracle .
this is the great challenge of b physics .
the author acknowledges a postdoc fellowship of the spanish ministry of education . this research is supported by grant fpa/2002 - 0612 of the ministry of science and technology . | i review the theoretical ideas and concepts along the line of factorization in the exclusive b decays . in order to understand the naive factorization ,
the effective field theories and the perturbative method of qcd are introduced and developed .
we focus our discussions on the large energy effective theory , the qcd factorization approach and the soft - collinear effective theory . |
one of the most fascinating arenas for testing general relativity in the neighborhood of a real astrophysical black hole is undoubtedly represented by sgr a * , the massive black hole ( mbh ) at the center of our galaxy , whose existence has been now strongly proven by the combination of radio , sub - mm , x - ray , and near infrared ( nir ) observations . while in the radio domain sgr a * appears as a steady and motionless compact source @xcite , its x - ray , sub - mm , and nir counterparts are variable since they exhibit outbursts of energy ( _ flares _ ) , which typically last for a few hours and occur several times a day @xcite .
these flares are likely due to energetic events arising very close to the central mbh , on a scale of a few schwarzschild radii , and they can be interpreted as due to emission from matter in relativistic orbits around sgr a*. moreover , by using large class telescopes , such as the eso very large telescope ( vlt ) and the keck telescopes , and thanks to the adaptive optics , it has been possible to observe a multitude of stars in the @xmath1 band ( the so - called s - stars ) moving in the gravitational potential of the galactic mbh @xcite .
the entire system is perfectly described by a single point mass and newton gravity .
very recently , @xcite have been able to determine the orbits of 28 s - stars , resting upon the results of 16 years of monitoring of their motions . in particular , it has been possible to observe the whole orbit of the famous star s2 , since it has completed a full revolution around sgr a*. the value of the galactic mbh mass , extrapolated from these improved data , is @xmath2 m@xmath3 . since the theory of general relativity predicts that the central mbh can act as a powerful gravitational lens on s - stars , some of them , depending on the knowledge of their orbital parameters , have been the object of careful investigations .
the strategy of this study is very clear .
in fact , knowing the position of the s - stars in space as a function of time , it is easy to predict the time of their periapse , that is the closest approach to sgr a * , and look for possible alignments between one of them and the lens , corresponding to the mbh , along the line of sight of an observer on the earth . in this way , it is possible to foresee at any time where to look for a secondary image , what its apparent magnitude should be , and , thanks to our knowledge of their orbits , easily predict the best time to observe them .
s2 was the first star to be studied as a possible source for gravitational effects due to the central mbh @xcite .
the light curve for its secondary image , as well as the first two relativistic images , were calculated in the schwarzschild black hole hypothesis .
the analysis was then enlarged by @xcite , who achieved the light curves for the secondary images of five more s - stars .
an important fact that emerges from the latter study is that the light curves are peaked around the periapse epoch , but two subpeaks may arise in nearly edge - on orbits , when the source is behind sgr a * ( _ standard lensing _ ) or in front of it ( _ retrolensing _ ) . in this respect
, the star s14 represents an emblematic case . in the present paper
we want to extend the previous analysis of @xcite to the new set of 28 s - stars , provided by @xcite .
we will see that , within these stars , s6 turns out to be a particularly intriguing source for gravitational lensing , since its secondary image is potentially observable by incoming interferometric instruments .
our aim is to calculate the position and the luminosity of the secondary images of the 28 s - stars as functions of time , generated by the mbh at the center of our galaxy .
the secondary image appears on the opposite side of the black hole with respect to the source and is thus formed by light rays passing behind the mbh . defining the deflection angle @xmath4 as the angle between the asymptotic directions of motion of the light rays , before and after the rendezvous with the black hole , the lens equation is @xcite @xmath5 where @xmath6 is the alignment angle of the source , that is the angle between the line connecting the source to the lens and the optical axis , defined as the line connecting the observer to the lens .
the geometry of the lensing configuration is such that @xmath6 runs from 0 ( perfect alignment ) to @xmath7 ( perfect anti - alignment ) .
@xmath8 is the impact angle for the light rays as seen from the observer , @xmath9 and @xmath10 are the distances between the observer and the lens , and the lens and the source , respectively .
@xmath9 is given by the distance of the sun to the galactic center , which is currently estimated to be @xmath11 kpc @xcite .
following @xcite , we obtain @xmath10 and @xmath6 as functions of time by an accurate modelling of the keplerian orbit of the s - stars around the mbh , taking into account the measured values of the orbital parameters .
for all the details , the reader is referred to @xcite . as explained in @xcite , the lens equation ( [ lens eq ] )
is the closest to the exact treatment of deflection by a spherically symmetric gravitational field .
the relative error in the position of the images committed by using eq . ( [ lens eq ] ) is at most of the order of @xmath12 , where @xmath13 is the gravitational radius of the black hole .
this number stays lower than @xmath14 in the worst cases .
the deflection angle is calculated assuming that the black hole is spherically symmetric , thus being described by a schwarzschild metric . since the secondary image of
an s - star demands a deflection angle in the whole range @xmath15 $ ] , we can not adopt the weak - field or the strong deflection limit approximation , but we have to calculate the exact deflection angle in terms of elliptic integrals @xcite .
finally , the magnification of a secondary image is given by the general formula @xcite @xmath16 where @xmath17 is the distance between the observer and the source .
in the nir band , the sgr a * stellar cluster has been originally observed with the eso s ntt starting from 1992 , then with the keck from 1995 , and with the vlt from 2002 . the good synergy between the adaptive optics and the 8m/10m - class telescopes @xcite allowed to push the astronomical imaging and the spectroscopy for point sources to a deep sensitivity , that is @xmath18 at an angular resolution of @xmath19 mas @xcite . thanks to these powerful instruments of astronomical investigation , it was possible to monitor the s - stars for more than 16 years and reconstruct for some of them their keplerian orbits around the galactic mbh @xcite . the orbital parameters of these 28 stars are summarized in table [ tbl-1 ] . following the steps described in section [ sec lens ] , we have calculated the light curves of the secondary images of the mentioned 28 stars .
the main features of these images are presented in table [ tbl-2 ] , allowing an immediate reading of the results . besides the values of the minimum magnitude for the secondary images and the epochs at which they peak , we also report the time spent by the secondary images when their luminosity is greater than half of the maximum , the alignment angle and the distance from the mbh of the source stars at the peak , and the maximal angular distance of the secondary images from the apparent shadow of the black hole ( whose angular radius is @xmath20 @xcite ) . among all the s - stars analyzed ,
we concentrate our attention on three particular cases : s6 , s17 , and s27 , characterized by a value of the orbit inclination very close to @xmath21 .
this feature qualifies them as very appealing candidates for gravitational lensing by the mbh . in figure [ fig 1 ] we report the light curves for the secondary images of the three mentioned stars ( left column ) and the alignment angles as functions of time ( right column ) .
s6 this star is indeed the most interesting one , since its orbit is nearly edge - on ( @xmath22 ) , and its eccentricity brings it quite close to the black hole ( up to 4800 @xmath23 ) .
the best alignment and anti - alignment times are very close to the periapse epoch and are responsible for two subpeaks , see figure [ fig 1]a . by examining the figure
[ fig 1]b , we notice that first we have the alignment peak ( @xmath24 ) , and then the anti - alignment peak ( @xmath25 ) .
what makes s6 particularly interesting is the high brightness attained by the secondary image . at the best alignment time , that will occur in 2062 , the secondary image has a magnitude @xmath0 , and the maximal angular distance from the apparent shadow of the mbh is about @xmath26 mas , according to the current estimates of the orbital parameters .
s17 the orbit of this star is nearly edge - on as well ( @xmath27 ) . yet , contrarily to the previous case , the eccentricity of s17 is very low and the periapse occurs in proximity of one of the nodes . for this reason , the alignment and anti - alignment peaks occur at opposite epochs in the period and neither of them is particularly bright ( figures [ fig 1]c and [ fig 1]d ) .
s27 from a gravitational - lensing point of view , the inclination of the orbit of this star is the most favorable among all the known s - stars ( @xmath28 ) .
moreover , the eccentricity of the orbit is very close to one .
however , the periapse occurs in correspondence to the anti - alignment epoch ( compare figure [ fig 1]e with figure [ fig 1]f ) . by reflex
, the anti - alignment peak is very tight and hardly visible in figure [ fig 1]e because s27 moves very fast at that time .
conversely , the alignment peak is not so bright ( @xmath29 ) as in the case of s6 , but anyway better than s14 and s17 .
it is interesting to note that the angular separation of the secondary image of s27 reaches almost 0.4 mas at the best alignment time , which puts this star in a very good position for possible observations .
thanks to the great technological advances earned by the modern big - class telescopes , it has been possible to have access to very remote zones of the milky way , like the neighborhood of its center , where the radio source sgr a * lies .
the discovery that most of the detected stars in this region move along tight keplerian orbits is a strong indication of the existence of a very compact object of more than @xmath30 m@xmath3 , currently recognized as a massive black hole , in the geometrical center of the milky way .
this great disclosure opens an amazing way of escape from the meanders of theoretical speculations and allows us to study gravitational lensing by a black hole on a real case and test general relativity in its strong field regime .
the idea of using the star s2 as a potential source to be lensed by the central mbh , was conceived originally by @xcite , and then acquired and rigorously developed by @xcite , who considered other s - stars as further possible sources for this phenomenology . in this paper , we extend the previous analysis to a new set of 28 s - stars provided by @xcite .
whereas most of these stars have undetectable secondary images , there are some cases in which the lensed images have a significant probability of being detected by incoming instruments , as can be deduced from table [ tbl-2 ] . among these , s6 represents the best case since its secondary image at the peak reaches @xmath0 , with an angular separation of 0.3 mas from the central black hole . actually , the perspectives for observing such an event are not so far from what the modern technology is developing . in fact , in 2013 the new second - generation vlti instrument , called gravity , will be scientifically operative at the eso paranal observational site @xcite .
this instrument , specifically designed to observe highly relativistic motions of matter close to the event horizon of sgr a * , will interferometrically combine the nir light collected by the four - units telescope of the vlt , exploiting all the potential of this extremely powerful eso observatory .
the resolution that will be obtained in the nir band is @xmath31 mas for objects that can be as faint as @xmath32 .
moreover , in its astrometric mode , gravity will have an accuracy of @xmath33 @xmath34 , allowing to study the motions of celestial objects to within a few times the event horizon size of the galactic mbh .
there are still many years before the secondary images of the best known s - stars will shine again ( 2047 for s14 , 2062 for s6 and s27 ) .
surely , new s - stars will be discovered and followed along their orbits , eventually providing even better candidates for gravitational lensing . in the meanwhile , we have no doubt that the astronomical community will fortify its stock of observational facilities with new advanced instruments , which will be able to catch this extraordinary kind of gravitational lensing events .
lcccccccc star & @xmath35 $ ] & @xmath36 & @xmath37 $ ] & @xmath38 $ ] & @xmath39 $ ] & @xmath40 $ ] & @xmath41 $ ] & @xmath1 + s1 & 0.508 @xmath42 0.028 & 0.496 @xmath42 0.028 & 120.82 @xmath42 0.46 & 341.61 @xmath42 0.51 & 115.3 @xmath42 2.5 & 2000.95 @xmath42 0.27 & 132 .
@xmath42 11 . &
14.7 + s2 & 0.123 @xmath42 0.001 & 0.88 @xmath42 0.003 & 135.25 @xmath42 0.47 & 225.39 @xmath42 0.84 & 63.56 @xmath42 0.84 & 2002.32 @xmath42 0.01 & 15.8 @xmath42 0.11 & 14 .
+ s4 & 0.298 @xmath42 0.019 & 0.406 @xmath42 0.022 & 77.83 @xmath42 0.32 & 258.11 @xmath42 0.3 & 316.4 @xmath42 2.9 & 1974.4 @xmath42 1 . &
59.5 @xmath42 2.6 & 14.4 + s5 & 0.25 @xmath42 0.042 & 0.842 @xmath42 0.017 & 143.7 @xmath42 4.7 & 109 .
@xmath42 10 . &
236.3 @xmath42 8.2 & 1983.6 @xmath42 2.5 & 45.7 @xmath42 6.9 & 15.2 + s6 & 0.436 @xmath42 0.153 & 0.886 @xmath42 0.026 & 86.44 @xmath42 0.59 & 83.46 @xmath42 0.69 & 129.5 @xmath42 3.1 & 2063 .
@xmath42 21 . & 105 .
@xmath42 34 . & 15.4 + s8 & 0.411 @xmath42 0.004 & 0.824 @xmath42 0.014 & 74.01 @xmath42 0.73 & 315.9
@xmath42 0.5 & 345.2 @xmath42 1.1 & 1983.8 @xmath42 0.4 & 96.1 @xmath42 1.6 & 14.5 + s9 & 0.293 @xmath42 0.052 & 0.825 @xmath42 0.02 & 81 .
@xmath42 0.7 & 147.58 @xmath42 0.44 & 225.2 @xmath42 2.3 & 1987.8 @xmath42 2.1 & 58 .
@xmath42 9.5 & 15.1 + s12 & 0.308 @xmath42 0.008 & 0.9 @xmath42 0.003 & 31.61 @xmath42 0.76 & 240.4 @xmath42 4.6 & 308.8 @xmath42 3.8 & 1995.63 @xmath42 0.03 & 62.5 @xmath42 2.3 & 15.5 + s13 & 0.297 @xmath42 0.012 & 0.49 @xmath42 0.023 & 25.5 @xmath42 1.6 & 73.1 @xmath42 4.1 & 248.2 @xmath42 5.4 & 2004.9 @xmath42 0.09 & 59.2 @xmath42 3.8 & 15.8 + s14 & 0.256 @xmath42 0.01 & 0.963 @xmath42 0.006 & 99.4 @xmath42 1 . &
227.74 @xmath42 0.7 & 339 .
@xmath42 1.6 & 2000.07 @xmath42 0.06 & 47.3 @xmath42 2.9 & 15.7 + s17 & 0.311 @xmath42 0.004 & 0.364 @xmath42 0.015 & 96.44 @xmath42 0.18 & 188.06 @xmath42 0.32 & 319.45 @xmath42 3.2 & 1992 .
@xmath42 0.3 & 63.2 @xmath42 2 . &
15.3 + s18 & 0.265 @xmath42 0.08 & 0.759 @xmath42 0.052 & 116 .
@xmath42 2.7 & 215.2 @xmath42 3.6 & 151.7 @xmath42 2.9 & 1996 .
@xmath42 0.9 & 50 .
@xmath42 16 . & 16.7 + s19 & 0.798 @xmath42 0.064 & 0.844 @xmath42 0.062 & 73.58 @xmath42 0.61 & 342.9 @xmath42 1.2 & 153.3 @xmath42 3 .
& 2005.1 @xmath42 0.22 & 260 .
@xmath42 31 . & 16 .
+ s21 & 0.213 @xmath42 0.041 & 0.784 @xmath42 0.028 & 54.8 @xmath42 2.7 & 252.7 @xmath42 4.2 & 182.6 @xmath42 8.2 & 2028.1 @xmath42 5.5 & 35.8 @xmath42 6.9 & 16.9 + s24 & 1.06 @xmath42 0.178 & 0.933 @xmath42 0.01 & 106.3 @xmath42 0.93 & 4.2 @xmath42 1.3 & 291.5 @xmath42 1.5 & 2024.9 @xmath42 5.5 & 398 .
@xmath42 73 . & 15.6 + s27 & 0.454 @xmath42 0.078 & 0.952 @xmath42 0.006 & 92.91 @xmath42 0.73 & 191.9 @xmath42 0.92 & 308.2 @xmath42 1.8 & 2059.7 @xmath42 9.9 & 112 .
@xmath42 18 . & 15.6 + s29 & 0.397 @xmath42 0.335 & 0.916 @xmath42 0.048 & 122 .
@xmath42 11 . & 157.2 @xmath42 2.5 & 343.3 @xmath42 5.7 & 2021 .
@xmath42 18 . & 91 .
@xmath42 79 . & 16.7 + s31 & 0.298 @xmath42 0.044 & 0.934 @xmath42 0.007 & 153.8 @xmath42 5.8 & 103 .
@xmath42 11 . & 314 .
@xmath42 10 .
& 2013.8 @xmath42 2.2 & 59.4 @xmath42 9.2 & 15.7 + s33 & 0.41 @xmath42 0.088 & 0.731 @xmath42 0.039 & 42.9 @xmath42 4.5 & 82.9 @xmath42 5.9 & 328.1 @xmath42 4.5 & 1967.9 @xmath42 6.5 & 96 .
@xmath42 21 . & 16 .
+ s38 & 0.139 @xmath42 0.041 & 0.802 @xmath42 0.041 & 166 .
@xmath42 22 . & 286 .
@xmath42 68 . & 203 .
@xmath42 68 . & 2003 .
@xmath42 0.2 & 18.9 @xmath42 5.8 & 17 .
+ s66 & 1.21 @xmath42 0.126 & 0.178 @xmath42 0.039 & 135.4 @xmath42 2.6 & 96.8 @xmath42 2.9 & 106 .
@xmath42 6.3 & 1782 .
@xmath42 23 . & 486 .
@xmath42 41 . & 14.8 + s67 & 1.1 @xmath42 0.102 & 0.368 @xmath42 0.041 & 139.9 @xmath42 2.3 & 106 .
@xmath42 6.1 & 215.2 @xmath42 4.8 & 1695 .
@xmath42 16 . & 419 .
@xmath42 19 . & 12.1 + s71 & 1.06 @xmath42 0.765 & 0.844 @xmath42 0.075 & 76.3 @xmath42 3.6 & 34.6 @xmath42 1.5 & 331.4 @xmath42 7.1 & 1646 .
@xmath42 251 . & 399 .
@xmath42 283 . & 16.1 + s83 & 2.79 @xmath42 0.234 & 0.657 @xmath42 0.096 & 123.8 @xmath42 1.3 & 73.6 @xmath42 2.1 & 197.2 @xmath42 3.5 & 2061 .
@xmath42 25 . & 1700 .
@xmath42 205 . & 13.6 + s87 & 1.26 @xmath42 0.161 & 0.423 @xmath42 0.036 & 142.7 @xmath42 4.4 & 109.9 @xmath42 2.9 & 41.5 @xmath42 3.7 & 1647 .
@xmath42 38 . & 516 .
@xmath42 44 .
& 13.6 + s96 & 1.55 @xmath42 0.209 & 0.131 @xmath42 0.054 & 126.8 @xmath42 2.4 & 115.78 @xmath42 1.93 & 231 .
@xmath42 9 . & 1624 .
@xmath42 34 . & 701 .
@xmath42 81 . & 10 .
+ s97 & 2.19 @xmath42 0.844 & 0.302 @xmath42 0.308 & 114.6 @xmath42 5 . &
107.72 @xmath42 3.15 & 38 .
@xmath42 52 . & 2175 .
@xmath42 88 . &
@xmath42 688 . & 10.3 + lcccccc star & @xmath43 $ ] & @xmath44 $ ] & @xmath45 & @xmath46 $ ] & @xmath47 $ ] & @xmath48 $ ] + s1 & 2130.34 & 4.9 & 32.2 & 54 & 30.8 & 2195 + s2 & 2018.15 & 0.095 & 26.8 & 43 & 45.4 & 127 + s4 & 2048.94 & 3.9 & 28.7 & 112 & 12.2 & 2868 + s5 & 2028.8 & 2.7 & 32.9 & 32 & 83.6 & 456 + s6 & 2062.65 & 0.051 & 20.8 & 316 & 3.56 & 464 + s8 & 2082.7 & 1.6 & 28.3 & 88 & 16.2 & 1335 + s9 & 2041.39 & 1.9 & 27.1 & 144 & 9.06 & 1833 + s12 & 2058.38 & 1.4 & 32.6 & 32 & 88.5 & 319 + s13 & 2060.3 & 34 & 36.4 & 31 & 90.8 & 1571 + s14 & 2047.55 & 0.073 & 23.5 & 136 & 9.46 & 230 + s17 & 2008.24 & 2.1 & 26.9 & 196 & 6.44 & 2942 + s18 & 2045.15 & 0.72 & 30.9 & 61 & 26.2 & 670 + s19 & 2262.63 & 1.3 & 29.9 & 87 & 16.5 & 1369 + s21 & 2027.23 & 1.1 & 32.3 & 49 & 36.5 & 617 + s24 & 2047.95 & 44 & 33.1 & 81 & 17.9 & 6225 + s27 & 2062.08 & 0.55 & 22.4 & 399 & 2.92 & 1397 + s29 & 2021.73 & 0.98 & 31.6 & 51 & 34.1 & 572 + s31 & 2013.9 & 0.56 & 31.7 & 31 & 91.2 & 189 + s33 & 2067.95 & 9.1 & 34.9 & 39 & 53.7 & 1637 + s38 & 2021.8 & 0.71 & 33.5 & 32 & 88.9 & 242 + s66 & 2254.09 & 66 & 36.7 & 43 & 45.4 & 8328 + s67 & 2035.14 & 116 & 34.6 & 40 & 51.5 & 9053 + s71 & 2059.51 & 8.3 & 31.6 & 100 & 13.9 & 4073 + s83 & 3647.18 & 129 & 35.6 & 50 & 35 . &
14435 + s87 & 2185.1 & 52 & 35.5 & 39 & 53.2 & 6508 + s96 & 2073.74 & 131 & 32.2 & 48 & 36.8 & 14049 + s97 & 2265.5 & 68 & 30.9 & 64 & 24.6 & 13897 + we acknowledge support for this work by miur through prin 2006 protocol 2006023491_003 , by research funds of agenzia spaziale italiana , by funds of regione campania , l.r .
n.5/2002 , year 2005 ( run by gaetano scarpetta ) , and by research funds of salerno university .
bozza , v. , & mancini , l. 2004 , , 611 , 1045 bozza , v. , & mancini , l. 2005 , , 627 , 790 clnet , y. , rouan , d. , gratadour , d. , et al .
2005 , , 439 , l9 darwin , c. 1959 , proc . of the royal soc . of london , 249 , 180 de paolis , f. , geralico , a. , ingrosso , g. , & nucita , a. a. 2003 , , 409 , 809 eckart , a. , genzel , r. , ott , t. , & schdel , r. 2002 , , 331 , 917 eckart , a. , baganoff , f. k. , schdel , r. et al .
2006 , , 450 , 535 eckart , a. , schdel , r. , garcia - marin m. , et al .
2008 , in press with , arxiv:0811.2753 eisenhauer , f. , genzel , r. , alexander , t. , et al .
2005 , , 628 , 246 eisenhauer , f. , perrin , g. , brandner , w. , et al .
2008 , to appear in the conference proceedings of spie astronomical instrumentation genzel , r. , schdel , r. , ott , t. , et al .
2003 , , 425 , 934 ghez , a. m. , morris , m. r. , becklin , e. e. , et al .
2001 , aspc , 228 , 309 ghez , a. m. , wright , s. a. , matthews , k. , et al .
2004 , , 601 , l159 ghez , a. m. , salim , s. , hornstein , s. d. , et al .
2005 , , 620 , 744 gillessen , s. , eisenhauer , f. , trippe , s. , et al .
2008 , in press with , arxiv:0810.4674 hamaus , n. , paumard , t. , mller , t. , et al .
2008 , in press with , arxiv:0810.4947 marrone , d. p. , baganoff , f. k. , morris , m. r. , et al .
2008 , , 682 , 373 ohanian , h. c. 1987 , am .
j. phys . , 55 , 428 reid , m. j. , menten , k. m. , trippe , s. et al .
2007 , , 659 , 378 schdel , r. , ott , t. , genzel , r. , et al .
2002 , , 419 , 694 schdel , r. , ott , t. , genzel , r. , et al .
2003 , , 596 , 1015 yusef - zadeh , f. , wardle , m. , heinke , c. , et al .
2008 , , 682 , 361 | the existence of a massive black hole in the center of the milky way , coinciding with the radio source sgr a * , is being established on more and more solid ground . in principle , this black hole , acting as a gravitational lens , is able to bend the light emitted by stars moving within its neighborhood , eventually generating secondary images . extending a previous analysis of the gravitational lensing phenomenology to a new set of 28 stars , whose
orbits have been well determined by recent observations , we have calculated all the properties of their secondary images , including time and magnitude of their luminosity peaks and their angular distances from the central black hole . the best lensing candidate is represented by the star s6 , since the magnitude of its secondary image at the peak reaches @xmath0 , with an angular separation of 0.3 mas from the central black hole , that is just at the borders of the resolution limit in the @xmath1 band of incoming astronomical instruments . |
the first two molecular states @xmath0 and @xmath1 are the isosinglet and the isotriplet states of the @xmath7@xmath8 bound system@xcite .
this binding requires a quark - spin hyperfine interaction in the over all @xmath9 @xmath9 @xmath10 @xmath10 system .
we will see that this binding is different from the particle exchange mechanisms that bind the rest of molecules of this report .
the exchanges of ref.@xcite that bind the @xmath7@xmath8 system are quark exchanges where the quark - spin hyperfine interaction leads to an attractive potential .
this attractive potential can only make states if the mesons of the fall apart mode @xmath9 @xmath10 - @xmath9 @xmath10 are below threshold .
thus we have states of @xmath7@xmath8 lying just below threshold in scalar channel(@xmath11 ) .
this is somewhat like the deuteron in the @xmath12 @xmath13 system .
the work of ref.@xcite has one flaw with regards to long - range color van der waals - type forces@xcite .
it has been pointed out that confining potentials of the type used in ref.@xcite leads to a long - range power - law residual potential between color singlets of @xmath14 between two mesons .
ref.@xcite calculates the potential between mesons to be given by @xmath15 this is to be compared with the coulomb force between charged mesons @xmath16 the r in equation 1 and equation 2 is given in fm .
let us compare the potential strength between them by setting them equal @xmath17 this occurs at the distance of 13.3 fm which is a very large distance .
the size of the scalar bound states are much smaller 2 fm at best .
this long range van der waals force implies that gluons are massless(like the photon ) and can travel to the edge of the universe in a virtual state(like photons in em - fields ) .
gluons can only try to travel to the edge if they are in a color singlet state ( glueballs or glueloops ) .
this would lead to an exponential cutoff @xmath18 where @xmath19 is the glueball mass .
sine mesons are much lighter than glueballs , and the pion is the lightest , it is the longest carrier of the strong force .
meson exchanges will be the binding force of the other hadron molecules of this report .
the report is organized in the following manner : sec .
1 is the introduction to @xmath0 and @xmath1 .
2 looks at particle exchange calculations and the generation of @xmath2 .
3 consider a dibaryon state @xmath3 and the similarity to @xmath2 .
this similarity predicts the @xmath6 state .
4 an exotic meson state @xmath20 which is seen in @xmath21 @xmath22 p - wave scattering @xmath4 is explained .
5 predicts another molecular state @xmath5 which should be found .
6 is the summary and discussion .
in order to bind @xmath7@xmath8@xmath22 together , we need to develop a dynamical theory that uses particle exchange mechanism not the quark exchange that led to the van der walls forces@xcite .
it is standard to break up the scattering of a three particle system into a sum of isobar spectator scatterings@xcite . to complete this task
we develop a unitary isobar model which has long - range particle exchange forces .
we will assume that the only interaction among the three particles are isobars decaying into two particles where one of these particle exchange with the spectator forming another isobar .
this one - particle - exchange(ope ) occurs in the @xmath23@xmath8 , @xmath24@xmath7 and @xmath25@xmath22 isobar systems .
see figure 1(a ) for the ope mechanism(note @xmath25 is the @xmath26 isobar which is its older name ) .
we choose as our dynamical framework the blankenbecler - sugar formalism@xcite which yields set of coupled integral equations for amplitudes @xmath27 ) , @xmath28 ) , @xmath29 ) , @xmath30 ) , @xmath31 ) , @xmath32 ) , @xmath33 ) , @xmath34 ) , and @xmath35 ) .
these amplitudes describe the isobar quasi - two - body processes @xmath36 , @xmath37 , @xmath38 , etc . , whose solution are lorentz invariant , and satisfy two- and three - body unitarity , and cluster properties . in operator formalism these equations have the structure(a schematic representation is show in figure 1(b ) ) , @xmath39 in equation 5 , @xmath40 is the overall c.m .
energy of the three - particle system .
the index @xmath41 which is summed over represent each isobar .
integration is over solid angles where each total angular momentum is projected out as an individual set of coupled equations .
momentum of the spectator which has the same magnitude as the isobar is integrated from zero to @xmath42 .
thus all effective masses of isobars are probed starting at the kinematic limit down to negative @xmath42 .
all details are given in ref.@xcite .
@xmath43 is the propagator of isobar @xmath41 and @xmath40 determines the kinematic limit where c.m .
momentum of the spectator and the isobar is zero . in figure 2
we show the breit - wigner shape of the @xmath23 propagator @xmath44 as a function of isobar mass(@xmath45 ) . in figure 3
we show the breit - wigner shape of the @xmath25 propagator @xmath46 as a function of isobar mass(@xmath47 ) .
equation 5 can be rewriting as @xmath48 this fredholm integral equation leads to a fredholm determinant as a function of @xmath40 for each partial wave or total @xmath49 projection .
we have solved this fredholm determinant for two @xmath50 states @xmath51 and @xmath52 .
the results of this analysis is shown in figure 4 .
we see no binding effect in the @xmath51 determinant , while in the @xmath52 channel there is a large effect around @xmath53 gev . at the energy of @xmath53 gev the @xmath23 and @xmath24 the peaks of figure 2 are just coming into play . since for @xmath52
they are in a s - wave they have maximum effect . in the @xmath51
these peak are suppressed by a p - wave barrier . around the mass @xmath53 gev
one can form a picture of the system being @xmath7@xmath8(@xmath25 ) molecule at the center of gravity with a light pion revolving in a p - wave orbit .
the momentum of the pion is such that at each half - revolution a @xmath23 or a @xmath24 is formed(see figure 5 ) .
the phase shift and production cross sections of this molecular state is explored in detail in ref.@xcite .
the dibaryon state interacts in three two - body scattering channels .
its mass is 2.15 gev and has a strong interaction resonance decay width of 100 mev .
it interacts in the @xmath54@xmath54 d - wave spin anti - aligned@xcite , @xmath55@xmath22 p - wave spin aligned@xcite , and @xmath56@xmath54 s - wave spin aligned@xcite .
the dibaryon system mainly resonates in the s - wave @xmath56@xmath54 mode with a pion rotating in a p - wave about a spin aligned @xmath54@xmath54 system which forms a isospin singlet .
the pion moves back and forth forming @xmath56 states with one nucleon and then the other(see figure 6 ) .
all three isospin states of the pion can be achieved in this resonance .
thus we can have @xmath57@xmath55 , @xmath58@xmath55 , and @xmath59@xmath55 states . if the pion is absorbed by any of the nucleons it under goes a spin flip producing a d - wave @xmath54@xmath54 system .
the resonance decays into @xmath54@xmath54 , @xmath22@xmath55 , or @xmath22@xmath54@xmath54 . in the last section we saw a meson system that had an analogous orbiting pion in a p - wave mode about a @xmath47 in a s - wave@xcite .
both systems have a similar lifetime or width of @xmath60 .100 gev@xcite . unlike the @xmath2 the @xmath3 has an isosinglet at the center of motion
. the @xmath7@xmath8 isosinglet state of sec .
1 could form the center of motion for an isotriplet molecular state @xmath6 .
the set of integral equation would be the same as in the @xmath2 case making a similar fredholm determinant . like for @xmath3 which had a @xmath61 decay mode
, one would expect that there would be a @xmath0 @xmath22 decay mode .
we can calculate the branching ration of @xmath2 to @xmath25@xmath22 from the dalitz plot calculated using equation 20 of ref.@xcite .
the ratio in the plot going into @xmath25@xmath22 is 22% .
the reason this mode is so small is because @xmath62 is much smaller than @xmath63@xcite . where as the ratios of @xmath64 and @xmath65 are one at resonance(see figure 2 and 3 ) . for the @xmath3 the @xmath61 branching
ratio is 25%@xcite .
we should expect that the branching of @xmath66 should be the same as @xmath67 .
suh - urk chung has claimed such a state has been observed@xcite .
in sec . 2 we explained the @xmath68(1420 ) seen in @xmath69@xcite . following the same approach we can demonstrate the possibility that the @xmath70(1400 ) is a @xmath71 molecule , where the @xmath69 in a relative s - wave with the other @xmath22 orbiting them in a p - wave . since the @xmath69 is resonating as the @xmath21(1295 ) , it is possible that the offshell @xmath72 would couple to the ground state @xmath21 , thus creating a @xmath73 p - wave decay mode . as was done in ref.@xcite , we need to arrange a set of born terms connecting all of the possible intermediate isobar states of the @xmath71 system ( @xmath21(1295)@xmath22 , @xmath25(980)@xmath74(770 ) , @xmath75(1270)@xmath8 or @xmath76@xmath7 ) .
we assume that the only interaction among the particles occurs through one - particle exchange ( ope ) , thus connecting the above isobar states ( figure 7 ) .
in order to completely derive the dynamics one would have to develop a true four - body scattering mechanism with ope born terms connecting two- and three - body isobar states .
we can take a short cut and use the three - body formalism developed in ref .
@xcite , if we note that the set of diagrams ( figure 8) could be summed using a true four - body formalism , and be replaced by the born term of figure 9 . here the @xmath25(980 ) is treated as a stable particle and the @xmath77 p - wave phase shift ( @xmath78 ) is assumed to be modified by the sum of terms in figure 8 . with this assumption ,
then binding can occur if we use the @xmath79 propagators for the @xmath21(1295)(see figure 10 ) and @xmath78(see figure 11 ) . in figure 11
we also show the unaltered p - wave phase shift ( @xmath74 ) .
figure 12 shows the final state enhancement times the @xmath21(1295 ) @xmath22 p - wave kinematics .
the bump is driven by the collision on the dalitz plot of the @xmath21(1295 ) breit - wigner ( figure 10 ) and the rapid increase of the @xmath77 p - wave phase shift ( figure 11 ) .
we have suggested the possibility that the @xmath70(1400 ) is a final state interaction for the @xmath80 system in a s - wave orbiting by a @xmath22 in a p - wave .
the @xmath73 decay mode is generated by the off shell appearance of the @xmath21 from the @xmath80 system ( @xmath51 ) .
our model thus predicts that a strong @xmath81 should be seen in the @xmath82 system at around 1.4 gev / c@xmath83 .
if the @xmath70(1400 ) is only seen in the @xmath73 channel then its hard to understand three facts about its production .
first , that the force between the @xmath21 and @xmath22 in a p - wave should be repulsive ( qcd ) @xcite .
this is not a problem if the @xmath73 is a minor decay mode .
second , why should the production be so small compared to the @xmath84 which has only a 14% branching to @xmath73
? one would think it should be produced in unnatural parity exchange not natural .
again this is not a problem if minor decay mode .
finally , it is reasonable to think that the largest decay amplitude would be the modes that have an @xmath25(980 ) in the final state .
however in ref.@xcite the same conclusion was initially drawn , except when one puts in all the numerical factors the @xmath25 modes become suppressed .
the explanation comes from the very powerful attraction of the kaons in the @xmath25 mode .
the isobar decay amplitude is proportional to @xmath85 both @xmath54 and @xmath86 are large numbers while the ratio is near one at the threshold(see last section and ref.@xcite ) .
thus the decay amplitude becomes proportional to @xmath87 .
we predict that the major mode could be @xmath88 p - wave having no @xmath74 peak ( work above ) forming a @xmath89 or a @xmath90 plus a @xmath8 or @xmath7 with overall @xmath91-parity minus .
the @xmath89 should more or less be a phase space distribution .
we saw in sec.1 and sec .
2 that the @xmath7@xmath8 system had attraction through the @xmath1 resonance . it seems only natural to investigate the possibility that a three - k molecule might exist .
this is only worthwhile if we consider only exotic quantum numbers .
the only exotic quantum number which can be be obtained is the isotopic spin .
thus a set of coupled equations for the @xmath7@xmath8@xmath7 system in a overall s - wave with isotopic spin of @xmath92 is created@xcite .
the fredholm determinate squared times of the equations is shown in figure 13 .
isopin spin @xmath92 implies there are four states @xmath93@xmath94@xmath93,@xmath93@xmath95@xmath93 , @xmath96@xmath94@xmath96 , and @xmath96@xmath95@xmath96 .
the @xmath93@xmath94@xmath93 is double charged .
there would also be a @xmath95@xmath96@xmath95 which is the anti - matter state of the @xmath93@xmath94@xmath93 .
these states are unique to this type of binding mechanism .
in this report we have discussed seven possible hadron molecular states .
these states are particles made out of hadrons that are held together by self interactions .
the seven molecules and their self interactions are explored .
the @xmath0 , @xmath1 relied quark exchange forces which made states of @xmath7@xmath8 lying just below threshold in scalar channel(@xmath11 ) .
this is somewhat like the deuteron in the @xmath12 @xmath13 system .
the @xmath2 , @xmath3 and @xmath4 molecular structure are held together by long range particle exchange mechanisms not the quark exchange that led to the van der walls forces@xcite .
these exchange mechanisms also predicts that two more states the @xmath5 and @xmath6 should be found . for the @xmath6
the set of integral equation would be the same as in the @xmath2 case making a similar fredholm determinant . like for @xmath3 which had a @xmath61 decay mode
, one would expect that there would be a @xmath0 @xmath22 decay mode .
dr . suh - urk chung has claimed such a state has been observed@xcite .
this research was supported by the u.s .
department of energy under contract no .
de - ac02 - 98ch10886 .
99 j. weinstein and n. isgur , phys . rev.lett .
48 ( 1982 ) 659 ; phys .
d 27 ( 1983 ) 588 ; phys .
d 41 ( 1990 ) 2236 . o.w . greenburg and h.j .
lipkin , nucl .
a370 ( 1981 ) 349 .
herndon , phys .
d 11 ( 1975 ) 3165 .
r. blankenbecler and r. sugar , phys .
142 ( 1966 ) 1051 .
r. longacre , phys .
d 42 ( 1990 ) 874 . r.a .
arndt _ et al .
c 76 ( 2007 ) 025209 .
oh _ et al .
c 56 ( 1997 ) 635 .
d. schiff and j. tran thanh van , nucl .
b5 ( 1968 ) 529 .
r. longacre , arxiv:1311.3609[hep - ph ] .
suh - urk chung(private communication ) .
t. barnes(private communication ) . | hadron molecules are particles made out of hadrons that are held together by self interactions . in this report
we discuss seven such molecules and their self interactions .
the @xmath0 , @xmath1 , @xmath2 , @xmath3 and @xmath4 molecular structure is given .
we predict that two more states the @xmath5 and @xmath6 should be found .
= 0.4 cm = 0.0 in = 0.0 in = 9.0 in = 6.5 in = 0.1 in + |
the transient turn - on dynamics of lasers is a topic which has received a considerable amount of attention over the years ( cf .
, e.g. , @xcite ) , not only for the fundamental understanding of its evolution but also , and especially , for the importance that it represents in data encoding in telecommunications with directly - modulated semiconductor lasers . while multiple quantum well @xcite ( or even quantum dot @xcite ) lasers have been proven to possess a peculiar modal dynamics
modelled with the help of mutual nonlinear coupling and noise @xcite , with a complex ginzburg - landau approach @xcite or based on multiscale analysis @xcite where the resulting model provides proof for a true phase instability @xcite , inexpensive , edge - emitting devices show a dynamics @xcite which is more appropriately characterized by cooperation , rather than competition @xcite . for these latter devices ,
we have recently shown @xcite that a realistic model for a multimode semiconductor laser , with experimentally matched parameters @xcite , predicts a slow , hidden dynamical mode - evolution governed by a master mode .
an overall agreement exists between our recent predictions and a wealth of experimental data ( e.g. @xcite ) since it has been time and again shown that a gradual line narrowing exists in the progress of the multimode laser transient ( cf . also previous theoretical calculations , e.g. , @xcite ) .
however , no specific experiments seem to have been carried out to quantify the physical features of the multimode dynamics in edge - emitting semiconductor lasers ( beyond its more technical aspects ) , and the physical origin of our predictions concerning the slow components of the dynamics remain so far unexplained . in this paper
, we analyze the deterministic dynamics of the multimode laser transient and determine the modal intensity distribution at the end of the fast transient starting from a below - threshold condition and leading into the slow dynamics @xcite with the characteristic oscillations of a class - b laser @xcite .
the * apparent * _ equilibrium solution _ , taking place at the end of the fast transient , in reality amounts to an inherent _ out - of - equilibrium one _ , when analyzed in terms of modal intensity distribution .
our interest for answering the questions related to this non - equilibrium intensity distribution is of a fundamental nature and extends , beyond strongly multimode semiconductor lasers , to all those other types of laser
e.g. fiber lasers @xcite or solid state lasers @xcite which are characterized by the simultaneous operation , at least in transient , of a large number of longitudinal modes . from the point of view of dynamical systems
, our model can be viewed as an ensemble of globally coupled modes possessing a common , stable attractor .
globally coupled systems have been , and are still , a topic receiving a large amount of attention given their potential for application to various fields ( cf . ,
e.g. , @xcite ) .
multimode lasers and laser arrays have been recognized very early on as interesting physical examples of globally coupled systems ( cf . , e.g. , @xcite ) . at variance with most other investigations , we do not study synchronization or the appearance of collective states , but the transient evolution towards a fixed point ( attractor ) by a collection of modes with unequal degrees of coupling to a global energy source . however , transient dynamics can become part of self - sustained dynamics if the system parameters are modulated on a time scale comparable to ( at least one ) of its internal constants .
the results are non - trivial and show the existence of multiple time scales and of collective dynamics which contribute to a fast evolution ( studied in this paper ) , which precedes the slow , hidden dynamics already presented elsewhere @xcite . the discussion is entirely cast in terms of laser physics , but , making abstraction from the direct physical meaning of the variables i.e. , identifying the carrier number as a global coupling field and the laser modes as oscillators nonlinearly coupled to that ( mean ) field , the results can be transposed to a generic dynamical system consisting of a large number of oscillators each individually coupled to a global field .
the model is briefly discussed in section [ modelsect ] and the transient evolution is analyzed in section [ analysis ] , which is divided into three subsections .
the first two provide a detailed analysis of the below - threshold , deterministic dynamics for the population where an analytical solution can be found for the carrier density ( section [ carrier - density ] ) and for the modal intensity distribution
describable in terms of approximate iterative analytical solutions ( section [ decoupled - modes ] ) .
this detailed analysis paves the way for the central point of the paper : the analysis of the strongly coupled ( multimode ) transient in the oscillatory regime ( section [ strong - coupling ] ) .
this regime , which displays the characteristic damped oscillations both in the carrier number and in the total intensity , connects the below - threshold dynamics with the slow modal one previously reported @xcite .
in particular , its final state corresponding to the disappearance of oscillations in the carrier density and in the total laser intensity is responsible for the ensuing , hidden modal dynamics @xcite .
the paper concludes with the analysis of the characteristics of the frequency spectrum emitted by the laser during the fast transient ( section [ spectra ] ) .
comments , a summary and conclusions are offered in section [ conclusions ] .
we resort to a standard model , where the physical constants are determined by comparing its predictions to experimental results @xcite , for the study of the fast transient response of a semiconductor laser to the sudden switch - on of its pump ( control parameter ) .
the details , numerical values and labeling choices for the simulations we perform have been published in @xcite .
the physical description is based on an ensemble of @xmath0 lasing modes , intensity - coupled to the carrier number @xmath1 : [ model ] @xmath2 i_j + \beta_j b n ( n+p_0 ) , \\
\label{modelb } \frac{dn(t)}{dt } & = \frac{j}{q } - r(n ) - \sum_j \gamma g_j(n ) i_j , \end{aligned}\ ] ] where @xmath3 is the intensity of each longitudinal mode of the electromagnetic ( e.m . )
field @xmath4 , @xmath5 is the number of carriers as a function of time , @xmath6 is the optical confinement factor , @xmath7 is the optical gain for the @xmath8-th lasing mode , @xmath9 is the photon lifetime in the cavity , @xmath10 is the fraction of spontaneous emission coupled in the @xmath8-th lasing mode , @xmath11 is the band - to - band recombination constant , @xmath12 is the intrinsic hole number in the absence of injected current , @xmath13 is the current injected into the active region , @xmath14 is the electron charge , and @xmath15 is the incoherent recombination term ( including radiative and nonradiative recombination ) which represents the global loss terms for the carrier number ( i.e. , the population inversion ) . the bracket multiplying @xmath16 in eq .
( [ modela ] ) represents the global gain ( effective modal gain minus losses ) for the @xmath8-th mode , while @xmath17 $ ] represents the effective mean contribution of the spontaneous emission to each mode .
@xmath18 represents the normalized current ( i.e. carrier number injected into the junction energy provided to the laser ) in eq .
( [ modelb ] ) , and the last term accounts for the global carrier number depletion due to the stimulated emission into all lasing modes .
the optical gain function , @xmath7 , contains information about the carrier number , @xmath19 , necessary to achieve transparency ( i.e. , no absorption : @xmath20 ) : @xmath21 \ , .\end{aligned}\ ] ] the parameters are : @xmath22 differential gain , @xmath23 gain compression factor ( multiplying the total intensity @xmath24 ) , the individual mode wavelength @xmath25 , the wavelength at the peak of the gain curve @xmath26 , and the full - width - at - half - maximum ( fwhm ) of the gain curve itself , @xmath27 .
see @xcite for further details . as in @xcite , we have allowed for @xmath28 modes to take part in the dynamics in order to include in the simulations modes extending over approximately about 1.5 times the fwhm of the gain curve ( cf .
@xcite for additional details ) .
this choice allows us to give a good description of the transient regime
the focus of this paper since the comparatively large contribution of the spontaneous emission to the side modes renders them an important element in the initial phases of the transient and plays a crucial role in the determination of the non - equilibrium distribution at the end of the _ fast transient _ , thereby considerably enlarging the transient power spectrum .
we numerically integrate the model equations , eqs .
( [ model ] ) , in response to a sudden switch of the injected current @xmath13 ( in form of a heaviside function ) , obtaining for the total laser intensity @xmath29 the response shown in fig .
[ time - trace ] ( curve ( b ) ) .
we notice the standard delay at turn - on for @xmath29 , relative to the application of the pump - switch at @xmath30 , followed by the usual relaxation oscillations with rapid convergence towards steady state at @xmath31 .
the transient , is however , composed of the sum of all individual transients and its apparent _ usual _ behaviour i.e. , that of a single - longitudinal mode laser @xcite is far from trivial .
[ time - trace ] also shows a selection of lasing modes .
the top two curves display the total intensity @xmath29 ( b ) and the carrier number @xmath1 ( a ) , while the lower curves represent the temporal evolution of selected modes near line center ( cf .
figure caption ) . in the fast transient ( @xmath32 )
, the individual modes all follow the same oscillatory behaviour , contributing to the global oscillation for @xmath29 ( and @xmath1 ) . however , we remark that @xmath33 for all modes ( even though @xmath34 ) , which represents the starting point for the slow dynamics @xcite .
( curve ( b ) ) , following a sudden switch - on of the injected current ( pump rate ) in form of a heaviside function centered at @xmath35 . initial current value : @xmath36 , final current value @xmath37 .
the threshold current for this laser is @xmath38 .
model with @xmath39 modes .
all other parameters as in @xcite .
all parameter values are kept constant throughout the paper . with these values of @xmath40 and
@xmath41 the intensity peak is reached at @xmath42 , while the steady state is attained at @xmath43 . here and in all subsequent figures _
( unless otherwise specifically noted in the figure caption ) _ the total intensity @xmath29 is divided by @xmath44 , the individual mode intensities by @xmath45 and the carrier number @xmath1 is divided by @xmath46 . curve ( a ) shows the carrier number as a function of time .
the other curves show the individual mode intensities for the central mode ( @xmath47 , ( c ) ) , and side modes close to line center ( blue side ) : @xmath48 ( d ) , @xmath49 ( e ) , @xmath50 ( f ) and @xmath51 ( g ) .
the symmetric modes , relative to the central one , are not shown but are identical .
notice that in the first phases of the transient ( peak ) curves ( c ) and ( d ) are superposed on this scale : they separate only at the second oscillation . ]
the fast transient , leading to the intensity peak and the usual damped oscillations , can be divided into two parts . a first one , where the carrier number @xmath1 can be decoupled from the modal intensities @xmath16 s , and a second one where due to the strong coupling the fully nonlinear system must be retained . in the decoupled regime
, the carrier number plays the role of an energy reservoir , virtually unaffected by the presence of the laser modes ( too weak to have an impact on the carriers ) .
thus , it is possible to find an analytical solution for its evolution starting from an initial condition towards a transient final state defined by the breakdown of this approximation ( section [ carrier - density ] ) . in this regime ,
the modal intensities are decoupled from one another ( section [ decoupled - modes ] ) and evolve under the action of the ( time - dependent ) energy reservoir , @xmath1 , whose behaviour has been analytically obtained with good accuracy in section [ carrier - density ] .
this approximate but accurate treatment allows us to define a threshold " for the multimode laser ( including spontaneous emission ) in analogy with a basic single mode laser model , where spontaneous emission is traditionally neglected @xcite .
the strongly coupled regime is analyzed in detail in section [ strong - coupling ] on the basis of the knowledge acquired in sections [ carrier - density ] and [ decoupled - modes ] and of _ ad hoc _ considerations .
the analysis of the fast transient is best conducted by choosing an initial state with the laser ( almost ) entirely off . a small amount of prebias ( @xmath52 , well below the lasing threshold @xcite which is placed around @xmath53 ) is useful since it provides an average contribution of the spontaneous emission to each mode , without having to wait for the carrier number to grow .
this choice is not restrictive and does not limit the validity of our results .
our analysis holds any time the laser starts from an injection current value below threshold , which corresponds to an initial energy repartition among modes which strongly differs from the final one , above threshold .
if instead the laser is already pumped above threshold and we suddenly change its pump to another value above threshold @xcite , the ensuing dynamics will reflect the slow modal intensity redistribution examined in @xcite .
a good understanding of the initial phases of the laser turn - on requires a description , even though approximate , of its evolution away from noise .
a good deal of work has been dedicated to an analytical or semi - analytical description of this question @xcite . in the following
, we briefly outline the ( standard ) derivation of an approximate analytical solution for ease of comparison with the full , numerical integration of the model .
the quantitative comparison will show that the analytical approximation provides excellent results even in the initial stages of the transient when the e.m .
field intensity reaches relatively high values , i.e. , exactly where one would expect the approximate solution to already fail .
the contribution of the spontaneous emission ( intrinsic noise ) to each mode is quite small . as a first approximation
, we can therefore consider that the initial phases of the transient are well described by the set of eqs .
( [ model ] ) without the coupling term between individual modal intensities @xmath16 and carrier number @xmath1 in eq .
( [ modelb ] ) . in this approximation
, we start by integrating the carrier number , eq .
( [ modelb ] ) , by variable separation : @xmath54 where @xmath55 with the constants given in @xcite . formal integration of the left - hand - side ( l.h.s . ) of eq .
( [ genintegral ] ) gives @xcite @xmath56 with the parameter values of the problem , eqs .
( [ coeffa][coeffd ] ) , only one root ( which we will denote @xmath57 ) of @xmath58 is real , while the other two are complex conjugate of each other @xmath59 defining @xmath60 and equivalently for @xmath61 , we expand the denominator of the formal solution ( r.h.s . of eq .
( [ formalsol ] ) ) @xmath62 \pm i [ 6 d \mathfrak{a } \mathfrak{b } + 2 c \mathfrak{b } ] \ , , \\
\label{def - omegas } & \equiv & \omega_r \pm i \omega_i \ , , \\ k & = & 2,3\end{aligned}\ ] ] where the @xmath63 ( @xmath64 ) sign in eqs .
( [ def - poly],[def - omegas ] ) corresponds to @xmath65 ( @xmath66 ) .
thus , the contributions to eq .
( [ formalsol ] ) coming from the two complex roots can be rewritten as @xmath67 with @xmath68 ( cf .
( [ defz],[defzstar ] ) ) . substituting into the full expression
, we finally obtain the solution for eq .
( [ genintegral ] ) : @xmath69 \frac{\omega_r}{|\omega|^2 } + \\ \label{nsolution } & & 2\left [ \operatorname{arg } \left\ { z \right\ } - \operatorname{arg } \left\ { z_i \right\ } \right ] \frac{\omega_i}{|\omega|^2 } \ , .\end{aligned}\ ] ] this relationship implicitly defines the approximate solution for the carrier number and gives the most practical way of numerically representing the analytical solution : treating time as the dependent variable , @xmath70 , the function can be straightforwardly plotted ( exchanging the horizontal and vertical axes allows one to restore the @xmath71 appearance of the function ) .
alternately , simple handling allows for a mathematical expression for @xmath71 which takes the form : @xmath72 where @xmath73 which can be used as an alternative definition of @xmath71 @xcite . and the complete carrier number ( dashed line ) @xmath5 resulting from the integration of the full model , eqs .
( [ model ] ) .
the inset shows the difference between the full solution @xmath5 and the analytical solution @xmath71 until the latter diverges away ( at @xmath74 ) .
notice that even extremely near the intensity peak ( cf .
[ time - trace ] ) the difference between the approximate and full solution remains modest ( @xmath75 ) . ] fig .
[ sols - n ] compares the analytical solution @xmath71 to the computed @xmath5 obtained from the numerical integration of the full system , eqs .
( [ model ] ) .
the agreement between the approximate , analytical solution and the full integration is excellent up until values of time which are quite close to the intensity peak ( occurring at @xmath76 ) , as shown in the inset .
the deviation is initially positive and remains below 0.5% ( in absolute value ) until @xmath77 .
the initial positivity of the relative deviation,@xmath78 , resides in the fact that the full solution requires a somewhat larger initial carrier number to support the different field modes , while in the approximate form no energy is lost to support them .
the switch in sign in the difference between the exact and the approximate value for the carrier number comes from the fact that once the different lasing modes grow sufficiently large , the carrier number saturates , instead of continuing its growth , as in the approximate expression .
the difference curve ( inset ) is numerically computed by integrating the full set of equations , eqs .
( [ model ] ) , and a set equivalent to the analytical solution , obtained by removing from eq .
( [ modelb ] ) the last term @xmath79 @xcite .
notice that at @xmath80 , the carrier number @xmath5 crosses its asymptotic value ( @xmath81 , cf .
[ time - trace ] ) . in a laser model without spontaneous emission such a value would define the laser threshold , separating the decoupled dynamics for the population inversion ( carrier number ) from the strongly coupled regime ( above threshold ) @xcite . here
, this value sets the upper limit to the carrier number for which we can consider its dynamics decoupled from that of the modal intensities ( when growing out of noise ) , thus we can extend the concept of threshold " even in the presence of spontaneous emission .
we begin this section with an overview of the modal dynamics during the transient evolution , regardless of the degree of coupling with the carrier number .
this allows us to get a general picture , which we then refine in the analysis of the decoupled regime ( this section ) and of the strongly coupled one ( next section ) .
the individual mode traces in fig .
[ time - trace ] show that at short times the lateral modes contribute more than their steady state share .
their peak ( cf . , e.g. , curves ( f ) and ( g ) corresponding to modes 5 and 7 places away from line center ) is much higher than the asymptotic intensity towards which they tend , contrary to those modes close to line center ( for the central mode even the peak intensity is lower than its steady state value ) @xcite .
this strongly suggests that a mechanism must exist for these modes to grow , in transient , well beyond their final value , while the very few central modes do not .
indeed , the accumulated intensity of the modes farthest away in the wings can exceed , in transient , that of the central mode , thus highlighting the impact that the side modes have on the initial phases of the dynamics . ,
( cf . text for details ) and ( b ) the sum of all side modes , @xmath82 , save for those near line center ( cf .
text for details ) .
the inset shows the initial phases of the transient and shows that the sum of the far modes ( c ) exceeds the intensity of the central mode ( a ) up until @xmath83 . ] a more complete illustration of this point is given by fig .
[ t113-sum ] , which compares the time - dependent intensity of the central mode curve ( a ) , to the cumulative contribution of the most distant twenty modes ( on each side , i.e. @xmath84 ) curve ( c ) ) , and to that of all modes save for the 17 central ones ( i.e.,@xmath85 curve ( b ) ) .
the inset shows that below threshold @xmath86 ( c ) gives a contribution larger than that of the central mode ( a ) until @xmath87 and that after a small maximum , at @xmath88 , it drops down again .
thus , the influence of the very far modes is of some importance only in the very first phases of the dynamics .
considering , though , that these modes are very far out in the wings and that their asymptotic value is order of magnitudes lower than that of the central mode , their strong influence on the dynamics up to threshold is an indicator of the essential difference between the transient intensity distribution and the asymptotic one ! the picture changes more dramatically if we include the lateral modes , @xmath82 , which carry little energy at steady state ( globally of the order of 3% of the total intensity ) , but which in transient give a cumulative peak larger than the one of the central mode ( curve ( b ) ) .
this contribution is rather striking .
indeed , it displays a faster growth than the central mode and deformed ( incomplete ) oscillations .
the latter are explained by the fact that the effective relaxation frequency is not constant over the whole set of modes ( cf .
@xcite for discussion on a slightly different model and section [ strong - coupling ] , in this paper ) , thus the sum over the ensemble deforms and attenuates the oscillations .
the substantial contribution of @xmath82 to the peak in transient is a consequence of the observation that in the initial phases of the dynamics the lateral modes contribute , relatively , much more than they do at steady state ( or at least on the long time scales ) .
in addition , we remark that the anticipated peak for @xmath82 contributes to a faster growth of the total intensity .
we now analyze the initial portion of the transient , i.e. , the decoupled regime .
the remarks concerning the transient leading to , but excluding , the intensity peaks ( and oscillations ) can be better understood with the help of the following approximate analysis . using the analytical ( approximate ) solution for @xmath71 ( eq .
( [ nsolution ] ) ) we can now decouple the modal intensities to obtain an approximate dynamical evolution : @xmath89 { ^di_j } + \\ \label{decoupledi } & & \beta_j b \mathcal{n}(t ) \left [ \mathcal{n}(t ) + p_0 \right ] \
, , \end{aligned}\ ] ] where the superscript @xmath90 denotes the approximate , decoupled variable .
the following auxiliary quantities simplify the notations : [ defcoeffjs ] @xmath91 \ , , \\
\mathcal{k}_j & = & \alpha_j n_0 + \frac{1}{\tau_p } \ , , \\ b_j & = & \beta_j b \ , , \end{aligned}\ ] ] where we have neglected the saturation term @xmath92 in the expression for @xmath7 , eq . ( [ coeff2 ] ) , since we are examining the transient portions in which the intensity is very small , @xmath23 being also very small ( cf .
table iii in @xcite ) .
this way , the modal intensities obey the set of m decoupled odes @xmath93 { ^di_j } + b_j \mathcal{n}(t ) \left [ \mathcal{n}(t ) + p_0 \right ] \
, , \end{aligned}\ ] ] which can be globally analyzed . over very short time intervals @xmath94 ( @xmath94 : @xmath95 , @xmath96 ( @xmath97 ) )
we can consider the two terms @xmath98 $ ] and @xmath99 $ ] as being constants , thus reducing eq .
( [ simplei ] ) to an ensemble of odes with constant coefficients .
in such a case , formal integration provides immediately a solution for the approximate problem , eq .
( [ simplei ] ) : @xmath100 } { \alpha_j \mathcal{n}(t_k ) - \mathcal{k}_j } + \\ \nonumber & & \left [ { ^di_j(t_k ) } + \frac{b_j \mathcal{n}(t_k ) \left [ \mathcal{n}(t_k ) + p_0 \right ] } { \alpha_j \mathcal{n}(t_k ) - \mathcal{k}_j } \right ] \\ \label{anal - int - shtime } & & e^{\left [ \alpha_j \mathcal{n}(t_k ) -
\mathcal{k}_j \right ] t^{\prime } } \ , , \\
\nonumber & = & -\frac{\mathcal{b}_j(t_k)}{\gamma_j(t_k ) } + \\ \label{simpler - approx - form } & & \left [ { ^di_j(t_k ) } + \frac{\mathcal{b}_j(t_k)}{\gamma_j(t_k ) } \right ] e^{\gamma_j(t_k ) \cdot t^{\prime } } \ , , \\ \label{def - gamma - j } \gamma_j(t_k ) & \equiv & \alpha_j \mathcal{n}(t_k ) - \mathcal{k}_j \quad ( < 0 ) \ , , \\
\label{def - beta - spont - em } \mathcal{b}_j(t_k ) & \equiv & b_j \mathcal{n}(t_k ) \left [ \mathcal{n}(t_k ) + p_0 \right ] \ , , \\ t_k & \equiv & k \times \delta t \ , , \\ k & = & 0 \ldots q \ , , \\ 0 \le & t^{\prime } & \le \delta t \ , , \\ { ^di_j(t_0 ) } & = & i_j(t_0 ) \ , , \end{aligned}\ ] ]
are the initial conditions for each modal intensity .
the inequality in parenthesis ( eq . ( [ def - gamma - j ] ) ) holds strictly in the range of validity of the current approximation . here
@xmath101 is a _ local _ time defined in each interval @xmath94 and is zeroed at the beginning of each time step .
the coefficients @xmath102 and @xmath103 @xcite are constant over a small time interval @xmath94 and are updated " , as time evolves , at the next value @xmath104 .
this , in itself , is not a limitation since their functional dependence is known and explicitly given by eqs .
( [ expl - an - sol],[expl - an - coeff],[defcoeffjs ] ) .
thus , eq . ( [ anal - int - shtime ] ) is a piecewise " , recursive solution over a discrete ensemble of times @xmath104 ( @xmath105 ) . at each time step , each @xmath106 starts to converge , for a short time , towards its locally asymptotic solution before being updated to the next time step .
the form of the approximate solution , eq .
( [ anal - int - shtime ] ) , highlights the role of the different coefficients : the spontaneous emission term , @xmath107 , controls the amplitude ( together with @xmath108 in the prefactor of eq .
( [ simpler - approx - form ] ) ) , while gain , @xmath109 , and cavity losses , @xmath110 , combined into @xmath108 play the role of an _ effective relaxation constant_. notice that @xmath108 results from the composition , eq .
( [ def - gamma - j ] ) , of the two functions shown in the inset of fig .
[ gamma - j ] ( notice the shift and multiplicative factor for curve b cf . figure caption ) which represent @xmath109 ( curve a ) and @xmath110 ( curve b ) .
the different local curvatures , combined with the multiplication of curve a by the carrier number @xmath5 ( or equivalently @xmath71 when using the approximate solution ) , differentiate the coefficient @xmath108 for the individual modes ( with the possibility of obtaining , in transient , positive values when curve a@xmath111 is larger than curve b
possible only outside the range of validity of the current approximation ) .
( in units of @xmath112 ) for selected modes : ( a ) j = 57 ; ( b ) j = 50 ; ( c ) j = 40 ; ( d ) j = 30 ; ( e ) j = 20 ; ( f ) j = 10 .
the solid lines correspond to the values of @xmath108 computed from the full set of equations , eqs .
( [ model ] ) , while the dashed lines ( which separate only for @xmath113 from the solid lines ) are computed for the analytical solution @xmath71 , eq .
( [ expl - an - sol],[expl - an - coeff ] ) .
the inset shows the modal dependence of @xmath109 ( curve ( a ) ) and @xmath110 ( curve ( b ) ) . for graphical purposes
we have plotted instead of @xmath110 its normalized and shifted version * k*@xmath114 . ] in order to better understand the behaviour of this approximate solution , we trace the coefficients @xmath108 and @xmath107 for selected modes in figs .
[ gamma - j ] and [ beta - j ] , respectively . fig .
[ gamma - j ] shows the evolution of @xmath108 as a function of time for a representative sample of modes . aside from the initial portions of the transient @xcite ( @xmath115 ) the larger @xmath108 ,
the closer is the corresponding mode to line center .
since @xmath108 s behave as effective relaxation constants , the figure immediately shows that the wing modes , with their larger ( negative ) values of @xmath116 , react on a shorter time scale than those modes close to line center .
this explains the numerical observation of fig .
[ t113-sum ] , where the wing modes grow faster than the central one , and partly accounts for the excess growth of these modes beyond their asymptotic values @xcite .
the constraint which forces the energy to be distributed over all modes ( cf .
discussion in section iva of @xcite ) , transfers the excess which can not be taken up by the stronger ( and slower ) modes to the faster wing modes during the transient . on the longer time scales , on which the central mode(s )
react(s ) , the energy is transferred back , allowing the side modes to relax to their asymptotic values .
notice that the agreement between the coefficients calculated from the analytical solution for the carrier number and from the full numerical solution holds extremely well until after threshold : in figs .
[ gamma - j ] and [ beta - j ] one can distinguish solid and dashed curves only in the top right hand portion of the figure . ) for selected modes : ( a ) j = 57 ; ( b ) j = 50 ; ( c ) j = 40 ; ( d ) j = 30 ; ( e ) j = 20 ; ( f ) j = 10 .
the solid lines correspond to the values of @xmath117 $ ] computed from the full set of equations , eqs .
( [ model ] ) , while the dashed lines ( which separate only for @xmath113 from the solid lines ) are computed for the analytical solution @xmath71 in place of @xmath5 , eq .
( [ expl - an - sol],[expl - an - coeff ] ) .
the labels ( a ) and ( b ) correspond , from top to bottom , to the top two sets of curves ( solid and dashed , respectively ) which can be distinguished only on the far right of the figure . ] in addition to the component ( @xmath118 in eq .
( [ simplei ] ) , with the substitutions , eqs .
( [ def - gamma - j],[def - beta - spont - em ] ) ) discussed so far , the temporal evolution of the individual modes includes the contribution of the spontaneous emission @xmath107 whose amplitude is plotted in fig .
[ beta - j ] ( dashed lines ) for a subset of modes , as a function of time ( thus , of carrier number , cf .
( [ nsolution ] ) ) and is compared to the amplitude ( solid lines ) for the full model , eqs .
( [ model ] ) .
its main feature is the similarity in its contribution to all modes ( maximum deviation , less than a factor 2 ) even at the end of the time interval ( largest @xmath71 ) .
the bundle of curves opens up with time , increasing by about one order of magnitude for the central mode ( @xmath47 ) somewhat less for the plotted mode farthest away from line center ( @xmath119 ) .
for the latter mode we remark that @xmath120 , thus maintaining a nearly constant , fast relaxation throughout the decoupled regime .
the fast , nearly constant , reaction time explains the sizeable growth of the wing modes , in spite of the smaller contribution which they receive from the spontaneous emission . the moderate change in @xmath121 ( and equivalent from the wing modes ) ensures a slight growth in the prefactor of eq .
( [ simpler - approx - form ] ) , thus a rather stable energy contribution of these modes below threshold . for comparison , in the units of fig .
[ gamma - j ] , @xmath122 ( cf .
hence , the effective relaxation rate changes considerably for the center mode and is eventually responsible for their predominance at the end of the below - threshold regime , even though the side modes maintain a level which is considerably larger than their asymptotic contribution . the approximate solution , eq .
( [ simpler - approx - form ] ) , shows the role played by @xmath123 , which is nothing else than the ratio between the coefficients shown in figs .
[ gamma - j],[beta - j ] .
thus , another way of interpreting the dynamics below threshold is to plot , fig .
[ i - incr ] , the r.h.s . of eq .
( [ simplei ] ) thus the intensity increment for the whole ensemble of modes at selected times ( i.e. , for selected values of @xmath71 ) , from the initial condition up to threshold . ) for the individual mode intensities ( r.h.s . of eq .
( [ simplei ] ) ) plotted at different times : ( a ) @xmath124 ( @xmath125 ) ; ( b ) @xmath126 ( @xmath127 ) ; ( c ) @xmath128 ( @xmath129 ) ; ( d ) @xmath130 ( @xmath131 ) ; ( e ) @xmath132 ( @xmath133 ) ; ( f ) @xmath134 ( @xmath135 ) ; ( g ) @xmath136 ( @xmath137 ) ; ( h ) @xmath138 ( @xmath139 ) . ] when the carrier number is well below transparency ( curves a - c ) the gain is nearly equal for all modes , with a somewhat higher level in the wings for the first two curves @xcite .
this implies that the modes far from line center start growing faster than the central ones . near transparency ,
curve ( d ) , the central modes start dominating and the combined effect of their somewhat larger intensity and of their increasing coefficients ( cf . figs .
[ gamma - j],[beta - j ] ) favors them . for the parameters of curve ( e ) the laser is already beyond transparency ( @xmath140 , @xmath141 ) and
the tendency for a strong differential growth becomes obvious and sharpens itself with growing values of @xmath142 .
equivalent curves showing the ratio ( cf .
( [ simpler - approx - form ] ) ) @xmath143 ( each component plotted in figs .
[ gamma - j],[beta - j ] ) look qualitatively very similar to those of fig .
[ i - incr ] , since they differ only for the intensity @xmath106 which multiplies @xmath108 ( the relative deformation is most visible for curve ( h ) , which is more strongly reshaped by the larger modal intensity variations ) . for a selected subset of modes : ( a ) @xmath144 ; ( b ) @xmath49 ; ( c ) @xmath145 ; ( d ) @xmath146 ; ( e ) @xmath147 ; ( f ) @xmath148 ; ( g ) @xmath149 ; ( h ) @xmath119 .
the ratios are computed with the decoupled system of equations , eqs .
( [ nsolution],[simplei ] ) . ] in spite of the clear increase in differential intensity growth between the central and the side modes displayed in fig .
[ i - incr ] the ratio remains finite up until threshold , as shown in fig .
[ ratio - int ] , where the amplitude of each intensity mode is displayed normalized to the intensity of the central mode .
[ ratio - int ] shows again that the gain is largest for the wing modes at the start of the transient ( curve ( h ) , @xmath119 , is the highest at @xmath150 ) , due to their lower absorption and that during the whole evolution in the decoupled regime the relative weight of the side modes largely exceeds their asymptotic value @xmath151 cf .
[ ratio - int ] to be compared to @xmath152 from table [ max - time ] ) .
this result is consistent with the numerical observation
obtained from the integration of the full model , eqs .
( [ model ] ) showing that at @xmath153 the cumulative intensity of the 20 modes farthest away from line center , @xmath86 , exceeds that of the central mode ( cf .
[ t113-sum ] ) . it also explains why at @xmath154 the intensity distribution is far from the asymptotic one ( fig . [ time - trace ] and @xcite ) : the wing modes at the beginning of the strongly coupled regime hold a very large relative amount of energy ( @xmath155 300 times its relative asymptotic value for @xmath156 ) . besides the numerical evidence of fig .
[ ratio - int ] , eq .
( [ simpler - approx - form ] ) shows the reason for the not - so - large differences among modes , below threshold . expanding the exponential in eq .
( [ simpler - approx - form ] ) to first order , we obtain an approximate form for the modal intensity : @xmath157 + \mathcal{b}_j(t_k ) \cdot t^{\prime } \ , , \end{aligned}\ ] ] thus its increment relative to the previous time step takes the form : @xmath158 the first term on the r.h.s .
( [ incrementij ] ) represents a relaxation ( @xmath159 , @xmath160 ) , while the second one is the modal intensity increment brought about by the ( average ) spontaneous emission contribution .
thus , the differential growth between center and wing modes remains moderate ( cf .
[ beta - j ] ) , which explains why even close to threshold the ratio of the wing modes to the center mode ( fig .
[ ratio - int ] ) remains very large compared to its asymptotic value .
notice that eq .
( [ incrementij ] ) indirectly determines a time step @xmath94 , since @xmath161 , @xmath162 @xcite . ;
( b ) @xmath163 ; ( c ) @xmath49 ; ( d ) @xmath145 ; ( e ) @xmath148 ; ( f ) @xmath149 ; ( g ) @xmath119 .
the solid lines display the result of the integration of the full model , eqs .
( [ model ] ) , while the dashed lines show , for comparison , the corresponding intensity values calculated from the decoupled model ( eqs .
( [ nsolution],[simplei ] ) ) .
inset a shows a detail of the temporal evolution around @xmath164 to see how the approximate and the full solution deviate from each other .
inset b shows three selected wing modes .
values for the intensity peaks and corresponding times are given in table [ max - time ] .
notice that in the main figure , as well as in inset a , curves ( a ) and ( b ) are practically indistinguishable . the arrow underneath the label in these two figures shows the order in which the labels have to be attributed to the ( closely ) adjacent curves . ]
[ comp - centremod ] shows the temporal evolution of a selection of modes ( those close to line center ) for the full integration ( eqs .
( [ model ] ) , solid lines ) and for the decoupled system ( eqs .
( [ nsolution],[simplei ] ) , dashed lines ) .
we notice that over the whole range of validity ( @xmath165 ) of the analytical solution for the carrier number , eq .
( [ nsolution ] ) , the decoupled approximation is very good .
inset a shows a detail of the temporal evolution for the same modes close to threshold : all curves lie extremely close to one another , confirming the validity of our approximation .
inset b shows the evolution for three wing modes ( cf .
caption ) : the agreement between full solutions ( solid lines ) and analytical approximations ( dashed lines ) remains excellent even beyond threshold for these modes .
( threshold crossing ) ; ( c ) @xmath166 ( total intensity peak ) ; ( d ) @xmath167 ( total intensity and carrier density relaxed to their asymptotic value ) : beginning of the slow transient @xcite ; ( e ) @xmath168 ( asymptotic intensity distribution ) .
the intensity values are plotted in logarithmic scale and are not rescaled according to the factors specified in fig .
[ time - trace ] . ]
we summarize the main result of this section in fig .
[ int - distr ] which shows how the intensity distribution among all modes at the end of the decoupled regime ( i.e. , the instant when the population carrier crosses its asymptotic value threshold for the first time ) does not differ very substantially from the initial distribution .
curve a ( black online ) shows the equilibrium intensity distribution at the initial condition the side modes carry more intensity thanks to their reduced absorption ( below transparency ) .
curve b ( red online ) shows the intensity distribution at threshold ( cf . above )
i.e. , @xmath77 .
even though the central modes carry a larger amount of intensity ( and the shape of the curve is inverted , compared to the initial condition , when comparing to the asymptotic , equilibrium , distribution ( curve d , blue online ) , the quantitative differences among modes at the first threshold crossing are nearly negligible . anticipating on the next section ,
we remark that even at the peak in the total intensity the intensity distribution among modes ( curve c , green online ) differs less from the threshold condition than from the asymptotic one .
in conclusion , in this section we have shown that the decoupled model gives a very good representation of the evolution of the modal content of the laser output in the below threshold region , i.e. , over a large part of the transient .
we have also seen , both numerically and analytically , how the side modes carry an amount of energy much larger than their asymptotic share .
once the system overcomes this threshold , the analytic approximation for the carrier number breaks down , the coupling becomes strong and only the full numerical solution can account for the remainder of the fast evolution towards the steady state for the two global variables ( @xmath1 , and @xmath29 @xcite ) .
the two preceding sections have analyzed in detail the portion of the dynamical evolution below the threshold value for the carrier number .
here we are going to investigate the next portion of the dynamics @xcite , which corresponds to the full nonlinear coupling between modes and which leads , eventually , into the time - independent total intensity with the residual slow evolution of the modal intensity distribution @xcite .
our analysis of the multimode high - intensity regime is based on a direct integration of the model equations and on the potential for the interpretation of the resulting observations which comes directly from the detailed analysis of the previous sections .
a direct - solution approach , such as the one adopted in @xcite for an analytical construction of the solution for short pulses in a single - mode device , would be very hard to adopt in our strongly multimode model . and onset of the strongly - coupled regime .
( a ) total intensity ; ( b ) @xmath47 ; ( c ) @xmath169 ; ( d ) @xmath170 ; ( e ) @xmath171 ; ( f ) @xmath156 ; ] fig .
[ error - i ] clearly shows the transition between the decoupled regime , where the deviation between the approximate and exact modal intensities is at most 5% ( at @xmath172 ) , and the fully nonlinear regime , where this difference rapidly diverges . the transition is clearly illustrated by fig . [ gamma - j ] which shows that during their growth some @xmath108 s become positive for a subensemble of modes ( near line center ) and oscillate together with the carrier number to relax @xcite to their ( negative @xcite ) asymptotic values ( not shown ) . when some @xmath173 the nature of eq .
( [ simpler - approx - form ] ) changes , since the growth of the @xmath106 s is no longer exclusively due to the growth of the prefactor but also to the exponential growth of ( some ) @xmath106 s , which eventually leads to a divergence from the solution of the full model , eqs .
( [ model ] ) . from the point of view of the solution of eqs .
( [ model ] ) the positivity of some @xmath108 s bears no weight . indeed , throughout the growth of the modal intensities ( @xmath174 , cf .
[ time - trace ] ) , the r.h.s . of eq .
( [ modela ] ) ( or , equivalently , eq .
( [ decoupledi ] ) ) remains positive due to the contribution of the spontaneous emission ( @xmath107 ) , independently on the sign of @xmath108 .
thus , the latter contributes quantitatively , but not qualitatively , to the growth of the modal intensities @xmath106 . .
intensity value at the first@xmath175 and second @xmath176 peak , @xmath177 , and time of occurrence , @xmath178 , @xmath179 time difference between the peaks , and asymptotic intensity @xmath180 for each laser mode @xmath8 ( from the full set of equations , ( [ model ] ) ) .
the last line , separated from the rest of the table , gives the equivalent information for the total intensity . [ cols="^,^,^,^,^,^,^",options="header " , ] a closer look at the modal dynamics ( solid lines in fig .
[ comp - centremod ] ) shows that all individual modes display the same generic behaviour ( peak with damped oscillations ) but that some differences appear in the time at which the peaks occur and in the shape ( frequency and damping ) of the oscillations .
table [ max - time ] gives the time @xmath181 at which the intensity maximum ( @xmath182 ) is reached for a selection of modes .
a clear anticipation in the first peak appears as the mode index moves away from line center ( @xmath183 ) , not unexpectedly after the discussion of the previous section based on the behaviour of the @xmath108 s ( fig .
[ gamma - j ] ) .
an inversion in this tendency , however , appears comparing @xmath184 and @xmath185 , hinting to a more complex behaviour in the strong - coupling regime .
( circles , left scale ) and amplitude of the second peak relative to the first peak , @xmath186 ( squares , right scale ) for a selection of laser modes .
notice that the central mode is at the far right , thus for picturing the full set of modes one should mirror - image the figure with respect to vertical axis passing through @xmath183 . for the total intensity ,
the equivalent quantities are ( from table [ max - time ] ) : @xmath187 , @xmath188 . ]
[ damping ] shows the ratio between the maximum peak amplitude for each mode and its corresponding asymptotic value ( circles , left scale ) . for the central mode
the peak amplitude @xmath189 is about half its steady state , @xmath190 , while in the far wings the peak ( e.g. , @xmath191 ) is only slightly higher than its steady state ( @xmath192 ) , showing that equilibrium for these modes is reached rather smoothly ( i.e. , with little overshoot and oscillations ) .
however , the peak amplitude grows very large for modes off line - center , but not too far from it , the largest ratio occurring for @xmath146 ( six places away from line center ) whose peak intensity @xmath193 .
this highlights the role of the ( close ) side modes in the strongly coupled regime , which temporarily carry a large amount of energy .
associated with this strong variation in the peak intensity , a remarkable , nearly anti - correlated , variation in the damping appears .
the ratio between the second and the first peak @xmath194 is represented by squares ( fig . [ damping ] , right scale ) ) . while at line center and far off in the wings the second peak is almost as large as the first ( @xmath195 ) , for the strongest overshoot we notice ( almost ) the largest damping ( with a slight shift , @xmath196 ) .
( diamonds , left scale ) , modal damped ( down - triangles ) , @xmath197 , and undamped ( up - triangles ) , @xmath198 , relaxation oscillation frequency ( right scale ) . notice that the central mode is at the far right , thus for picturing the full set of modes one should mirror - image the figure with respect to vertical axis passing through @xmath183 .
the equivalent quantities for the total intensity are : @xmath199 , @xmath200 , @xmath201 . ] assuming a simple dependence for the damped oscillations , of the form @xmath202 , we can estimate the modal damping constant @xmath203 in the strongly coupled regime , together with the oscillation frequency @xmath197 directly from the data , using the peak amplitudes ( @xmath182 , @xmath204 ) and times ( @xmath205 ) at which they occur . with the help of the standard expression for the undamped frequency @xmath206 of a damped oscillator , we can further estimate @xmath207 .
the corresponding results are shown in fig .
[ freqs ] for the damping @xmath203 ( diamonds , left scale ) and for the damped ( down - triangles ) , @xmath197 , and undamped ( up - triangles ) , @xmath198 , frequencies ( right scale for both ) . a resonance - like behaviour appears clearly in the relaxation oscillation frequency , with a maximum at mode @xmath208 for @xmath206 ( at @xmath209 for @xmath210 ) , and a sizeable variation in frequency @xmath211 ( i.e. , @xmath212 ) .
the contribution due to the modal damping @xmath203 is clearly visible in the graph and increases the frequency , at its maximum , by @xmath213 , while shifting it away from line center ( by one mode ) .
the combination of the contribution of all modes lends the total intensity ( fig .
[ time - trace ] ) features ( cf . captions of figs .
[ damping ] , [ freqs ] ) which are a combination of those of different modes .
looking at the damping , @xmath214 or at the relative height of the first peak , @xmath215 , the values are similar to those of mode @xmath216 , the relaxation oscillation frequencies , @xmath217 , are consistent with mode @xmath49 , while the second peak is less strongly damped , as for a mode placed between @xmath218 and @xmath219 .
we therefore conclude that the dynamics in the portion of the fast transient corresponding to the strongly - coupled regime is dominated by a group of modes close to , but not at , line - center . for the variable @xmath220 in both numerator and denominator of the vertical axis .
( a ) @xmath221 ; ( b ) @xmath171 ; ( c ) @xmath222 ; ( d ) @xmath223 ; ( e ) @xmath51 ; ( f ) @xmath50 ; ( g ) @xmath224 ; ( h ) @xmath47 .
mode @xmath163 shows the same evolution as mode @xmath47 ( in this figure ) slightly displaced towards higher values ( not shown for simplicity ) . ]
an equivalent graphical illustration is given by fig . [ norm - peaks ] which shows the transient evolution for each mode , normalized to its asymptotic value .
the modes far in the wings show a very early and gradual growth with little overshoot and oscillations ( the decay is masked by the normalization to their asymptotic state , cf .
@xcite ) . coming closer to line center we notice a gradual delay in the growth , accompanied by a sharp , and very large , peak and a gradual relaxation towards the asymptotic intensity level .
the last mode to grow is the central one , whose peak , together with the first side mode , is lower than its steady state .
the relaxation for the central modes is extremely slow and extends well beyond the figure range @xcite .
[ norm - peaks ] , together with the analysis of the previous sections , convincingly demonstrates the reason for the slow dynamics discussed in @xcite .
the transient evolution from the initial , below - threshold condition takes place first through a regime where the modal growth is not strongly differentiated between mode center and wings ( i.e. , below threshold ) .
when the laser finally passes threshold , the central ( slower ) modes have to grow away from an intensity distribution which is far from the asymptotic one ( fig .
[ int - distr ] ) .
this intensity configuration favours those lateral modes which are sufficiently close to line center to have a strong growth , but which at the same time possess faster time constants , which allow them to more rapidly exploit the excess population inversion .
thus , we find a strongly out - of - equilibrium intensity distribution at the peak ( curve c in fig .
[ int - distr ] ) , and it is not surprising that when the total intensity and the carrier number have relaxed to their respective asymptotic values ( at @xmath225 for the initial conditions chosen in this paper , fig .
[ time - trace ] ) the modal repartition of the intensity be still strongly very far from its asymptotic configuration . in this way , we clearly understand the reason for the slow dynamics reported in @xcite .
the analytical and numerical treatment of the transient has given us a good description of the modal evolution from the initial state ( laser nearly off ) to the point where the two global variables ( @xmath1 and @xmath29 ) have attained equilibrium .
we now look at the time - resolved transient emission spectrum , defined as the frequency interval over which the modal intensity passes a chosen percentage of the total output intensity , @xmath226 , where @xmath227 stands for full spectral and @xmath228 for width , while @xmath229 fixes the chosen level in db .
we remark that , contrary to what found for the slow dynamics @xcite , the transient spectra are only marginally affected by the modal arrangement under the gain line ( even vs. odd number of modes in the simulation cf .
@xcite for a discussion ) .
we will therefore discuss only the modal placement used so far . of the laser evolution during the transient .
solid lines @xmath230 , dashed lines @xmath231 .
the level in @xmath232 is defined , as in @xcite , as the modal intensity which passes the preset level compared to the instantaneous total intensity . ]
[ all - spectra ] shows the full spectral width ( @xmath233 ) , resolved in time , calculated at -40 db ( @xmath234 , solid line ) and at -20 db ( @xmath235 , dashed line ) .
the frequency edge " , marked by each line in the figure , is obtained by linear interpolation between the frequencies of the two modes between which the @xmath226 criterion is satisfied .
the -40 db level captures the full evolution of the spectral content during the transient ; we recognize that in the very first phases of the transient ( @xmath236 ) none of the modes passes this mark ( the spectral width starts at this point ) @xcite because the intensity is widely distributed over a large number of modes , rather than being carried by a few modes around line center . in other words ,
no frequency component is sufficiently strong to reach one - hundredth of the total intensity ( equivalent to -40 db ) . since we are integrating over slightly more than 100 modes
, this implies that the intensity distribution , in this first phase of the transient , is sufficiently homogeneous that no spectral feature emerges from the spectrum .
this is in agreement with our previous findings which show a rather homogeneous intensity distribution among modes in the decoupled regime .
the maximum width is obtained between @xmath237 and @xmath238 and corresponds to approximately 47 modes passing the @xmath239 level ( i.e. , the -40 db mark ) .
the @xmath234 continues to decrease beyond @xmath240 and @xmath235 makes its appearance at @xmath241 , grows until @xmath242 and then initiates its slow convergence to its asymptotic value , as discussed in @xcite . in light of the results of the previous section the wide spectral width ( @xmath234 ) does not come as a surprise and quantitatively shows the degree of modal spread of the laser intensity in the initial phases of the transient . on the other hand , the large modal coverage has a direct impact on the validity of models which are obtained by truncating the full one , @xmath243 , to a subset of modes around line center @xcite .
the results of this paper have been obtained on a semiconductor laser model where specific choices have been made to describe relaxation processes to match a specific experimental device ( a trench buried heterostructure bulk semiconductor laser ) .
we have selected this model to have a tested , realistic comparison with an actual device @xcite , but the relevance of the understanding that we have gained in the process extends beyond the specific details of this particular system .
numerical tests have shown @xcite that only minor quantitative details change when replacing the fixed gain line with a variable , dynamic one @xcite , or when substituting the parabolic profile ( eq . [ coeff2 ] ) with a gaussian one @xcite , or even when exchanging the linear gain ( eq . [ coeff2 ] ) with a logarithmic one @xcite .
more sophisticated modeling choices can be made for the gain line @xcite , but on the basis of the previous remarks , we expect the general results we have obtained in this paper to hold at least for most kinds of multi - longitudinal mode semiconductor lasers .
the kind of interaction among modes ( coupling to a mean field ) is at the origin of our numerical observations and of the understanding we have gained , independently of the modeling details .
thus , we expect our predictions to apply to any other kind of laser where an intensity - based intermodal interaction dominates .
candidates for such behaviour are all those lasers where diffusion in the gain medium washes out any trace of spatial modulation resulting from the interference among the modal fields and/or lasers with a partial , but not dominant , inhomogeneous broadening
. our modeling of the below threshold region ( decoupled regime , sections [ carrier - density ] , [ decoupled - modes ] ) may deliver useful physical information on the dynamics of carriers and optical emission in cavity light emitting diodes ( cleds ) .
two particular subgroups of cleds may benefit from this modeling : superluminescent diodes , which are very sensitive to parasitic cavity effects and may suffer from parasitic lasing , and edge emitting leds ( eleds ) , which are the led equivalent of an edge - emitting semiconductor laser , exactly the kind of device which is the object of the model we have investigated @xcite . finally , we recall what stated in the introduction about the generalization of the model to a set of oscillators , possessing a stable attractor and mutually coupled through a mean field .
the dynamics of such systems , including the presence of a master mode dominating the slow evolution , will qualitatively match the one described in our paper , when externally driven , irrespective of the details of the model . in summary
, we have numerically investigated the fast transient dynamics of a multimode semiconductor laser , in response to a sudden turn - on by a switch of a parameter ( pump ) , modeled by m odes for the modal intensities , with coupling occurring through the population inversion ( i.e. , the carrier number ) .
the dynamics have been proven to separate into two regimes : ( a ) one where the carrier number and the modal intensities can be decoupled and for which approximate analytical solutions can be found ; and ( b ) the strongly coupled ( or fully nonlinear ) one where all variables are interdependent in their evolution .
our analysis gives a clear physical explanation and a quantitative illustration of the origin of the strong deviation for the modal intensity distribution at the onset of equilibrium ( @xmath244 ) for the global variables ( carrier number @xmath1 and total intensity @xmath245 ) which gives rise to the slow dynamics @xcite .
choosing a very low initial bias allows us to investigate the various regions of the transient .
practically useful bias values , such as those of @xcite , will skip the transient evolution below transparency , thus shortening and simplifying the dynamics .
sokolovskii , v.v .
dudelev , e.d .
kolykhalova , a.g.deryagin , m.v .
maximov , a.m. nadtochiy , v.i .
kuchinskii , s.s .
mikhrin , d.a .
livshits , e.a .
viktorov , and t. erneux , appl .
. lett . * 100 * , 081109 ( 2012 ) .
the choice of using @xmath247 is entirely arbitrary and eqs .
( [ expl - an - sol],[expl - an - coeff ] ) can be written in terms of @xmath248 as well , since in the polynomial terms only @xmath247 s modulus or real part enter into the equation , while in the exponential form , a simple sign change restores the form with @xmath248 in place of @xmath247 .
we have checked that the numerical solution obtained from this decoupled set of equations coincides with the analytical one within the numerical precision , for the same initial condition ( in practice deviations of a few parts per thousand are tolerable ) .
the advantage of using the numerical result comes from the synchronicity of the points for the two traces : computing both with the help of the set of odes provides synchronous ensembles of points , while those originating from eq .
( [ nsolution ] ) are given at times @xmath249 which can not be synchronized ( since @xmath249 is the dependent variable in this expression ) .
one should define two different kinds of coefficients for @xmath116 and @xmath250 , depending on whether they are defined as a function of @xmath1 or of @xmath142 . due to the practically identical results ( cf . figs .
[ gamma - j ] , [ beta - j ] ) , originating from the almost coincidence of @xmath1 and @xmath142 over the whole interval of validity of the decoupled approximation ( cf .
[ sols - n ] ) , for the sake of simplicity we refrain from introducing additional symbols . the point at which all curves for @xmath108 cross corresponds to @xmath20 , which represents the carrier number for which the gain medium is bleached ( transparency ) . below this value ,
the wing modes experience less absorption , since they are far off - resonance , thus their corresponding @xmath108 s are larger . for @xmath251
the situation is reversed , since the gain is larger near line center . the larger intensity increment below transparency
is explained , for the wing modes , by their lower absorption .
notice that for the given lineshape , eq .
( [ coeff2 ] ) , for a couple of modes at the extrema of the interval there may be a ( temporary ) sign inversion , which bears , however , no practical consequences on the numerics .
since the radius of convergence of the exponential function is infinite , one could use a less restrictive condition .
however , the magnitude of @xmath94 is appropriately set this way and is consistent with the one used in fig .
[ ratio - int ] .
an alternative way of defining the upper limit for the range of parameters ( or time interval ) for which the dynamical decoupling between carrier number and modal intensities holds is the following : evaluate steady states for the variables @xmath1 and @xmath16 in the decoupled model ( eqs .
( [ nsolution],[simplei ] ) and impose the condition that @xmath254 @xmath255 . while well below transparency three distinct real solutions exist for the steady state condition ( eq . ( 3b ) in @xcite , adapted to the decoupled system ) , close to threshold only one real one , which we will call @xmath256 , can be found .
thus , the steady state intensity takes the form ( for all modes ) @xmath257 , thus @xmath258 iff @xmath259 hence , the decoupled regime holds only for @xmath260 and the instant @xmath261 for which @xmath262 ( in particular @xmath183 ) defines the upper limit to this regime ( either by defining @xmath261 or the corresponding value @xmath263 ) .
+ thus , for the decoupled regime to hold , the effective relaxation constant must remain negative at all times for all modes .
as soon as one @xmath264 ( @xmath183 , specifically ) the approximation can no longer be used . | we analyze the fast transient dynamics of a multi - longitudinal mode semiconductor laser on the basis of a model with intensity coupling .
the dynamics , coupled to the constraints of the system and the below - threshold initial conditions , imposes a faster growth of the side modes in the initial stages of the transient , thereby leading the laser through a sequence of states where the modal intensity distribution dramatically differs from the asymptotic one . a detailed analysis of the below - threshold
, deterministic dynamical evolution allows us to explain the modal dynamics in the strongly coupled regime where the total intensity peak and relaxation oscillations take place , thus providing an explanation for the modal dynamics observed in the slow , hidden evolution towards the asymptotic state ( cf .
phys .
rev .
a * 85 * , 043823 ( 2012 ) ) .
the dynamics of this system can be interpreted as the transient response of a driven , globally coupled ensemble of nonlinear modes evolving towards an equilibrium state .
since the qualitative dynamics do not depend on the details of the interaction but only on the structure of the coupling , our results hold for a whole class of globally , bilinearly coupled oscillators .
( all figures in color online ) . |
the interplay between mechanical and electronic effects in carbon nanostructures has been studied for a long time ( e.g. , @xcite ) .
the mechanics in those studies invariably enters within the context of continuum elasticity .
one of the most interesting predictions of the theory is the creation of large , and roughly uniform pseudo - magnetic fields and deformation potentials under strain conformations having a three - fold symmetry @xcite .
those theoretical predictions have been successfully verified experimentally @xcite .
nevertheless , different theoretical approaches to strain engineering in graphene possess subtle points and apparent discrepancies @xcite , which may hinder progress in the field .
this motivated us to develop an approach @xcite which does not suffer from limitations inherent to continuum elasticity .
this new formulation accommodates numerical verifications to determine when arbitrary mechanical deformations preserve sublattice symmetry .
contrary to the conclusions of ref .
@xcite , with this formulation one can also demonstrate in an explicit manner the absence of @xmath3point dependent gauge fields on a first - order theory ( see refs .
@xcite and @xcite as well ) .
the formalism takes as its only direct input _
atomistic data as the data obtained from molecular dynamics runs .
the goal of this paper is to present the method , making the derivation manifest .
we illustrate the formalism by computing the gauge fields and the density of states in a graphene membrane under central load .
the theory of strain - engineered electronic effects in graphene is semi - classical .
one seeks to determine the effects of mechanical strain across a graphene membrane in terms of spatially - modulated pseudospin hamiltonians @xmath4 ; these pseudospin hamiltonians @xmath5 are low - energy expansions of a hamiltonian formally defined in reciprocal space . under `` long range '' mechanical strain
( extending over many unit cells and preserving sublattice symmetry @xcite ) @xmath4 also become continuous and slowly - varying local functions of strain - derived gauges , so that @xmath6 . within this first - order approach ,
the salient effect of strain is a local shift of the @xmath1 and @xmath2 points in opposite directions , similar to a shift induced by a magnetic field @xcite . in the usual formulation of the theory @xcite , this dependency on position leads to a _ continuous _ dependence of strain - induced fields @xmath7 and @xmath8 .
such continuous fields are customarily superimposed to a discrete lattice , as in figure [ fig : f1 ] @xcite .
when expressed in terms of continuous functions , a pseudospin hamiltonian @xmath4 is defined down to arbitrarily small spatial scales and it spans a zero area . in reality , however , the pseudospin hamiltonian can only be defined per unit cell , so it should take a single value at an area of order @xmath9 ( @xmath10 is the lattice constant in the absence of strain ) .
this observation tells us already that the scale of the mechanical deformation with respect to a given unit cell is inherently lost in a description based on a continuum model . for this reason ,
it is important to develop an approach which is directly related to the atomic lattice , as opposed to its idealization as a continuum medium . in the present paper we show that in following this program one gains a deeper understanding of the interrelation between the mechanics and the electronic structure of graphene .
indeed , within this approach we are able to quantitatively analyze whether the proper phase conjugation of the pseudospin hamiltonian holds at each unit cell .
the approach presented here will give ( for the first time ) the possibility to explicitly check on any given graphene membrane under arbitrary strain if mechanical strain varies smoothly on the scale of interatomic distances .
consistency in the present formalism will also lead to the conclusion that in such scenario strain will not break the sublattice symmetry but the dirac cones at the @xmath1 and @xmath2 points will be shifted in the opposite directions @xcite .
clearly , for a reciprocal space to exist one has to preserve crystal symmetry , so that when crystal symmetry is strongly perturbed , the reciprocal space representation starts to lack physical meaning , presenting a limitation to the semiclassical theory .
the lack of sublattice symmetry observed on actual unit cells on this formulation beyond first - order continuum elasticity may not allow proper phase conjugation of pseudospin hamiltonians at unit cells undergoing very large mechanical deformations .
nevertheless this check can not proceed and hence has never been discussed on a description of the theory within a continuum media , because by construction there is no direct reference to actual atoms on a continuum . as it is well - known
, it is also possible to determine the electronic properties directly from a tight - binding hamiltonian @xmath11 in real space , without resorting to the semiclassical approximation and without imposing an _ a priori _ sublattice symmetry .
that is , while the semiclassical @xmath12 is defined in reciprocal space ( thus assuming some reasonable preservation of crystalline order ) , the tight - binding hamiltonian @xmath11 in real space is more general and can be used for membranes with arbitrary spatial distribution and magnitude of the strain .
in addition , contrary to the claim of ref .
@xcite , the purported existence of @xmath3point dependent gauge fields does not hold on a first - order formalism @xcite .
what we find instead , is a shift in opposite directions of the @xmath1 and @xmath2 points upon strain @xcite .
the continuum theories of strain engineering in graphene , being semiclassical in nature , require sublattice symmetry to hold @xcite .
one the other hand , no measure exists in the continuum theories @xcite to test sublattice symmetry on actual unit cells under a mechanical deformation .
for this reason , sublattice symmetry is an implicit assumption embedded in the continuum approach . and
@xmath13 upon strain ( @xmath14 ) .
relative displacements of neighboring atoms lead to modified lattice vectors ; the choice of renormalized lattice vectors will be unique _ only _ to the extent to which sublattice symmetry is preserved : @xmath15.,scaledwidth=45.0% ] to address the problem beyond the continuum approach , let us start by considering the unit cell before ( fig .
[ fig : f2](a ) ) and after arbitrary strain has been applied ( fig .
[ fig : f2](b ) ) . for easy comparison of our results
, we make the zigzag direction parallel to the @xmath16axis , which is the choice made in refs .
@xcite and @xcite .
( arbitrary choices of relative orientation are clearly possible ; in ref .
@xcite we chose the zigzag direction to be parallel to the y - axis . ) the lattice vectors before the deformation are given by ( fig .
[ fig : f2](a ) ) : @xmath17 @xmath18 ( note that @xmath19 , and @xmath20 . ) after mechanical strain is applied ( fig .
[ fig : f2](b ) ) , each local pseudospin hamiltonian will only have physical meaning at the unit cells where : @xmath21 condition ( [ eq : applicabilitycondition ] ) can be re - expressed in terms of changes of angles @xmath22 or lengths @xmath23 for pairs of nearest - neighbor vectors @xmath24 and @xmath25 [ @xmath26 is shown in thick solid and @xmath27 in thin dashed lines in fig .
[ fig : f2](b ) ] : @xmath28\cdot \hat{k}\right),\ ] ] where @xmath29 is a unit vector along the z - axis , @xmath30 is the sign function ( @xmath31 if @xmath32 and @xmath33 if @xmath34 ) , and : @xmath35 even though in the problems of practical interest the deviations from the sublattice symmetry do tend to be small @xcite , it is important to bear in mind that the sublattice symmetry _ does not hold a priori _ @xcite .
it is therefore important to have a method to quantify such deviations and check whether the sublattice symmetry holds at the problem at hand . forcing the sublattice symmetry to hold from the start
amounts to introducing an artificial mechanical constraint on the membrane which is not justified on physical grounds @xcite . for this reason
the method we propose is discrete and directly related to the actual lattice ; it does not resort to the approximation of the membrane as a continuum medium @xcite .
being expressed in terms of the actual atomic displacements , our formalism holds beyond the linear elastic regime where the first - order continuum elasticity may fail .
the continuum formalism is recovered as a special case of the one presented here in the limit when @xmath36 . in the absence of mechanical strain ,
the reciprocal lattice vectors @xmath37 and @xmath38 are obtained by standard methods : we define @xmath39 , with @xmath40 and @xmath41 given in eq .
( [ eq : defa ] ) and shown in fig .
[ fig : f2](a ) .
the reciprocal lattice vectors @xmath42 are related to the lattice vectors by @xcite : @xmath43 with the choice we made for @xmath40 and @xmath41 we get : @xmath44 as seen in fig .
[ fig : f3](a ) the @xmath3points on the first brillouin zone are defined by : @xmath45 and : @xmath46 points ( @xmath1 and @xmath2 ) move in the opposite directions .
the dashed hexagon in ( b ) represents the boundary of the first brillouin zone in the absence of strain.,scaledwidth=45.0% ] the relative positions between atoms change when strain is applied : @xmath47 ( @xmath48 , and @xmath49 ( @xmath14 ) . for negligible curvature , one may assume that @xmath50 ( and similar for the primed displacements @xmath51 ) .
we present here a formulation of the theory strictly valid for in - plane strain ( it would also be valid for membranes with negligible curvature ) .
we wish to find out how reciprocal lattice vectors change to first order in displacements under mechanical load . in order for reciprocal lattice vectors to make sense at each unit cell , eqn .
[ eq : applicabilitycondition ] must hold . in terms of numerical quantities
one would need that @xmath22 and @xmath23 are all close to zero .
in that case we set @xmath52 for j=1,2 , and continue our program . for this purpose
we define : @xmath53 or in terms of ( two - dimensional ) components : @xmath54 the matrix @xmath55 changes to @xmath56 , and we must modify @xmath57 so that eqn
. still holds under mechanical load .
to first order in displacements @xmath58 becomes : @xmath59 by comparing eqns .
( 7 ) and , the reciprocal lattice vectors in fig .
[ fig : f3](b ) must be renormalized by : @xmath60 we note that the existence of this additional term is quite evident when working directly on the atomic lattice , but it was missed in ref .
@xcite , where the theory was expressed on a continuum .
let us now calculate some shifts of the @xmath3points due to strain .
for example , @xmath61 ( @xmath62 in fig .
[ fig : f3](a ) ) requires an additional contribution , which we find by explicit calculation to be : @xmath63 and using eqn .
( 10 ) one immediately sees that @xmath64 , so that the @xmath1 ( @xmath61 ) and @xmath2 ( @xmath65 ) points shift in opposite directions , as expected @xcite .
equation gives a condition for which the mechanical strain that varies smoothly on the scale of interatomic distances does not break the sublattice symmetry @xcite . on the other hand , arbitrary strain breaks down to some extent the periodicity of the lattice , and `` short - range '' strain can be identified to occur at unit cells where @xmath22 and @xmath23 cease to be zero by significant margins .
this observation provides the rationale for expressing the gauge fields without ever leaving the atomic lattice : when @xmath66 at each unit cell a mechanical distortion can be considered `` long - range , '' and the first - order theory is valid .
the process to lay down the gauge terms to first order is straightforward .
local gauge fields can be computed as low energy approximations to the following @xmath67 pseudospin hamiltonian : @xmath68 with @xmath69 , and @xmath70 .
we defer discussion of the diagonal terms for now . keeping exponents to first order we have : @xmath71 the exponent
is next expressed to first - order : @xmath72.\end{aligned}\ ] ] carrying out explicit calculations , one can see that : @xmath73=0.\ ] ] for example , at @xmath74 we have : @xmath75=0 .
\left[1+\frac{4i\pi(\delta \tau_{1x}+\delta \tau_{2x}+\delta \tau_{3x})}{9a_0}\right](1+e^{\frac{2\pi i}{3}}-e^{\frac{\pi i}{3}}),\ ] ] with phasors adding up to zero .
similar phasor cancelations occur at every other @xmath3point . the term linear on @xmath76 on eqn .
[ eq : cancellation ] cancels out the fictitious @xmath3point dependent gauge fields proposed in ref .
@xcite , which originated from the term linear on @xmath77 on this same equation .
this observation constitutes yet another reason for the formulation of the theory directly on the atomic lattice . with this
we have demonstrated that gauges will not depend explicitly on @xmath3points , so we now continue formulating the theory considering the @xmath61 point only @xcite .
equation takes the following form to first order at @xmath61 in the low - energy regime : @xmath78 with the first term on the right - hand side reducing to the standard pseudospin hamiltonian in the absence of strain .
the change of the hopping parameter @xmath79 is related to the variation of length , as explained in refs .
@xcite and @xcite : @xmath80 this way eqn . becomes : @xmath81 with @xmath82 $ ] , and @xmath83 .
the parameter @xmath84 can be expressed in terms of a vector potential : @xmath85 @xmath86 . this way : @xmath87.\end{aligned}\ ] ] we finally analyze the diagonal entries in eqn .
, which are given as follows @xcite : @xmath88 and @xmath89 these entries represent the scalar deformation potential which we take to linear order in the average bond increase @xcite .
we next establish how the theory based on a continuum relates to the present formalism . in the absence of significant curvature ,
the continuum limit is achieved when @xmath90 ( for @xmath91 ) .
we have then ( cauchy - born rule ) : @xmath92 , where @xmath93 are the entries of the strain tensor . this way eqn . becomes : @xmath94 as expected @xcite .
equation confirms that if the zigzag direction is parallel to the @xmath16axis the vector potential we have obtained is consistent with known results in the proper limit @xcite . besides representing a consistent first - order formalism , the present approach is exceptionally suited for the analysis of `` raw '' atomistic data obtained , for example , from molecular dynamics simulations as there is no need to determine the strain tensor explicitly : the relevant equations ( [ eq : asdiscrete ] , [ eq : esa ] , [ eq : esb ] ) take as input the changes in atomic positions upon strain . within the present approach
@xmath95 space - modulated pseudospinor hamiltonians can be built for a graphene membrane having @xmath96 atoms .
we finish the present contribution by briefly illustrating the formalism on two experimentally relevant case examples .
the developments presented here are motivated by recent experiments where freestanding graphene membranes are studied by local probes @xcite .
( one must keep in mind , nevertheless , that the theory provided up to this point is rather general . )
it is an established fact that graphene membranes will be naturally rippled due to a number of physical processes , including temperature - induced ( i.e. , dynamic ) structural distortions @xcite , and static structural distortions created by the mechanical and electrostatic interaction with a substrate , a deposition process @xcite , or line stress at the edges of finite - size membranes @xcite . in reference
@xcite it is argued that the rippled texture of freestanding graphene leads to observable consequences , the strongest being a sizeable velocity renormalization . in order to demonstrate such statement , one must take a closer look at the underlying mechanics of the problem .
the model @xcite assumes that a graphene membrane is originally pre - strained ( in bringing an analogy , one would say that the membrane would be an `` ironed tablecloth '' ) , so that curvature due to a single wrinkle directly leads to increases in interatomic distances .
those distance increases directly modify the metric on the curved space . in practice
, an external electrostatic field can be used to realize such pre - strained configuration @xcite . in improving the consideration of the mechanics beyond first - order continuum elasticity ,
let us consider what happens if this pre - strained assumption is relaxed ( in continuing our analogy , the rippled membrane in fig .
[ fig : f4](a ) would then be akin to a `` wrinkled tablecloth prior to ironing '' ) : how do the gauge fields look in such scenario ? with our formalism , we can probe the interrelation between mechanics and the electronic structure directly . in figure
[ fig : f4](a ) we display a graphene membrane with three million atoms at 1 kelvin after relaxing strain at the edges .
the strain relaxation proceeds by the formation of ripples or wrinkles on the membrane .
this initial configuration is already different to a flat ( `` pre - strained '' ) configuration within the continuum formalism , customarily enforced prior to the application of strain . _
the ripples must be `` ironed out '' before any significant increase on interatomic distances can occur : _
`` isometric deformations '' lead to curvature without any increase on interatomic distances @xcite ( in continuing our analogy , this is usually what happens with clothing ) .
we believe that a local determination of the metric tensor from atomic displacements alone will definitely be useful in continuing making a case for velocity renormalization @xcite ; this is presently work in progress @xcite .
the local density of electronic states is obtained directly from the hamiltonian of the membrane in configuration space @xmath11 , and shown in fig .
[ fig : f4](b ) . when compared to the dos from a completely flat membrane , no observable variation on the slope of the dos appears , and hence , no renormalization of the fermi velocity either .
one can determine the extent to which nearest - neighbor vectors will preserve sublattice symmetry in terms of @xmath97 and @xmath23 , eqns .
( 4 - 6 ) .
we observe small and apparently random fluctuations on those measures in fig .
[ fig : f4](c ) : @xmath98 1% and @xmath99 .
we display the deformation potential in figure [ fig : f4](d ) in terms of the average ( @xmath100 ) and difference ( @xmath101 ) between @xmath102 and @xmath103 ( eqns .
( [ eq : esa ] ) and ( [ eq : esb ] ) ) at any given unit cell : @xmath104 both quantities are of the order of tens of mevs .
the ripples lead to the random - looking pseudo - magnetic field shown in fig .
[ fig : f4](e ) , reminiscent of the electron density plots created by random charge puddles @xcite .
we next consider how strain by a sharp probe modifies the results in fig .
[ fig : f4 ] . in what follows we consider a central extruder creating strain on the freestanding membrane . for this
, we placed the membrane shown in fig .
[ fig : f4 ] on top of a substrate ( shown in blue / light gray in fig . [
fig : f5](a ) ) with a triangular - shaped hole ( in green / dark gray in fig .
[ fig : f5](a ) ) .
the membrane is held fixed in position when on the substrate , and pushed down by a sharp tip at its geometrical center , down to a distance @xmath105=10 nm . as indicated earlier , sublattice symmetry is not exactly satisfied right underneath the tip , where @xmath97 and @xmath23 take their largest values ( fig .
[ fig : f5](c ) ) .
while @xmath23 still displays some fluctuations , this is not the case for @xmath22 ( the scale for @xmath22 is identical to that from fig .
[ fig : f4](c ) ) .
the large white areas tells us that fluctuations on @xmath97 are wiped out upon load as the extruder removes wrinkles .
this observation stems from the lattice - explicit consideration of the mechanics .
we have presented a detailed discussion of the problem along these lines @xcite .
we found that for small magnitudes of load a rippled membrane will adapt to an extruding tip isometrically .
this observation is important in the context of the formulation with curvature @xcite , because in that formulation there is the assumption that distances between atoms increase as soon as graphene deviates from a perfect 2-dimensional plate .
the gauge fields given in fig .
[ fig : f5](c - d ) reflect the circular symmetry induced by the circular shape of the extruding tip @xcite . .
the locations where the dos is computed are shown in the insets ( the most symmetric line patterns are displayed in yellow).,scaledwidth=48.0% ] we finish the discussion by probing the local density of states at many locations in fig .
[ fig : f6 ] , which may relate to the discussion of confinement by gauge fields @xcite .
@xmath106 was was not included in computing dos curves .
some generic features of dos are clearly visible : ( i ) near the extruder , the deformation is already beyond the linear regime , and the dos is indeed renormalized for locations close to the mechanical extruder @xcite .
( ii ) a sequence of features appear on the dos farther away from the extruder .
because the field is not homogeneous and perhaps due to energy broadening we are unable to tell a central peak . as indicated on the insets , the plots on fig .
6(b ) and 6(d ) are obtained along high - symmetry lines ( the colors on the dos subplots correspond with the colored lines on the insets ) .
for this reason they look almost identical , and the three sets of curves ( corresponding to the dos along different lines ) overlap . due to lower symmetry , the ldos in fig .
6(a ) and 6(c ) appear symmetric in pairs , with the exception of the plots highlighted in gray .
( the light v-shaped curve in all subplots is the reference dos in the absence of strain ) .
ldos curves complement the insight obtained from gauge field plots .
hence , they should also be reported in discussing strain engineering of graphene s electronic structure , particularly in situations where gauge fields are inhomogeneous .
we presented a novel framework to study the relation between mechanical strain and the electronic structure of graphene membranes .
gauge fields are expressed directly in terms of changes in atomic positions upon strain . within this approach , it is possible to determine the extent to which the sublattice symmetry is preserved .
in addition , we find that there are no @xmath3dependent gauge fields in the first - order theory . we have illustrated the method by computing the strain - induced gauge fields on a rippled graphene membrane with and without mechanical load .
in doing so , we have initiated a necessary discussion of mechanical effects falling beyond a description within first - order continuum elasticity .
such analysis is relevant for accurate determination of gauge fields and has not received proper attention yet . +
* acknowledgments * + we acknowledge conversations with b. uchoa , and computer support from hpc at arkansas ( _ razorii _ ) , and xsede ( tg - phy090002 , _ blacklight _ , and _ stampede _ ) . m.v .
acknowledges support by the serbian ministry of science , project no . | we present a new first - order approach to strain - engineering of graphene s electronic structure where no continuous displacement field @xmath0 is required .
the approach is valid for negligible curvature .
the theory is directly expressed in terms of atomic displacements under mechanical load , such that one can determine if mechanical strain is varying smoothly at each unit cell , and the extent to which sublattice symmetry holds . since strain deforms lattice vectors at each unit cell ,
orthogonality between lattice and reciprocal lattice vectors leads to renormalization of the reciprocal lattice vectors as well , making the @xmath1 and @xmath2 points shift in opposite directions . from this observation
we conclude that no @xmath3dependent gauges enter on a first - order theory . in this formulation of the theory
the deformation potential and pseudo - magnetic field take discrete values at each graphene unit cell .
we illustrate the formalism by providing strain - generated fields and local density of electronic states on graphene membranes with large numbers of atoms .
the present method complements and goes beyond the prevalent approach , where strain engineering in graphene is based upon first - order continuum elasticity . |
the theory of polymers near surfaces is a very important subject for theoretical investigation .
the main reason is the very broad and deep theoretical research possible for these types of systems .
modern theories have been developed and brought to a very high standard @xcite .
on the other hand there is a large demand on practical interest on studying polymers near surfaces .
one of them is surface protection .
the theory on polymer brushes @xcite is a typical example of these type of applications .
polymers are attached to surfaces to protect the surface from further adsorption of , e.g. , biological active molecules such as proteins .
these molecules need to be pretended to from all kind of surface interactions in order to save their biological function @xcite .
the ( technical ) problem with polymer brushes is , that a large surface coverage is needed to protect the surface in a most effective way .
polymers attached chemically or physically at one end at the surface form brushes , i.e. , the chains are extended and the brush height @xmath0 follows a scaling law @xmath1 ( in good solvent ) , where @xmath2 is the area per chain an is thus related to the grafting density . @xmath3 the degree of polymerization of the chains .
the problem is to set up a small value of @xmath2 , or , correspondingly a large grafting density . for low values of the grafting density the chains behave as `` mushrooms '' and
the surface protection is incomplete . in earlier papers
it has been shown by one of us , that the use of branched chains in much more effective @xcite .
chains , polymers and polymeric fractals with a larger connectivity seem to be more appropriate to protect surfaces more effectively . indeed due to their connectivity
their occupied area is larger and it turns out that these systems behave more like single chain fractals .
typical many body effects , such as occur in conventional polymer brushes do not play a significant role . in the present paper
we suggest a different route of surface protection by using microgels and branched structure .
microgels had become an important tool in designing polymeric nano structures .
these systems can be synthesized @xcite with different structures .
indeed the structure of these microgels can range from a fractal state , i.e. , a branched self - similar polymer with a large connectivity , up to almost hard and highly crosslinked spheres .
such systems are well designed to study the transition form polymer to colloid behavior by variation of the structure and the crosslinking state .
we have shown earlier @xcite , that fractal polymers and gels can interpenetrates each other and screen excluded volume forces , wherever their connectivity is low . in terms of the spectral dimension this is the case , when it is lower than a critical value , i.e. @xmath4 . thus fractals with lower spectral dimensions screen their excluded volume , whereas polymeric fractals with larger spectral dimension saturate .
then they form soft balls , which can not interpenetrate each other and are well separated from each other @xcite .
this state has an analogon in the case of linear polymers .
polymer melts in two dimensions correspond to a saturated state .
the individual chains are separated from each other and form on average disks on a hexagonal lattice @xcite . for surface protection in three dimensions it would thus be more effective to use fractal microgels with a large connectivity .
then the surface coverage is ruled by the single gel behavior .
alternatively crosslinked gels can be used to have the same effect .
sufficiently crosslinked gels can not interpenetrate each other .
the adsorption behavior can be studied then by the single gel adsorption .
the significant parameters are either the connectivity ( spectral dimension @xmath5 ) or the crosslink number @xmath6 . in the following we emphasize mainly on the crosslinked gels rather on the self similar connected polymeric fractals .
nevertheless we will consider both cases below .
the paper is organized as follows . in section
[ 2 ] we introduce the model of self crosslinked polymer chains , which form the microgels . in section
[ 3 ] we repeat briefly the scaling behavior of ideal microgels before we consider in section [ 4 ] the effects of excluded volume in the bulk .
sections [ 5 ] and [ 6 ] treat the adsorption behavior of the gels close to attracting walls using simple scaling arguments . in section [ 5 ] we will also make some remarks on fractal type gels .
flexible interacting macromolecule modeled usually by the edwards hamiltonian in three dimensional space .
the edwards hamiltonian consists of two parts , i.e.,the gaussian connectivity of monomers @xmath7 and the self avoidance between monomers @xmath8 thereby @xmath9 is the excluded volume of the monomers , @xmath3 the degree of polymerization , and @xmath10 the kuhn length .
the chain configurations are determined by monomer coordinates @xmath11 , where @xmath12 labels all monomers @xmath13 .
the edwards hamiltonian is sufficient to describe a free self avoiding walk chain . to study the properties of a microgel crosslinks have to be introduced .
the most obvious statistical representation of a microgel is a self crosslinked single chain .
such a situation has been studied many years ago by the manchester group in three papers @xcite .
the static properties have been computed by variational techniques . in this paper
we choose a different route .
let us therefore consider a microgel as a self crosslinked polymer of roughly spherical structure , which can be visualized as given in fig.([one ] ) .
in this paper we want to describe the crosslinks such that a continuous transition between the free chain and the fully crosslinked state can be represented in the same model .
this has been motivated by our earlier work , where soft crosslinks have been introduced .
the method of using soft crosslinks is able to interpolate from the free chain to the hard microgel .
the state in between is a new kind of branched chain , whose scaling properties have been already described @xcite . to be more precise
let us introduce @xmath6 permanently crosslinked monomers , where each of them is characterized by pair of randomly chosen monomer coordinates @xmath14 that form a permanent crosslink .
in fact , the whole set of crosslinks @xmath15 determine the random connectivity of the micro network .
these definitions and proposals allow us to formulate the partition function of the microgel , i.e. , @xmath16 the partition function describes the gaussian network with self avoiding interactions , and takes into account the total connectivity of the certain crosslink configuration c. if the delta constraint for the permanent crosslinks is represented by a soft gaussian function , i.e. , @xmath17 , the problem can be solved exactly .
thus we model the @xmath18-function by gaussian distribution with width @xmath19 in limit @xmath20 and formulate the hamiltonian of the crosslinked chain by @xmath21
although we have shown in previous publications that the partition function can be solved for any value of @xmath22 exactly it is useful to rederive the results previously by the use of scaling arguments .
these consideration will reproduce the exact results apart from prefactors . to do
so it is useful to recall that the classical random walk contains two elastic contributions , one for stretching and one for compression .
the addition of both yield a free energy @xmath23 and by minimization the size of gaussian chain @xmath24 is recovered .
let us shortly repeat the results for later use . for the case of soft crosslinks ,
i.e. , whenever @xmath22 is within the range @xmath25 we have for the relevant part of flory free energy @xmath26 and minimization yields the branched polymer regime @xmath27 the appearance of the typical branched polymer exponent @xmath28 is not surprising in the ideal case , since the constraint can be visualized as springs .
then the connectivity is changed .
the branched polymer regime in the range of @xmath29 was confirmed also by the exact solution of the crosslinked chain problem , see @xcite .
moreover it agrees also with the corresponding branched polymer @xcite in the case of hard crosslinks , whenever @xmath30 the crosslink term must be differently estimated .
to do so the crosslink term can be estimated by the size of the random walk through crosslinks .
then the relevant part of flory free energy is then given by @xmath31 minimization of the free energy provides the size of the microgel by @xmath32 as also given by the exact results @xcite .
formally the latter result can be found from a special choice for @xmath22 from the corresponding result for soft gels ( [ soft ] ) , but note that the way of estimating the free energy contribution of the crosslinks is estimated very differently .
thus there can be a different prefactor , which is not accessible by scaling .
the exact values for the radius of gyration of the non - interacting but crosslinked chains have been computed exactly in @xcite , where the numerical prefactors can be found .
although we have been able to compute the size of the microgel exactly whenever the interactions are not present the self avoiding case appears very difficult . the only possibility for the problem at this stage is to use flory estimates for size @xmath33 .
let us first consider the case of soft crosslinks .
the distance constraint that forces two arbitrarily chosen polymer segments together shrinks the chain
. the shrinkage costs entropy penalty which balances with the distance constraint .
the use of the pseudo potentials allows us , however , to cast this in a simple flory free energy to @xmath34 and minimization yields the size of the swollen soft microgel @xmath35 the result is very intriguing .
although the gaussian chain size scales the same way as the branched polymer , the swelling behavior produces another excluded volume exponent , resulting in a different swelling behavior as branched chains . in the latter case
the swollen branched chain is characterized by @xmath36 . on the other hand
it can also be seen that the pure scaling in terms of the variable @xmath37 always present in the gaussian case is destroyed .
this becomes clear , since the excluded volume introduces interactions .
a similar estimate of the size can be carried out in the case of hard crosslinks . here
the elastic term stemming from the pseudo crosslink potential can be estimated the same way as in the case of non interacting networks , which was given by a random walk through the crosslinks .
the total excluded volume energy remain the same , because it depends only on the total amount of monomers and not on the special connectivity .
then minimization of the free energy @xmath38 yields swollen c*-micro gels of size @xmath39 the size might appear small compared to what is expected intuitively .
this comes from the fact that the crosslinks has been chosen totally randomly . in many theories of macroscopic networks
the choice of the crosslink pairs is guided by the conformation of the excluded - or random walk chain , i.e. , the often terminated `` zeroth replica '' @xcite .
for completeness we mention that the number of crosslinks in excluded volume gels can not be arbitrarily large .
a natural limit of the crosslink number is given by the condition that the size must be larger than a fully collapsed ball of space filling density , i.e. @xmath40 . the latter condition yields the upper limit for the number of crosslinks to be @xmath41 .
let us first study the adsorption of ideal microgels near a flat surface by naive scaling arguments . to do so
, we repeat the scaling idea of de gennes for arbitrarily flexible objects of arbitrary connectivity , but selfsimilarly linked . the connectivity of of such gaussian fractal networks can be described by the spectral dimension @xmath5 .
the total number of monomers is therefore @xmath42 , and their ideal size is given by @xmath43 , yielding a fractal dimension of @xmath44 .
thus @xmath45 counts the number of monomers through a linear dimension through the fractal object .
this way of description includes the wellknown cases of linear chains for @xmath46 and for randomly branched polymers , i.e. , @xmath47 .
a simple way of looking at the adsorption conditions is to compare the free energy penalty of confinement of the gaussian structure near the wall with the gain of energy by adsorbing a certain number of monomers .
@xmath48 for gaussian chains the confinement is simply given by by @xmath49 .
here we have used the symbol @xmath50 for the height of adsorbed layer , @xmath51 is the number of adsorbed monomers , and @xmath52 is the gain of energy per @xmath53 . following de gennes book @xcite
we can reproduce the result given there for the linear ideal chain . to do so we must first estimate the number of adsorbed monomers at the surface , we assume that the surface is penetrable for a moment .
this yields immediately @xmath54 the main problem is to estimate the confinement free energy for gaussian fractals with a larger connectivity compared to the linear chain .
this is definitely not just the inverse of the gaussian free energy of stretching , i.e. , @xmath55 , because upon stretching only the monomers in the shortest path are taking part on the deformation , whereas upon confinement the total number of monomers are concerned .
the corresponding confinement free energy must then be of the form @xmath56 this result contains the special case of linear chains , @xmath57 , and the latter agrees with the classical confinement free energy .
for the ideal gaussian structures the height of the adsorbed layer scales as @xmath58 the same result can be found by a blob argument . for selfsimilarly branched polymers a blob model can be used @xcite .
the manifold is confined to a height @xmath50 . inside blobs of diameter
@xmath50 the branched chain is gaussian , i.e. , @xmath59 , where @xmath60 is the number of monomers inside the blob .
thus the number of blobs is given by @xmath61 .
the confinement free energy is proportional to the number of blobs , i.e. , @xmath62 . employing the same scaling argument as above yields immediately @xmath63 which agrees with the above result , eq.([hhh ] ) . in any case
, the above arguments only yield the behavior of ideal chains near the surface . for excluded volume chains and
excluded volume manifolds the gaussian elastic entropy penalty must be replaced by the confinement energy .
to do so , the manifold can be put between two plates of distance @xmath50 .
this procedure yields similar results along to those derived in de gennes book @xcite . to do so , the extension of the manifold between two parallel plates must be computed .
it is given by @xmath64 @xcite .
this result is consistent with the linear saw chain between two plates . for @xmath46
the two dimensional chain of blobs is recovered .
the confinement free energy can be only a function of the ration of the size of the manifold and the distance between the plates , i.e. , @xmath65 , where @xmath33 is the size of the self avoiding manifold @xmath66 .
the confinement free energy is then easily found from the condition that the free energy must be an extensive quantity .
thus it must scale as @xmath67 , which yields @xmath68 replacing the elastic free energy in the scaling argument above by the confinement energy , yields the physically sensible result @xmath69 which is now independent of the molecular weight for any manifold .
moreover for @xmath46 the linear chain result is recovered .
moreover the result bears an interesting point in it : the parameter @xmath52 is the gain of energy at adsorption of one segment per thermal energy @xmath70 .
thus it is physically reasonable that this parameter is sufficiently small , i.e. , @xmath71 .
thus a significant change of the behavior can be expected if the exponent in @xmath50 is larger and smaller than one .
it is interesting to note that this happens at @xmath47 which is close to the spectral dimension of randomly branched chains or accidentally for percolation clusters .
thus randomly connected manifolds of large connectivity adsorb weaker than objects of low connectivity , as linear chains .
this is physically intuitively clear since the number of accessible sites become smaller for increasing connectivity . in the following section we will use the same strategy to discuss the adsorption behavior of microgels
as we have seen in the last paragraph on the previous section we have to construct the free energy of confinement .
we use the same concept of gels between two plates for energy cost of squeezing .
we have seen in the first two sections that we can distinguish between soft and hard microgels by the value of the parameter @xmath22 . for both cases
we expect physically different behavior .
let us first consider soft gels between plates .
to do so we have to determine the size of the gel parallel to the plates .
the relevant parts of the free energy is given by @xmath72 where we have argued that the elastic confinement and the `` anisotropic '' excluded volume term balance each other .
this yields immediately @xmath73 and the corresponding confinement energy to @xmath74 note that the confinement free energy is determined by the fact that it must be proportional to the total number of monomers , since the free energy is extensive . the same procedure can be employed for hard gels .
the relevant free energy takes a similar form as before , apart from the elastic part of the energy .
@xmath75 the size of the gel parallel to the plates is given for completeness .
it scales as @xmath76 . to find the confinement free energy the same argument yields
@xmath77 an important observation is that for @xmath78 ( no crosslink ) linear chain is recovered .
finally we are in the position to discuss the adsorption behavior of the gels . to begin with ,
we employ the same scaling arguments as given in the case of self - similar polymeric fractal .
thus we have to consider the competition between confinement , or the entropy penalty of confinement and the energy gain by adsorption .
this results in a total free energy of the general form @xmath79 we just summarize the results to be brief .
first for soft gels we find @xmath80 similarly for hard gels @xmath81 note that the latter equation contains the free chain result for @xmath78 , i.e. , if no crosslinks are present .
we see that in both cases the height of the adsorbed layer depends on the number of crosslinks in a significant and characteristic way .
the results are in accordance with the physical intuition .
the soft microgels adsorb more easily , because these objects are more flexible .
this is also shown by the different exponents of the interaction energy @xmath52 .
with use of flory - approximation we had investigated behavior of microgels .
the complexity of the distribution function of the non interacting network prevented us to use more refined methods , as they are well known in the case of linear polymer chains .
nevertheless we got results which are reasonable and could be checked by experimental methods , at least in their tendency .
the cases worked out here have been relevant for penetrable surfaces , i.e. , interfaces .
a direct comparison to hard surfaces is not possible , since the number of monomers close to the surface can not be determined by eq .
( [ mono ] ) .
the case of linear chains in half space has been studied in detail , see e.g. @xcite for a general reference .
crossover exponents and new critical points determine the physics . in the present case of microgels and polymeric manifolds
a similar treatment appears very difficult , since the `` bare propagator '' has a very complicated structure , although it is exactly known @xcite .
nevertheless we expect that the principal statements can be compared at least qualitatively with experiments . to study the adsorption behavior we first had to calculate the size of the microgel in solution .
this has been carried out by employing the flory arguments .
the basis for the reliability of the result has been their agreement withe comparison of exact calculations without excluded volume in bulk .
then generalization opened the determination of sizes of microgels with excluded volume interaction in bulk systems and near adsorbing flat surfaces .
moreover we made straightforward generalizations to fractal type microgels , which could be described by the spectral dimension .
the results show a transition from polymeric type of adsorption behavior for soft microgels , i.e. , with small number of crosslinks , or alternatively low spectral dimension , to colloidal adsorption , whenever the crosslink number is large and their coupling is strong .
we have seen that the results have been presented from considerations on single gels .
this has also experimental interest on surface protection .
a layer of adsorbed gels of height @xmath50 at a surface is a single gel problem . unlike linear polymer chains the gels
can no longer interpenetrate each other .
thus microgels and branched polymers appear more effective in surface protection .
another interesting question is also the interplay between vulcanization and adsorption . in the case we studied so far
, we had assumed preformed gels and fractals which had then brought to the interacting surfaces .
the other case , i.e. , the vulcanization in presence of interacting walls would lead to new types of gels with new structures which depend on the strength of the surface - monomer interaction @xcite .
lun des auteurs ( tav ) remercie pierre - gilles de gennes et elie raphal pour leurs remarques interessantes sur ce travail .
nous remercions aussi lun des critiques pour avoir attir notre attention sur une methode lgante dobtantion du `` blob '' , qui a permis dameliorer la qualit de ce papier .
e. eisenriegler , _ polymers at surfaces _ , world scientific , singapore , 1994 p.g .
de gennes , _ simple views on condensed matter _ , world scientific , singapore , 1992 j. rhe , preprint , 1997 t.a .
vilgis , p. haronska , macromol .
theory simul .
, * 4 * , 111 , ( 1995 ) t.a .
vilgis , p. haronska , m. benhamou , j. phys.ii , * 4 * , 2139 , 1994 e. zhulina , t.a .
vilgis , macromolecules , * 28 * , 1008 , ( 1995 ) m. antonietti , d. ehrlich , k.j .
flsch , h. sillescu , m. schmidt , p. lindner , _ macromolecules _ ,
* 22 * ( 1989 ) 2802 m. antonietti , k.j .
flsch , h. sillescu , t. pakula , _ macromolecules _ , * 22 * ( 1989 ) 2812 m. antonietti , w. bremser , k.j .
flsch , h. sillescu , _ makromol .
, makromol .
_ , * 30 * ( 1989 ) 81 m. antonietti , ch .
rosenauer , _ macromolecules _ , * 24 * ( 1991 ) 3434 t.a .
vilgis , physica a * 153 * , 341 , ( 1988 ) p. haronska , t.a .
vilgis , j. chem .
phys . , * 102 * , 6586 , ( 1995 ) i. carmesin , k. kremer , j. phys .
( france ) , * 51 * , 915 , ( 1990 ) g.allen , j. burgess , s.f .
edwards , d.j .
walsh , proc .
a , * 334 * , 453 , 465 , 477 , ( 1973 ) m.p .
solf , t.a .
vilgis , j. phys .
a , * 28 * , 6655 , ( 1995 ) m.p .
solf , t.a .
vilgis , j. phys .
i ( france ) , * 6 * , 1451 , ( 1996 ) m.p .
solf , t.a .
vilgis , phys .
e , * 55 * , 3037 , ( 1997 ) t.a .
vilgis , macromol .
theory simul .
, in press , ( 1997 ) p.g .
de gennes , _ scaling concepts in polymer physics _ , cornell university press , ithaca , 1979 r. t. deam , s.f .
edwards , proc .
a * 260 * , 31 , ( 1976 ) m.a .
cohen stuart , j.t.f .
keurentjes , b.c .
bonekamp and j.g.e.m .
fraaye , colloids surfaces * 17 * , 91 , ( 1986 ) m. abel , r. lipowsky , unpublished | the behavior of microgels near surfaces and their adsorption is studied by simple scaling theory .
two different types of microgels can be studied , i.e. , fractal type microgels and randomly crosslinked polymer chains . in the first case
the gel can be described mainly by introducing a spectral dimension .
the second type requires more attention and uses the number of crosslinks as parameter .
the main result is that soft gels with weakly coupled crosslinks and a low number of crosslinks adsorb much better than hard gels , with many crosslinks .
similar results for fractal gels and branched polymer are presented .
fractal gels with low connectivity adsorb easier than gels with a large connectivity dimension .
we discuss also consequences on surface protection by microgels . |
in the framework of loop quantum cosmology ( lqc)@xcite one can introduce phenomenological hamiltonian @xcite @xmath4 ^ 2 \nonumber \\ & + & \frac{1}{2 } \frac { \pi_{\phi}^2 } { { |p|}^{3/2 } } + { |p|}^{3/2}v(\phi ) . \label{model}\end{aligned}\ ] ] here effects of the quantum holonomies have been introduced .
the classical limit corresponds to the case @xmath5 , what gives @xmath6 .
the parameter @xmath7 is in general function of the canonical variable @xmath8 . in our considerations
we choose @xmath9 where @xmath10 .
it was shown that this leads to the proper classical limit and can be treated as a unique choice in the loop quantization class @xcite .
based on the hamilton equation @xmath11 we can derive equations of motion for the canonical variables @xmath12 . the variables @xmath13 can be related to the standard frw variables @xmath14 .
subsequently , with use of the hamiltonian constraint @xmath15 we can derive modified friedmann equation @xmath16 where @xmath17 and @xmath18 the equation of motion of the scalar field @xmath19 holds classical form @xmath20 the special limit of the above theory is the case @xmath21 , then @xmath22 .
it means that solution of the equation ( [ friedmann ] ) traces the mean value @xmath23 ( for rigorous verification of the effective scenario in lqc see ref .
@xcite ) . it is however not necessarily the true for the general case with @xmath24 .
however , even when we are not sure whether theory is truly effective one , it is interesting to investigate phenomenological consequences of the models with potentials .
such studies have been performed for the broad class of potentials . in ref .
@xcite model ( [ model ] ) with constant potential ( @xmath25 ) has been solved analytically .
power law , cyclic and bicyclic potentials have been numerically studied in ref .
recently model with exponential potential has been studied , exhibiting sudden singularity behaviour @xcite .
other examples can be found in ref .
@xcite . in this letter
we perform restriction for the scalar field energy density and pressure @xmath0 with @xmath26 const . in this situation scalar field
can mimic any matter with equation of state @xmath27 where @xmath26 const and @xmath1 $ ] .
based on this restriction we derive expression for the scalar field potential which fulfills this condition .
we repeat these considerations for the case of phantom field @xmath28 .
subsequently we highlight possible applications of the model .
in particular we consider scalar field perturbations and indicate its non - gaussian features .
this can be important from the point of observational constraints for the model .
namely since the obtained potential is non - quadratic , interaction between different modes is present , leading to the nonlinear effects in the cmb radiation .
therefore presented model can be potentially verified with astronomical observations .
we show that scalar field with derived potential is monotonic function of coordinate time .
this is important property since in quantum cosmological models we are looking for a well - defined intrinsic time .
this is necessary since the hamiltonian constraint @xmath29 leads to the time coordinate independent wheeler - dewitt equation @xmath30
. therefore wave function of the universe @xmath31 does not depend on time explicitly and some new intrinsic time must be introduced to trace quantum evolution .
monotonic scalar field is good candidate .
namely performing canonical quantisation we replace @xmath32 .
then matter part of the hamiltonian is @xmath33 now @xmath19 plays a role of time in analogy with the klein - gordon equation .
energy density and pressure of the homogeneous scalar field are expressed as follows @xmath34 in this section we consider restriction @xmath0 where @xmath2 const .
this assumption leads to the expression @xmath35 which gives @xmath36 with use of equations ( [ friedmann]),([fieldequation ] ) and considered condition we obtain @xmath37 the above equation has solution in the form @xmath38 } \label{potential}\ ] ] where @xmath39 .
it is worth to mention that for @xmath40 we recover @xmath41 which is well - known classical limit . in fig .
[ pot ] we draw potential ( [ potential ] ) for the different values of parameter @xmath42 . .
in the figure the value @xmath43 was assumed.,width=264 ] particular situation corresponds to the case @xmath44 @xmath45 .
field can get such an equation of state only in the limit @xmath46 . therefore this particular case must be excluded from considerations .
namely , we can not mimic pure cosmological constant with time dependent scalar field .
we must to stress that potential in the form @xmath47 has already appeared in the different context in the loop cosmology .
namely , it was obtained by singh @xcite for the scaling solutions in lqc dual to those in randall - sundrum cosmology .
these considerations were performed for the model with a self - interacting scalar field and matter with a fixed equation of state @xmath42 .
in this section we show analytical solution of the model with potential ( [ potential ] ) . for the considered case we can additionally derive equation @xmath48 with use of continuity equation @xmath49 ( or equivalently equation ( [ fieldequation ] ) )
we directly obtain @xmath50 comparing this to the expression ( [ rhov ] ) with ( [ potential ] ) we obtain @xmath51 \right)^{\frac{2}{3(1+w)}}. \label{solp}\ ] ] similar solution , however not in such an explicit form , has already been derived in ref .
@xcite . to obtain time @xmath52 dependence we use expression @xmath35 with potential ( [ potential ] ) to obtain @xmath53 = 1 + 6\pi g \rho_c ( 1+w)^2 ( t - t_0)^2 . \label{phi_t}\ ] ] this leads to @xmath54 this solution can be also found in ref .
@xcite . in case of the @xmath55 solution
has exponential form and can be found in the paper @xcite . in fig .
[ bounce ] we show time dependence of the canonical variable @xmath8 for different values of parameter @xmath42 . for the different values of parameter @xmath42.,width=264 ] resulting dynamics is singularity free bounce .
the time derivative of the field @xmath19 can be obtained from the equation ( [ phi_t ] ) and has a form @xmath56 from this we see that @xmath19 is monotonic function of time ( growing or decreasing ) .
choosing growing solution we have @xmath57}{\sqrt{6\pi g(1+w)}}. \label{solphit}\ ] ] we draw this function in fig .
[ field ] . for the different values of parameter @xmath42.,width=264 ] to complete above reasoning we verify obtained solutions in appendix [ sec : appendix ] .
namely we show there how to solve equations of motion without imposing the condition @xmath26 const in equation of state . instead of this
we solve equations from scratch assuming the form of potential ( [ potential ] ) .
in this section we perform analogous consideration for the case of phantom field @xcite . for this kind of field energy density and pressure
are expressed as follows @xmath58 from this we see that @xmath59 .
the parameter @xmath42 in the equation of state for the phantom field takes values @xmath3 .
phantom matter was introduced in loop quantum cosmology in ref .
subsequently phantom field in the loop quantum cosmology has been studied in ref . @xcite . also the models with interacting phantom field
have been investigated @xcite .
based on the continuity equation we derive equation of motion @xmath60 performing calculations in the same way like in the section [ sec : multi ] we obtain @xmath61 } \label{potential2}\ ] ] where @xmath39 .
solution for the scalar phantom field has a form @xmath62}{\sqrt{-6\pi g(1+w)}}. \label{phiphantom}\ ] ] it is monotonically growing function too .
therefore phantom field with potential ( [ potential2 ] ) can be treated as internal time .
the canonical parameter @xmath8 depends on value of the @xmath19 as follows @xmath63 \right)^{\frac{2}{3(1+w)}}.\ ] ] rewriting equation ( [ phiphantom ] ) to the form @xmath64 = 1 + 6\pi g \rho_c ( 1+w)^2 ( t - t_0)^2\ ] ] one can obtain explicit time dependence of the parameter @xmath8 .
we show this function in fig .
[ phan ] . for the different values of parameter @xmath42.,width=264 ]
we see that , in contrast to the bouncing behaviour , here universe shrinks to zero for @xmath65 . to investigate this state we consider time dependence of the energy density of the phantom field .
we have @xmath66 what is shown in fig . [ rho ] . for the different values of parameter @xmath42.,width=264 ]
we see that despite the variable @xmath8 tends to zero energy density decreases too .
in the previous part of this letter we have considered models with a homogeneous scalar field .
now we are going to take into account perturbations .
we consider simplified model where scalar field is perturbed while background geometry holds its homogeneity .
moreover we aim rather to indicate further applications of the potential ( [ potential ] ) than to perform full analysis .
therefore considerations presented should be seen as an invitation to the more detailed studies .
we split the scalar field @xmath67 as @xmath68 here @xmath69 is homogeneous mode defined as @xmath70 where @xmath71 is some fixed fiducial volume .
then equation for the background part is @xmath72 therefore solutions of equations of motion for the background part are those derived in section [ sec : analytical ] with @xmath73 .
first order perturbation of the full equation of motion leads to @xmath74 which is valid for @xmath75 .
here we have defined the mass term @xmath76 where @xmath77 = 1 + 6\pi g \rho_c ( 1+w)^2 t^2 . \label{phi_t}\ ] ] performing fourier transform @xmath78 and introducing conformal time @xmath79 we can write equation for the mode functions @xmath80 as @xmath81u({\bf k } , \eta)=0 . \label{equ}\ ] ] here we have defined effective mass @xmath82 . \label{meffdef}\ ] ] based on the equation ( [ solp ] ) we can write solution for the scale factor as follows @xmath83 applying this to the equation ( [ meffdef ] ) we obtain @xmath84 \label{meff2}\end{aligned}\ ] ] where @xmath85 the full analysis of the equation ( [ equ ] ) with @xmath86 given above is behind the scope of this letter .
however we can immediately investigate its behaviour in the pre - bounce phase @xmath87 . in this limit @xmath88 where we have changed time for conformal .
advanced and normalised solution of the equation ( [ equ ] ) in considered limit is @xmath89 where @xmath90 this solution is valid for @xmath91 or equivalently for @xmath92 \cup \{1\}$ ] . therefore only in the small range .
power spectrum of perturbations is then given as @xmath93 where super - horizontal approximation @xmath94 has been used .
it is worth to mention that scale invariant spectrum @xmath95 is recovered only for @xmath55 .
therefore duality investigated in ref .
@xcite is broken here due to additional mass term . in the range @xmath96 ( @xmath97 ) we can obtain solutions by the replacement @xmath98 .
then for @xmath99 we have @xmath100 as discussed in ref .
@xcite .
presented solutions are valid in the pre - bounce phase .
later when modes cross the bounce solutions of the equation of motion have a different form . despite the fact that size of horizon diverges at the turning point ( @xmath101 @xmath102 @xmath103 ) the solutions evolve smoothly without any anomalies . in the context of loop cosmology such
a behaviour has been recently approved for gravitational waves based on the analytical and numerical calculations @xcite .
interesting generic feature of the bouncing cosmologies is suppression of power of the low energy models @xcite .
therefore we can expect damping of the low energy modes also for the model studied here .
however it can depend whether condition @xmath75 for validity of the linear approximation is conserved during the whole evolution .
if this condition is not satisfied then nonlinear effects should be also taken into account .
otherwise suppression of power of the low energy modes should be easily obtained .
such an effect is typically presented as a possible explanation of the low cmb multipoles suppression @xcite .
however it seems that fine tuning must occur to match energy scales of the bounce with the cosmological ones at which suppression of the cmb modes is observed .
these issues have to be addressed in the further considerations . scalar field with potential ( [ potential ] )
is non - gaussian . namely modes with different wave numbers
interact one another @xcite .
since observations suggest that primordial perturbations were nearly gaussian @xcite the cmb nonlinearities can in principle be used to restrict models with interacting potentials . in our case potential of interaction
has non - polynomial form therefore non - gaussian features of the field should be significant .
however since we do not know how many perturbations originate from the short quantum phase ( where model studied here can be applied ) we can not infer about their impact on the cmb nonlinearities . in principle contribution from the phase of bounce
can be affected by the succeed phase of inflation .
detailed studies of this issues are however beyond the scope of this letter .
we devote for these issues another paper @xcite while now concentrate only on some preliminary estimations of nonlinear effects .
the lowest order interacting term is given by @xmath104 where @xmath105 this term becomes important when dominates the mass one .
therefore ratio @xmath106 can be seen as a measure of non - gaussianity .
here we have defined parameter @xmath107 we show this function in fig .
[ xi ] . for @xmath108.,width=264 ] at the points where @xmath109 non - gaussian part becomes dominant since the mass vanishes there . in the limits
@xmath110 the parameter @xmath111 stabilises .
above results indicate that non - gaussian effects can be in fact produced .
especially interesting are points where @xmath109 , then nonlinear effects are dominant .
these regions could be an effective source of the cosmological non - gaussianity .
it is therefore in principle possible to constraint this model from cmb nonlinearities .
this could bring observational bounds on parameter @xmath112 or equivalently barbero - immirzi parameter @xmath113 in this specific model .
however to prove this supposition further detailed analysis is required . in principle bispectrum and
resulting parameter of nonlinearities @xmath114 have to be determined .
in this letter we have considered self - interacting scalar field theory in the universe with holonomy effects resulting from the loop quantum cosmology .
we have performed constraint for the scalar field in the form @xmath0 where @xmath2const .
based on this restriction we have derived expression for the scalar field potential which fulfills this condition . this condition can be fulfilled for @xmath115 $ ] .
the case @xmath55 is excluded .
such a state can be obtained only in the case @xmath46 and can not be mimicked with @xmath116 .
we have solved considered model analytically .
subsequently , we have performed analogous investigations for the phantom field @xmath117 .
potentials for scalar and phantom fields are related by the symmetry @xmath118 .
we have shown how barotropic matter can be introduced in lqc in the well - defined manner . in this approach matter
content is mimicked by the scalar field with derived potential ( [ potential ] ) .
we have shown that field @xmath19 is monotonic function of time in both considered cases .
this result indicates that scalar field ( as well as phantom field ) with derived potentials can be moreover treated as intrinsic time .
the models therefore seems to be good candidate to the further purely quantum considerations .
we have performed preliminary studies of the perturbations of the scalar field with obtained multi - fluid potential . in the pre - bounce phase
super - horizontal modes are typically damped .
namely the power spectrum is @xmath100 .
however to verify that this spectrum survives during the bounce phase further numerical analysis should be performed .
if super - horizontal spectrum does not change qualitatively this can lead to the suppression of the low cmb multipoles .
we have also indicated that non - gaussian perturbations are produced .
further analysis of this issue may provide observational bounds on the model .
therefore beside the purely theoretical results as the method of introducing barotropic mater in lqc also phenomenological consequences were found .
author would like to thank to anonymous referee for useful remarks .
this work was supported in part by the marie curie actions transfer of knowledge project cocos ( contract mtkd - ct-2004 - 517186 ) .
in this appendix we are going to verify obtained solutions ( [ solpt],[solphit ] ) solving from scratch equations of motion ( [ friedmann],[fieldequation ] ) with potential ( [ potential ] ) . without loose of generality
we assume here @xmath43 .
therefore we consider a theory with potential in the form @xmath119 } \label{ap0}\ ] ] where @xmath120 . here
we do not assume what is the interpretation of the parameter @xmath121 .
it is just some fixed parameter and its meaning will be clarified later .
equation ( [ fieldequation ] ) with potential ( [ ap0 ] ) takes the form @xmath122 where to simplify notation we have defined @xmath123 . to solve this equation we postulate solution in the form @xmath124 where @xmath125 is some unknown constant . applying this solution , we rewrite equation ( [ ap1 ] ) to the form @xmath126\frac{1}{3}. \label{ap3}\ ] ] now let us calculate expression for the energy density ( [ defrhophi ] ) .
with use of ( [ ap2 ] ) we obtain @xmath127.\ ] ] applying this to the modified friedmann equation ( [ friedmann ] ) we have @xmath128 \label{ap5 } \left(\beta^2 + 2v_*\alpha^2 \right)}{4\alpha^4 ( 1+\beta^2t^2)}.\ ] ] one can easily verify than equations ( [ ap3 ] ) and ( [ ap5 ] ) are equivalent only if @xmath129 therefore we obtain @xmath130}{\sqrt{6\pi g(1+w')}}. \label{ap6}\ ] ] moreover equation ( [ ap3 ] ) can be written now as @xmath131 with solution @xmath132 now one can interpret the parameter @xmath121 .
namely let us calculate expression for the coefficient @xmath133 applying potential ( [ ap0 ] ) together with solution ( [ ap6 ] ) we find @xmath134 it is now transparent that solutions found in section [ sec : analytical ] are recovered .
thanks to relation ( [ ap7 ] ) we have @xmath135 , therefore all found trajectories follow the @xmath2 const .
a. ashtekar , t. pawlowski and p. singh , phys .
* 96 * ( 2006 ) 141301 [ arxiv : gr - qc/0602086 ] .
a. ashtekar , t. pawlowski and p. singh , phys .
d * 73 * ( 2006 ) 124038 [ arxiv : gr - qc/0604013 ] .
a. ashtekar , t. pawlowski and p. singh , phys .
d * 74 * ( 2006 ) 084003 [ arxiv : gr - qc/0607039 ] .
a. corichi and p. singh , phys .
d * 78 * ( 2008 ) 024034 [ arxiv:0805.0136 [ gr - qc ] ] .
v. taveras , phys .
d * 78 * ( 2008 ) 064072 [ arxiv:0807.3325 [ gr - qc ] ] .
j. mielczarek , t. stachowiak and m. szydlowski , phys .
d * 77 * ( 2008 ) 123506 [ arxiv:0801.0502 [ gr - qc ] ] .
p. singh , k. vandersloot and g. v. vereshchagin , phys .
d * 74 * ( 2006 ) 043510 [ arxiv : gr - qc/0606032 ] .
t. cailleteau , a. cardoso , k. vandersloot and d. wands , arxiv:0808.0190 [ gr - qc ] .
j. mielczarek and m. szydlowski , jcap * 0808 * ( 2008 ) 014 [ arxiv:0803.1742 [ hep - th ] ] .
m. artymowski , z. lalak and l. szulc , arxiv:0807.0160 [ gr - qc ] .
p. singh , phys .
d * 73 * ( 2006 ) 063508 [ arxiv : gr - qc/0603043 ] .
r. r. caldwell , phys .
b * 545 * ( 2002 ) 23 [ arxiv : astro - ph/9908168 ] .
p. singh , m. sami and n. dadhich , phys .
d * 68 * ( 2003 ) 023522 [ arxiv : hep - th/0305110 ] .
m. p. dabrowski , t. stachowiak and m. szydlowski , phys .
d * 68 * ( 2003 ) 103519 [ arxiv : hep - th/0307128 ] .
m. sami , p. singh and s. tsujikawa , phys .
d * 74 * ( 2006 ) 043514 [ arxiv : gr - qc/0605113 ] .
d. samart and b. gumjudpai , phys .
d * 76 * ( 2007 ) 043514 [ arxiv:0704.3414 [ gr - qc ] ] .
b. gumjudpai , arxiv:0706.3467 [ gr - qc ] .
x. fu , h. yu and p.
wu , arxiv:0808.1382 [ gr - qc ] .
p. wu and s. n. zhang , jcap * 0806 * ( 2008 ) 007 [ arxiv:0805.2255 [ astro - ph ] ] .
d. wands , phys .
d * 60 * ( 1999 ) 023507 [ arxiv : gr - qc/9809062 ] .
a. riotto , arxiv : hep - ph/0210162 .
j. mielczarek , arxiv:0902.2490 [ gr - qc ] .
y. s. piao , b. feng and x. m. zhang , phys .
d * 69 * ( 2004 ) 103520 [ arxiv : hep - th/0310206 ] .
j. mielczarek , jcap * 0811 * ( 2008 ) 011 [ arxiv:0807.0712 [ gr - qc ] ] . c. r. contaldi , m. peloso , l. kofman and a. linde , jcap * 0307 * ( 2003 ) 002 [ arxiv : astro - ph/0303636 ] .
n. bartolo , e. komatsu , s. matarrese and a. riotto , phys .
* 402 * ( 2004 ) 103 [ arxiv : astro - ph/0406398 ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , arxiv:0803.0547 [ astro - ph ] .
j. mielczarek , ( in preparation ) | the scalar field can behave like a fluid with equation of state @xmath0 , where @xmath1 $ ] . in this letter
we derive a class of the scalar field potentials for which @xmath2 const .
scalar field with such a potential can mimic ordinary matter , radiation , cosmic strings , etc .
we perform our calculations in the framework of the loop cosmology with holonomy corrections .
we solve the model analytically for the whole parameter space .
subsequently , we perform similar consideration for the model with a phantom field ( @xmath3 ) . we show that scalar field is monotonic function in both cases .
this indicates that it can be treated as a well - defined internal time for these models .
moreover we perform preliminary studies of the scalar field perturbations with this potential .
we indicate that non - gaussian features are present admitting for the possible observational constraints of the model . |
facial expressions are a primary modality to understand the emotional status of an individual .
the expression provides a useful contextual clue for social communication @xcite .
however , individuals do not always clearly reveal their facial expressions .
when an individual reveals an ambiguous facial expression , a human may have an experience to compare his / her expression with other expressions observed in past in order to extract their facial expression differences .
the related evidence is found in the literature of brain sciences . according to @xcite , an individual can discern various facial expressions by recalling the memorized face shapes of a shown person .
the neural pathways for detecting changeable aspects of faces ( e.g. , eye movements and emotional expressions ) and for memorizing the unique face shape are separately distributed @xcite .
these two processes are interacted in the core system of the brain @xcite .
we attempt to utilize a reference face image that indicates the memorized unique face in the brain to discriminate a facial expression input in a deep neural network framework ( see figure [ fig : overview_concept ] ) .
we assume that an expression factor can be extracted from the _ contrastive characteristics _ between the given image and the reference image . the reference image for an individual identity , however , is not always available in the wild .
we start from the assumption that there is a generative artificial neural network that can be used to infer a reference image from the given facial expressions .
if a single image is given , the reference image is generated using the generative ( encoder - decoder ) networks .
the next required process is to model the _ contrastive characteristics _ mentioned above using deep neural networks .
one of main concerns is to find out how to extract or encode the contrastive features between the reference image and the given expression image . in our proposed networks , two representation models are included , for 1 ) disentangling of expressions and 2 ) explaining of contrastive representation with a supervised setting . in general , deep networks disentangle multiple variation factors of an input image .
several unknown or unintended factors are revealed in the networks and a useful factor is selected by a proper objective . in this paper
, we attempt to disentangle directly the intended factor : a facial expressive factor .
hence , disentangling of expression is conducted in two steps .
first , through learning a generator network that estimates a reference expression image , expressive factors can be eliminated .
this estimated reference image is used to measure a expressive representation by comparing with the original expression image in the feature space . in a later part
, disentangling is assisted by contrastive metric learning and a supervised reconstruction . from the approach in the literature , gradual changes of facial expressions are utilized to extract temporal information along the multiple frames .
this multiple images ( video ) based model has abundant information of the expression transition , which can be used for the recognition . in this paper , we focus on exploring a representation from a pair of a generated frame and a given frame .
our proposed framework could be easily extended to utilize multiple frames as well . in this paper
, we attempt to answer to a few questions quantitatively and qualitatively : 1 ) is a generated reference image useful for the discriminative task ?
2 ) how generative networks are controlled by contrastive metric learning for a discriminative purpose ?
3 ) how does facial generation affect expression recognition ? the main contributions of this paper are as follows : * we combine encoder - decoder networks and convolutional neural networks into a unified network that simultaneously learns to generate , compare , and classify samples on a dataset . *
we show that the contrastive representation trained with contrastive metric learning and a supervised reconstruction is useful to achieve a better discriminative performance for a facial expression recognition task . *
we show that the proposed method outperforms the state - of - the - art methods including the multiple images based approach in terms of facial expression recognition accuracy even when a single image is utilized in a test phase .
facial expression recognition has been studied over decades .
several different approaches exist that are based on local feature extraction , facial action units ( faus ) , temporal information , and convolutional neural networks .
the local feature - based methods such as the gabor filter , lbp , hog , and bow are the most common and widely studied to extract good visual features @xcite . in the fau
based methods @xcite , faus are detected and analyzed to classify an expression .
this is mainly based on the facial action coding system ( facs ) proposed by paul ekman @xcite .
temporal information - based methods @xcite utilize multiple images .
these methods , however , achieve _ limited _ recognition accuracy performance because the designed features lost some information . to overcome the insufficient representations of the hand - crafted features ,
deep learning based methods have been recently adopted .
an ensemble of two deep networks models that handle temporal information including appearance and geometric features has been proposed @xcite . a simple convolutional neural network has been used to analyze the fau in the learned filter of the networks @xcite . to obtain discriminative spatiotemporal representation
, facial action parts detection is performed using 3d - cnn @xcite .
however , it shows limited performance when compared to the state - of - the - art methods .
this is because those cnn - based methods still could not show a good enough representation of a facial expression .
another deep learning framework has been proposed to take advantage of the discriminative and generative models for realizing a better generalization performance .
traditionally in generative networks such as the autoencoder , a popular approach is that the entire stack of encoders is finetuned using pre - trained autoencoders in a layer - wise manner for discriminative purposes .
recently , a generative model was simultaneously learned with a discriminative model . in generative adversarial networks ( gans )
@xcite , the generative model is learned against an adversary and a discriminative model that learns to determine whether a sample is from the model distribution or data distribution .
the stacked what - where auto - encoders ( swwaes ) @xcite integrate discriminative and generative learning pathways and provide a unified approach to supervised , semi - supervised , and unsupervised learning . in this paper , we deploy a generative model with discriminative learning as well .
we are mainly focusing on investigating a contrastive representation of a facial expression that is optimized with appropriate objectives .
consider an input image matrix @xmath0 and a reference image matrix @xmath1 that are elements of a given set of image matrices @xmath2 .
the corresponding expression label is denoted by @xmath3 and @xmath4 respectively . in a real world
, an expressive face might be changed from a reference ground face ( due to emotional changes that incur facial muscle movements @xcite ) .
we define a relationship between two images with a hidden factor denoted by @xmath5 formally as follows : @xmath6 where the addition indicates operations for facial expression change is added at the equation ( [ eq : additive_epsilon ] ) : @xmath7 . in this paper , we omit the term @xmath8 for a simplicity in the notation . ] . as a facial expression
is not always apparently represented as an absolute value , a quantity of expression change obtained by comparing with a reference image might be useful .
an expression image with a very small change could be recognized via difference maps ( e.g. , a pixel - wise distance and optical flows ) . as a human keeps a neutral - like or less - expressive face most of time , that face image
could be considered as the reference image .
a representation of a difference between expression images can appear in various ways .
a simple approach is to compare image pixels of the faces .
however , owing to distortions between the images ( e.g. , distortions by an affine transform ) , comparing the images at the pixel level is not effective . for example
, a small translation in the image level might return large pixel - wise errors even though a human face shows no expression changes .
the representation of the difference ( contrastive representation " ) can be better extracted at the feature level , but not at the pixel level .
the feature - wise representation can offer an invariance towards distortions ( e.g. , translation , scale , or rotation ) .
we employ a contrastive representation in the networks to extract a latent difference factor between expressions .
consider a pair of images @xmath9 , where @xmath0 is an input sample and @xmath1 is a reference sample .
let @xmath10 ( abbreviation of an encoder ) be a transform function used to map an input matrix to the embedding space ( @xmath11 ) . in the transformed space , a latent factor in the feature level ( @xmath12 ) can be represented as follows : @xmath13 where @xmath14 is an element - wise distance formulation and @xmath15 . in this paper
, we adopt a distance @xmath16 of the @xmath17-th element of the feature vector , @xmath18 ) for @xmath14 . in the contrastive representation of expressiveness , "
we expect that other factors ( e.g. , individual s identity , pose , and etc . ) than the expression will be eliminated .
the contrastive representation @xmath12 is used for a discriminative task . the reference face
image , such as less - expressive face image of an individual , may not be available in the test phase . in this paper , therefore , we propose to generate the reference face using convolutional encoder - decoder networks . to estimate a reference face transformed from an expressive face , we apply the concept of the denoising auto - encoder ( dae ) . in the dae ,
a term corresponding to corruption , i.e. , a gaussian distribution , added to the original input is eliminated via learning is corrupted into @xmath19 using the known conditional distribution @xmath20 in order to train the autoencoder to estimate the reverse conditional @xmath21 @xcite . in this section
, we assume that the term corresponding to corruption should not be limited to a specific probability distribution . there might be a latent model ( or unknown transform ) that makes a face with a certain expression appear to be a reference face . without a definition of the latent distribution , in this paper , the model is represented using encoder - decoder networks . by disentangling facial expressive factors in feature learning , hence , information that is irrelevant or of negligible use for the discriminative purposes
could be discarded @xcite . in this section ,
we show how the generated reference image can be used in deep networks .
multiple objectives are adopted to optimize parameters of the networks to generate a good contrastive representation .
as shown in figure [ fig : overview ] ( f ) and [ fig : overview_detail ] , a loss function ( @xmath22 ) of the proposed networks consists of three kinds of objectives .
formally , the loss function is written as follows : @xmath23 where @xmath24 denotes a discriminative loss function , @xmath25 denotes a contrastive loss function , and @xmath26 denotes a reconstruction loss function .
@xmath27 indicates a weight .
the main purpose of the proposed networks is to classify a facial expression in the given input . for the discriminative objective @xmath24
, we adopt the cross entropy loss function which is widely used for the classification task .
consider a pair of features @xmath28 , @xmath29 extracted from encoder layers @xmath30 , where a subscript @xmath31 at @xmath30 indicates the second encoder layers ( layers shown in figure [ fig : overview_detail ] ) .
a contrastive representation feature @xmath32 where @xmath14 is the element - wise distance and @xmath33 is used for the classification task . for learning a contrastive representation ,
two learning objectives are deployed in the proposed networks : the first objective is contrastive metric learning ( @xmath25 ) to enlarge or to diminish the distance between the two feature vectors , and the second is reconstruction learning ( @xmath26 ) for a better representation .
hence , the two objective functions are designed to jointly assist the classification task for realizing a good generalization performance .
[ [ loss - for - contrastive - metric - learning - in - feature - space . ] ] loss for _ contrastive metric learning _ in feature space .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the objective of the loss @xmath25 is to optimize a similarity between two features @xmath34 according to an expression label . if the expression labels of @xmath35 and @xmath36 are not identical , the function optimizes to obtain dissimilar features within a predefined margin ; if the expressions are identical , it optimizes to similar features .
hence , the contrastive loss @xcite is adopted for @xmath25 in a feature space as follows : @xmath37 where @xmath38 if the labels of a pair @xmath39 are not the same , @xmath40 otherwise , @xmath41 is a similarity measure , and @xmath42 is a margin . a feature space is defined as ( @xmath43 ) at the encoder layers . [
[ loss - for - generation - and - representation . ] ] loss for generation and representation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the main objectives of the loss @xmath26 are two - fold : one is to generate a reference image , and the other is to supplement to represent a good contrastive feature in the embedding layer ( @xmath44 ) .
hence , a reconstruction loss ( @xmath26 ) can be represented as a weighted summation of three terms as follows : @xmath45 where of the subscripts of @xmath46 , the first one @xmath47 indicates the stage for a generation ( @xmath48 ) or a reconstruction ( @xmath49 ) ( shown in figure [ fig : overview_detail ] ) .
the second subscript @xmath50 , indicates a target : @xmath51 for a _
_ r__eference image , and @xmath52 for an _
_ i__nput image .
@xmath53 where @xmath54 is learned to estimate @xmath1 .
in this section , we describe the experiments conducted to compare the proposed method with the state - of - the - arts on three publicly available face expression databases ( ck+ , mmi , and oulu - casia ) that are widely adopted in the literatures @xcite .
all models used in different databases share exactly the same architecture ( shown in figure [ fig : overview_detail ] ) , including encoder - decoder networks depicted in table [ table : ae_detail ] .
all parameter settings are shared through the databases with the same value .
the encoder - decoder networks in table [ table : ae_detail ] are pre - trained with the reconstruction task using the casia - webface database @xcite , and three convolutional layers in the encoder are adopted at @xmath55 ( @xmath56 ) of the proposed generative - contrastive networks ( gcnet ) shown in the figure [ fig : overview_detail ] .
the baseline cnn consisting of three convolutional layers and two inner - product ( fc ) layers are pre - trained with the identification task using the same database , and convolutional layers are adopted at @xmath57 ( @xmath58 ) . during the training of the proposed networks , the learning rate at layers of the decoder networks is set to @xmath59 during fine - tuning . the number of outputs at the first fully - connected layer ( inner - product ) is empirically determined by @xmath60 where we set @xmath61 , @xmath62 .
this is intended that a dimensionality of the vector decreases smoothly as the number of ( conv./pool ) layers increases .
@xmath63 is related to a pooling size ( @xmath64 ) at each layer .
the dropout is applied before this fully - connected layer with a ratio of @xmath65 . after the fc - layer
, a softmax layer is connected with the number of outputs corresponding to the number of classes .
we arbitrarily set @xmath66 for each loss function .
the maximum iteration is set to @xmath67 .
our models are trained with ` nesterov ' optimization using an ` inverse ' learning policy , a base learning rate of @xmath68 , a momentum of @xmath69 , a gamma term of 0.75 , a weight decay of @xmath70 , and a mini - batch size of @xmath71 .
the proposed network model is implemented on _ python _ and the deep learning framework and run using the nvidia tesla k80 gpu . to avoid over - fitting , we applied data augmentation during the training phase .
we used input images on a gray level ( 1 channel ) where a facial region is cropped , normalized based on 5 points ( eyes , the end of a nose , and two ends of lips ) and resized into @xmath72 .
the resized image is cropped again with the size of @xmath73 at a random location .
each cropped image is manipulated using 2d affine transform such as scaling , rotation , and intensity multiplication , in addition to random flipping . [
[ ck - database ] ] ck+ database @xcite + + + + + + + + + + + + + + + + + + + + this database is widely adopted in the benchmark for facial expression recognition tasks .
this database consists of 593 sequences with 123 individuals .
the images are captured expression transitions from a neutral face to peak facial expression acted by an individual .
the 327 valid sequences with 118 individuals that maintain discrete emotion labels such as anger , contempt , disgust , fear , happy , sad , and surprise " are adopted for an experiment .
we divide the valid sequences into ten different subsets with individual - independent way . according to individual
i d in the database , individuals are grouped by sampling in i d ascending order with ten even intervals first .
one subset out of ten subsets is used for validation ( test ) , the remains are used for training .
this procedure is repeated ten times .
this 10-fold cross - validation follows the previous works @xcite .
[ [ mmi - database ] ] mmi database @xcite + + + + + + + + + + + + + + + + + + + + this database consists of 312 sequences from 30 individuals with six basic expressions ( contempt included in the ck+ database is excluded ) .
we selected 205 sequences captured in a front view .
each sequence starts from a neutral face , and shows a peak expression within a single expression type in the middle of the sequence . at the end
, it returns to a neutral face again . as a peak expression frame number is not given , we selected it manually .
similar to the ck+ database settings , we divided the mmi database into ten different individual independent subsets . consequently
, 10-fold cross validation was conducted .
this database includes individuals who pose expressions non - uniformly , wear glasses / caps , and have mustaches / head movements .
therefore , the facial expression recognition task is relatively challenging .
moreover , the small number of sequences and individuals makes it difficult to achieve a good generalization performance .
this database could be suitable to measure the recognition performance in realistic situations when compared to other databases .
[ [ oulu - casia - vis - database ] ] oulu - casia vis database @xcite + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + this database consists of 480 image sequences with 80 individuals .
this database is captured under the visible ( vis ) normal illumination conditions and is a subset of oulu - casia nir - vis database .
each individual poses six basic expressions similar to mmi database .
similar to the ck+ database , the sequence starts from a neutral face and ends with peak facial expression within a same emotion category .
as done with the two databases above , individual - independent 10-fold cross - validation is conducted .
.details of the convolutional encoder - decoder layers @xcite embedded in the proposed networks .
an encoder part consists of three convolutional layers ( conv . ) which is followed by batch normalization ( bnorm ) , relu , and max poooling layers .
correspondingly , a decoder part consists of three de - convolutional ( transposed convolutional ) layers . in a conv and deconv .
layers , ( @xmath74 , @xmath75 ) indicates that there is 32 sets of 5@xmath765 filters
. in maxpool and maxunpool layers , ( @xmath74 ) indicates a pooling window size . [ cols="^",options="header " , ] [ [ small - expression - images ] ] small expression images + + + + + + + + + + + + + + + + + + + + + + + in figure [ fig : ambiguous_images ] , several examples of recognition errors on the small expression images are shown .
our proposed method shows approximately twice less recognition errors than baseline cnn method .
however , both the baseline cnn and the proposed method still have a limitation to recognize the small ( or ambiguous ) expressions .
[ [ visualization - in - the - feature - space ] ] visualization in the feature space + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to observe a discriminative distribution of the extracted features , we visualized the feature vectors from the first layer of the fully - connected layers of the proposed networks of the baseline cnn and our proposed networks .
we visualize the 384 dimensional feature vectors using t - sne @xcite .
the feature points of original images are scattered within a narrow region .
the point distribution of the baseline cnn forms partially overlapped clusters .
the proposed network features are clustered well to discriminate individual expression further .
[ [ patterns - of - the - learned - filters ] ] patterns of the learned filters + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we observe the characteristics of the filters learned in the proposed networks . as shown in figure [ fig : learned_filter ] , the encode filters learned by contrastive metric learning , ( b ) , has more gabor like edge and blob detection filters than ( a ) .
the decoder filters for the expression reconstruction , ( e ) , show a simpler patterns than that for the reference image generator , ( c ) , and the reconstruction decoder of a neutral image , ( d ) , as shown in figure [ fig : learned_filter ] . [ [ visualization - of - the - response - maps ] ] visualization of the response maps + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we observe the response maps resulted from generation and reconstruction layers of the proposed networks to understand what the networks have been conducted in the test phase . in figure
[ fig : response_map_ck+ ] , a generated reference image , a reconstructed neutral image , and a reconstructed image of a given expression are shown .
the generated reference image is affected by reconstruction and contrastive metric learning .
in this paper , we proposed facial expression recognition method based on contrastive representation learning .
the contrastive representation is calculated in the embedding layer of deep networks by comparing a single given image with a reference image .
the reference image is generated by deep generative ( encoder - decoder ) networks .
this approach is useful especially if an expressive depth of an emotional face is varied among individuals , expressions , or situations . in our proposed networks
, we attempted to disentangle a facial expressive factor directly .
disentangling of expression is conducted in two steps : 1 ) learning of a reference face by a generator network and 2 ) learning of a contrastive representation with a combination of contrastive and reconstruction objectives .
extensive experiments were conducted on three face expression databases that are publicly available and widely adopted in the literature .
the proposed method outperforms the known state - of - the arts , including both single image and multiple - image based methods .
this study could be extended to effectively detect and recognize small changes of facial expressions from sequential images .
m. s. bartlett , g. littlewort , m. frank , c. lainscsek , i. fasel , and j. movellan . recognizing facial expression : machine learning and application to spontaneous behavior . in _
2005 ieee computer society conference on computer vision and pattern recognition ( cvpr05 ) _ , volume 2 , pages 568573 .
ieee , 2005 .
y. bengio , l. yao , g. alain , and p. vincent .
generalized denoising auto - encoders as generative models . in _ proceedings of the 26th international conference on neural information processing systems _ , nips13 , pages 899907 , 2013 .
i. goodfellow , j. pouget - abadie , m. mirza , b. xu , d. warde - farley , s. ozair , a. courville , and y. bengio .
generative adversarial nets . in _ advances in neural information processing systems 27
_ , pages 26722680 . 2014 .
r. hadsell , s. chopra , and y. lecun .
dimensionality reduction by learning an invariant mapping . in _
2006 ieee computer society conference on computer vision and pattern recognition ( cvpr06 ) _ , volume 2 , pages 17351742 .
ieee , 2006 .
s. jain , c. hu , and j. k. aggarwal .
facial expression recognition with temporal modeling of shapes . in _
computer vision workshops ( iccv workshops ) , 2011 ieee international conference on _ , pages 16421649 .
ieee , 2011 .
h. jung , s. lee , j. yim , s. park , and j. kim .
joint fine - tuning in deep neural networks for facial expression recognition . in _ proceedings of the ieee international conference on computer vision _ , pages 29832991 , 2015 .
p. khorrami , t. paine , and t. huang .
do deep neural networks learn facial action units when doing expression recognition ? in _ proceedings of the ieee international conference on computer vision workshops _ ,
pages 1927 , 2015 .
m. liu , s. li , s. shan , and x. chen .
au - aware deep networks for facial expression recognition . in _ automatic face and gesture recognition ( fg ) , 2013 10th ieee international conference and workshops on _ , pages 16 .
ieee , 2013 .
m. liu , s. li , s. shan , r. wang , and x. chen . deeply learning deformable facial action parts model for dynamic expression analysis . in _ asian conference on computer vision _ , pages 143157 .
springer , 2014 .
m. liu , s. shan , r. wang , and x. chen . learning expressionlets on spatio - temporal manifold for dynamic facial expression recognition . in _ proceedings of the ieee conference on computer vision and pattern recognition _ , pages 17491756 , 2014 .
p. lucey , j. f. cohn , t. kanade , j. saragih , z. ambadar , and i. matthews .
the extended cohn - kanade dataset ( ck+ ) : a complete dataset for action unit and emotion - specified expression . in _ 2010 ieee computer society conference on computer vision and pattern recognition - workshops _ , pages 94101 .
ieee , 2010 .
m. valstar and m. pantic .
induced disgust , happiness and surprise : an addition to the mmi facial expression database . in _ proc .
3rd intern .
workshop on emotion ( satellite of lrec ) : corpora for research on emotion and affect _
, page 65 , 2010 .
z. wang , s. wang , and q. ji . capturing complex spatio - temporal relations among facial muscles for facial expression recognition . in _ proceedings of the ieee conference on computer vision and pattern recognition _ ,
pages 34223429 , 2013 .
l. zhong , q. liu , p. yang , b. liu , j. huang , and d. n. metaxas .
learning active facial patches for expression analysis . in _
computer vision and pattern recognition ( cvpr ) , 2012 ieee conference on _ , pages 25622569 .
ieee , 2012 . | as the expressive depth of an emotional face differs with individuals , expressions , or situations , recognizing an expression using a single facial image at a moment is difficult .
one of the approaches to alleviate this difficulty is using a video - based method that utilizes multiple frames to extract temporal information between facial expression images . in this paper
, we attempt to utilize a generative image that is estimated based on a given single image . then , we propose to utilize a contrastive representation that explains an expression difference for discriminative purposes .
the contrastive representation is calculated at the embedding layer of a deep network by comparing a single given image with a reference sample generated by a deep encoder - decoder network .
consequently , we deploy deep neural networks that embed a combination of a generative model , a contrastive model , and a discriminative model . in our proposed networks ,
we attempt to disentangle a facial expressive factor in two steps including learning of a reference generator network and learning of a contrastive encoder network .
we conducted extensive experiments on three publicly available face expression databases ( ck+ , mmi , and oulu - casia ) that have been widely adopted in the recent literatures .
the proposed method outperforms the known state - of - the art methods in terms of the recognition accuracy . |
condensates of resonantly pumped exciton - polaritons in semiconductor microcavities constitute a novel and exciting system for the study of fundamental physical properties of superfluids out of equilibrium @xcite , and for future device applications @xcite . being quantum superpositions of light and matter they are privileged candidates for the realization of the next generation of optical devices @xcite , for example for quantum information technologies . from the point of view of the fundamental physical properties ,
particularly interesting is the case of a coherently pumped polariton superfluid in presence of defects . here , in contrast to the corresponding equilibrium case , a weak residual drag force is always present even at extremely high polariton densities @xcite .
nevertheless , properties paradigmatical of an equilibrium superfluids , such as frictionless flow of polariton bullets @xcite , quantized vortices and metastable persistent flow @xcite , and the appearance and disappearance of cerenkov - like waves @xcite have been recently observed in coherently driven exciton - polaritons . in view of the potential device applications ,
especially important is the unique versatility of the polaritonic system , which , combined with its high non - linear properties , have been already demonstrated to produce parametric scattering @xcite and bistability @xcite . here
, the implementation of logic operations and gates comes in a natural way : by manipulating the non - linear properties of the system using several lasers which frequencies , angles of incidence , and intensities can be freely varied externally . in this paper
we investigate a new realization of a two - component polariton system coherently driven by two lasers with independently tunable frequencies , angles of incidence and intensities .
firstly , we study the stability of the two polariton components when the two laser intensities are varied . differently from the case of a single laser pump , where the system can only be bistable , we disclose a rich phase diagram , where either one , two or three stable states can coexist at given pumping conditions .
then we suggest possible easily tunable multistable hysteresis cycles when the two pumping lasers are varied up and down in intensities .
alternatively , a multicomponent polariton fluid can be realized by considering the polarization degrees of freedom .
multistability of different polariton spin states has been recently proposed theoretically @xcite and confirmed experimentally @xcite by the observation of three stable spin states for a given excitation condition . for the case of two polarized components , multistability in space
have also been theoretically proposed @xcite and experimentally observed @xcite .
for the system with two pumping lasers presented here , the same kind of spatial multistability is expected but with much more complex features . additionally , superimposed to the spatial multistability
, interference fringes will appear due to the difference in frequency and momentum of the two pumping lasers .
the analogy with the two - component polarized case , suggests that , aside from the interest in investigating multistability , two - component polariton condensates obtained with independent lasers can also be used to realize switches @xcite and memories .
the paper is organized as follows : in section ii , we present the model used to describe the steady state behavior of polaritons excited by two continuous - wave lasers with different frequencies , wave - vectors and intensities . the results obtained within a linear response framework
are shown in section iii . in this section
we study the number of the possible solutions , their nature and possible kerr or parametric instabilities associated to them . as a consequence of the presence of multiple stable solutions ,
different cycles of hysteresis can be produced by varying , along different paths , the intensities of the two pumping lasers .
finally , section iv contains the conclusions drawn from our analysis .
the dynamics of resonantly - driven microcavity polaritons @xcite can be described via a gross - pitaevskii equation for coupled cavity ( @xmath0 ) and exciton ( @xmath1 ) fields generalized to include decay and resonant pumping ( @xmath2 ) : @xmath3 \begin{pmatrix } \psi_x \\ \psi_c \end{pmatrix}\ ; . \label{eq : model}\ ] ] the repulsive ( @xmath4 ) exciton - exciton interaction induces a non - linear dynamics of the eigenmodes of the single polariton hamiltonian ( lower and upper polariton , @xmath5 ) : @xmath6 here , we assume the cavity dispersion to be quadratic , @xmath7 , with @xmath8 ( @xmath9 is the bare electron mass ) , we will neglect the exciton dispersion and consider the case of zero detuning at normal incidence , @xmath10 .
the rabi frequency @xmath11 @xmath12mev@xmath13 $ ] and the excitonic and photonic decay rates , @xmath14 @xmath12mev@xmath13 $ ] are chosen in the range of experimental values .
because of the continuous decay , a stationary state requires a continuous injection of photons . here , we consider two continuous - wave laser fields , @xmath15 with independently tunable frequencies @xmath16 and momenta @xmath17 , which can be experimentally changed by changing the laser angle of incidence with respect to the growth direction .
we study the mean - field solutions of eq .
@xmath18 and their stability with respect to small fluctuations within a linear response analysis .
substituting the expression into we obtain 4 contributions , two of which oscillate at the main frequencies @xmath19 and @xmath20 and the additional two at the replica ( or satellite state ) frequencies @xmath21 and @xmath22 , where @xmath23 .
similarly to what is done in the opo regime @xcite where replica states in addition to the pump signal and idler states are neglected , here , we consider only the terms oscillating at the main frequencies @xmath19 and @xmath20 .
later , see eq .
, we analyse the dynamical stability of the two - pump - frequency solution against the weak population of satellite states @xmath24 via parametric scattering processes . through the paper , we will consider only dynamically stable two - pump - frequency solutions . in this approximation , we obtain the following mean - field equations for @xmath25 : @xmath26\psi_{1_x}^{ss}+ \frac{\omega_r}{2}\psi_{1_c}^{ss}=0 \\ [ \omega_c({\bf k}_1)-\omega_1-i\kappa_c]\psi_{1_c}^{ss}+ \frac{\omega_r}{2}\psi_{1_x}^{ss}+f_1 = 0 \\ [ \omega_x-\omega_2-i\kappa_x+g_{21}]\psi_{2_x}^{ss}+ \frac{\omega_r}{2}\psi_{2_c}^{ss}=0 \\ [ \omega_c({\bf k}_2)-\omega_2-i\kappa_c]\psi_{2_c}^{ss}+ \frac{\omega_r}{2}\psi_{2_x}^{ss } + f_2 = 0 \ ; , \end{cases } \label{eq : stead}\ ] ] where @xmath27 with @xmath28 .
note that the repulsive interaction term between excitons in different states is two times larger the interaction term between excitons in the same mode , resulting in a non - uniform blue - shift .
the mean - field system of equations can have up to 9 solutions , i.e. 6 solutions more than in the case of one pumping laser , but , as discussed below , only a maximum of 3 solutions are stable
. the dynamical stability of the two - pump - frequency mean - field solution can be established by adding small fluctuations , @xmath29 + \\
e^{-i\omega_2 t } \left[e^{i{{\mathbf k}}_2\cdot{{\mathbf r } } } \psi^{ss}_{2_{x , c } } + \theta_{2_{x , c}}({{\mathbf r , t}})\right ] \ ; , \label{eq : fluct}\end{gathered}\ ] ] where the fluctuation fields can be divided into particle - like and hole - like excitations @xmath30 $ ] .
expanding eq . up to linear terms in @xmath31
, we obtain 4 terms oscillating at frequencies @xmath32 and @xmath33 , which we neglect , and 4 terms oscillating at @xmath24 . in other words
, we are checking the stability of our solution , where only the two states with frequencies @xmath16 are occupied , against the weak population of the satellite states @xmath24 which can be populated by parametric scattering processes .
the fact that we consider only linear terms in @xmath34 and @xmath35 implies that we can obtain only the threshold conditions for such parametric processes , as well as the nature of the instability , whether of kerr - type or parametric - type see later .
the equations for @xmath34 and @xmath35 can be written as an eigenvalue equation rearranging the excitations into an 8-component vector @xmath36 : @xmath37 \mathbb{u}_{{{\mathbf k } } } = 0\ ; . \label{eq : eigen}\ ] ] here matrices @xmath38 with @xmath39 are given by @xmath40 and @xmath41 are given by @xmath42 with @xmath43 being the total excitonic density . at given values of the pumping strength @xmath44 and @xmath45 ,
the solutions of the mean - field equations are stable if all the eight eigenvalues ( @xmath46 , @xmath47 ) of eq .
have negative imaginary part for every value of the momentum @xmath48 .
for some choices of the system parameters , we find that the number of stable solutions can be larger than one . in the case of one pumping laser , the typical @xmath49-like shape dependence of the polariton field intensity on the pump strength , also referred to as optical bistability , can be explained in terms of the non - linear blue - shift induced by the polariton - polariton interaction @xcite . when the laser frequency is well above the bare lower polariton dispersion , @xmath50 , and the pump intensity increases from low values , the polariton population remains small because it is hard for the laser to inject polartions with a different energy .
however , increasing the pump power , the blue - shift pulls the polariton energy towards resonance with the pump causing the population to grow superlinearly and eventually to abruptly jump to a high value when the pump intensity reaches a critical value @xmath51 . in the opposite situation ,
when the laser intensity is decreased from high values , the polariton energy is blue - detuned close to the pumping laser frequency and , therefore , the cavity is efficiently filled by the laser even at low pumping intensities . in this case the polariton population jumps down back at low polariton densities for a value @xmath52 of the pump strength lower than @xmath51 .
the two jumps at different values of the pump intensity cause therefore a hysteresis cycle .
as explained below , in the case of two - component fluids , the situation becomes even richer .
[ mev@xmath53 .
white , green , yellow and black regions correspond to respectively zero , one , two or three stable solutions . in panels
i and ii @xmath54 @xmath55m@xmath56 and @xmath57 @xmath55m@xmath56 while in panels iii and iv @xmath58 @xmath55m@xmath56 and @xmath57 @xmath55m@xmath56 . in the left panels ( i and iii )
@xmath59 @xmath12mev@xmath13 $ ] , while in the right panels ( ii and iv ) @xmath60 @xmath12mev@xmath13 $ ] .
the horizontal black lines lies at the three fixed values of @xmath61 corresponding to the three panels of fig .
[ fig : cuts ] , while the blue diagonal line is the path used to plot fig .
[ fig : hyst ] .
3d panels : plots of @xmath62 [ mev ] as a function of @xmath63 with parameters equal to panel i. stable solutions with higher populations are shown in green , stable solution with the second higher population in yellow and third stable solution with lower population in black .
all the solutions are shown in the left panel .
since the upper green branches hide a yellow upper branch , the right panel shows only the yellow and black solutions.,title="fig : " ] [ mev@xmath53 .
white , green , yellow and black regions correspond to respectively zero , one , two or three stable solutions . in panels
i and ii @xmath54 @xmath55m@xmath56 and @xmath57 @xmath55m@xmath56 while in panels iii and iv @xmath58 @xmath55m@xmath56 and @xmath57 @xmath55m@xmath56 . in the left panels ( i and iii )
@xmath59 @xmath12mev@xmath13 $ ] , while in the right panels ( ii and iv ) @xmath60 @xmath12mev@xmath13 $ ] .
the horizontal black lines lies at the three fixed values of @xmath61 corresponding to the three panels of fig .
[ fig : cuts ] , while the blue diagonal line is the path used to plot fig .
[ fig : hyst ] .
3d panels : plots of @xmath62 [ mev ] as a function of @xmath63 with parameters equal to panel i. stable solutions with higher populations are shown in green , stable solution with the second higher population in yellow and third stable solution with lower population in black .
all the solutions are shown in the left panel .
since the upper green branches hide a yellow upper branch , the right panel shows only the yellow and black solutions.,title="fig : " ] we fix both laser frequencies to be blue detuned with respect to the bare polariton dispersion : @xmath64 @xmath12mev@xmath13 $ ] , with @xmath54 @xmath55m@xmath56 and @xmath57 @xmath55m@xmath56 .
we plot in panel i of fig .
[ fig:2d3d ] the phase diagram showing the regions with a different number of stable solutions ( either one , two or three ) as a function of the two rescaled pumping intensities @xmath65 [ mev@xmath53 . in order to understand better the structure of this phase diagram
, we show in fig .
[ fig : cuts ] the total exciton density @xmath66 when the pump intensities @xmath61 is kept constant at different values and @xmath67 is varied . when the constant pump @xmath68 has a small value ( fig .
[ fig : cuts ] top left panel ) , the dependence of the population on the varying pump intensity @xmath67 is similar to the one - fluid case showing bistability with an @xmath49-like shape . for higher values of the constant pump @xmath61 ( fig . [
fig : cuts ] lower left panel ) , the number of possible solutions increases but just two are found to be stable .
finally when @xmath61 is further increased the set of possible solutions further goes up but only a maximum of @xmath69 are found to be stable .
@xmath12mev@xmath13 $ ] ( red dotted curves unstable solutions , black lines stable solutions ) for fixed pump intensities as a function of @xmath67 , for @xmath70 @xmath12mev@xmath71}$ ] ( top left ) , @xmath72 @xmath73 $ ] ( bottom left ) , and @xmath74 @xmath12mev@xmath75 $ ] ( right ) . points @xmath76 ( @xmath77 ) , @xmath78 correspond to the cases when the lower ( upper ) branch of the stability curve becomes unstable ( see fig . [
fig : imaginary1 ] and [ fig : imaginary2 ] ) .
point @xmath79 does not have a counterpart in the one - fluid case and correspond to the cases when the second high branch of the stability curve becomes unstable ( see fig .
[ fig : imaginary2 ] ) . ]
the coexistence of three solutions , corresponding to the black regions of fig .
[ fig:2d3d ] , can be understood as follows : when the two pump intensities increase from low values , the polariton population is small because it s energy is far below the laser frequencies , resulting in one stable solution . in the opposite situation ,
when the intensity of one of the lasers decreases from high values , the polariton population is high and its dispersion is significantly blue detuned with respect to the bare one .
such blue - shift can be sustained by any of the two lasers , thus giving two additional stable solutions for the same values of the pump intensities .
therefore a maximum of three stable solutions can be expected .
this is also evident while considering the partial densities for particular values of the pump strength at which three stable solutions are present ( black region of fig . [ fig:2d3d ] ) , e.g. @xmath80 @xmath12mev@xmath75 $ ] and @xmath81 @xmath12mev@xmath75 $ ] . here
, the solution with lower total polariton density corresponds to partial densities @xmath82 @xmath12mev@xmath13 $ ] and @xmath83 @xmath12mev@xmath13 $ ] .
the other two solutions correspond to a high value of just one of the two partial populations : @xmath84 @xmath12mev@xmath13 $ ] and @xmath85 @xmath12mev@xmath13 $ ] in one case and @xmath86 @xmath12mev@xmath13 $ ] and @xmath87 @xmath12mev@xmath13 $ ] in the other .
note that this situation is similar to the case of two - component condensates obtained with two spins .
however , while in the spin - dependent case the two lasers pump the two spin populations with different intensities but at the same angle and energy , here the two pumps are independent also in angle and in energy .
this analogy is also visible in eq . .
however , the difference between our system of equations and the spin - dependent case is that here the interaction between different components is twice the interaction between particles in the same component . in panel
ii of fig .
[ fig:2d3d ] we plot the phase diagram for the same parameters as in panel i but with the two pumping lasers @xmath88 @xmath12mev@xmath13 $ ] blue detuned with respect to the bare lp branch .
we see that the effect of the increased detuning is simply to stretch the phase diagram . since the two pumps are further apart from the lp branch , it is more difficult to inject polaritons into the cavity , and thus the need for higher pump intensities . in panel
iv of fig .
[ fig:2d3d ] we show that a similar phase diagram can be obtained by changing @xmath48 vector of pump 1 from @xmath89 to @xmath90 @xmath55m@xmath56 .
we observe that the multistability is quite robust with respect to the choice of the parameters and , therefore , it should be within an experimental reach .
an interesting configuration is plotted in panel iii of the same figure . here a region with no stable solutions appears in the central part of the plot ( white region ) .
the instability of this region can be understood by noting that with a pump at @xmath91 and just slightly blue detuned from the lp and a pump at @xmath92 , close to the inflection point of the lp , it is easy to satisfy phase matching conditions for parametric scattering processes .
for this set of parameters the system is in a configuration unstable towards the population of satellite states by scattering processes .
for the other three sets of parameters , shown in the remaining three panels of fig .
[ fig:2d3d ] , it is also possible to find regions of the phase diagram where no solutions are stable .
these are the regions where the proposed solution , given by eq .
[ eq : meanf ] , where only the two frequency states , @xmath19 and @xmath20 , are populated , is not a stable solution because satellite states start also to be populated our analysis giving the threshold for this to happen . .
the three panels correspond to points @xmath76 with @xmath78 of fig .
[ fig : cuts ] , where the lower part of the stability curves become unstable . in blue ( red ) the parts corresponding to the scattering of two particles with @xmath93 ( @xmath94 ) . ]
to further discuss the stability of the system with respect to small perturbations , we plot the dispersion of the imaginary part of the excitation eigenfrequency @xmath95 for several points of the stability curves shown in fig . [
fig : cuts ] .
we start with the cases where the lower branch of the stability curve became unstable at points @xmath76 .
for very small values of @xmath61 the imaginary part of the dispersion ( top panel of fig .
[ fig : imaginary1 ] ) shows two peaks for given values of @xmath48 .
one peak lies at higher value @xmath96 @xmath55m@xmath56 and one peak at lower value @xmath97 @xmath55m@xmath56 with @xmath98 .
this two - peaks structure is a precursor of a parametric instability due to the scattering between two particles in the component of the condensate with momentum @xmath99 .
this situation corresponds exactly to the case of one component fluids . when the pump intensity @xmath61 is slightly increased ( middle panel ) we observe 6 other peaks appearing in the imaginary part of the dispersion .
two of these new peaks are such that @xmath98 ( blue lines ) while the other four can be combined to identify two different scattering processes with @xmath100 ( red curves ) .
this more complicated structure of the imaginary parts of the eigenvalues is consistent with the fact that with two components a richer mechanisms of scattering might occur .
when the intensity of @xmath61 is further increased ( lower panel ) , still 4 different scattering may occur but , in this case , it is the scattering between two particles with @xmath101 that induce the instability of the system . .
the three panels correspond to points @xmath77 with @xmath78 and @xmath79 of fig .
[ fig : cuts ] , where the higher part of the stability curve becomes unstable . ] for the transition from stable to unstable regions of the higher branch of the stability curve we plot the dispersion of the imaginary part of the excitation eigenfrequency @xmath95 for points @xmath77 in fig .
[ fig : imaginary2 ] . in analogy with the case of fluids with one component , for low intensities of pump 2 ,
the imaginary part of the dispersion shows a peak at the wavevector of pump 1 , a clear precursor of a kerr instability .
when the intensity of pump 2 is increased new peaks appear in the imaginary part .
in the case of point @xmath102 two new peaks are precursors of a parametric instability for the state with @xmath103 @xmath104 $ ] even if the mechanism responsible for the instability of the solution is still of kerr type .
it is only when the pump intensity @xmath61 is further increased ( point @xmath105 ) that the two peaks at @xmath106 and @xmath107 @xmath104 $ ] became more important and the mechanism of instability is of the parametric type . finally in the lower right panel ( corresponding to point c ) of fig .
[ fig : imaginary2 ] a single peak at @xmath93 is the precursor of a kerr - type instability that ends the region with three stable solutions in the right panel of fig .
[ fig : cuts ] .
( color online ) hysteresis cycles of @xmath108 [ dimensionless ] as a function of @xmath67 for different values of @xmath61 stable solutions are in black , while the hysteresis cycle performed by the system is in brown .
panel a : @xmath109 as in the blue line of fig .
[ fig:2d3d ] .
panel b : @xmath110 @xmath12mev@xmath75 $ ] as in the horizontal black line of fig .
[ fig:2d3d ] . ]
multistability also manifests itself in a hysteresis loop for the populations and emission intensities obtained with a cycle of first increasing and later decreasing the pumping intensities . here
, the presence of three stable solutions gives more complicated loops than the ones obtained for a bistable system in a single - component polariton fluid . in order to study this aspect
, we calculate the exciton emission intensity at a given frequency @xmath111 normalized to the total exciton emission intensity , @xmath112 , along a closed path of varying pumping intensities .
the panels a ) and b ) of fig . [
fig : hyst ] respectively show the hysteresis cycles of @xmath113 when the two pump intensities change along either the blue or the higher horizontal black line of fig .
[ fig:2d3d ] . in panel
a ) , one starts from a low value of @xmath67 taking @xmath114 @xmath12mev@xmath75 $ ] so that the population of state @xmath115 is much higher than the population of state @xmath116 , i.e. @xmath117 . increasing @xmath67 the two populations smoothly evolve until @xmath118 @xmath12mev@xmath75 $ ] and @xmath119 @xmath12mev@xmath75 $ ] . at this point @xmath61
is too weak to sustain high population densities in state 2 and , therefore , the system jumps to a new stable configuration , in which the populations of both states are low , i.e. @xmath120 .
a further increase of @xmath67 produces a smooth evolution of the two populations until @xmath121 @xmath12mev@xmath75 $ ] when the system jumps to a third configuration with a population in state 1 much higher than in state 2 , i.e. @xmath122 .
when we revert the variation of the pumping intensities along the same path , the jumps to states corresponding to intermediate and low values of @xmath113 are shifted to the left of the ones just described for increasing @xmath67 . the multistable hysteresis loop shown in fig .
[ fig : hyst]a ) is related to the fact that the two pumping lasers are at different pumping angles , @xmath123 , and pumping frequencies , @xmath111 .
therefore the jumps from low to high population for each component appear at different values of the pumping intensities , producing the multistable behavior of @xmath113 .
a similar situation occurs when the system evolves along a path on which one of the pumping intensities remains constant , while the other varies ( black horizontal line at @xmath74 of panel 1 in fig [ fig:2d3d ] ) , as shown in fig .
[ fig : hyst]b ) . starting the with @xmath124 , @xmath113 increases smoothly from zero following the lower branch up to @xmath125 @xmath12mev@xmath75 $ ] . at this point @xmath113 jumps from values of the order of @xmath126 to @xmath127 ,
corresponding to a population of state 1 being smaller but non - negligible compared to state 2 .
as @xmath67 further increases up to @xmath128 @xmath12mev@xmath75 $ ] , @xmath113 again jumps abruptly to values of the order of @xmath129 . in the reverse process , @xmath67 is decreased completing the loop .
also in this case , the jumps from high to low values of @xmath113 are shifted to the left because of the different angles and energies at which the lasers are pumping the two component of the fluid .
it is worth noting that the length and the height of the different plateaus of the hysteresis loops can be efficiently tuned by carefully choosing the path followed by the intensities , angles and frequencies of the pumping lasers .
to summarize , we have studied the stability of a two - component exciton - polariton fluid under resonant excitation of two pumping lasers with independently tunable frequencies , angles of incidence and intensities .
we have studied the effect of the detuning between the laser pump and the bare lp branch , and discussed the different kinds of instability that might occur for different values of the parameters .
we have shown that , even though the kind of instabilities are the same as in the one component case ( kerr or parametric ) , here the interplay between different instabilities in the two components of the fluid , can produce a much richer picture . moreover ,
we have shown that since each component of the fluid jumps between stable states of its population at a different values of the pumping intensity , the system sustain multistable hysteresis loops that can be easily modulated by changing the parameters of the pumping lasers . finally , we believe that , due to the wide range of parameters for which the system is multistable , and due to the increased number of degrees of freedom with respect to the spin - dependent case , multiatability and hysteresis loops should be within an experimental reach . therefore , the novel system of two - component fluid is a promising candidate for the realization of optical switches and memories .
we thank i. carusotto and d. whittaker for discussions .
this research has been supported by the spanish mec ( mat2008 - 01555 , qoit - csd2006 - 00019 ) and cam ( s-2009/esp-1503 ) .
f.m.m . acknowledges financial support from the program ramn y cajal . | we study the stability of a multicomponent exciton - polariton fluid under resonant excitation within the linear response approximation of a generalized gross - pitaevskii equation .
we show that , two spatially homogeneous and independently tunable pumping lasers produce , for the same values of the system parameters , up to three stable solutions .
three - stability is understood by noting that the cavity can be either little or highly populated and , in this second case , the largest part of the population lies in either one of the two components .
moreover , we discuss the different kinds of instabilities appearing at different pumps intensities and compare them with the case of one - component fluids . finally , we show that easily tunable multistable hysteresis loops can be performed by the system . |
let @xmath0 be a closed surface and @xmath1 be a riemannian metric on @xmath0 .
then the laplace - beltrami operator @xmath2 acts on the space of smooth functions on @xmath0 by the formula @xmath3 it is known that the spectrum of @xmath2 is discrete and consists only of eigenvalues .
moreover , the multiplicity of any eigenvalue is finite and the sequence of eigenvalues tends to infinity .
let us denote this sequence by @xmath4 where the eigenvalues are written with their multiplicities . for a fixed @xmath0 the following quantities can be considered as functionals on the space of all riemannian metrics on @xmath0 , @xmath5 several recent papers @xcite deal with finding supremum of these functionals in the space of all riemannian metrics on @xmath0 .
an upper bound for @xmath6 in terms of genus of @xmath0 was provided in the paper @xcite and the existence of such a bound for @xmath7 was shown in the paper @xcite .
the exact upper bounds are known for a limited number of functionals : @xmath8 ( see @xcite ) , @xmath9 ( see @xcite ) , @xmath10 ( see @xcite ) , @xmath11 ( see @xcite ) , @xmath12 ( see @xcite ) .
we refer to the introduction to the paper @xcite for more details .
the functional @xmath7 depends continously on @xmath1 but this functional is not differentiable . however , it is known that for an analytic deformation @xmath13 of the initial metric @xmath1 there exist the left and right derivatives of @xmath14 with respect to @xmath15 , see e.g. the papers @xcite .
this is a motivation for the following definition .
a riemannian metric @xmath1 on a closed surface @xmath0 is called an _ extremal _
metric for a functional @xmath7 if for any analytic deformation @xmath13 such that @xmath16 the following inequality holds , @xmath17 a metric @xmath1 is called a _
metric for a functional @xmath7 if for any metric @xmath18 on @xmath0 @xmath19 a question whether there exists smooth maximal metric is not trivial .
for example there is no smooth maximal metric for @xmath12 ( see @xcite ) .
the list of known extremal metrics is longer than the list of known exact upper bounds for @xmath7 , but until now their maximality has not been studied . in the present paper
we investigate the maximality of the known extremal metrics .
the list of currently known extremal metrics follows .
* metrics on the otsuki tori @xmath20 were studied in the paper @xcite .
* metrics on the lawson tori and klein bottles @xmath21 were studied in the paper @xcite .
* metrics on the surfaces @xmath22 bipolar to lawson surfaces were studied in the paper @xcite .
* metrics on the bipolar surfaces @xmath23 to otsuki tori were studied in the paper @xcite . in further description a klein bottle
is denoted by @xmath24 .
the definitions of these surfaces are given in the following sections .
the main result of the present paper is the following theorem .
there are no maximal metrics among the metrics ( a)-(d ) except for @xmath25 .
[ maintheorem ] the metric on the lawson bipolar klein bottle @xmath25 is maximal for the functional @xmath26 , see @xcite .
we also prove the following proposition .
the metric on the clifford torus is extremal for an infinite number of functionals @xmath7 , but it is not maximal for any of them .
[ cliffordtheorem ] the extremality of the clifford torus for an infinite number of functionals @xmath7 is known , but to the best of author s knowledge has not yet been published . in the present paper
we fill this gap . in the following description
we use the notations @xmath27 , @xmath28 and @xmath29 for the elliptic integrals of the first , second and third kind respectively , see @xcite , @xmath30 the paper is organized in the following way . in section [ bound ]
we prove lower bounds for @xmath31 and @xmath32 .
these bounds are used throughout the paper in order to prove the non - maximality of metrics ( a)-(d ) . in section [ connection ]
we recall a connection between extremal metrics and minimal submanifolds of the unit sphere .
section [ otsukidef ] contains a discription of otsuki tori as an @xmath33-invariant minimal submanifolds of @xmath34 of cohomogeneity 1 . sections [ otsukiestimate ] , [ lawsonestimate ] , [ biplawsonestimate ] , [ bipotsukiestimate ] are dedicated to estimates for extremal metrics ( a)-(d ) respectively and this finishes the proof of theorem [ maintheorem ] .
finally , section [ cliffordestimate ] contains the proof of proposition [ cliffordtheorem ] .
the aim of this section is to prove the following proposition ( compare with corollary 4 in the paper @xcite ) .
[ bound ] one has the following inequalities , @xmath36 @xmath37 where @xmath28 stands for the elliptic integral of the second kind .
[ lowerbound ] let @xmath0 be a compact smooth riemannian manifold of dimension @xmath38 .
let us pick two distinct points @xmath39 .
for @xmath40 we define * @xmath41 union of open geodesic balls of radius @xmath42 about @xmath43 and @xmath44 , * @xmath45 , * @xmath46 . here
the number @xmath42 is chosen to be less than @xmath47 of injectivity radius of @xmath0 and less than @xmath47 of a distance between @xmath43 and @xmath44 if @xmath43 and @xmath44 lie in the same connected component of @xmath0 .
we say that manifold @xmath48 is obtained from @xmath0 by adding a handle across @xmath49 if * @xmath50 is isometrically embedded in @xmath48 ; * there exists a diffeomorphism @xmath51\times \mathbb{s}^{n-1}$ ] such that @xmath52\times\mathbb{s}^{n-1}\right).\ ] ] let us denote by @xmath53 and @xmath54 the laplace spectrum of @xmath0 and @xmath48 respectively .
chavel and feldman in their paper @xcite obtained a sufficient condition for convergence @xmath55 as @xmath42 tends to @xmath56 . in order to formulate this condition
we need to give the following definition . for any
compact connected riemannian manifold @xmath57 of dimension @xmath38 , the isoperimetric constant @xmath58 is defined by @xmath59 where @xmath60 stands for @xmath61-dimensional riemannian measure , and @xmath62 ranges over all compact @xmath63-dimensional submanifolds of @xmath57 such that they divide @xmath57 into @xmath64 open submanifolds @xmath65 , @xmath66 each having boundary @xmath62 .
assume that @xmath48 is connected for any @xmath42 and there exists a constant @xmath67 such that @xmath68 for all @xmath69 .
then @xmath70 for all @xmath71 [ ch1 ] the assumption in the previous theorem implies @xmath72 by picking @xmath73 . in the same paper existence of such @xmath48 is verified for any surface @xmath0 and almost any pair of points @xmath43 , @xmath44 .
let @xmath0 be a compact @xmath64-dimensional riemannian manifold , @xmath74 be its gaussian curvature and @xmath75 .
then @xmath76 is open and dense in @xmath0 .
suppose that @xmath77 and one of the following possibilities occur : * @xmath0 is connected , * @xmath0 has two connected components and @xmath78 lie in different connected components
. then @xmath48 can be constructed so that assumption of theorem [ ch1 ] holds .
in particular , @xmath79 as @xmath80 . [ ch2 ] let us remark that chavel and feldman considered only the case of a connected manifold @xmath0 .
however , their arguments could be extended almost without changes to the non - connected case as stated above .
consider the flat equilateral torus @xmath81 .
after suitable rescaling of the metric we have @xmath82 and @xmath83 .
for the euclidean sphere @xmath84 of volume @xmath85 one also has @xmath83 .
let us take @xmath86 copies of @xmath84 denoted by @xmath87 , @xmath88 .
thus for @xmath89 we have @xmath90 and therefore @xmath91 .
consecutive application of theorem [ ch2 ] yields the existance of the sequence @xmath48 , diffeomorphic to torus , such that @xmath92 as @xmath42 tends to @xmath56 .
this observation completes the proof of the first inequality .
the second inequality can be proved in the same fashion .
the only difference is that instead of @xmath81 one has to use lawson bipolar klein bottle @xmath25 ( see section [ biplawsonestimate ] for a defintion ) .
it was proven in the paper @xcite that @xmath93 . by a suitable rescaling of the metric on @xmath25
, one can assume that @xmath94 and then apply construction of the previous paragraph .
let @xmath95 be a minimal immersion in the unit sphere with canonical metric @xmath96 .
we denote by @xmath2 the laplace - beltrami operator on @xmath0 associated with the metric @xmath97 .
let us introduce the weyl s eigenvalues counting funcion @xmath98 the following theorem provides a general approach to finding smooth extremal metrics .
let @xmath99 be a minimal immersion in the unit sphere @xmath100 endowed with the canonical metric @xmath96 .
then the metric @xmath97 on @xmath0 is extremal for the functional @xmath101 .
[ th1 ] therefore , one can start with minimal submanifold @xmath102 of the unit sphere then compute @xmath103 and the metric induced on @xmath102 by this immersion is extremal for the functional @xmath104 .
however , for a given minimal submanifold there is no algorithm for computing the exact value of @xmath103 .
nevertheless , this approach was succesfully realized by penskoi in the papers @xcite for metrics ( a),(b ) as well as by the author in the paper @xcite for metrics ( d ) .
some ideas of this approach was partially used in the paper @xcite for metrics ( c ) .
let @xmath0 be a [ connection ] riemannian manifold equipped with a metric @xmath105 and let @xmath106 be a compact group acting on @xmath0 by isometries . for every point @xmath107
let us denote by @xmath108 the stability subgroup of @xmath109 . for two points @xmath110
we say that @xmath111 if @xmath112 for some @xmath113 .
the orbit @xmath114 is the _ orbit of principal type _ if for any point @xmath115 one has @xmath111 .
let @xmath116 stand for the union of all orbits of principal type , then @xmath116 is an open dense submanifold of @xmath0 ( see @xcite ) .
moreover , @xmath117 carries a natural riemannian metric @xmath1 defined by the formula @xmath118 , where @xmath119 are tangent vectors at @xmath120 and @xmath121 are tangent vectors at a point @xmath122 such that @xmath123 and @xmath124 are orthogonal to the orbit @xmath125 and @xmath126 .
let @xmath127 be a @xmath106-invariant immersed submanifold , i.e. a manifold equipped with an action of @xmath106 by isometries such that @xmath128 for any @xmath129 .
a _ cohomogeneity _ of a @xmath106-invariant immersed submanifold @xmath102 is the number @xmath130 , where @xmath131 is the dimension of the orbits of principal type .
let us define for @xmath120 a volume function @xmath132 by the formula @xmath133 . also for each integer @xmath134 let us define a metric @xmath135 .
let @xmath136 be a @xmath106-invariant immersed submanifold of cohomogeneity @xmath61 , and let @xmath117 be equipped with the metric @xmath137 .
then @xmath136 is minimal if and only if @xmath138 is minimal .
[ hsiangth ] otsuki tori [ otsukidef ] were introduced by otsuki in the paper @xcite .
let us recall the concise description by penskoi from the paper @xcite
. for more details see section 1.2 of the paper @xcite .
consider the action of @xmath33 on the three - dimensional unit sphere @xmath139 given by the formula @xmath140 where @xmath141 is a coordinate on @xmath33 .
the space of orbits @xmath142 is the closed half - sphere @xmath143 , @xmath144 where a point @xmath145 corresponds to the orbit @xmath146 .
the space of principal orbits @xmath147 is the open half sphere @xmath148 let us introduce the spherical coordinates in the space of orbits , @xmath149 since we look for minimal submanifolds of cohomogeneity 1 , the hsiang - lawson s metric is given by the formula @xmath150 an immersed minimal @xmath33-invariant two - dimensional torus in @xmath34 such that its image by the projection @xmath151 is a closed geodesics in @xmath147 endowed with the metric ( [ otsukimetric ] ) is called an _ otsuki torus_.
the following proposition was proved in the paper @xcite .
except one particular case given by the equation @xmath152 , otsuki tori are in[penskoiprop ] one - to - one correspondence with rational numbers @xmath153 such that @xmath154 by @xmath20 we denote the otsuki torus corresponding to @xmath153 . following the paper
@xcite we reserve the term `` otsuki tori '' for the tori @xmath20 . in order to fix notations
we give a sketch of the proof of proposition [ penskoiprop ] .
let us use the standard notation for the coefficients of the metric ( [ otsukimetric ] ) , @xmath155 as we know the velocity vector of a geodesic has a constant length .
suppose this length equals 1 .
then this assumption as well as the equation of geodesics for @xmath156 provides the following two equations , @xmath157 @xmath158 where @xmath159 is the minimal value of @xmath160 on the geodesic . then the geodesic is situated in the annulus @xmath161 .
we choose a natural parameter @xmath15 such that @xmath162 .
let us denote by @xmath163 the difference between the value of @xmath164 corresponding to @xmath165 and the closest to it value of @xmath164 corresponding to @xmath166 .
it is clear that @xmath167 the geodesic is closed iff @xmath168 .
the rest of the proof follows from the following properties of the function @xmath163 , see the paper @xcite , * @xmath163 is continuous and monotonous on @xmath169 $ ] , * @xmath170 and @xmath171 . according to the paper @xcite
, the metric on an otsuki torus @xmath20 is extremal [ otsukiestimate ] for the functional @xmath173 .
the goal of this section is to prove the following proposition . for @xmath174 ,
such that @xmath175 and @xmath176 , the following inequality holds , @xmath177 [ otsukiestimateprop ] in order to prove proposition [ otsukiestimateprop ] we have to prove several auxiliary propositions . for @xmath178
such that @xmath168 one has @xmath179 [ valueotsuki ] let us use the notations of proposition [ penskoiprop ] . as we know
, @xmath180 where @xmath181 .
therefore , the length of the segment on the geodesic @xmath182 between the closest points with @xmath165 and @xmath183 is equal to @xmath184 , where @xmath185 let us express @xmath186 in terms of elliptic integrals , @xmath187 here the following changes of variables were used , @xmath188 since the maps @xmath189 and @xmath190 are isometries , the length of the geodesic @xmath182 is equal to @xmath191 . by proposition 13 from the paper @xcite , @xmath172 is equal to the doubled length of the geodesic @xmath182 . for @xmath192
$ ] one has the following inequality , @xmath193 [ mainlemma ] let us expand the left hand side using the definitions of @xmath194 and @xmath74 , @xmath195 since the integrand is negative on @xmath196 and positive on @xmath197 , one has @xmath198 let us introduce the notation @xmath199 the function @xmath200 is non - decreasing and @xmath201 for any @xmath202 .
in particular , @xmath203 . [ phiprop ]
one has @xmath204 [ ineqotsuki ] * remark .
* let us remark that during the preparation of the manuscript inequality ( [ inotsuki ] ) appeared in the paper @xcite .
let us recall the following formulae for the derivatives of elliptic integrals , @xmath205
@xmath206 let us introduce a notation @xmath207 .
one obtains @xmath208 now the monotonicity of the function @xmath200 follows from proposition [ mainlemma ] . for
the proof of the second part , let us go back to formula ( [ dphi ] ) .
one has @xmath209 this finishes the proof of proposition [ phiprop ] .
the function @xmath210 is increasing on the interval @xmath211 . in the paper @xcite the following formula
was proved , @xmath212 using formulae ( [ dpi ] ) one obtains the following formula , @xmath213 let us recall the notation @xmath214 . then one has @xmath215
@xmath216 moreover , by formula ( [ dphi ] ) one has @xmath217 the inequality @xmath218 and proposition [ mainlemma ] imply the inequality @xmath219 for @xmath220 $ ] one has @xmath221 [ cor1 ] using the tables of elliptic integrals , e.g. the book @xcite , one obtains the inequality @xmath222 the rest of the proof follows from the monotonicity of the function on the left hand side . for @xmath223 $ ] one
has @xmath224 [ prop1 ] by formula ( [ domega ] ) for @xmath223 $ ] one has @xmath225 in the last inequality we used the facts that @xmath27 is increasing function and @xmath28 as well as @xmath226 are decreasing functions .
the table of the elliptic integrals in the book @xcite provides the inequality @xmath227 which completes the proof .
we want to prove that @xmath228 where @xmath168 .
this inequality is equivalent to the following one @xmath229 since @xmath168 , it is sufficient to prove that @xmath230 since @xmath231 , the application of corollary [ cor1 ] provides inequality ( [ mainineq ] ) for @xmath220 $ ] . in order to prove inequality for
@xmath232 $ ] let us note that by proposition [ phiprop ] @xmath233 for some @xmath234 .
moreover , @xmath235 or @xmath236 therefore , inequality ( [ mainineq ] ) follows from the inequality @xmath237 or the inequality @xmath238 the last inequality easily follows from proposition [ prop1 ] .
a lawson tau - surface @xmath21 is an immersed surface in the sphere @xmath34 [ lawsonestimate ] defined by the double - periodic immersion of @xmath239 given by the formula @xmath240 it was introduced by lawson in the paper @xcite .
he also proved that for each pair @xmath241 , such that @xmath242 and @xmath243 , the surface @xmath21 is a distinct compact minimal surface in @xmath34 .
let us assume that @xmath243 then if both @xmath244 and @xmath61 are odd then @xmath21 is a torus , we call it a lawson torus . otherwise @xmath21 is a klein bottle , we call it a lawson klein bottle .
let @xmath21 be a lawson surface .
then the induced metric on @xmath21 is an extremal metric for the functional @xmath245 , where @xmath246 + m + k - 1 , \label{j}\ ] ] @xmath247 if both @xmath248 are odd and @xmath249 otherwise .
the corresponding value of the functional is @xmath250 let @xmath251 be defined by formula ( [ j ] ) .
if @xmath21 is a lawson torus , then @xmath252 if @xmath21 is a klein bottle , then @xmath253 [ lawsonprop ] it is sufficient to obtain the inequality @xmath254 let us remark that the function @xmath255 is positive on the interval @xmath256 $ ] .
indeed , @xmath257 let us divide both sides of inequality ( [ lawsonineq ] ) by @xmath244 and denote by @xmath109 the ratio @xmath258 $ ] . since @xmath259\geqslant \left[\frac{m+k}{2}\right]\geqslant\left[\frac{m+1}{2}\right ] > \frac{m}{2},\ ] ] one has that inequality ( [ lawsonineq ] ) follows from the positivity of @xmath260
let @xmath261 be a minimal immersion .
[ biplawsonestimate ] a gauss map @xmath262 is defined pointwise as the image of the unit normal in @xmath34 translated to the origin in @xmath263 .
then the exterior product @xmath264 is an immersion of @xmath102 in @xmath265 .
lawson proved in the paper @xcite that this immersion is minimal .
the image @xmath266 is called a bipolar surface to @xmath102 .
let us denote by @xmath267 the bipolar surface to the surface @xmath21 .
lapointe proved in the paper @xcite that * if @xmath268 then @xmath267 is a torus carrying the extremal metric for the functional @xmath269 and @xmath270 * if @xmath271 then @xmath267 is a torus carrying the extremal metric for the functional @xmath272 and @xmath273 * if @xmath274 then @xmath267 is a klein bottle carrying the extremal metric for the functional @xmath275 and @xmath276 if @xmath277 then the following inequality holds @xmath278 if @xmath279 then the following inequality holds @xmath280 if @xmath274 and @xmath281 then the following inequality holds @xmath282 [ biplawsonprop ] in order to prove the first inequality it is sufficient to prove that @xmath283 it is well - known that @xmath284 for @xmath285 $ ] .
this implies that it is sufficient to prove that @xmath286 this inequality holds for @xmath287 .
the statement for @xmath288 follows from the fact that @xmath289 is a clifford torus and @xmath290 . in the same way , in order to prove the second inequality in proposition [ biplawsonprop ] it is sufficient to prove that @xmath291 this inequality holds for @xmath292 .
the third inequality is equivalent the following one @xmath293 since @xmath294 it is sufficient to prove that @xmath295 this inequality holds for @xmath296 . for the exceptional case @xmath297 one
verifies the third inequality explicitly using the tables of elliptic integrals in the book @xcite .
in the paper @xcite the following proposition was proved .
[ bipotsukiestimate ] the bipolar surface @xmath23 to an otsuki torus @xmath20 is a torus .
if @xmath298 is odd then the metric on bipolar otsuki torus @xmath23 is extremal for the functional @xmath299 and @xmath300 .
if @xmath298 is even then the metric on bipolar otsuki torus @xmath23 is extremal for the functional @xmath301 and @xmath302.[bipotsuki ] if @xmath298 is even then the following inequality holds , @xmath303 if @xmath298 is odd then one has the following inequality , @xmath304 [ bipotsukiprop ] if @xmath298 is even , then we have @xmath305 we used the inequalities @xmath306 and @xmath307 in order to prove the last inequality . in the same way ,
if @xmath298 is odd , then we have @xmath308 now it is easy to see that propositions [ otsukiestimateprop ] , [ lawsonprop ] , [ biplawsonprop ] , [ bipotsukiprop ] together with proposition [ lowerbound ] imply theorem [ maintheorem ] .
let us represent the clifford torus as a flat torus with the square lattice with edges equal to @xmath309 .
in this case [ cliffordestimate ] the laplace - beltrami coinsides up to a sign with the classical two - dimensional laplace operator .
therefore , using the separation of variables one obtains that the eigenfunctions are @xmath310 where @xmath311 and @xmath312 .
then , the eigenvalues are equal to @xmath313 , @xmath314 , @xmath315 , @xmath316 respectively .
it is easy to check that the set of functions @xmath324 form an isometrical immersion of clifford torus in the unit sphere .
the same is true for the set @xmath325 and the set @xmath326 where @xmath327 .
therefore , according to theorem [ th1 ] , the metric on the clifford torus is extremal for the functionals @xmath328 , where @xmath329 with @xmath330 , and @xmath331 .
so , it is sufficient to prove that @xmath335 and this inequality holds for @xmath336 . and
for @xmath337 holds the inequality @xmath338 . this inequality can be obtained by the direct enumeration of all possible values of @xmath339 .
this completes the proof of proposition [ cliffordtheorem ] .
99 a. el soufi , h. giacomini , m. jazar , a unique extremal metric for the least eigenvalue of the laplacian on the klein bottle .
_ duke math . j. _ * 135*:1 ( 2006 ) , 181 - 202 .
preprint ` arxiv : math/0701773 ` .
a. el soufi , s. ilias , riemannian manifolds admitting isometric immersions by their first eigenfunctions .
_ pacific .
_ * 195*:1 ( 2000 ) , 91 - 99 .
a. el soufi , s. ilias , laplacian eigenvalues functionals and metric deformations on compact manifolds .
_ j. geom .
phys . _ * 58*:1 ( 2008 ) , 89 - 104 .
preprint ` arxiv : math/0701777 `
. j. hersch , quatre proprits isoprimtriques de membranes sphriques homognes .
_ c. r. acad .
paris sr a - b _ * 270 * ( 1970 ) , a1645-a1648 .
d. jakobson , n. nadirashvili , i. polterovich , extremal metric for the first eigenvalue on a klein bottle .
_ * 58*:2 ( 2006 ) , 381 - 400 .
preprint ` arxiv : math/0311484 ` .
n. korevaar , upper bounds for eigenvalues of conformal metrics .
_ j. differential geom .
_ * 37*:1 ( 1993 ) , 79 - 93 .
p. li , s .-
yau , a new conformal invariant and its applications to the willmore conjecture and the first eigenvalue of compact surfaces .
math . _ * 69*:2 ( 1982 ) , 269 - 291 .
n. nadirashvili , berger s isoperimetric problem and minimal immersions of surfaces .
. anal _ * 6*:5 ( 1996 ) , 877 - 897 .
n. nadirashvili , isoperimetric inequality for the second eigenvalue of a sphere .
_ j. differential geom . _ * 61*:2 ( 2002 ) , 335 - 340 .
p. c. yang , s .-
yau , eigenvalues of the laplacian of compact riemann surfaces and minimal submanifolds .
_ ann . scuola norm .
sci . _ * 7*:1 ( 1980 ) , 55 - 63 .
a. v. penskoi , extremal spectral properties of otsuki tori . to appear in _
matematicsche nachrichten_. preprint ` arxiv:1108.5160 ` .
s. bando , h. urakawa , generic properties of the eigenvalue of laplacian for compact riemannian manifolds .
_ thoku math . j. _ * 35*:2 ( 1983 ) , 155 - 172 .
m. berger , sur les premires valeurs propres des varts riemanniennes , _ compositio math . _
* 26 * ( 1973 ) , 129 - 149 .
a. v. penskoi , extremal spectral properties of lawson tau - surfaces and the lam equation .
_ moscow math .
j. _ * 12*:1 ( 2012 ) , 173 - 192 .
preprint ` arxiv:1009.0285 ` .
h. lapointe , spectral properties of bipolar minimal surfaces in @xmath340 .
_ differential geom .
appl . _ * 26*:1 ( 2008 ) , 9 - 22 .
preprint ` arxiv : math/0511443 ` .
m. a. karpukhin , spectral properties of bipolar surfaces to otsuki tori . to appear in
_ j. of spectral theory_. preprint ` arxiv:1205.6316 ` p. byrd , m. friedman .
_ handbook of elliptic integrals for engineers and scientists , _ springer verlag , new york - heidelberg - berlin , 1971 .
b. colbois , a. el soufi , extremal eigenvalues of the laplacian in a conformal class of metrics : the `` conformal spectrum '' , _ ann .
of global analysis and geometry _ * 24 * 2003 , 337 - 349 .
preprint ` arxiv : math/0409316 `
. i. chavel , e. a. feldman , spectra of manifolds with small handles , _ comment .
helvetici _ * 56 * ( 1981 ) , 83 - 102 .
d. montgomery , h. samelson , c. t. yang , exceptional orbits of highest dimension , _ ann . of math . _
* 64 * ( 1956 ) , 131 - 141 .
t. otsuki , minimal hypersurfaces in a riemannian manifolds of constant curvature .
_ * 92*:1 ( 1970 ) , 145 - 173 .
hsiang , h. b. lawson , minimal submanifolds of low cohomogeneity .
_ j. diff .
geom . _ * 5 * ( 1971 ) , 1 - 38 .
z. hu , h. song , on otsuki tori and their willmore energy .
_ j. of math .
anal . and appl .
_ * 395*:2 ( 2012 ) , 465 - 473 .
h. b. lawson , complete minimal surfaces in @xmath34 .
_ ann . of math .
_ * 92 * ( 1970 ) , 335 - 374 . | el soufi - ilias theorem establishes a connection between minimal submanifolds of spheres and extremal metrics for eigenvalues of the laplace - beltrami operator .
recently , this connection was used to provide several explicit examples of extremal metrics .
we investigate the maximality of these metrics and prove that all of them are not maximal .
+ _ 2010 mathematics subject classification .
_ 58e11 , 58j50 , 35p15 .
+ _ key words and phrases . _
extremal metric , otsuki tori , lawson tau - surfaces , bipolar surface . |
object tracking plays a crucial role for successful implementation of various kinds of vision applications such as surveillance , human robot interaction , activity recognition , navigation of intelligent vehicles , and the like . to ensure robust performance
, object tracking almost always struggles against dynamic natural scenes .
one of the toughest challenges confronted by any tracking algorithm is to locate a target robustly in the presence of outliers resulting from partial occlusion or background clutter .
an outlier can be defined as a statistical observation different in any physical value ( for example , color , shape , position , motion vector , and so on ) from the others belonging to a target in question .
outliers make it hard not only to locate a target but also to update its object model correctly . in the object tracking domain , there are roughly two different approaches to deal with outliers .
one approach is to use a robust motion estimator@xcite or a robust similarity measure@xcite , designed to preserve its original performance to some extent in the presence of outliers .
an obvious problem of this approach is that trackers are prone to drift when the ratio of outliers exceeds a estimator - specific threshold .
for example , median flow tracker @xcite is based on the assumption that the occluded portion of a target is less than 50 percent of its whole area .
the other approach identifies and removes outliers explicitly to minimize their influence on target localization .
an inherent problem in this kind of approach is how to identify outliers faithfully and how to restore inliers promptly from disocclusion .
previous works @xcite mainly focus on the detection of occlusion and disocclusion , showing reactive behavior of target tracking depending on instantaneous scene structure . in this paper
we consider outliers as important as inliers for object tracking and propose a motion estimation algorithm based on concurrent tracking of inliers and outliers .
our approach is quite different from most of the existing methods which simply discard detected outliers and pay little or no attention to employing them as an important source of information for motion estimation .
the rest of the paper is organized as follows . in section [ sec : concurrent_tracker ] , we present our approach and give a detailed description of its individual parts . in section [ sec : experiment ] , we evaluate the performance of our proposed tracker and compare the result to other state - of - the - art methods on benchmark tracking data sets .
finally , in section [ sec : conclusion ] , we conclude the paper .
our tracker is similar to median - flow tracker @xcite in estimating target motion based on optical flow and making use of pyramidal implementation of the lucas - kanade tracker @xcite .
the proposed tracker accepts as input a pair of images @xmath0 , @xmath1 and a bounding box @xmath2 surrounding a target in question , either determined manually or determined automatically by a target detection algorithm prior to target tracking .
the proposed tracker outputs an updated estimate of the target location in the form of a bounding box @xmath3 .
a set of grid points @xmath4 is initialized uniformly across the bounding box @xmath2 ( figure [ fig : basic_tracker]a ) . for these points ,
optical flow is computed by lucas - kanade tracker .
let @xmath5 be a set of estimated flows .
unreliable flows are then identified based on the forward - backward ( fb ) error criterion introduced in @xcite . for evaluation of fb error
, the proposed tracker computes the optical flow backwardly from @xmath1 to @xmath0 for the set of tracked points @xmath6 , which generates a set of backward flows @xmath7 .
the fb error is computed by employing euclidean distance between @xmath8 and @xmath9 and the flows whose fb error is larger than a predefined threshold @xmath10 are considered as unreliable and eventually filtered out .
we will call the grid points @xmath11 associated with the remaining reliable flow _ matched _ points and the others _ unmatched _ points .
note that the tracked points may fail to be matched during optical flow estimation by the lucas - kanade tracker .
we do not distinguish such unmatched points from the points filtered out and simply call both of them _ unmatched _ points .
considering only in - plane rotation , scale change , and translation , a rigid transformation @xmath12 is then estimated from the matched motion flows by ransac @xcite estimator .
we adopt rigid transformation because more general transformation such as full affine transformation or homography is more error - prone for most tracking tasks .
finally the bounding box @xmath3 is obtained by applying the estimated transformation to @xmath2 : @xmath13 figure [ fig : basic_tracker ] illustrates the tracking procedure of the proposed tracker . in figure
[ fig : basic_tracker]b , unreliable flows which are filtered out by fb test with a threshold value @xmath14 is depicted in black while the matched flows are depicted in green or red according to whether it is classified as inlier ( green ) or outlier ( red ) by ransac estimator .
the basic flow tracker described in section [ ssec : basic_tracker ] has a limited performance for several reasons although it is robust to outliers to some extent .
for example , the tracker may fail even in the presence of a small fraction of outliers if the outliers have motion flows similar to those of the inliers and not filtered out since a small drift can lead a tracker to fail for long - term operation .
this situation is not rare in actual tracking tasks .
in addition , the tracker may fail if the number of outliers exceeds the number of inliers and the outliers form consistent motion flows as in the case of partial occlusion . in such case
the tracker will drift along a direction of the outlier motion .
a more severe problem is that the tracker may not recognize a tracking failure when target motion estimation is carried out in a normal manner .
we solve the aforementioned problems by extend the functionality of the basic flow tracker to track inliers and outliers separately at the same time , which we call the basic flow tracker with the extended functionality a _ concurrent tracker _ ( or simply a tracker in the remaining part of this paper ) .
let @xmath11 be tracking points and @xmath15 be the states of the tracking points at time @xmath16 , where @xmath17 initially , the tracking points @xmath18 are initialized by uniform grid points across the bounding box @xmath19 and their states are set to be _ inlier _
( @xmath20 ) .
the concurrent tracker accepts a pair of images @xmath0 , @xmath1 and a bounding box @xmath2 and a model state @xmath21 and outputs @xmath3 and @xmath22 , where @xmath23 .
the tracking points @xmath11 always are reinitialized by the uniform grid points within the bounding box @xmath2 .
the tracker firstly estimates optical flows and then filters out unreliable flows as in the basic flow tracker of section [ ssec : basic_tracker ] .
let @xmath24 be the set of unmatched points ; @xmath25 be the set of matched points ; and @xmath26 be the set of matched flows .
the set of matched points @xmath27 are further divided into two sets of matched inliers @xmath28 and matched outliers @xmath29 .
similarly @xmath30 is divided into @xmath31 and @xmath32 .
next , the tracker estimates two rigid transformations of @xmath33 for inlier flows @xmath31 and @xmath34 for outlier flows @xmath32 by applying ransac estimator respectively to the two sets of matched flows .
if @xmath33 is estimated successfully ( ii.e .
, in case the number of matched inlier flows is sufficiently large for estimating the rigid transformation and the estimated transformation is supported by a number of matched inlier flows more than a predefined threshold ) , the tracker updates the bounding box as follows : @xmath35 if it fails to estimate @xmath33 , the tracker fails ( ii.e .
, in the event of the insufficient number of matched flows for estimating the rigid transformation or the estimated transformation is supported by too small number of inliers ) and stops all the remaining steps , and simply returns the previous state : @xmath36 @xmath37 in case we succeed to estimate both of @xmath33 and @xmath34 , the states of tracking points are updated as follows .
first , those flows far away from the estimated transformation are determined as unreliable and filtered out .
let @xmath38 and @xmath39 be the remaining flow points after the filtering , given by @xmath40 @xmath41 where @xmath42 is residual given by @xmath43 and @xmath44 and @xmath45 are the estimated variance of the inlier residuals and outlier residuals respectively , and @xmath46 and @xmath46 are constants . the points in @xmath38 and @xmath39 can be regarded as error - free and the model state is updated based on these points .
the model state is updated according to the following rules : 1 .
@xmath47 is set to be _ outlier _ if @xmath48 .
@xmath47 is set to be _ inlier _ if @xmath49 and @xmath50 .
@xmath47 is set to be _ outlier _ , otherwise . in actual implementation , before the state update
, the inlier points in @xmath38 are smoothed by median filtering to remove noise and then dilated .
the median filtering and dilation is applied to a rectangular grid after mapping the points on the rectangular grids , which gives a smoothed version of @xmath38 .
the outlier points in @xmath39 are also smoothed by median filtering but not dilated . after applying smoothing
, some inlier points in @xmath38 and outlier points in @xmath39 may overlap ; in this case , outlier points always have a higher priority according to the update rules above . in case
it fails to estimate @xmath34 , the model state is updated according to the following rules : 1 .
@xmath47 is set to be _ inlier _ if @xmath49 .
@xmath47 is set to be _ outlier _ , otherwise .
concurrent observation of inlier motion and outlier motion provides a powerful cue for the restoration of target inliers from disocclusion . in case inlier motion and outlier motion
are successfully estimated and the two estimated motions are sufficiently different from each other , one of the following two cases holds .
one case is that the target is currently occluded by an occluding object and the target and the occluding object have different motion .
the other case is that current target roi ( region of interest ) includes part of its background and the target is moving while the target roi contains part of the background . in either case , if a large motion difference is detected , restoration conditions are checked to determine the current outliers as inliers .
although the primary goal of this step is to restore disoccluded target region as inliers , the inliers currently misclassified as outliers due to classification errors can also be restored through the step .
although a motion difference between inliers and outliers is detected from the step above , if the difference is not sufficiently large , there is a risk of wrong restoration which leads to drift or tracking failure .
for example , if outliers have a large motion variance , some outliers can be classified as inliers .
therefore , it is very important to measure motion difference reliably .
a simple checking based on euclidean distance between two average motion vectors of inliers and outliers is not adequate .
it is because pure rotation or scale change without translation gives zero mean motion and euclidean distance does not reflect motion variance .
therefore , in order to estimate a statically meaningful motion difference , we measure distance of two rigid transformations instead of comparing motion vectors directly . the distance is measured based on mahalanobis distance under the assumption that the motion vectors follow gaussian distribution .
mahalanobis distance between inlier motion and outlier motion is estimated as follows : @xmath51 @xmath52 @xmath53 where @xmath54 is the estimated distance between two transformations , @xmath33 is the estimated transformation from the matched inlier motions between the previous frame and current frame and @xmath34 is the estimated transformation from the matched outlier motions respectively .
@xmath11 is the location of the tracking points in the previous frame and @xmath55 is the estimated common variance of the two distributions of inliers and outlier motions . in ( [ eq : common_sigma ] ) , @xmath44 ( @xmath45 ) is the variance of the residuals of the matched inliers ( outliers ) with respect to @xmath33 ( @xmath34 ) and it is computed by @xmath56 where @xmath57 is the number of matched inliers .
the condition for inlier restoration is satisfied when the estimated mahalanobis distance @xmath58 , estimated distance of the two transformations @xmath59 , and the ratio of outliers are larger than predefined thresholds respectively : @xmath60 in case the condition @xmath61 is satisfied , we classify the matched outliers whose overall motion is more close to @xmath33 than @xmath34 and its residual from @xmath33 is less than a predefined threshold as inliers .
finally , state of tracking points is updated as follows : @xmath62 where @xmath63 @xmath64 in our implementation , the parameters are empirically set to be @xmath65 , @xmath66 , @xmath67 , and @xmath68 .
figure [ fig : motion_restoration ] shows an example of inlier restoration based on motion cue .
the left figure shows the state of inliers and outliers before applying restoration and the right figure shows an updated state after restoration . in this example , estimated measures were @xmath69 , @xmath70 , and 12 inliers were restored .
one problem of the restoration precedure using motion difference described in subsection [ ssec : restore_motion ] is that the restoration is performed only when a significant motion difference occurs .
this restoration condition can cause problems in two respects .
firstly , inliers may be misclassified as outliers by various causes such as abrupt illumination change , 3d motion , and deformation of target as well as occlusion . in that case
the restoration condition can not be met if there is no external occluding object or included background . secondly ,
as the restoration is delayed until the ratio of outliers exceeds some threshold by ( [ eq : condition ] ) , the rate of unrecovered inliers possibly exceeds true outliers when the size of an occluding object is relatively small as illustrated in figure [ fig : restoration_problem ] , which leads to wrong estimation of outlier motion , @xmath34 . in figure
[ fig : restoration_problem ] , a moving car crosses the road and the tracking points occluded by a street lamp turn into outliers rapidly .
in this case , however , disoccluded tracking points can not be recovered as the majority of outliers consists of misclassified inliers and thus they will give a motion estimate similar to inlier motion . to overcome these difficulties , as a complementary step , we perform an additional step of inlier restoration by using an appearance reference model after the restoration step based on motion difference .
the restoration procedure using a reference model is as follows .
the appearance reference model is initialized by the initial target region of the image and the state of grid points of the reference model all set to be inliers .
we first compute optical flow between the reference model and the estimated target region in a current frame for the grid points of the reference model .
next , we check whether the grid points of the reference model succeeds to match or not and set the state of current tracking point to be inlier if the match succeeds and the state of matched grid point of the reference model is inlier
. frame to frame motion estimation based on optical flow is prone to drift in the presence of background interference , occlusion , fast illumination changes , and deformation of a target .
even a minute drift at early stage can lead a tracker to totally fail in long term tracking .
the appearance reference model described earlier is used not only to restore inliers but also to compensate for possible motion drift of the tracker .
let @xmath71 be the matched point pairs between the reference model and the current frame in the previous inlier restoration step , where @xmath72 is the inlier points in the reference model whose state is inlier and @xmath73 is the matched point in the current frame from the estimation of optical flow . if there exists a rigid transformation and the number of matched points is larger than a predefined threshold , the motion drift is compensated by updating current target location to be the transformed location of the reference location by the estimated transformation .
a subtle and well - known problem caused from the use of a reference model is how to update the model to adapt to the appearance change .
fixing the reference model with the initial model does not suffer from drift or occlusion but can not adapt to the change of the target . on the other hand , too frequent update of the model may lose its basic role as a reference .
our strategy is that the reference model is updated only when the mean intensity or scale of the current target estimate is different from the reference model significantly more than 10 percent and the ratio of outliers is less than a predefined threshold , @xmath74 .
the actual update of the reference model is done simply by replacing it with the current target region and the state of the tracking points .
the idea of compensation of motion drift by using an appearance reference model is motivated from the work by zhang iet.al .
, @xcite , where a klt feature tracker is combinatorially used to detect occlusion and remedy drift of color - based particle filter tracker .
.test dataset . [ cols="^,^",options="header " , ] table [ table : runtime ] shows the average runtime of the proposed trackers on the test dataset .
we are able to observe that even with @xmath75 tracker it runs at more than 100 frames per seconds without using any hardware acceleration , which is rarely observable in other trackers .
this real time performance makes the proposed method more attractive and portable . in this experiment
we demonstrate the performance of our concurrent tracker ( @xmath75 ) by comparing it with several recent state of the art trackers including : mil tracker ( mil ) @xcite , pn tracker ( pn ) @xcite , and locally orderless tracker ( lot ) @xcite . in case of mil tracker ,
scale of the detected bounding boxes are adjusted to be the same with the scale of ground truth for fair comparison as mil tracker does not adopt to scale change .
all code come from the original authors .
the quantitative comparisons are shown in table [ table : accuracy_comp ] and table [ table : success_comp ] . the best performance
are shown in * bold*. the results show that our concurrent tracker overall shows significant improvement of tracking performance especially in terms of tracking accuracy compared with state - of - the - art algorithms
. it also is worth to note that the performance improvement is most significant for the video sequences that involve partial or severe occlusions like ft_face , lot_human , lot_shop , mil_coke11 , mil_dollar , and oal_dudek , demonstrating robustness against partial occlusions of the proposed tracker . in cases of prost_board , prost_box , tld_car , tld_volkswagen sequences
for which pn tracker @xcite obtains best score , the sequences contain challenging intervals where target objects disappear completely from camera view .
different from pn tracker , current implementation of the proposed tracker does not have detection functionality and failed to recover from such full disappearances .
in this paper we proposed a new tracking algorithm based on concurrent tracking of inliers and outliers . by tracking inliers and outliers concurrently , we are able to minimize possible misclassification errors regarding target state ( occlusion / disocclusion ) , giving more reliable tracking performance .
the experimental results shows that our tracker can track targets reliably without drift in various challenging scenarios , confirming the effectiveness of the proposed tracking paradigm .
one limit of our tracker is that it is hard to be recovered from the tracking failures due to moving target during full occlusion .
our future work is to extend the proposed tracker to cope with such failures by combining it with other detectors . | in object tracking , outlier is one of primary factors which degrade performance of image - based tracking algorithms . in this respect , therefore , most of the existing methods simply discard detected outliers and pay little or no attention to employing them as an important source of information for motion estimation .
we consider outliers as important as inliers for object tracking and propose a motion estimation algorithm based on concurrent tracking of inliers and outliers .
our tracker makes use of pyramidal implementation of the lucas - kanade tracker to estimate motion flows of inliers and outliers and final target motion is estimated robustly based on both of these information .
experimental results from challenging benchmark video sequences confirm enhanced tracking performance , showing highly stable target tracking under severe occlusion compared with state - of - the - art algorithms .
the proposed algorithm runs at more than 100 frames per second even without using a hardware accelerator , which makes the proposed method more practical and portable . |
a natural way to investigate dynamical chiral symmetry breaking is via considering the spectral difference of chiral partners . in @xcite , chiral partner sum rules for heavy - light quark mesons in the spin 0 and spin 1 channels
have been presented . for heavy - light quark mesons ,
a crucial step in the evaluation of the operator product expansion is the cancellation of infrared divergences in virtue of the introduction of non - normal ordered condensates .
thereby , the absorption of infrared divergent wilson coefficients into the condensates at non - zero densities or temperatures requires additional renormalization relations . in @xcite , in - medium relations which are consistent to the vacuum case have been given .
however , only the pseudoscalar case has been investigated there and , moreover , a general proof to all orders in the mass dimension is still missing .
chiral partner sum rules may , therefore , serve as a consistency check of the relations derived in @xcite . in view of the dynamical chiral symmetry breaking of quantum chromodynamics and order parameters of the symmetry restoration ( phase ) transition , qcd sum rules for heavy - light quark
pseudoscalar mesons provide the amplification of the numerical impact of the chiral condensate in virtue of the heavy quark mass . hence , four - quark condensates , which are of mass dimension 6 , are numerically not as important as in the case of mesons consisting of light quarks only . however , in order to determine their numerical impact one has to take them into account .
the evaluation of wilson coefficients for four - quark condensates in case of heavy - light quark sum rules is an elaborate task .
a common approximation is the vacuum saturation hypothesis , which leads to a factorization of four - quark condensates into squares of the chiral condensate .
once applied , the factorization significantly simplifies the calculation .
however , as only certain combinations of four - quark condensates may serve as order parameters of chiral symmetry the factorization of all four - quark condensates in terms of the chiral condensate is not necessarily consistent .
therefore , the extension of mass dimension 5 in - medium heavy - light quark qcd sum rules to mass dimension 6 requires the consistent evaluation of four - quark condensate contributions . with respect of this task ,
chiral partner sum rules provide assistance as only chirally odd condensates do not cancel out and the number of terms which have to be calculated is significantly reduced .
we consider the currents [ eq : def_current_2_flavor ] @xmath0 and the corresponding causal correlators [ eq : def_corr_2_flavor ] @xmath1 \rangle \ : , \\ \label{eq : def_corr_2_flavor_1 } \pi^{\rm ( v , a)}_{\mu\nu}(q ) & = i \int { \mathrm{d}^{4 } \hspace{-1pt}x } e^{iqx } \langle { \rm t } \left [ j^{\rm ( v , a)}_\mu(x ) j^{\rm ( v , a)\dagger}_\nu(0 ) \right ] \rangle \ : , \end{aligned}\ ] ] where t@xmath2 $ ] denotes time - ordering and @xmath3 means gibbs averaging @xcite . in the rest frame of the nuclear medium , i.e. @xmath4 , and for mesons at rest , i.e. @xmath5 , the ( axial- ) vector correlator can be decomposed as @xmath6 with @xmath7 and @xmath8 written manifestly covariant .
the transversal projection for a vanishing light quark mass is given by @xmath9 with @xmath10 .
in lowest - order perturbation theory the operator product expansion within the background field method @xcite for the difference of chiral partner correlators reads [ eq:3 ] @xmath11 { \text{tr}_\text{d}}[s_1(p ) ] \right . \\
+ \frac12 { \text{tr}_\text{d}}[s_2(p+q)\sigma_{\mu\nu } ] { \text{tr}_\text{d}}[s_1(p ) \sigma^{\mu\nu } ] \left .
+ { \text{tr}_\text{d}}[s_2(p+q ) \gamma_5 ] { \text{tr}_\text{d}}[s_1(p)\gamma_5 ] \right\ } { \hspace{-0pt } : } \rangle \ : , \end{gathered}\ ] ] @xmath12 \right ) q_1
\\ + \overline{q}_2 \gamma \partial^{\vec\alpha_n } \left ( { \text{tr}_\text{d } } [ \gamma s_1(-q ) ] \right ) \overrightarrow{d}_{\vec\alpha_n } q_2 { \hspace{-0pt } : } \rangle \ ] ] @xmath13 { \text{tr}_\text{d}}[s_1(p ) ] \right .
\\ \left
. - { \text{tr}_\text{d}}[s_2(p+q ) \gamma_5 ] { \text{tr}_\text{d}}[s_1(p)\gamma_5 ] \right\ } { \hspace{-0pt } : } \rangle \ : , \end{gathered}\ ] ] @xmath14 \right ) q_1
\\ + \overline{q}_2 \gamma \partial^{\vec\alpha_n } \left ( { \text{tr}_\text{d } } [ \gamma s_1(-q ) ] \right ) \overrightarrow{d}_{\vec\alpha_n } q_2 { \hspace{-0pt } : } \rangle \ : , \end{gathered}\ ] ] where @xmath15 denotes trace w.r.t
. color or dirac indices .
@xmath16 is the covariant derivative and an arrow indicates whether it acts to the left or right . for sake of a concise notation
we have defined @xmath17 ( with an analog notation for the partial derivative ) .
the quark propagator is denoted by @xmath18 | \omega \rangle$ ] , @xmath19 is the ground state of the strong interaction , @xmath20/2 $ ] and @xmath21 means normal ordering w.r.t . the perturbative vacuum @xmath22 .
. stems from the application of wicks theorem to in lowest order of the perturbative expansion .
@xmath23 , where @xmath24 corresponds to the term where the quark fields of the currents are contracted to propagators and @xmath25 to the term with 2 not contracted quark fields .
there is no flow of hard momentum in the term with four not contracted quark fields .
hence , it does not contribute . in @xcite
, it has been proven that @xmath26 = 0 $ ] for @xmath27 , @xmath28 and @xmath29 .
hence , a superficial view on may tempt to the conclusion that @xmath24 is zero for chiral partner operator product expansions of heavy - light quark meson currents .
on the other hand it is clear that only matrix elements of chirally odd operators may enter the ope , whereas gluon condensates are chirally even . in this sense ,
the cancellation of @xmath24 is in line with naive expectations .
but the introduction of non - normal ordered condensates via @xmath25 would introduce gluon condensates together with infrared divergences . taking this as a heuristic argument for non - zero @xmath24 raises the question of the precise cancellation of these terms and terms added by introducing
non - normal ordered condensates . indeed , from the in - medium ope of d mesons up to mass dimension 5 @xcite it is known that the medium specific divergences are canceled because of the renormalization of @xmath30 .
but this term is chirally even ( formally , it is the matrix element of a vector current ) and does not enter .
moreover , it is known that introducing non - normal ordered condensates , in order to cancel infrared divergent wilson coefficients of gluon condensates , leads to additional finite gluon contributions .
clearly , these have to cancel out in case of chiral partner sum rules .
again , this can be taken as a heuristic argument that only those mass divergences can remain in @xmath24 which are cancelled by chirally odd condensates .
two questions have to be answered
. do all infrared divergences cancel out ? and does the renormalization procedure introduce chirally even condensates ? thus , a careful analysis is mandatory to prove that the obtained results are infrared stable and that the renormalization procedure is consistent .
the quark propagator in fixed - point gauge @xmath31 @xcite can be written as @xmath32 with @xmath33 @xmath34 denotes a derivative operator which arises due to the fourier transform of the perturbation series for the quark propagator in coordinate space from the gluonic background field @xmath35 ; @xmath36 with @xmath37 where @xmath38/g = g_{\mu\nu}^a t^a$ ] is the gluon field strength tensor and @xmath39 is the coupling .
@xmath40 are the generators of the color group and @xmath41 . for the contributions to up to mass dimension 5 , which restricts the quark propagator to next - to - next - to leading order and the gluon field to lowest order ,
the following traces are evaluated _ & = , & _ & = 0 , & _ & = 0 , [ eq : s0_gamma_5_trace ] + _ & = -i g _ ^ , & _ & = 0 , & _ & = 0 , + _ & = 8 g^2 ^ ^ _ .
it is not necessary to consider traces of the second order quark propagator with @xmath42 and @xmath43 .
up to mass dimension 5 they can only be multiplied with their lowest order counterparts , which are zero .
a combination of second - order and first - order propagator leads to mass dimension 6 terms . using @xmath44 { \text{tr}_\text{d}}\left[s_2 \gamma_5 \sigma^{\mu\nu}\right ] = { \text{tr}_\text{d}}\left[s_1 \sigma_{\mu\nu}\right ] { \text{tr}_\text{d}}\left[s_2 \sigma^{\mu\nu}\right ]
$ ] the following contributions have to be considered @xmath45 { \text{tr}_\text{d}}[s^{(0)}_d(p ) ] + { \text{tr}_\text{d}}[s^{(2)}_c(p+q ) ] { \text{tr}_\text{d}}[s^{(0)}_d(p ) ] \biggr . \\
\left .
+ { \text{tr}_\text{d}}[s^{(0)}_c(p+q ) ] { \text{tr}_\text{d}}[s^{(2)}_d(p ) ] + \frac12 { \text{tr}_\text{d}}[s^{(1)}_c(p+q)\sigma_{\mu\nu } ] { \text{tr}_\text{d}}[s^{(1)}_d(p ) \sigma^{\mu\nu } ] \right\ } { \hspace{-0pt } : } \rangle \ : .\end{gathered}\ ] ] due to @xmath46 = { \text{tr}_\text{d}}[s^{(1 ) } \gamma_5 ] = 0 $ ] for arbitrary quark masses , there is no @xmath42 contribution up to this mass dimension .
likewise @xmath47 = { \text{tr}_\text{d}}[s^{(1 ) } ] = 0 $ ] .
lorentz invariance requires that terms with only one gluon field are zero .
therefore , a first - order quark propagator must be combined with a propagator of at least the same order .
keeping both quark - masses @xmath48 finite these four terms give rise to the following integrals @xmath49 \\
+ \frac { g^2}2 \frac{8 m_c}{[(p+q)^2-m_c^2]^4 } \frac{4 m_d}{p^2-m_d^2 } p^\mu p^\alpha g^a_{\mu\nu } g^{b \phantom{\mu}\nu}_{\phantom{b } \alpha } { \text{tr}_\text{c}}\left[t^a t^b\right ] \\
+ \frac { g^2}{2 } \frac{4 m_c}{(p+q)^2-m_c^2 } \frac{8 m_d}{[p^2-m_d^2]^4 } p^\mu p^\alpha g^a_{\mu\nu } g^{b \phantom{\alpha}\nu}_{\phantom{b } \alpha } { \text{tr}_\text{c}}\left[t^a t^b\right ] \\
+ \frac { g^2}{4 } \frac{4 m_c}{[(p+q)^2-m_c^2]^2 } \frac{4 m_d}{[p^2-m_d^2]^2 } g^a_{\mu\nu } g^{b \mu\nu } { \text{tr}_\text{c}}\left[t^a t^b\right ] { \hspace{-0pt } : } \rangle \ : .\end{gathered}\ ] ] analyzing the integrals in euclidean space in terms of the integral @xmath50 reveals the following results .
the first term is @xmath51 .
the second term generates two terms which are @xmath52 and @xmath53 , respectively .
the last term is @xmath54 .
these terms are all zero in the limit @xmath55 . on the other hand ,
the third term does not vanish for @xmath55 .
it causes a term @xmath56 , which diverges with @xmath57 .
projection of the lorentz indices leads to @xmath58 where @xmath59 is the medium four velocity , @xmath60 and @xmath61 .
note that the diagonal elements of are vacuum specific , whereas the medium specific contribution is traceless .
the integral of the third term of can be evaluated in euclidean space : @xmath62 ^ 4 } \frac{(p - q)_\mu ( p - q)_\alpha}{p^2-m_c^2 } \\
\stackrel{\rm w.r.}{\longrightarrow } \frac 23 \frac{m_c m_d}{(4\pi)^2 } \left ( \frac{g_{\mu\alpha}}{2 } { \rm i}_{3,0,2}(q^2,m_d^2,m_c^2 ) + 2 q_\mu q_\alpha { \rm i}_{3,2,3}(q^2,m_d^2,m_c^2 ) \right ) \ : , \end{gathered}\ ] ] where @xmath63 has been used . by virtue of @xmath64 the corresponding integral of the second term in can be derived .
the limit @xmath55 for both terms in is [ eq : ir_divergences_limits ] @xmath65 where reveals the famous infrared singularity from vacuum d meson sum rules .
as the medium specific contribution to is traceless with respect to lorentz indices and due to the vanishing of the second term in there is no medium specific infrared divergent term .
hence , the only terms that have to be absorbed into the condensates by virtue of the introduction of non - normal ordered condensates are vacuum specific .
this is in line with the cancellation of @xmath30 in which would have to absorb the medium specific divergences . moreover , infrared divergent terms which enter through the medium specific gluon condensate , therefore , must enter the @xmath66 or @xmath67 parts of the quark propagator .
owing to , and the limit @xmath55 of in euclidean space is @xmath68 { \text{tr}_\text{c , d}}[s^{(2)}_d(p ) ] { \hspace{-0pt } : } \rangle \stackrel{\rm w.r.}{\longrightarrow } - \frac i6 i{\langle : \hspace{-3pt } \frac{\alpha_s}{\pi } g^2 \hspace{-3pt } : \rangle}{\rm i}_{3,0,2}(q^2,m_d^2,m_c^2 ) m_c m_d \\
\shoveright { = - \frac i6 i{\langle : \hspace{-3pt } \frac{\alpha_s}{\pi } g^2 \hspace{-3pt } : \rangle}\frac{m_c}{m_d } \frac{1}{q^2+m_c^2 } \ : , } \end{gathered}\ ] ] where it has been used that the lowest order quark propagator is a unit in color space and the additional imaginary unit stems from the wick rotation ( w.r . ) .
the normal ordered chiral condensate in lowest order of the light quark mass enters via the following expression @xcite @xmath69 \ : .\ ] ] here , @xmath70 $ ] is the wilson coefficient of the chiral condensate . expressing the normal ordered condensate by the non - normal ordered condensate @xmath71 { \hspace{-0pt } : } \rangle\ ] ] leads to @xmath72 + \frac 12 { \text{tr}_\text{d}}\left [ s_c(q ) \right ] i \int \frac{{\mathrm{d}^{4 } \hspace{-1pt}p}}{(2\pi)^4 } \langle { : \hspace{-0pt}}{\text{tr}_\text{c , d}}\left [ s_d(p ) \right ] { \hspace{-0pt } : } \rangle \ : .\ ] ] despite the striking similarity , revealed by the projection onto elements of the clifford basis in , of the third term in which has to be canceled and the additional term in , a general proof can not be given and the necessity for a precise evaluation is obvious .
evaluation of up to order @xmath73 in @xmath74-scheme gives @xmath75 with the renormalization scale @xmath76 .
insertion into and in virtue of the limit @xmath55 the anticipated cancellation of infrared divergences in terms of light quark masses is revealed @xmath77 - \frac 16 \frac{m_c}{m_d } \frac{1}{q^2+m_c^2 } { \langle : \hspace{-3pt } \frac{\alpha_s}{\pi } g^2 \hspace{-3pt } : \rangle}\ : .\ ] ] adding and the infrared divergent term cancels out .
furthermore , eq . and the explicit evaluation of the renormalization of normal ordered condensates @xcite shows , that in chiral partner sum rules up to and including mass dimension 5 in the limit @xmath55 only the chiral condensate mixes with other condensates by virtue of introducing non - normal ordered condensates .
thus , apart from the term which cancels the infrared divergence no additional chirally even terms enter and the final operator product expansion is chirally odd .
so far , the investigation was carried out for the spin 0 case .
fortunately , as the only formal difference between the spin 0 and spin 1 case is the cancellation of the projections of quark propagators onto the tensor @xmath43 in the spin 1 case and a global factor @xmath78 , see , from the previous evaluation it is clear that the terms of interest are the same in both cases up to mass dimension 5 .
we now turn our attention to four - quark contributions in .
the typical four - quark condensates for the @xmath79 corrections are of the form @xmath80 where @xmath81 is a dirac matrix and @xmath82 is a matrix in flavor and color space ( respectively , @xmath83 is the number of flavors and colors ) .
@xmath84 denotes an n - fold tensor in minkowski space constructed by suitable combinations or contractions of @xmath85 such that is a lorentz scalar , in particular even w.r.t .
@xmath86 is the levi - civita symbol , a pseudotensor
. their diagrammatical representation can be found in fig .
[ fig:4q_quark_prop ] for the contributions with soft gluon momenta and in fig .
[ fig:4q_gluon_prop ] for contributions of hard gluon momenta .
these can be grouped into such ones which are already present in the vacuum sum rule and others which are specific for the medium . for the medium specific condensates
the medium velocity @xmath87 enters @xmath84 .
it should be emphasized that four - quark condensates ( cf .
@xcite for a complete classification ) enter in different combinations the light vector meson ( cf .
@xcite ) or the nucleon sum rule ( cf .
@xcite ) .
soft gluon momentum contributions arise from non - local quark fields or higher order quark propagators . expanding non - local quark fields up to third order in the covariant derivative and using the equations of motion for the gluon field strength tensor gives rise to the diagrams depicted in figs .
[ fig:4q_quark_prop_sub1 ] and [ fig:4q_quark_prop_sub2 ] .
inserting quark propagators with a next - to - leading order gluon line attached and using the equations of motion for the gluon field strength tensor leads to the diagram shown in fig .
[ fig:4q_quark_prop_sub3 ] .
soft gluon contributions in chiral partner sum rules are proportional to the medium velocity @xmath87 and are therefore absent in vacuum .
contributions with hard gluons arise from the @xmath88 perturbative correction .
there are vacuum as well as medium condensates .
in summary , we try to extend the chiral partner qcd sum rules derived in @xcite in zero order of @xmath79 .
higher order of @xmath79 are related to higher - dimensional condensates .
most notable the four - quark condensates , which are of mass dimension 6 , enter such difference sum rules .
abandoning the vacuum saturation hypothesis introduces a large number of different vacuum and medium four - quark condensates , but it is necessary to keep a clean separation of chirally odd and even terms .
furthermore , gluon condensates of the form @xmath89 are of mass dimension 6 as well .
possible infrared divergences linked to them and their cancellation for finite temperatures or densities have to be investigated carefully .
a. i. bochkarev and m. e. shaposhnikov , nucl .
b * 268 * , 220 ( 1986 ) .
m. a. shifman , a. i. vainshtein and v. i. zakharov , nucl .
b * 147 * , 385 ( 1979 ) .
v. a. novikov , m. a. shifman , a. i. vainshtein and v. i. zakharov , fortsch .
phys . * 32 * , 585 ( 1984 ) .
t. hilger , r. schulze and b. kmpfer , j. phys .
g * 37 * , 094054 ( 2010 ) .
r. thomas , t. hilger and b. kmpfer , nucl .
phys . a * 795 * , 19 ( 2007 ) .
r. thomas , t. hilger and b. kmpfer , prog . part .
* 61 * , 297 ( 2008 ) .
r. thomas , t. hilger , s. zschocke and b. kmpfer , aip conf .
proc . * 892 * , 274 ( 2007 ) .
r. thomas , phd thesis , technische universitt dresden , 2008 . | recently , in @xcite qcd sum rules for chiral partners in the open - charm meson sector have been presented at nonzero baryon net density or temperature up to and including mass dimension 5 . referring to this ,
details concerning the cancellation of infrared divergences are presented and important technical and conceptional ingredients for an incorporation of four - quark condensates beyond factorization and of other mass dimension 6 condensates are collected . |
amorphous magnetic wires are of interest for their actual and potential use , e. g. in sensors @xcite . as it is almost usual in micromagnetism , their properties depend on the technique of sample preparation and thermal treatment . then , they are a good subject for computational science . the problem of the stray field and/or magnetization distribution of the wire have been treated either by means of a purely computational methods , where the wire volume is divided into magnetically polarized units @xcite , or within the dipolar or similar model approximation @xcite .
the latter approach is particularly suitable for magnetic arrays , formed as sets of parallel microwires perpendicular to the array @xcite .
other authors used an approximation of a homogeneous interaction field @xcite . as it was pointed out by velazquez et al .
@xcite , the homogeneous field does not work for short interwire distances .
moreover , in fact the spatial distribution of the magnetization near the wire ends remains unknown ; and it is precisely in this area where the remagnetization process starts . as a consequence , our knowledge on the subtle process of the switching of the magnetic moment in the wire advances with difficulties , despite an obvious interest motivated by technology .
recent progress in the field is due to an application of some phenomenological theories , and by the obtained accordance of the results with experimental data @xcite . although this procedure improved much our understanding of the underlying physical mechanisms , still questions seem to be not much less valuable than answers . here
we are interested in the bistable hysteresis loops for various orientations of the wires with respect to each other and to the applied field .
the motivation is the potential flexibility of the hysteresis loops , which we expect to result when we modify mutual positions and orientations of the wires . on the other hand , it is desirable to confront the experimental data with the model calculations .
even simplified , the model provides a reference point , and the departures between model and reality tells us where an established picture fails .
our method here is to use a simplified model of the interaction , where the wires are magnetized homogeneously and the wire diameter is set to zero .
this leaves the stray field of a wire reduced to the field created by two magnetic charges , placed at the wire ends .
the approximation of an infinitely thin wire is justified as long as the wire - wire distance is much larger than the wire diameter .
the resulting formula is more refined than the approximation of a homogeneous field , and much easier to use than the micromagnetic simulation .
this advantage allows to calculate the interaction field for any mutual position and/or orientation of the wires . despite of its practical ease
, there is an additional methodological argument for the assumption on the homogeneous distribution of the magnetization .
let us suppose the case of two parallel wires , which form two edges of a rectangle .
this system has been investigated by several authors @xcite .
the experimental results on the hysteresis clearly prove that the stray field of the wires act as to prefer the antiparallel orientations of their magnetic moments .
it is reasonable to expect that the stray field of one wire acting at another wire should be evaluated near the end of the latter wire ; it is at this point where the remagnetization process starts .
then , every deviation from the homogeneous distribution of the magnetization weakens the interaction field , what makes the model uneffective .
let us add that in our numerical simulations reported from section 3 we are going to evaluate the interaction field not near the wires ends , but at the wires ends .
this simplification is expected not to change results remarkably as long as the wires are not parallel .
we are interested in two questions : i ) which geometrical configurations of the wires allow for the bistability , ii ) what kinds of the hysteresis loops appear for various wire configurations ? the former question appears as a consequence of the fact , that for some mutual positions and orientations of the wires , the effective ( external + interaction ) field causing the large barkhausen jump at one wire end changes its sign at a point along of the wire ; then the domain wall is stopped at this point and the bistability is lost . in this case
some more complex behaviour of the hysteresis loop can be expected ; for clarity we limit the subject of this text to the bistable case . the second question is a necessary introduction to consider magnetic properties of sets of magnetostatically interacting elements .
such systems are of potential interest e.g. for submicrometer magnetic dots @xcite .
we hope that the answer , even simplified , will activate the search of spatial configurations of the wires which can be useful for other new applications .
the condition applied here can seem to be too restrictive , because in fact the bistability can still be observed if the point at the wire where the field along the wire is zero may vanish at an increasing applied field before the domain wall reaches it . here , however , we are not going to rely on these dynamical effects .
we limit our interest to the cases where the bistability is observed even in the limit of small frequency , and the field amplitude only as large as to activate the switching process . on the other hand , it is known that at field below the order of 1 a / m the domain wall motion is much slower .
however , this effect does not influence the shape of the hysteresis loop as long as the period 1/@xmath9 of the applied sinusoidal field is larger than the time of the remagnetization .
our discussion is limited to this case .
it is known that the wire properties depend on the applied tensile stress @xcite . for the purposes of this work , most important
agent is the stress dependence of the switching ( coercive ) field .
we should repeat here that the condition of the bistability considered in this text is applied for the effective field ( applied field + interaction ) which triggers the domain wall motion .
the enhancement of the switching field by the tensile stress demands an increase of the external field , what influences the bistability condition .
then we consider the tensile stress as an important and nontrivial factor which can change the system properties .
the layout of this text is as follows . in the next section
the data are provided on the experimental dependence of the switching field of the wire on the applied tensile stress . in section 3
we outline the question on the bistability of interacting wires , and we ilustrate this question by an analytical solution in the simplest case of parallel wires . in section 4 ,
conditions of the bistability are investigated numerically .
this section contains also some examples of the hysteresis loops of various bistable systems , including the case with the applied tensile stress .
final conclusions close the text .
in order to capture the dependence of the switching field vs the tensile stress the axial hysteresis loop of as - cast wire of the nominal composition @xmath10 , the diameter @xmath11 @xmath12 and the length of @xmath13 @xmath14 was measured by means of the conventional induction method @xcite .
the frequency of the external sine - like field was equal to @xmath15 @xmath16 and the amplitude of was set to @xmath17 @xmath18 .
one end of a wire was fixed to the sample holder , while to the other one a mechanical loading was attached in order to apply the additional tensile stress @xmath19 during the magnetic measurement .
the mechanical loading was being changed from zero to @xmath20 @xmath21 . obtained values of the switching field @xmath22
are shown in fig . 1 .
the wires considered here are iron - based straight pieces of amorphous alloy , @xmath23=10 cm long , with diameter of 2@xmath24=120 micrometers .
their magnetization is @xmath25=5.6@xmath26 a / m .
this kind of wires are well worked out in the literature @xcite .
the magnetic pole at each end contains the magnetic charge @xmath27= @xmath280.0063 am .
the switching ( coercive ) field is about 10 a / m .
suppose that a center of such a wire is placed at the coordination center , and the wire is parallel to oz - axis .
in the surrounding space , there are planes where the vertical component of the stray field is equal to the applied field , equal in this case to the switching field .
if a system of two wires is to be bistable , the second wire can not go through such a plane .
if it does , the domain wall in the second wire is stopped . in the ( r - z ) plane ,
the condition that the field created by one magnetic pole plus the external field @xmath29 cancel with the field created by another magnetic pole is @xmath30 where @xmath31 for @xmath29=0 , the equation can be untangled . expressing all lengths in units of @xmath32 , what is a half of the wire length , and
after some simple manipulation we get the equation @xmath33 one of the consequences of this condition is that parallel wires of exactly the same length and very close to each other cease their bistability if they are mutually shifted along their length .
let us consider the case when they are not shifted ; then they form edges of a rectangle .
the axial component of the stray field created at an end of one wire at distance @xmath34 from the closer end of another wire is @xmath35 where @xmath36 is the interwire distance .
this component is maximal at the point @xmath34 where @xmath37 , and its value at the maximum varies with @xmath36 approximately as @xmath38 .
if @xmath39 , the contribution to the stray field from other wire end is negligible .
however , this evaluation relies on an assumption that the remagnetization starts at the point at the wire where the stray field from another wire has a maximum .
this assumption remains not justified .
also , the obtained values of the interaction field much excess the observed value , which is of order of 10 a / m for the distance @xmath36 of 1 mm @xcite .
this can mean , that in fact the switching starts at smaller value of @xmath34 , i.e. closer to the wire end .
anyway , the evaluation @xmath40 seems to better reflect the experimental data of ref .
@xcite , than the formula used in ref .
@xcite for microwires @xmath41 at least for small interwire distance @xmath36 .
the latter formula can be derived as the interaction between the opposite ends of the wires .
we should add that the authors of ref .
@xcite used is to test the stray field dependence on the wire length .
the stress dependence of the point @xmath34 where the remagnetization starts could explain the observed variation of the interaction field with stress @xcite .
the values of the interaction field , observed in this experiment , are as large as 500 a / m .
this order of magnitude can be reproduced by the formula @xmath38 , given above .
then , the observed difference between the conventional and the cold - drawn wires can be discussed in terms of the position of the point , where the remagnetization starts .
in figs . 2 - 8 we show numerical solutions of the same problem for various mutual orientations of the wires .
the value of the applied field is taken as to get the remagnetization .
this means that the effective field at one end of the wire is equal to the switching field . here
again , the question is : will the domain wall reach the other wire end ? marked areas in the figures give the values of the parameters , where the answer is `` no '' .
let us assume that the magnetic field is along the oz axis .
the spatial configurations are : wire a along the field , wire b in the same plane xz forms an angle @xmath42 ( fig .
2 ) ; one wire along the field , another perpendicular , also in the plane xz but with y different by 2 mm ( fig .
3 ) ; two wires in the parallel planes xz with distance 2 mm one from another , both wires form angles @xmath43 with the field ( fig .
4 ) ; two wires in the same plane xz , one perpendicular to the field , another forms angle @xmath44 with the former , as clock hands ( fig .
5 ) ; two wires in the same plane xz , one parallel to the field , another forms angle @xmath44 with the former , as clock hands ( fig .
6 ) ; two wires in the same plane xz form the same angle with the field , rotating as bird wings ( fig .
7 ) ; three wires along three edges of a tetrahedron , form the same angles with the field ( fig . 8) . in the case shown in fig .
6 we include also the tensile stress , as large as to enhance the switching field @xmath45 from 10 to 15 a / m .
this is shown to change remarkably the area where the bistability is absent .
we also show some parameters of the obtained hysteresis loops .
the width of the horizontal part of the loops which is responsible for the wire - wire interaction @xcite , is denoted as @xmath46 .
the right - shift of the loop center along the field axis is denoted as @xmath47 , and the width of the loop - as @xmath48 . in several cases ,
the results reflect obvious properties of the systems .
for example , for the wires almost perpendicular to the applied field , the hysteresis loop must be very wide .
we note that when the wires are not parallel to the applied field , the measured and calculated @xmath46 ceases its simple interpretation as the interaction field , but depends also on the orientation of wires with respect to the external field .
to summarize main shortcomings of our approach : _
i ) _ the wire - wire interaction is approximated with an assumption on the uniform wire magnetization , _ ii ) _ our considerations are limited to the case of strict bistability , where the domain wall moves till the end of the wire at the same external field as it started to move . as it was remarked above , the former approximation can be justified if the distance between wires is much larger ( at least one order of magnitude ) than the wire diameter .
the second limitation excludes many cases which can be of interest . in these cases the dynamics of domain walls depends on the field amplitude and frequency @xcite , and the numerical solution of the equation of motion of the domain wall seems to be the only proper method .
even with the drawbacks listed above , the results of the model calculations point out that spatial configurations of bistable wires offer a rich variety of hysteresis loops .
the parameters of the loops can be controlled by the modification of mutual positions of the wires , the angles formed by the wires and the applied magnetic field and of the applied tensile stress .
the obtained hysteresis loop reveal stable states , which remain until the external field is enhanced above some critical values .
then , these states encode the information on previous conditions applied to the wire .
+ k. k. and p. g. are grateful to julian gonzlez and arcady zhukov for helpful discussions and kind hospitality .
00 m. vzquez and a. hernando , j. phys .
d * 29 * ( 1996 ) 939 .
m. vzquez , physica b * 299 * ( 2001 ) 302 .
k. mohri , t. uchiyama , l. p. shen , c. m. cai and l. v. panina , j. magn .
magn . mater .
* 249 * ( 2002 ) 351 .
r. hertel , j. magn .
magn . mater .
* 249 * ( 2002 ) 251 .
h. forster , t. schrefl , w. scholz , d. suess , v. tsiantos and j. fidler , j. magn .
magn . mater .
* 249 * ( 2002 ) 181 .
j. velzquez , k. r. pirota and m. vzquez , ieee trans . magn .
* 39 * ( 2003 ) 3049 .
r. varga , k. l. garcia , m. vzquez and p. vojtanik , phys .
* 94 * ( 2005 ) 17201 .
a. kaczanowski and k. kuakowski , micr . eng .
* 81 * ( 2005 ) 317 .
j. velzquez , c. garcia , m. vzquez and a. hernando , phys .
b * 54 * ( 1996 ) 9903 . l. c. sampaio , e. h. c. p. sinnecker , g. r. c. cernicchiaro , m. knobel , m. vazquez and m. velazquez , phys .
b * 61 * ( 2000 ) 8976 .
t. a. savas , m. farhoud , h. i. smith , m. hwang and c. a. ross , j. appl .
* 85 * ( 1999 ) 6160 .
r. p. cowburn and m. e. welland , science * 287 * ( 2000 ) 1466 .
a. mitra and m. vzquez , j. phys .
* 23 * ( 1990 ) 228 .
m. vzquez , j. gonzlez and a. hernando , j. magn .
magn . mater .
* 53 * ( 1986 ) 323 .
chen , n. m. dempsey , a. hernando and m. vzquez , j. phys .
* 28 * ( 1995 ) 1022 . c. gomez - polo , m. vzquez and d
chen , appl .
. lett . * 62 * ( 1993 ) 108 .
p. gawroski , z. zhukov , j. m. blanco , j. gonzlez and k. kuakowski , j. magn .
* 290 - 291 * ( 2005 ) 595 .
g. krupiska , a. zhukov , j. gonzlez and k. kuakowski , comp .
* 36 * ( 2006 ) 268 . | long - range magnetostatic interaction between wires strongly depends on their spatial position .
this interaction , combined with applied tensile stress , influences the hysteresis loop of the system of wires through the stress dependence of their coercive fields . as a result
, we obtain a set of stable magnetic states of the system , dependent on the applied field , applied stress and mutual positions of the wires .
these states can be used to encode the system history .
* stable states of systems of bistable magnetostrictive wires against applied field , applied stress and spatial geometry * + p. gawroski@xmath0 , a. chizhik@xmath1 , j. m. blanco@xmath2 and k. kuakowski@xmath3 + _ @xmath4 faculty of physics and applied computer science , agh university of science and technology , al .
mickiewicza 30 , 30 - 059 krakw , poland + @xmath5 departamento fisica de materiales , facultad de quimica , upv , 1072 , 20080 san sebastian , spain + @xmath2 departamento fisica aplicada i , eupds , upv / ehu , plaza europa , 1 , 20018 san sebastian , spain + @xmath6e - mail : [email protected] + @xmath7e - mail : [email protected] + @xmath8e - mail : [email protected] + _ _ pacs numbers : _
75.50.kj , 75.60.ej , 75.80.+q _ keywords : _ micromagnetism , amorphous wires , hysteresis , magnetostatics |
in this note we consider the continuum limit of a field theory defined in a large box of size @xmath0 . in the limit @xmath1 , all discrete states apart from bound states become continuum states . according to most textbooks , quantum field theory
is supposed to be well understood in this limit .
we show , however , that the limiting procedure is subtle and unless proper care is exercized the usual prescriptions may give incorrect results . in particular ,
we show an example where the charge of the system seems to be a continuous function of the background field whereas it should be integer - valued .
an example which we have discussed in previous papers ( which are referred to as cdi @xcite and cd2 @xcite ) is provided by the second quantized dirac theory in the presence of a one dimensional four - vector potential vanishing at spatial infinity . the usual limiting procedure for counting states in the continuum limit
is given by @xmath2 where @xmath3 is the scattering phase shift for a particle of momentum @xmath4 in that limit . yet as we ( and many others ) have argued the vacuum charge @xmath5 of the system defined in the normal way by the spectral asymmetry @xmath6 which is obviously integer - valued in a box of finite size @xmath0 is given in the continuum limit using eq .
( [ sum ] ) by @xmath7 where @xmath8 and @xmath9 refer to electron and positron scattering phase shifts and @xmath10 and @xmath11 are the number of positive and negative energy bound states .
since in one - dimension ( see cdi ) @xmath12 this implies for the potential @xmath13 that @xmath14 which is a continuous function of @xmath15 the correct result is calculated using eq .
( [ spec ] ) in cd2 to be @xmath16\ ] ] where @xmath17 denotes integral part of , which is obviously an integer .
the erroneous result of eq .
( [ err ] ) for the vacuum charge for this model is found in many places besides cdi , for example in refs .
@xcite and @xcite . the same error may also be responsible for similar results where other quantised quantities such as baryon charge or angular momentum seem to be given non - quantised values @xcite for no apparent reason . in cdi following barton @xcite we counted states and showed that the number of both positive and negative energy states is unchanged when a potential is switched on from zero thus the vacuum charge is still zero in the presence of a small potential : there is a bound state but the number of continuum states
has decreased by one . as the potential increases in strength @xmath5 changes by one according to eq .
( [ spec ] ) whenever a state crosses @xmath18
so why is the erroneous result obtained and how can it be avoided ?
we now turn to these points . in order to quantize the model consistently with standard anticommutation relations
it is essential that eigenfunctions be normalised to unity .
we shall see that to ensure correct normalisation we have to include an unfamiliar normalisation factor @xmath19 which reduces to unity as @xmath20 ; in fact @xmath21 .
it turns out that the expression for @xmath22 involves the phase shifts in a way reminiscent of the incorrect result of eq .
( [ err ] ) .
we obtain a new expression for the charge density differing from the conventional one by terms of order @xmath23 . we find a finite change in the vacuum charge ( defined as the space integral of the vacuum charge density ) induced by the spectral asymmetry residing in a certain region of space of the system .
it is crucial to get this @xmath24 behaviour correct if we wish to sum over states and convert sums to integrals via @xmath25 there are topologically nontrivial models where the dirac particle is coupled to a soliton ; the attending zero modes induce non - integer values for the vacuum charge @xcite .
however in problems with trivial topology such as ours @xmath5 is integral .
the potential @xmath26 is symmetric and is taken to vanish for @xmath27 we take the system in a box of length @xmath28 with periodic boundary conditions @xmath29 wavefunctions are classified according to the magnitude of the wavevector outside the well and parity .
positive energy solutions outside the potential take the form
@xmath30 @xmath31 the subscript + in the phase shifts @xmath32 refers to the energy sign .
similar expressions are valid for negative energy states @xmath33 , @xmath34 provided we replace @xmath35 by @xmath36 and change notation from @xmath37 @xmath38 @xmath39 @xmath40 to @xmath41 @xmath42 @xmath43 @xmath44 .
we also quote for future reference the form of the even bound state wavefunction outside the well @xmath45 we insist on the normalization @xmath46 for all eigenstates of the hamiltonian . for the bound state the appropriate value of @xmath47 to ensure correct normalization depends on the detailed behaviour of the potential .
we rederive a result originally due to stone @xcite which itself is based on a problem in landau and lifshitz @xcite .
we start with the dirac equation ; the argument is equally valid for either positive or negative energy solutions , thus @xmath48@xmath49 and left - multiply by @xmath50 @xmath51 write the dirac equation for @xmath52 , take the hermitian conjugate and right - multiply by @xmath53@xmath54 subtract ( [ e3 ] ) from ( [ e2 ] ) to get @xmath55 integrating over @xmath56 from @xmath57to @xmath58 : @xmath59 _ { z_1}^{z_2}=\left ( e_{k^{\prime } } -e_k\right ) % tcimacro{\dint } % beginexpansion \displaystyle \int % endexpansion _ { z_1}^{z_2}u_k^{\dagger } u_{k^{\prime } } dz\ ] ] take @xmath60 in the above equation and divide by @xmath61 : @xmath62 _ { z_1}^{z_2}=% \frac{de}{dk}% % tcimacro{\dint } % beginexpansion \displaystyle \int % endexpansion
_ { z_1}^{z_2}u_k^{\dagger } u_kdz\ ] ] or @xmath63 _ { z_1}^{z_2}=% \frac ke% % tcimacro{\dint } % beginexpansion \displaystyle \int % endexpansion _ { z_1}^{z_2}u_k^{\dagger } u_kdz\ ] ] this is the key equation .
its power lies in the fact that to evaluate the left hand side for @xmath64 it suffices to use the asymptotic expressions eqs .
( [ ee ] , [ eo ] ) for the wavefunctions where only the phase shifts appear .
note that the normalisation condition ( [ n1 ] ) together with periodic boundary conditions entail restrictions on @xmath4 .
apply relation ( [ e6 ] ) at the endpoints @xmath65 .
then in the left hand side we only need the asymptotic expressions eqs .
( [ ee ] , [ eo ] ) and on the right hand side we can use eq.([n1 ] ) to set the integral equal to unity .
the evaluation of the left hand side simplifies because of the periodic boundary conditions .
we thus obtain @xmath66 and therefore @xmath67 for @xmath0 large .
note that the quantities @xmath68 vanish both when @xmath69 and in the limit @xmath1 .
( [ enoe ] ) is the main result of this note .
we focus on the charge density @xmath70 of an eigenstate @xmath71 ( of definite parity and sign of energy ) corresponding to a particular wavevector @xmath72 @xmath73 can be written down for the scattering states for @xmath74 by using the asymptotic forms ( [ ee ] ) and ( [ eo ] ) of the wavefunctions .
the factors @xmath75 provide a spatially uniform contribution to the charge density equal to @xmath76 in the absence of the potential the first two terms in eqs .
( [ d5 ] , [ d7 ] ) would still be there .
we wish to calculate @xmath83 defined as the part of @xmath5 residing outside the well .
( since @xmath5 itself vanishes this charge is cancelled exactly by an opposite charge residing inside the well . )
we see that the continuum contribution to the vacuum charge outside the well resulting from the last term in ( [ d5 ] , [ d7 ] ) is given by ( where the phase shifts @xmath85 refer as before@xcite to the sum of the even and odd phase shifts ) , an expression which is very similar to eq .
( [ err ] ) and which will in general give non - integral values for @xmath86 to get the total continuum contribution we should integrate over the terms that depend on the mass explicitly .
the final result is in general non - integral since there is no reason why the charge inside or outside a particular region of space should be integer - valued .
we would like to thank gabriel barton , fred goldhaber and mike birse for their help and encouragement .
one of us ( ac ) wishes to thank professor volovik and the low temperature laboratory of helsinki university of technology for their hospitality and eu human capital and mobolity visitor programme chgect94 - 0069 for its support . | we investigate a commonly used formula which seems to give non - integral vacuum charge in the continuum limit .
we show that the limit is subtle and care must be taken to get correct results . |
the analysis of intensity ratios of collisionally excited lines ( cels ) is the standard method for deriving ionic abundances in ionized nebulae .
almost all the available determinations of the gas - phase galactic abundance gradients from regions observations are based on this kind of analysis @xcite .
the use of cels has the disadvantage that their intensity strongly depends on the assumed electron temperature and on the presence of temperature variations or fluctuations over the observed volume of the nebula .
the most dramatic effect of temperature fluctuations is that ionic abundances are underestimated if these fluctuations are present in the objects but not considered in the abundance calculations @xcite .
recombination lines ( rls ) of heavy element ions are very faint and have been difficult to observe in nebular spectra .
fortunately , the availability of high efficiency echelle spectrographs in large aperture telescopes makes feasible the measurement of rls for an increasing number of nebular objects .
several autors have obtained o@xmath0/h@xmath1 and c@xmath0/h@xmath1 from the intensity of and rls for the brightest regions of the galaxy ( peimbert et al .
1993 ; esteban et al .
1998 , 1999a , b , 2004 , hereafter e04 ; tsamis et al . 2003 ; garca - rojas et al .
2004 , hereafter g04 ) ; all of them have found that abundance determinations from rls are systematicaly larger than those obtained using cels , independently of the ion considered .
these differences , as well as other discrepancies in the nebular properties derived , can be consistently accounted for assuming a _ @xmath6 _ ( mean square temperature variation over the observed volume ) in the 0.020.05 range .
@xcite were the first authors who estimated the galactic c and o abundance gradients from rls in regions , but only with three objects and covering a rather small range of galactocetric distances ( from 6.4 to 8.4 kpc , assuming the sun at 8 kpc ) .
this gradient was later revised by @xcite and g04 , these last authors included abundances for an additional region at 7.46 kpc ( ngc 3576 ) . in this letter , we present new values of the c and o gradients obtained from very deep echelle spectra of eight galactic regions located at galactocentric distances ranging from 6.3 to 10.4 kpc ( assuming the sun at 8 kpc ) . some of our sample nebulae were previously observed by our group with other instrumentation the orion nebula @xcite , m 8 @xcite , and m 17 @xcite , but they have been re - observed for this project .
data of some of the objects included in this paper have been already published : ngc 3576 ( g04 ) and orion nebula ( e04 ) .
the observations were made on 2002 march 11 and 12 and on 2003 march 29 , 30 , and 31 with the ultraviolet visual echelle spectrograph , @xcite , at the very large kueyen telescope in cerro paranal observatory ( chile ) .
we used the standard settings in the two arms of the spectrograph , covering the region from 3100 to 10360 . the instrumental setup and
the technical aspects of the observations were the same in all cases and are described in detail in g04 and e04 .
the results included in this letter correspond to a single slit position in each nebula corresponding to the brightest part of the object .
the atmospheric dispersor corrector was used during the observations in order to avoid the effects of differential atmospheric refraction along the slit .
the total exposure time was very different for each case depending on the surface brightness of the nebulae , ranging from 300 to 600 seconds for the orion nebula and from 2700 to 9612 seconds for s 311 .
in all the cases , the total exposure time was achieved adding three or more individual spectra in each spectral zone .
the slit width was set to 3.0 `` and the slit length was set to 10 '' in the blue arm and to 12 `` in the red arm in the cases of the orion nebula and ngc 3576 and 3.0''@xmath710 " in both arms for the rest of the objects .
the effective spectral resolution at a given wavelength is @xmath8 .
the spectra were reduced using the iraf echelle reduction package , following the standard procedure .
the standard star eg 274 was observed for flux calibration in the cases of the orion nebula and ngc 3576 ( both observed in 2002 ) .
the rest of the objects were observed in 2003 and flux calibrated with the standard stars : hd 49798 , eg 274 , and cd @xmath332 9927 .
line intensities were measured as described in g04 and e04 with the splot routine of the iraf package .
table [ lines ] presents the extinction - corrected intensity of the emission lines relevant to the results concerning this letter , except for orion nebula and ngc 3576 , for which complete line lists and abundance data have been published in e04 and g04 , respectively .
the complete line lists of the rest of the objects will be published elsewhere .
the reddening coefficient , @xmath9 , was determined by fitting iteratively the observed and theoretical balmer decrement following the predictions by @xcite and is included in table [ abund ] .
we have used the reddening law of @xcite for all the objects listed in table [ lines ] except in the cases of m8 and m 16 for which we have used the reddening law of @xcite following the prescriptions of @xcite and @xcite .
the errors associated with the line intensities have been estimated following g04 .
in table [ abund ] , we present the adopted electron density , @xmath10 , and temperature , @xmath11 , for each object
. the adopted @xmath10 corresponds to the mean of the values obtained from [ ] , [ ] , [ ] , and [ ] and [ ] in several of the objects , weighted by their corresponding uncertainties . in the case of
@xmath11 we have adopted two values , one corresponding to the high ionization potential ions , @xmath12 , which is calculated as the weighted mean of the temperature obtained from [ ] , [ ] , and [ ] ( only temperatures from [ ] were available for m 16 and m 20 ) , and other to the low ionization potential ones , @xmath13 , which corresponds to the weighted mean of the values obtained for [ ] and [ ] . in any case ,
intensity ratios of rls are almost independent on @xmath10 and @xmath11 and they are almost unaffected by the values assumed for these parameters .
the temperature and density determinations were carried out with the iraf task temden of the package nebular .
more details on the determination of the physical conditions can be found in g04 and e04 .
the adopted values for @xmath10 , @xmath12 , and @xmath13 are included in table [ abund ] .
we have adopted @xmath12 for the derivation of c@xmath0/h@xmath1 and o@xmath0/h@xmath1 ratios and @xmath13 for o@xmath1/h@xmath1 .
we have used the effective recombination coefficients of @xcite for , of @xcite for , and of @xcite for .
the average of the abundances obtained with the effective recombination coefficients of @xcite and @xcite has been used to obtain the o@xmath1/h@xmath1 ratio .
we have used the intensity of the brightest line ( 4267 , see figure [ cii ] ) to derive the c@xmath0/h@xmath1 ratio .
other much fainter lines of @xmath14 and @xmath15 transitions of have been observed in all nebulae except ngc 3603 and s311 .
these lines give abundances in complete agreement with those obtained with 4267 .
the only rls of observed in our spectra are those of multiplet 1 .
however , there are several bright sky lines in that spectral region and only 7771.94 can be measured .
unfortunately , a satisfactory deblending of the line from sky lines was not possible in some of the objects due to the radial velocity of the nebulae : orion nebula , m 16 , m 17 , and s 311 .
for these four nebulae as well as for ngc 3603 ( for which lines were not detected ) we have determined the o@xmath1/h@xmath1 ratio from the intensity of [ ] lines using nebular and assuming an appropriate value of @xmath6 .
the @xmath6 adopted for each object is also included in table [ abund ] .
this parameter has been determined from the weighted mean of the different _
_ values obtained from the comparison of ionic abundances obtained from cels and rls of the same ions , the comparison of @xmath11 obtained from cels and from the balmer and paschen discontinuities , as well as a maximum likelihood method applied to the intensity of the brightest lines .
details on the derivation of @xmath6 can be found in g04 and e04 . in the case of the o@xmath0 abundances , the values quoted in table [ abund ]
are the weighted means of the o@xmath0/h@xmath1 ratios obtained from the sum of the intensities of the individual lines of the less case - dependent multiplets observed in each nebula following the prescriptions given in g04 and e04 .
we have used multiplets number 1 , 2 , 10 and the @xmath16 transitions in the cases of orion nebula ( e04 ) and ngc 3576 ( g04 ) , multiplets 1 , 2 , and 10 in the case of m 8 and m 17 , and multiplet 1 in m 16 , m 20 , ngc 3603 , and s 311 .
the lines of multiplet 1 are much brighter than those of the other multiplets and dominate the final mean value , therefore , we have only included the line intensities of multiplet 1 in table [ lines ] , the rest of the line intensities of m 8 and m 17 will be published elsewhere .
once we have the ionic abundances , we can derive the total abundances of the elements . in the case of o
, we have information of all the relevant ionic species that are expected to be present in the ionized nebulae : o@xmath1 and o@xmath0 . the total o abundance will be simply the sum of the o@xmath1/h@xmath1 and o@xmath0/h@xmath1 ratios .
in the case of c we have only information about the c@xmath0/h@xmath1 ratio . since we do not expect a significant presence of c@xmath17 in the sample ( see garnett et al .
1999 ) , the only important contribution is of c@xmath1 , which does not show emission lines in the spectral range observed . to obtain the final c / h abundance we have used the ionization correction factor ( icf ) for the presence of c@xmath1 derived by @xcite from photoionization models ( see their figure 2 ) .
this icf indicates that the c@xmath1/c@xmath0 ratio goes from [email protected] in the case of ngc 3603 to [email protected] in the case of m20 , the two nebulae with the extreme values of excitation .
figure [ grad ] shows the c / h , o / h , and c / o ratios derived for the eight nebulae of the sample .
the galactocentric distances of the objects have been taken from the stellar photometric distances quoted in the survey of galactic star - forming complexes of @xcite and the work on distances to open clusters of @xcite .
the sun is assumed to be at a galactocentric distance of 8 kpc .
simple least - squares linear fits to the data given in figure [ grad ] give the following gradients : these gradients and the data used for their derivation correspond to the composition of the ionized gas - phase of the interstellar medium ( ism ) .
@xcite estimated that c and o abundances of the orion nebula should be increased about 0.1 dex and 0.08 dex , respectively , to obtain the current gas+dust composition of the local ism .
we do not know if these depletion factors for the orion nebula can be applied to all the sample objects or if they depend on the particular ionization conditions of the nebulae . in any case
, the relatively small dispersion of the c and o abundances for a given distance suggests that the possible differences in the depletion factors should not be large .
therefore , in the absence of further information , we recommend to apply the depletion factors estimated by @xcite to all the objects prior to compare our abundances with the gas - phase ones predicted by chemical evolution models .
the value of the o gradient we derive is shallower than those obtained by @xcite ( @xmath30.080 dex kpc@xmath5 ) and @xcite ( @xmath30.070 dex kpc@xmath5 ) but similar to the value of @xmath30.040 dex kpc@xmath5 obtained by @xcite and that estimated by @xcite ( @xmath30.051 dex kpc@xmath5 ) .
all these determinations of the literature are based on the standard analysis based on the intensity of cels , therefore , their result depend very much on the exact value of the @xmath11 assumed for each object . in our case ,
the o gradient we obtain from cels and assuming @xmath6=0 is found of the form @xmath2log(o / h ) = @[email protected] dex kpc@xmath5 , which is very similar to the gradient we derive from rls .
the ne , ar and and s gradients -which are expected to follow the behavior of o- determined by martn - hernndez , van der hulst , & tielens ( 2003 , and references therein ) from iso data are also consistent with shallower values of the slope .
our c gradient is clearly much steeper than that we obtain for o. its slope is also steeper than the one obtained by @xcite for b stars ( @xmath30.070 dex kpc@xmath5 ) .
the c / o gradient is an important constraint for chemical evolution models and the radial star formation history of the galactic disk because the nucleosynthetic origin of both elements is expected to be quite different .
we find that the c / o ratio increases with metallicity .
our c / o gradient is similar to that obtained by @xcite for two nearby spiral galaxies : m 101 and ngc 2403 ( @xmath30.04 and @xmath30.05 dex kpc@xmath5 , respectively ) , from the analysis of cels in the uv domain and is also similar to that obtained by @xcite for galactic b stars ( @xmath30.050 dex kpc@xmath5 ) .
models of galactic chemical evolution to explain the observed gradients are presented by carigi et al .
c and o are the most important `` biogenic elements '' and their gradients are paramount , in the study of the galactic habitable zone ( e.g. lineweaver , fenner & gibson 2004 and references therein ) as well as to determine the co to h@xmath19 conversion factor ( e.g. strong et al . 2004 and references therein ) .
we are grateful to the referee , dick henry , for his careful review of this letter .
ce and jg would like to thank the members of the instituto de astronoma , unam , for their always warm hospitality .
this work has been partially funded by the spanish ministerio de ciencia y tecnologa ( mcyt ) under project aya2001 - 0436 .
mp received partial support from dgapa unam ( grant in114601 ) .
mtr received partial support from fondap(15010003 ) , a guggenheim fellowship and fondecyt(1010404 ) .
mr acknowledges support from mexican conacyt project j37680-e .
lc s work is supported by conacyt grant 36904-e .
c@c@cccccc 3726.03 & [ o ii ] & 155@xmath46 & 151@xmath45 & 46@xmath42 & 147@xmath46 & 40@xmath42 & 165@xmath47 + 3728.82 & [ o ii ] & 122@xmath45 & 106@xmath44 & 46@xmath42 & 173@xmath47 & 25@xmath41 & 194@xmath48 + 4267.15 & c ii & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 4363.21 & [ o iii ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 4638.86 & o ii & [email protected] & [email protected] & [email protected] & ... & 0.057 : & [email protected] + 4641.81 & o ii & [email protected] & [email protected] & [email protected] & 0.030 : & [email protected] & [email protected] + 4649.13 & o ii & [email protected] & [email protected] & [email protected] & 0.036 : & [email protected] & [email protected] + 4650.84 & o ii & [email protected] & [email protected] & [email protected] & 0.017 : & [email protected] & [email protected] + 4661.63 & o ii & [email protected] & [email protected] & [email protected] & 0.018 : & [email protected] & [email protected] + 4673.73 & o ii & ... & ... & 0.022 : & ... & ... & ...
+ 4676.24 & o ii & ... & [email protected] & [email protected] & ... & 0.040 : & ... + 4861.33 & h i & 100@xmath43 & 100@xmath43 & 100@xmath43 & 100@xmath43 & 100@xmath43 & 100@xmath43 + 4958.91 & [ o iii ] & [email protected] & 33@xmath41 & 114@xmath43 & [email protected] & 180@xmath45 & 43@xmath41 + 5006.84 & [ o iii ] & 82@xmath43 & 96@xmath43 & 335@xmath410 & 59@xmath42 & 533@xmath416 & 126@xmath44 + 7771.94 & o i & 0.013:@xmath20 & [email protected] & [email protected]@xmath20 & [email protected] & ... & ... + l@cccccccc r@xmath21 ( kpc ) & 6.34@xmath20 & 6.41@xmath22 & 6.75@xmath20 & 7.19@xmath20 & 7.46@xmath22 & 8.40@xmath20 & 8.65@xmath20 & 10.43@xmath22 + c(h@xmath23 ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + n@xmath24 ( @xmath25 ) & 1120@xmath4220 & 1600@xmath4200 & 470@xmath4120 & 270@xmath460 & 2800@xmath4400 & 8900@xmath4200 & 5150@xmath4750 & 310@xmath480 + t@xmath26 ( k ) & 7650@xmath4250 & 8040@xmath4130 & 8050@xmath4150 & 7800@xmath4300 & 8500@xmath4150 & 8320@xmath440 & 9050@xmath4200 & 9050@xmath4200 + t@xmath27 ( k ) & 8350@xmath4200 & 8450@xmath4150 & 8870@xmath4300 & 8400@xmath4200 & 8500@xmath4150 & 10000@xmath4400 & 11400@xmath4700 & 9500@xmath4250 + @xmath6 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + c@xmath0/h@xmath1 @xmath7 10@xmath28 & 25@xmath42 & 20@xmath41 & 53@xmath44 & 15@xmath42 & 28@xmath44 & 22@xmath41 & 30@xmath45 & 10@xmath41 + 12+log(c / h ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + o@xmath29/h@xmath1 @xmath7 10@xmath28 & 44@xmath410@xmath30 & 34@xmath45 & 10@xmath42@xmath30 & 42@xmath47 & 13@xmath43 & 14@xmath44@xmath30 & 4@xmath41@xmath30 & 25@xmath45@xmath30 + o@xmath0/h@xmath1 @xmath7 10@xmath28 & 20@xmath42 & 17@xmath41 & 48@xmath42 & 10@xmath45 & 42@xmath45 & 37@xmath41 & 49@xmath46 & 11@xmath41 + 12+log(o / h ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + | we present results of deep echelle spectrophotometry of eight galactic regions located at galactocentric distances between 6.3 and 10.4 kpc .
the data have been taken with the very large telescope ( vlt ) ultraviolet echelle spectrograph ( uves ) in the 3100 to 10360 range .
we have derived c@xmath0 and o@xmath0 abundances from recombination lines for all the objects , as well as o@xmath1 abundances from this kind of lines for three of the nebulae .
the intensity of recombination lines is almost independent on the assumed electron temperature as well as on the possible presence of spatial temperature variations or fluctuations inside the nebulae .
these data allow the determination of the gas - phase c and o abundance gradients of the galactic disk , of paramount importance for chemical evolution models .
this is the first time the c gradient is derived from a so large number of regions and for a so wide range of galactocentric distances .
abundance gradients are found of the form @xmath2log(o / h ) = @[email protected] dex kpc@xmath5 , @xmath2log(c / h ) = @[email protected] dex kpc@xmath5 , and @xmath2log(c / o ) = @[email protected] dex kpc@xmath5 . |
rydberg - rydberg interactions are very interesting for use in mesoscopic quantum manipulations .
the extremely strong interactions between two rydberg atoms have been proposed to entangle large numbers of atoms in a non - trivial manner using the phenomenon of blockade @xcite . when a mesoscopic sample is illuminated with narrowband lasers tuned to a rydberg state
, only one atom at a time can be excited if the rydberg - rydberg interaction exceeds the linewidth of the laser - atom coupling .
therefore the mesoscopic cloud of atoms behaves as an effective 2-level system , with the upper level being a single collective excitation . in order to attain the strongest possible rydberg blockade , it is desirable to operate under conditions where the rydberg - rydberg interaction is not the usual @xmath2 or @xmath3 van - der - waals interactions , but instead is resonantly enhanced by `` fster @xmath4 '' processes @xcite such as @xmath5 that lead to isotropic @xmath0 long - range behavior when the @xmath6 states are degenerate with the @xmath7 states .
dramatic enhancements of collisional interactions due to such resonant couplings have been demonstrated for rydberg excitation in a magneto - optical trap@xcite .
the quantum nature of these types of interactions was recently used to resolve molecules 12 nm apart in an organic solid @xcite . due to the high density of rydberg states , there are typically many candidate levels for such fster processes .
an important consideration for quantum information applications of rydberg blockade is the size of cloud required .
the spatial extent of the cloud must be small enough for excitation of a single atom anywhere in the cloud to block the excitation of every other atom in the cloud . even with the great strength of rydberg - rydberg interactions , the required size of the clouds is likely to be less than 10 microns for fast ( 1 mhz ) operations .
only recently have atom clouds of such small size been produced in the laboratory@xcite . even in the case that sub - micron mesoscopic samples are realized experimentally , there are other applications of such samples that benefit from the sample being as large as possible .
for example , we have recently proposed single - photon sources , single - photon detectors , and quantum state transmission @xcite using mesoscopic rydberg blockade . in these applications
one would like the cloud to be many wavelengths in size in order that the diffraction limited light fields not occupy too large a solid angle .
for this to be effective requires the blockade to operate over clouds of several microns in extent .
the purpose of this paper is to examine some issues that arise in the application of fster processes to rydberg blockade .
in particular , when the quantum numbers for the states are of the form @xmath8 we show that one or more of the @xmath9 states have @xmath10 .
these `` fster zero '' states then only exhibit the usual van - der - waals long - range behavior and will therefore set limits on the attainable cloud sizes for quantum manipulations using rydberg blockade . only when @xmath11 are there no fster zero states .
recent experiments @xcite have observed strong suppression of excitation in very large clouds that is strong evidence for blockade but do not address whether or not the blockade is complete as required for quantum information applications . in the following ,
we first ( section [ sec : background ] ) present a more detailed background discussion of the importance of strong isotropic rydberg - rydberg interactions for quantum manipulations using rydberg blockade .
we then illustrate in section [ sec : foster ] how the fster process accomplishes this .
section [ sec : zeros ] presents the main result of this paper : for many possible fster processes there exists a linear combination of atomic sublevels that have @xmath10 .
this result is extended to the important case of fine - structure interactions in section [ sec : finestructure ] .
we conclude with more implications of this work .
the basic physics behind the dipole - blockade concept is illustrated in figure [ entangle ] .
suppose we have @xmath12 atoms in an atomic ground state @xmath13 .
we envision exciting the atoms to a rydberg state @xmath14 of energy @xmath15 using a narrowband laser . as shown in the figure , excitation of one of the two atoms to the rydberg state
is allowed while excitation of the second atom is off - resonant and therefore suppressed .
addition of more atoms changes the effectiveness of this `` blockade '' but does not change the basic physics of the situation .
excitation of more than one atom is energetically suppressed as long as the interaction between each pair of atoms in the ensemble exceeds the bandwidth of the laser - atom excitation . neglecting spontaneous emission or other decohering effects
, when subject to dipole blockade and a continuous light field the atoms will undergo coherent rabi oscillations between states @xmath16 and @xmath17 . .
] to see how this process can be used to generate interesting entanglement , we simply imagine driving the atom pair in the @xmath18 state with a @xmath19 pulse .
the atoms will then be in a coherent superposition of states @xmath13 and @xmath14 @xmath20 that can not be written as a product of two individual wavefunctions and is therefore entangled . to avoid the inevitable decoherence of the unstable rydberg state , the entanglement can be usefully transferred to a second ground state @xmath21 with a @xmath19-pulse from a second laser tuned to the @xmath22 transition .
again , this concept can be extended to a collection of @xmath23 atoms without loss of generality .
the final wavefunction is a symmetric entangled superposition of @xmath24 atoms in state @xmath13 and one atom in state @xmath21 .
there are several intriguing characteristics of this process .
first , the entanglement is generated between the internal states of the atom ; any motion of the atom is unimportant .
this means that it is not necessary for the atoms to be localized in the ground state of a confining potential , so the method does not require coherence of the external degrees of freedom of the atoms .
there is also no motional constraint on the speed at which the process can occur .
second , the value of the rydberg - rydberg interaction @xmath25 is not important to first order , as long as it is much larger than the bandwidth of the light - atom interaction .
this implies that the atoms can be at random distances @xmath26 from each other , as long as they are not too far apart . finally , since the blockade mechanism suppresses excitation of multiple rydberg atoms , the atoms never actually experience the strong rydberg - rydberg forces , avoiding heating .
key to the entanglement process is the requirement that the rydberg - rydberg frequency shift be large compared to the bandwidth of the light - atom interaction .
if this is violated , the fidelity of the entanglement operations will be compromised . in a mesoscopic sample ,
insufficient blockade in any possible excitation channel is sufficient to cause copious production of excited atoms .
we now examine the rydberg - rydberg interactions to see under what conditions this will be problematic .
we wish to consider the interactions between two like rydberg atoms that lead to blockade .
normally , in the absence of an external electric field these interactions are of the van - der - waals forms @xmath27 or @xmath28 .
since an electric field mixes states of opposite parity together , a rydberg atom can have a permanent dipole moment , leading to a much stronger and longer range @xmath29 interaction .
the interaction between two such atoms a and b is the familiar @xmath30 where @xmath31 is the angle between the interatomic separation @xmath32 and the electric field @xmath33 .
the fact that @xmath34 makes this interaction huge for two rydberg states .
however , it has the undesirable feature for dipole - blockade that it vanishes at @xmath35 , allowing for excitation of rydberg atom pairs located at this angle with respect to each other .
it is possible to have an isotropic rydberg atom - atom interaction of comparable strength to this if there is a degeneracy in the energy - level spectrum for a pair of atoms .
for example , inspection of figure [ rbelevels ] shows that the 50s state of rb is nearly symmetrically placed between the 49p and 50p states .
this means that the 50s+50s state of a rydberg atom pair is nearly degenerate with the 49p+50p state . neglecting the influence of other nearby states , the eigenstates of the two - atom system
are linear combinations of 50s+50s and 49p+50p , with energy shifts proportional to @xmath29 . using the methods described below , we find that the rydberg - rydberg potential energy curves are given by @xmath36 where @xmath37 mhz @xmath38m@xmath39 and the ss - pp energy defect is @xmath40 mhz .
estimates similar to this have been previously presented by protsenko _
et al._@xcite . partially compensates for the non - zero energy defect . ] a plot of @xmath41 is shown in figure [ pot50s ] . if there were perfect fster degeneracy ( @xmath42 ) the dipole - dipole interaction would exceed 10 mhz out to distances of 10 microns
this would be very promising for realizing dipole blockade in quite large clouds .
however , the energy defect for real rb atoms reduces the interaction substantially , becoming the usual van - der - waals form @xmath43 at large distances . here
we should note that we are restricting ourselves to mixing of states that are nearly resonant .
other states , such as 51p+48p in this example , have larger energy defects ( @xmath44 ghz for 51p+48p ) and often much smaller matrix elements as well . adding up many such states could ( and should ) be done in second order perturbation theory and
will alter the @xmath28 part of the potential somewhat but not change the overall conclusions . for the rest of this paper we will continue to neglect these other states .
we will also neglect the @xmath27 quadrupole - quadrupole interactions .
the strength of the rydberg - rydberg interactions varies strongly with principal quantum number .
the radial matrix elements scale as @xmath45 , so @xmath46 scales as @xmath47 .
since the energy defect @xmath48 scales as @xmath49 , the van - der - waals interactions scale as @xmath50 .
thus the range of the rydberg blockade increases as roughly @xmath51 , so that @xmath52 has nearly twice the range as @xmath53 , as illustrated in figure [ pot50s ] .
increasing @xmath54 comes with increased sensitivity to stray electric fields , black - body radiation , and so forth . in practice , states up to @xmath55 should be stable enough to perform fast quantum operations@xcite .
however , it would be preferable to find states with smaller energy defects and therefore larger range at smaller @xmath54 . given the lack of perfect fster resonances for the s - state , one should consider the possibility of fster processes in other angular momentum states where the energy defects may be smaller than they are for the @xmath56 states .
interesting new properties of the fster process arise when we consider higher angular momentum states . for example , consider again the s and p states of rb , but this time with laser excitation of the p - states , so that the relevant fster degeneracy is , say , 50p+50p@xmath5750s+51s with an energy defect @xmath58930 mhz .
suppose also that the atoms are subject to linearly polarized light polarized along the laboratory @xmath59-axis .
we shall show that there is no fster blockade for this situation . for simplicity ,
let us assume that two atoms are aligned with @xmath60 . in this case
, consider the wavefunction @xmath61 ( @xmath62 ) whose matrix element of @xmath63 is zero with the @xmath64 states : @xmath65 and so the only long - range interaction will be a van - der - waals interaction with the comparatively far off - resonant @xmath66 and @xmath67 states
. note that @xmath68 is strongly coupled to the light through its @xmath69 part . with strong light coupling and weak dipole - dipole coupling ,
we conclude that the rb p - states will not experience long - range dipole blockade .
these conclusions are not changed when @xmath70 is rotated away from @xmath59 .
if one takes the quantization axis along @xmath70 then @xmath68 stays of the same form , but each of the three parts of @xmath68 will contribute to the light - atom interaction .
for the rest of the paper we shall take the quantization axis for the atomic basis states to be along @xmath70 .
another very interesting possibility from figure [ rbelevels ] is the nearly degenerate combination 48d+48d@xmath5750p+46f with an energy defect of only @xmath71 mhz ( neglecting fine structure for now ) .
this has the potential for much stronger fster interactions at large distance as compared to the @xmath64 states . in this case
there is also a wavefunction with zero coupling via @xmath63 to the p+f manifold : @xmath72 where interchange - symmetric kets are defined in terms of the quantum numbers @xmath73 of the individual atoms as @xmath74 , and @xmath75 .
we shall label such states as fster zero states . whereas the fster zero state of equation [ zeroppss ]
can be deduced in a straightforward way essentially by inspection once the matrix elements of s+s with the three p+p combinations are calculated , the d+d fster zero state of equation [ zeroddpf ] is more subtle since its matrix elements with each of the three @xmath76 states must vanish .
thus we now discuss in more generality the conditions for fster zero states to exist for the process @xmath77 we will assume without loss of generality that @xmath78 .
the fster zero state , if it exists , is written as a linear combination @xmath79 , so the condition @xmath80 gives @xmath81 which is effectively a generalization of equation [ zeroppss2 ] to the case where there are multiple possible final states .
there are three cases of interest . for @xmath82 and @xmath83 ( of
which d+d@xmath57p+f is an example ) there are @xmath84 equations in the @xmath85 unknowns @xmath86 .
but the reflection symmetry @xmath87 means that @xmath88 of the equations are the same , leaving @xmath89 equations in @xmath85 unknowns and therefore a solution exists . for the case
@xmath90 the same argument holds .
the final case , with @xmath91 , has equation [ eqzeros ] with @xmath92 equations in @xmath85 unknowns and therefore no fster zero state .
these results can also be understood from the point of view of the molecular symmetries of the problem . following the analysis of marinescu@xcite , the fster zero states have the molecular symmetries
@xmath93 and @xmath94 depending on their triplet or singlet spin character . for @xmath95 , the orbital exchange - symmetric kets as defined above are not eigenstates of the operator @xmath96 that reflects the wavefunction about a plane containing the interatomic axis .
taking into account this symmetry , we define kets @xmath97 which have molecular parity ( @xmath98 ) if @xmath99 ) and reflection symmetry eigenvalues @xmath100 , giving rise to states of traditional molecular designation @xmath101 . considering triplet states , and assuming @xmath102 , there are @xmath103 @xmath104 @xmath105 states , @xmath103 @xmath106 @xmath107 states , @xmath108 @xmath109 @xmath110 states , and @xmath108 @xmath111 @xmath112 states .
thus for the @xmath113 problem , there are three @xmath93 states with @xmath66 character but only two of @xmath114 character .
therefore it is always possible to find a linear combination of the three @xmath93 @xmath66 states that has zero coupling to the two @xmath114 @xmath93 states .
the reasoning is identical for the @xmath94 symmetry . to summarize ,
the isotropic @xmath0 dipole - dipole interaction generated by the fster process @xmath115 will have states with @xmath10 unless @xmath91 .
the above analysis has emphasized the @xmath93 and @xmath94 states that are degenerate in the absence of overlap .
similar reasoning shows that for the case @xmath116 the @xmath117 and @xmath118 states have no fster zeros , nor do states of @xmath119 . in general ,
all of these states are coupled to the light field , but the @xmath93 and @xmath94 states destroy the complete blockade .
in general , the fine stucture interaction can not be neglected . at @xmath53
the rb p - state fine structure splitting is about 800 mhz and the d splitting is 100 mhz , so that at micron - scale distances there will be strong fine - structure mixing by the dipole - dipole interaction . at long enough range , where the dipole - dipole interaction is smaller than the fine - structure splitting
, we can use the same type of arguments as above to analyze the problem .
let us consider the fster process @xmath120 we are mostly interested in states with total @xmath121 .
as before , the @xmath122 states are symmetric linear combinations @xmath123 . for alkali atoms with half - integer @xmath124
there are @xmath125 such states . on the other hand ,
there are ( assuming @xmath126 ) @xmath127 @xmath128 states .
half of these are removed from consideration due to the @xmath129 symmetry .
thus the system of equations for the fster zero amplitudes @xmath130 has @xmath125 equations in @xmath131 unknowns .
a normalizable solution will exist only for @xmath132 .
it follows that potential fster processes such as @xmath133 will have zeros .
the inclusion of fine structure brings new possibilities , however , since states of different @xmath134 can have the same value of @xmath124 .
for example , @xmath135 has an energy defect of only -15 mhz at @xmath136 and should not have any fster zeros .
the hamiltonian matrix has the structure @xmath137 where the interaction submatrix @xmath138 and @xmath139 mhz @xmath38m@xmath39 . for @xmath42 , the eigenvalues of @xmath140 corresponding to the @xmath66 states are @xmath141 .
two of the eigenvalues are quite small ( @xmath142 ) , leading to poor rydberg blockade .
the potential curves for @xmath143 mhz are shown in figure [ fig : finestructure ] . the nearly flat potential shows that while there are no fster zeros for this case , the rydberg blockade is still strongly suppressed at large distances even though the energy defect is very small . even in the presence of fine structure ,
the blockade is still poor if @xmath144 .
potential curves near the @xmath145 asymptote of rb@xmath146 .
the nearly flat curve coming from the @xmath145 asymptote implies suppressed dipole blockade . ]
the primary result of this paper is that the very long - range @xmath0 interactions produced by resonances @xmath147 have states with @xmath10 unless @xmath148 .
this strongly reduces the number of possibilities for attaining high fidelity dipole - blockade in mesoscopic atom samples .
one solution is to rely instead on quadrupole - quadrupole ( @xmath27 ) or second - order dipole - dipole ( 1/@xmath149 ) interactions to acheive blockade at high values of @xmath54 . as pointed out recently by the connecticut group @xcite the interactions under these conditions
can be quite strong .
another possibility for attaining strong dipole blockade is to tune @xmath148 resonances with an electric field .
for example , the great sensitivity of @xmath150 states to electric fields can tune @xmath151 into resonance at modest fields . at resonance ,
the long range potentials for this case are @xmath152 and thus should lead to quite strong rydberg blockade at long range .
the author gratefully recognizes very helpful conversations with other members of the wisconsin amo physics group , and with c. goebel and r. cote .
this work was supported by the national science foundation , nasa , and the army research office .
_ calculation of dipole - dipole matrix elements _
let @xmath134 , @xmath108 , and @xmath153 be the angular momenta involved in the fster process @xmath154 since the dipole - dipole interaction @xmath155 expressed in a coordinate system with @xmath156 aligned with @xmath32 , is symmetric on atom interchange ( @xmath157 ) , the interchange - symmetric state @xmath158mixes only with the symmetric combinations @xmath159 the matrix element of @xmath63 is therefore @xmath160 we use the wigner - eckart theorem to write this explicitly as @xmath161 where @xmath162 and the reduced matrix elements are given in terms of radial integrals of @xmath14 as @xmath163 for the calculation of matrix elements in this paper , we have used the l - dependent core potentials of marinescu _ et .
al _ @xcite and obtained the radial wavefunctions by numerov integration of the schrdinger equation .
energy levels were calculated using the recent quantum defect determinations of li _
et al._@xcite . _ reflection symmetry _
the dipole - dipole matrix element @xmath164is proportional to @xmath165 where the clebsch - gordan symmetry @xmath166 has been used along with the specific parities @xmath167 . inspection of equation [ cgsum ] then shows that @xmath168
10 m. d. lukin , m. fleischhauer , r. cote , l. m. duan , d. jaksch , j. i. cirac , and p. zoller , `` dipole blockade and quantum information processing in mesoscopic atomic ensembles '' , phys .
lett . * 87 * , 037901 ( 2001 ) .
s. peil , j. v. porto , b. l. tolra , j. m. obrecht , b. e. king , m. subbotin , s. l. rolston , and w. d. phillips , `` patterned loading of a bose - einstein condensate into an optical lattice '' , phys .
a * 67 * , 051603(r ) ( 2003 ) .
k. singer , m. reetz - lamour , t. amthor , l. g. marcassa , and m. weidemuller , `` spectral broadening and suppression of excitation induced by ultralong - range interactions in a cold gas of rydberg atoms '' , physics/0402113 ( 2004 ) .
w. li , i. mourachko , m. w. noel , and t. f. gallagher , `` millimeter - wave spectroscopy of cold rb rydberg atoms in a magneto - optical trap : quantum defects of the ns , np , and nd series '' , phys .
a * 67 * , 052502 ( 2003 ) . | rydberg states of atoms are of great current interest for quantum manipulation of mesoscopic samples of atoms .
long - range rydberg - rydberg interactions can inhibit multiple excitations of atoms under the appropriate conditions .
these interactions are strongest when resonant collisional processes give rise to long - range @xmath0 interactions .
we show in this paper that even under resonant conditions @xmath1 often vanishes so that care is required to realize full dipole blockade in micron - sized atom samples . |
how can one best bound the complexity of an algebraic set in terms of the complexity of its defining polynomials ? over the complex numbers ( or any algebraically closed field ) , bzout s theorem @xcite bounds the number of roots , for a system of multivariate polynomials , in terms of the degrees of the polynomials . over finite fields ,
weil s famous mid-20@xmath11 century result @xcite bounds the number of points on a curve in terms of the genus of the curve ( which can also be bounded in terms of degree ) .
these bounds are optimal for dense polynomials
. for sparse polynomials , over fields that are not algebraically closed , these bounds can be much larger than necessary .
for example , descartes rule @xcite tells us that a univariate real polynomial with exactly @xmath2 monomial terms always has less than @xmath12 real roots , even though the terms may have arbitrarily large degree .
is there an analogue of descartes rule over finite fields ? despite the wealth of beautiful and deep 20@xmath11-century results on point - counting for curves and higher - dimensional varieties over finite fields , varieties defined by sparse _ univariate _
polynomials were all but ignored until @xcite ( see lemma 7 there , in particular ) . aside from their own intrinsic interest , refined univariate root counts over finite fields are useful in applications such as cryptography ( see , e.g. , @xcite ) , the efficient generation of pseudo - random sequences ( see , e.g. , @xcite ) , and refined estimates for certain exponential sums over finite fields ( * ? ? ?
* proof of theorem 4 ) .
for instance , estimates on the number of roots of univariate tetranomials over a finite field were a key step in establishing the uniformity of the _ diffie - helman distribution _
* proof of thm . 8 , sec .
4 ) a quantitative statement important in justifying the security of cryptosystems based on the discrete logarithm problem .
we are thus interested in the number of roots of sparse univariate polynomials over finite fields .
the polynomial @xmath13 having two terms and exactly @xmath0 roots in @xmath14 might suggest that there is no finite field analogue of descartes rule .
however , the roots of @xmath13 consist of @xmath15 and the roots of @xmath16 , and the latter roots form the unit group @xmath17 . for an arbitrary binomial @xmath18
$ ] with @xmath19 and @xmath20 and @xmath21 nonzero , the roots consist of @xmath15 ( if @xmath22 ) and the roots of @xmath23 .
note that the number of roots of @xmath23 in @xmath14 is either @xmath15 or @xmath24 .
in the latter case , the roots form a coset of a subgroup of @xmath5 . for polynomials with three or more terms ,
the number of roots quickly becomes mysterious and difficult , and , as we shall demonstrate in this paper , may exhibit very different behaviors in the two extreme cases where ( a ) @xmath0 is a large power of a prime , and ( b ) @xmath0 is a large prime .
to fix notation , we call a polynomial @xmath25 $ ] with @xmath26 and @xmath27 for all @xmath28 a _ ( univariate ) @xmath2-nomial_. the best current upper bounds on the number of roots of @xmath6 in @xmath14 , as a function of @xmath0 , @xmath2 , and the coset structure of the roots of @xmath6 , can be summarized as follows , using @xmath29 for set cardinality : [ thm : z ] let @xmath1 $ ] be any univariate @xmath2-nomial with degree @xmath3 and exponent set @xmath30 containing @xmath15
. set @xmath31 , @xmath32 , @xmath33 , and let @xmath34 denote the maximum cardinality of any coset ( of any subgroup of @xmath35 contained in @xmath36 . then : + 0 .
( special case of ( * ? ? ?
* thm . 1 ) ) @xmath37 .
1.1 ) @xmath36 is a union of no more than @xmath38 cosets , each associated to one of two subgroups @xmath39 of @xmath5 , where @xmath40 , @xmath41 , and @xmath42 can be determined within @xmath43 bit operations .
1.2 ) for @xmath44 we have @xmath45 and , if we have in addition that @xmath0 is a square and @xmath46 , then @xmath47 .
( see ( * ? ? ?
2.2 & 2.3 ) ) for any @xmath48 we have @xmath49 .
+ [ 1]furthermore , @xmath50 .
@xmath51 for any fixed @xmath48 , _ dirichlet s theorem _ ( see ,
e.g. , ( * ? ? ?
* thm . 8.4.1 , pg .
215 ) ) implies that there are infinitely many prime @xmath0 with @xmath52 . for such pairs
@xmath53 the bound from assertion ( 0 ) is tight : the roots of @xmath54 are the disjoint union of @xmath55 cosets of size @xmath56 .
( there are no @xmath57-cosets for this @xmath2-nomial . )
however , assertions ( 1 ) and ( 3 ) tell us that we can get even sharper bounds by making use of the structure of the cosets inside @xmath36 .
for instance , when @xmath44 and @xmath46 , assertion ( 2 ) yields the upper bound @xmath58 , which is smaller than @xmath59 for @xmath60 . while assertion ( 3 )
might sometimes not improve on the upper bound @xmath61 , it is often the case that @xmath34 is provably small enough for a significant improvement .
for instance , when @xmath62 and @xmath63 for all @xmath64 , we have @xmath65 and then @xmath66 .
[ thm : opt ] let @xmath67 with @xmath48 and @xmath68 prime . if @xmath69 then the polynomial @xmath70 has @xmath71 and exactly @xmath72 roots in @xmath14 .
furthermore , if @xmath73 , then the polynomial @xmath74 has @xmath75 , @xmath76 , and exactly @xmath77 roots in @xmath14 .
theorem [ thm : opt ] is proved in section [ sec : trick ] below .
the polynomials @xmath78 show that the bounds from assertions ( 1 ) and ( 3 ) of theorem [ thm : z ] _ are within a factor of @xmath79 of being optimal _ , at least for @xmath80 and a positive density of prime powers @xmath0 .
note in particular that @xmath81 shows that assertion ( 2 ) of theorem [ thm : z ] is _ optimal _ for square @xmath0 and @xmath46 .
( see ( * ? ? ?
1.3 ) for a different set of extremal trinomials when @xmath0 is an odd square . )
the second family @xmath82 reveals similar behavior for a different family of prime powers .
optimally bounding the maximal number of roots ( or cosets of roots ) for the case of _ prime _
@xmath0 is already more subtle in the trinomial case : we are unaware of any earlier lower bound that is a strictly increasing function of @xmath0 .
( note , for instance , that @xmath83 vanishes on just cosets of roots , assuming @xmath84 mod @xmath85 . ) in section [ sec : harder ] , we shall prove the following theorem .
[ thm : grh ] for any @xmath86 and prime @xmath87 , consider @xmath88 as an element of @xmath89 $ ] .
then @xmath90 , and there exists an infinite family of primes @xmath68 such that the trinomial @xmath91 has exactly @xmath92 distinct roots in @xmath9 where this theorem is proved by using results from algebraic number theory on the splitting of primes in number fields .
in particular , we use a classic estimate of lagarias , montgomery , and odlyzko @xcite ( reviewed in section [ sec : harder ] below ) on the least prime ideal possessing a frobenius element exhibiting a particular galois conjugacy class .
the latter result is in turn heavily based on the effective chebotarev density theorem of lagarias and odlyzko @xcite .
cohen @xcite applied an effective function field analogue of the chebotarev density theorem to trinomials of the form @xmath95 , where @xmath21 varies in @xmath9 and @xmath68 goes to infinity .
his results tell us that when @xmath68 is large , there always exist @xmath96 so that @xmath95 splits completely over @xmath9 , and the least such @xmath68 satisfies @xmath94 unconditionally .
this gives the existence of @xmath97 ( and infinitely many primes @xmath68 ) with @xmath98 having many roots in @xmath9 , but gives no information on how to easily find such @xmath21 .
the novelty of theorem [ thm : grh ] is thus an even simpler family of trinomials with number of roots in @xmath9 increasing as @xmath68 runs over an infinite sequence of primes .
assuming grh , the rate of increase matches cohen s lower bound .
could the existence of trinomials @xmath6 with @xmath46 and , say , @xmath99 roots in @xmath9 ( as one might conjecture from theorem [ thm : opt ] ) be obstructed by @xmath68 not being a square ? we feel that @xmath68 being prime is such an obstruction and , based on experimental results below , we conjecture the following upper bound : ( see also ( * ? ? ?
1.5 & sec .
4 ) for other refined conjectures and heuristics in this direction . )
it is a simple exercise to show that , to compute the maximal number of roots in @xmath9 of trinomials with @xmath46 , one can restrict to the family of trinomials in the conjecture . for any @xmath107 ,
let @xmath108 be the least prime for which there exists a univariate trinomial @xmath109 with @xmath110 and exactly @xmath92 distinct roots in @xmath111 .
we did a computer search to find the values of @xmath108 for @xmath112 .
they are ... in particular , for each @xmath113 , a full search was done so that the trinomial @xmath109 below has the least degree among all trinomials over @xmath111 having exactly @xmath92 roots in @xmath111 .
( it happens to be the case that , for @xmath113 , we can also pick the middle degree monomial of @xmath109 to be @xmath114 . ) by rescaling the variable as necessary , we have forced @xmath115 to be among the roots of each of the trinomials below .
it is easily checked via the last part of assertion ( 3 ) of theorem [ thm : z ] that @xmath116 for each @xmath113 .
the least prime @xmath117 for which there is a trinomial @xmath118 with @xmath119 and exactly @xmath120 roots in @xmath121 is currently unknown ( as of july 2016 ) .
better and faster code should hopefully change this situation soon .
jingguo bi ; qi cheng ; and j. maurice rojas , _ `` sub - linear root detection , and new hardness results , for sparse polynomials over finite fields , '' _ proceedings of issac ( international symposium on symbolic and algebraic computation , june 2629 , boston , ma ) , pp .
6168 , acm press , 2013 . ran canetti ; john b. friedlander ; sergei konyagin ; michael larsen ; daniel lieman ; and igor e. shparlinski , _
`` on the statistical properties of diffie - hellman distributions , '' _
israel j. math .
120 ( 2000 ) , pp .
2346 .
alexander kelley , _ `` roots of sparse polynomials over a finite field , '' _ in proceedings of twelfth algorithmic number theory symposium ( ants - xii , university of kaiserslautern , august 29 september 2 , 2016 ) , to appear .
also available as math arxiv preprint 1602.00208 .
jeff lagarias and andrew odlyzko ,
_ `` effective versions of the chebotarev density theorem , '' _ algebraic number fields : @xmath123-functions and galois properties ( proc .
durham , durham , 1975 ) , 409464 , academic press , london , 1977 .
david eugene smith and marcia l. latham , _ the geometry of ren descartes , _ translated from the french and latin ( with a facsimile of descartes 1637 french edition ) , dover publications inc .
, new york ( 1954 ) . | suppose @xmath0 is a prime power and @xmath1 $ ] is a univariate polynomial with exactly @xmath2 monomial terms and degree @xmath3 . to establish a finite field analogue of descartes rule , bi , cheng , and rojas ( 2013 ) proved an upper bound of @xmath4 on the number of cosets in @xmath5 needed to cover the roots of @xmath6 in @xmath5 . here , we give explicit @xmath6 with root structure approaching this bound : for @xmath0 a @xmath7-st power of a prime we give an explicit @xmath2-nomial vanishing on @xmath8 distinct cosets of @xmath5 .
over prime fields @xmath9 , computational data we provide suggests that it is harder to construct explicit sparse polynomials with many roots .
nevertheless , assuming the generalized riemann hypothesis , we find explicit trinomials having @xmath10 distinct roots in @xmath9 . |
quantum transport in low - dimensional systems has been a topic of interest within the past few decades due to its potential applicability in the field of nanoscience and nanotechnology .
exploitation of the spin degree of freedom adds a possibility of integrating memory and logic into a single device , leading to remarkable development in the fields on magnetic data storage application , device processing technique , quantum computation @xcite , etc .
naturally a lot of attention has been paid to study spin transport in low - dimensional systems both from experimental @xcite and theoretical @xcite points of view .
the understanding of electronic localization in low - dimensional model quantum systems is always an interesting issue .
whereas , it is a well established fact that in an infinite one - dimensional ( @xmath0d ) system with random site potentials all energy eigenstates are exponentially localized irrespective of the strength of randomness due to anderson localization @xcite , there exists another kind of localization , known as wannier - stark localization , which results from a static bias applied to a regular @xmath0d lattice , even in absence of any disorder @xcite . till date a large number of works have been done to explore the understanding of anderson localization and scaling hypothesis in one- and two - dimensional systems @xcite .
similarly , wannier - stark localization has also drawn the attention of many theorists @xcite as well as experimentalists @xcite .
for both these two cases , viz , infinite @xmath0d materials with random site energies and @xmath0d systems subjected to an external electric field , one never encounters any _ mobility edge _ i.e. , energy eigenvalues separating localized states from the extended ones , since all the eigenstates are localized .
but there exist some special types of @xmath0d materials , like quasi - periodic aubry - andre model and correlated disordered systems where mobility edge phenomenon at some particular energies is obtained @xcite .
although the studies involving mobility edge phenomenon in low - dimensional systems have already generated a wealth of literature @xcite there is still need to look deeper into the problem to address several interesting issues those have not yet been explored .
for example , whether the mobility edges can be observed in some other simple @xmath0d materials or the number of mobility edges separating the extended and localized regions in the full energy band of an @xmath0d material can be regulated , are still to be investigated . to address these issues in the present article we investigate two - terminal spin dependent transport in a @xmath0d mesoscopic chain composed of magnetic and non - magnetic atomic sites in presence of external electric field . to the best of our knowledge
, no rigorous effort has been made so far to explore the effect of an external electric field on electron transport in such a @xmath0d magnetic - non - magnetic superlattice geometry .
here we show that , depending on the unit cell configuration , a @xmath0d superlattice structure subjected to an external electric field exhibits multiple mobility edges at different values of the carrier energy .
we use a simple tight - binding ( tb ) framework to illustrate the model quantum system and numerically evaluate two - terminal spin dependent transmission probabilities through the superlattice geometry based on the green s function formalism . from our exact numerical analysis
we establish that a sharp crossover from a completely opaque to a fully or partly transmitting zone takes place which leads to a possibility of tuning the electron transport by gating the transmission zone .
in addition to this behavior we also show that the magnetic - non - magnetic superlattice structure can be used as a pure spin filter for a wide range of energy .
these phenomena enhance the prospect of such simple superlattice structures as switching devices at multiple energies as well as spin filter devices , the design of which has significant impact in the present age of nanotechnology . with an introduction in section i
, we organize the paper as follows . in section ii , first we present the model , then describe the theoretical formulation which include the hamiltonian and the formulation for transmission probabilities through the model quantum system .
the numerical results are illustrated in section iii and finally , in section iv , we draw our conclusions .
let us start with fig .
[ chain ] where a @xmath0d mesoscopic chain composed of magnetic and non - magnetic atomic sites is attached to two semi - infinite @xmath0d non - magnetic electrodes , namely , source and drain .
the chain consists of @xmath1 ( @xmath1 being an integer ) number of unit cells in which each unit cell contains @xmath2 and @xmath3 numbers of magnetic and non - magnetic atoms , respectively .
both the chain and side - attached electrodes are described by simple tb framework within nearest - neighbor hopping approximation .
the hamiltonian for the entire system can be written as a sum of three terms as , @xmath4 the first term represents the hamiltonian for the chain and it reads @xmath5 \label{eqn2}\ ] ] where , @xmath6 @xmath7 @xmath8 ; + @xmath9 + @xmath10 = @xmath11 .
+ + here , @xmath12 refers to the on - site energy of an electron at the site @xmath13 with spin @xmath14 ( @xmath15 ) , @xmath16 is the nearest - neighbor hopping strength , @xmath17 ( @xmath18 ) is the creation ( annihilation ) operator of an electron at the @xmath13th site with spin @xmath14 and @xmath19 is the strength of local magnetic moment where @xmath20 for non - magnetic sites .
the term @xmath10 corresponds to the interaction of the spin of the injected electron with the local magnetic moment placed at the site @xmath13 .
the direction of magnetization in each magnetic site is chosen to be arbitrary and specified by angles @xmath21 and @xmath22 in spherical polar co - ordinate system for the @xmath13th atomic site . here
, @xmath21 represents the angle between the direction of magnetization and the chosen @xmath23 axis , and @xmath22 represents the azimuthal angle made by the projection of the local moment on @xmath24-@xmath25 plane with the @xmath24 axis . in presence of bias voltage @xmath26 between the source and drain an electric field
is developed , and therefore , the site energies of the chain becomes voltage dependent .
mathematically we can express it as @xmath27 , where @xmath28 is the voltage independent term .
the voltage dependence of @xmath29 reflects the bare electric field in the bias junction as well as screening due to longer range electron - electron interaction . in the absence of such screening
the electric field varies uniformly along the chain and it reads @xmath30 , where @xmath31 corresponds to the total number of atomic sites in the chain . in our present work , we consider both the linear and screened electric field profiles . as illustrative example , in fig .
[ potprofile ] we show the variation of voltage dependent site energies for three different electrostatic potential profiles for a chain considering @xmath32 atomic sites and describe the nature of electronic localization for these profiles in the forthcoming section .
the second and third terms of eq .
[ eqn1 ] describe the tb hamiltonians for the @xmath0d semi - infinite non - magnetic electrodes and the chain - to - electrode coupling .
these hamiltonians are written as follows .
@xmath33 \right ] \label{eqn3}\ ] ] and , @xmath34 + \tau_d[\mbox { \boldmath $ c$}_n^{\dag } \mbox { \boldmath $ c$}_{n+1 } + h.c . ] .
\label{eqn4}\end{aligned}\ ] ] the summation over s and d in eq .
[ eqn3 ] implies the incorporation of both the two electrodes , viz , source and drain . @xmath35 and @xmath36 stand for the site energy and nearest - neighbor coupling , respectively .
the electrodes are directly coupled to the chain through the lattice sites @xmath0 and @xmath31 , and the coupling strengths between these electrodes with the chain are described by @xmath37 and @xmath38 , respectively . to obtain
spin resolved transmission probabilities of an electron through the source - chain - drain bridge system , we use green s function formalism .
the single particle green s function operator representing the entire system for an electron with energy @xmath39 is defined as , @xmath40 where , @xmath41 . following the matrix form of and
the problem of finding in the full hilbert space can be mapped exactly to a green s function @xmath42 corresponding to an effective hamiltonian in the reduced hilbert space of the chain itself and we have , @xmath43 where , @xmath44 these and are the self - energies introduced to incorporate the effect of coupling of the chain to the source and drain , respectively . using dyson equation the analytic form of the self energies
can be evaluated as follows , @xmath45 where , @xmath46 .
following fisher - lee relation , the transmission probability of an electron from the source to drain is given by the expression , @xmath47 .
\label{eqn9}\ ] ] where , @xmath48@xmath49 s are the coupling matrices representing the coupling between the chain and the electrodes and they are defined as , @xmath50 .
\label{eqn10}\ ] ] here , @xmath51 and @xmath52 are the retarded and advanced self - energies associated with the @xmath53-th ( @xmath54 ) electrode , respectively . finally , we determine the average density of states ( ados ) , @xmath55 , from the following relation , @xmath56\right ] .
\label{equ11}\ ] ] in what follows we limit ourselves to absolute zero temperature and use the units where @xmath57 . for the numerical calculations we set @xmath58 , @xmath59 , @xmath60 for the magnetic sites , @xmath61 , @xmath62 , @xmath63 and @xmath64 .
the energy scale is measured in unit of @xmath16 .
throughout our numerical calculations we assume that the magnetic moments are aligned along @xmath65 direction ( @xmath66 ) , which yields vanishing spin flip transmission probability , viz , @xmath67 , across the bridge system .
the net transmission probability is therefore a sum @xmath68 , and the origin of this zero spin flipping can be explained from the following arguments .
the operators @xmath69 @xmath70 and @xmath71 @xmath72 associated with the term in the tb hamiltonian eq .
[ eqn2 ] are responsible for the spin flipping , where being the pauli spin vector with components @xmath73 , @xmath74 and @xmath75 for the injecting electron . in our present model
since we consider that all the magnetic moments are aligned along @xmath65 direction , the term @xmath76 gets the form @xmath77 , and accordingly , the hamiltonian does not contain @xmath73 and @xmath74 and so @xmath69 and @xmath71 do not appear , which leads to the vanishing spin flip transmission probability across the @xmath0d chain .
below , we address the central results of our study i.e , the possibility of getting multiple mobility edges in @xmath0d magnetic - non - magnetic superlattice geometries and how such a simple model quantum system can be used as a perfect spin filter for a wide range of energy . in fig .
[ localization1 ] we show the variation of total transmission probability @xmath78 along with the average density of states for a @xmath0d magnetic - non - magnetic superlattice geometry considering a linear bias drop .
here we consider a @xmath79-site chain in which each unit cell contains one magnetic and four non - magnetic sites and the results are shown for three different bias voltages .
for the particular case when the chain is free from external electric field i.e. , @xmath80 electronic conduction through the bridge takes place for the entire energy band as shown in fig .
[ localization1](a ) which predicts that all the energy eigenstates are extended in nature .
the situation becomes really very interesting when the superlattice geometry is subjected to an external electric field .
it is illustrated in figs .
[ localization1](b ) and [ localization1](c ) . from these spectra
we notice that there are some energy regions for which the transmission probability completely drops to zero which reveals that the eigenstates associated with these energies are localized , and they are separated from the extended energy eigenstates .
thus , sharp mobility edges are obtained in the spectrum , and , the total number of such mobility edges separating the extended and localized regions in a superlattice geometry in presence electric field strongly depends on the unit cell configuration and it can be regulated by adjusting the number of magnetic and non - magnetic sites .
this phenomenon describes the existence of multiple mobility edges in a superlattice geometry under finite bias condition .
now if the fermi energy is fixed at a suitable energy zone where @xmath78 drops to zero an insulating phase will appear , while for the other case , where @xmath78 is finite , a metallic phase is observed and it leads to the possibility of controlling the electronic transmission by gating the transmission zone .
the width of the localized regions between the band of extended regions increases with the strength of the electric field as clearly shown by comparing the spectra given in figs .
[ localization1](b ) and [ localization1](c ) , and , for strong enough field strength almost all energy eigenstates are localized . in that particular limit metal - to - insulator transition will no longer be observed .
the above results are analyzed for a particular ( linear ) variation of electric field along the chain . to explore the sensitivity of getting metal - to - insulator transition on the distribution of electric field , in figs .
[ localization2 ] and [ localization3 ] we present the results for two different screened electric field profiles taking the identical chain length . from the spectra we clearly observe that the width of the localized region gradually disappears with the flatness of the electric field profile in the interior of the bridge system .
if the potential drop takes place only at the chain - to - electrode interfaces , i.e. , when the potential profile becomes almost flat along the chain the width of the localized region almost vanishes and the metal - to - insulator transition is not observed , as is the case for the zero bias limit . finally , we illustrate how such a simple magnetic - non - magnetic superlattice geometry can be utilized as a perfect spin filter for a wide range of energy in absence of any external electric field . as illustrative example , in fig .
[ filter1 ] we present the transmission probabilities for up and down spin electrons together with the average density of states as a function of energy for a @xmath0d magnetic - non - magnetic superlattice geometry . from the spectra we observe that the up and down spin electrons follow two different channels while traversing through the superlattice geometry , since the spin flipping
is completely blocked for this configuration .
this splitting of up and down spin conduction channels is responsible for spin filtering action and the total number of these channels strongly depends on the unit cell configuration . from figs .
[ filter1](a ) and ( b ) we clearly see that for a wide range of energy for which the transmission probability of up spin electrons drops to zero value , shows non - zero transmission probability of down spin electrons .
therefore , setting the fermi energy to a suitable energy region we can control the transmission characteristics of up and down spin electrons , and , a spin selective transmission is thus obtained through the bridge system .
before we end , we would like to point out that since the overlap between the up and down spin conduction channels depends on the magnitudes of the local magnetic moments , we can regulate the spin degree of polarization ( dop ) simply by tuning the strength of these magnetic moments and for a wide range of energies it ( dop ) almost reaches to @xmath81 .
thus , our proposed magnetic - non - magnetic superlattice geometry is a very good example for designing a spin filter .
to conclude , in the present work we investigate in detail the spin dependent transport under finite bias condition through a @xmath0d magnetic - non - magnetic superlattice geometry using green s function formalism .
we use a simple tb framework to describe the model quantum system where all the calculations are done numerically . from our exact numerical analysis
we predict that in such a simple @xmath0d magnetic - non - magnetic superlattice geometry multiple mobility edges separating the localized and extended regions are obtained in presence of external electric field and the total number of mobility edges in the full energy spectrum can be controlled by arranging the unit cell configuration .
this phenomenon reveals that the superlattice geometry can be used as a switching device for multiple values of fermi energy .
the sensitivity of metal - to - insulator transition and vice versa on the electrostatic potential profile is thoroughly discussed .
finally , we analyze how such a superlattice geometry can be utilized in designing a tailor made spin filter device for wide range of energies . setting the fermi energy at a suitable energy zone
, a spin selective transmission is obtained through the bridge system .
all these predicted results may be utilized in fabricating spin based nano electronic devices .
the results presented in this communication are worked out for absolute zero temperature .
however , they should remain valid even in a certain range of finite temperatures ( @xmath82k ) .
this is because the broadening of the energy levels of the chain due to the chain - to - electrode coupling is , in general , much larger than that of the thermal broadening @xcite .
s. aubry and g. andr , in _ group theoretical methods in physics _ ,
annals of the israel physical society vol . 3 , edited by l. horwitz and y. neeman ( american institute of physics , new york , 1980 ) , p. 133 . | based on green s function formalism , the existence of multiple mobility edges in a one - dimensional magnetic - non - magnetic superlattice geometry in presence of external electric field is predicted , and , it leads to the possibility of getting a metal - insulator transition at multiple values of fermi energy .
the role of electric field on electron localization is discussed for different arrangements of magnetic and non - magnetic atomic sites in the chain .
we also analyze that the model quantum system can be used as a perfect spin filter for a wide range of energy . |
quantum fourier transform is the most important tool of quantum computation @xcite , and one of the algorithms which can bring about exponential speedup .
shor s algorithm , hidden subgroup problem and solving systems of linear equations all make use of quantum fourier transform @xcite .
quantum fourier transform utilizes the superposition of quantum state , whereby the required time and space for computation can be notably reduced from @xmath0 to @xmath1 .
hence , the implementation of quantum fourier algorithm is crucial to exponential speedup in quantum computation williams , deutsch .
a novel method to simulate quantum entanglement using classical fields modulated with pseudorandom phase sequences was introduced in ref .
@xcite , which can realize tensor product in quantum computation and the representation of arbitrary quantum state @xcite . at the same time , ref .
@xcite also introduced a new concept , pseudorandom phase ensemble , to simulate a quantum ensemble .
however , it is necessary to implement similar quantum fourier transform algorithm to implement some current quantum algorithms . therefore , in this paper
, we , at first utilize this method to implement the simulation of quantum fourier transform , then investigate the required computational resources , and at last take three kinds of fields as an example to verify this algorithm .
generally , quantum fourier transform takes as input a vector of complex numbers , @xmath2 , and output a new vector of complex numbers @xmath3 as following:@xmath4 this calculation involves the additions and multiplications of @xmath5 complex numbers , leading to an increase of computational complexity with the increase of the number of vector components .
classically , the most effective algorithm , fast fourier transform is in time @xmath6 . on the contrary
, the quantum fourier transform can be defined as a unitary transformation on @xmath7 qubits @xcite , which is:@xmath8furthermore , the quantum fourier transform of arbitrary state @xmath9 can be expressed as:@xmath10
\left\vert k\right\rangle , \nonumber\end{aligned}\]]where @xmath11 .
then , we expand @xmath12 into:@xmath13where the coefficients satisfy the following equation:@xmath14 in quantum fourier transform , after the hadamard gate and controlled - phase gate , we can obtain the final state @xmath15 of quantum fourier transform:@xmath16there are n hadamard gates and @xmath17 controlled - phase gates on n qubit registers , which means the quantum fourier transform takes @xmath18 basic gate operations .
nevertheless , the quantum fourier transform can not output precise result of final states directly , but the probability of every state by repeated measurements , which can output the final result of fourier transform at a certain accuracy @xcite .
@xcite , a way to simulation of quantum states was introduced , which utilize the properties of pseudorandom sequence to modulate classical optical fields into different quantum states with different pseudorandom sequences .
the formal product states for these fields can be a simulation to arbitrary quantum states in the pseudorandom phase ensemble model @xcite . to further illustrate this method ,
the following is a brief introduction .
there are two orthogonal modes ( polarization or transverse ) of a classical field , which are denoted by @xmath19 and @xmath20 , respectively .
thus , a qubit state @xmath21 can be expressed by the mode superposition , where latexmath:[$|\alpha superposition states span a hilbert space . choosing any @xmath1 ppss from the set @xmath23 over @xmath24 to modulate @xmath1 classical fields , we can obtain the states expressed as follows:@xmath25 according to the properties of ppss and hilbert space
, we can define the inner product of any two fields @xmath26 and @xmath27 .
we obtain the orthogonal property in our simulation,@xmath28where @xmath29 are the @xmath30-th units of @xmath31 and @xmath32 , respectively .
the orthogonal property supports the construction of the tensor product structure of the multiple states .
a formal product state @xmath33 for the @xmath34 classical fields is defined as being a direct product of @xmath35@xmath36 according to the definition , @xmath1 classical fields of eq .
( [ eq7 ] ) can be expressed as the following states:@xmath37 as mentioned in @xcite , a general form of @xmath38 for @xmath1 fields can be constructed from eq .
( [ eq7 ] ) using a gate array model,@xmath39 where @xmath40 .
then , the formal product state ( [ eq9 ] ) can be written as@xmath41 further , we can obtain each item of the superposition of @xmath42 as follows:@xmath43 according to the closure property @xcite , the phase sequences @xmath44 of @xmath45 remain in the set @xmath46 , which means @xmath47 .
therefore , we obtain the formal product state @xmath33 spans a hilbert space with the basis @xmath48 and can be expressed as follows:@xmath49 , \label{eq14}\]]where @xmath50 denotes a total of @xmath51 coefficients . ref . @xcite
further proposed the ensemble - averaged reduced states in ensemble model , which utilize the closure and balance property of pseudorandom phase sequence to remove some terms in the formal product states after ensemble averaging , whereby the required arbitrary quantum states can be obtained , including quantum entanglement states .
considering a quantum state @xmath33 expressed by @xmath1 classical fields , according to the equation ( [ eq14 ] ) , we obtain its formal product states as following:@xmath52where @xmath53 is the same as that in eq .
( [ eq14 ] ) .
after quantum fourier transform , this state involves into @xmath54where @xmath55 according to the definition eq .
( [ eq1 ] ) , the relation between the coefficients @xmath45 and @xmath56 of these two states have to be satisfied as eq .
( [ eq5 ] ) . to obtain the relation between these coefficient
, we design the following algortithm : \(1 ) selected a basis state @xmath57 of @xmath58 ; \(2 ) apply the following controlled - phase transformation on every field of @xmath59 according to the specific value of bits in the selected basis state:@xmath60 \(3 ) apply hadamard gate on these fields , after which we obtain:@xmath61 \(4 ) apply the mode selection gate on these fields according to the specific values in @xmath57 , after which the mode of every field is identical to the corresponding value in @xmath62 , e.g. , if @xmath63 , @xmath64 , while if @xmath65 , @xmath66 ; \(5 ) apply mode detection on these fields and obtain the m matrix .
then we can obtain the corresponding coefficient @xmath67 using the method in ref .
@xcite .
the above algorithm can be summarized as the following block diagram in fig .
[ fig1 ] . at last
, we can analysis the computational complexity : there are @xmath1 fields in @xmath33 after @xmath1 controlled - phase gates , @xmath1 hadamard gates , @xmath1 mode selection operations and finally @xmath68 correlation detection in mode detection .
hence , the total number of operations is in @xmath18 , which is the same as that in quantum fourier transform .
however , the result we obtain is with certainty but not with probability like the case in quantum fourier transform .
in pseudorandom phase ensemble , we utilize the characteristic of pseudorandom phase sequence to define the ensemble - averaged reduced state @xcite .
the balance property of pseudorandom phase sequences @xcite : with the exception of @xmath69 , any sequence of the set @xmath70 satisfies@xmath71where @xmath72 is the @xmath30-th phase unit in pseudorandom sequence @xmath73 . due to this property ,
several terms of the formal product state can be reduced under ensemble averaging , enabling us to utilize classical fields to simulation arbitrary quantum states . according to ref .
@xcite , the reduced state can be defined as@xmath74where @xmath75 , which is the sum of all phase sequences of classical fields .
then , we discuss about the quantum fourier transform of ensemble - averaged reduced states . from equation ( [ eq5 ] ) , we obtain the coefficients of quantum fourier transform satisfies@xmath76 in these equations , the combinations of @xmath77 and @xmath78 satisfy the following relations:@xmath79 } \label{eq23}\ ] ] obviously , these terms also satisfy the balance property of pseudorandom sequence . hence , the ensemble - averaged reduced states can also be used in the states after quantum fourier transform .
then , we can obtain the fourier transform:@xmath80 at last , we show the equivalence of ensemble - averaged reduced states in quantum fourier transform .
according to ref .
@xcite , three pseudorandom phase sequences @xmath81 are required to implement the simulation of the quantum states consisting of three particles . modulated with these phase sequence
, three classical optical fields can be expressed as following:@xmath82after proper gate array operation @xcite , we can obtain arbitrary quantum states which can be expressed as following:@xmath83 according to the algorithm in sec .
[ sec3.2 ] , we obtain : \(1 ) apply controlled - phase gates on three classical fields respectively : @xmath84 where @xmath85 .
\(2 ) hadamard transformation@xmath86
\(3 ) calculate coefficients ( 3.1 ) when @xmath87 and @xmath88,@xmath89then obtain the corresponding coefficients @xmath90 and @xmath91:@xmath92@xmath93 ( 3.2 ) when @xmath94 and @xmath95,@xmath96then obtain the corresponding coefficients @xmath97 and @xmath98:@xmath99@xmath100 ( 3.3 ) when @xmath101 and @xmath102,@xmath103then obtain the corresponding coefficients @xmath104 and @xmath105:@xmath106@xmath107 ( 3.4 ) when @xmath108 and @xmath109,@xmath110then obtain the corresponding coefficients @xmath111 and @xmath112:@xmath113@xmath114at last , we obtain the transform matrix of all coefficients as following:@xmath115we will utilize the above algorithm to apply quantum fourier transform on several kinds of states in the following : \(1 ) product state in quantum mechanics , the product state of three particles is @xmath116 .
we can expressed these three fields as equation ( [ eq25 ] ) , except for normalization constant . in ref .
[ 6 ] , the formal product state of this state can be expressed as:@xmath117except for normalization constant and overall phase factor , that is , the sum of three pseudorandom sequence , there is not difference between this state and the product state of three particles . from ref .
@xcite , utilizing the concept of pseudorandom phase ensemble and the properties of pseudorandom sequence , we obtain the ensemble - averaged reduced state:@xmath118 using the above algorithm , we can easily obtain the coefficients of the fourier transform of this state , @xmath119 , while the other terms is 0 . then we obtain:@xmath120which is identical to the quantum fourier transform , except for the normalization constant .
\(2 ) ghz state in quantum mechanics , ghz state is biggest entanglement state in the system of the three particles .
this state is of great importance since it can verify the entanglement criterion in the correlation measurement of quantum entanglement . from ref .
@xcite , we can obtain the following form of three fields by proper transformation:@xmath121the formal product state of these three fields can be expressed as:@xmath122 \nonumber\end{aligned}\]]from ref .
@xcite , utilizing the concept of pseudorandom phase ensemble and the properties of pseudorandom sequence , we obtain the ensemble - averaged reduced state:@xmath123similarly , except for normalization constant and overall phase factor , the state is identical to ghz state . using the above algorithm , we can easily obtain the coefficients of the fourier transform of this state respectively:@xmath124@xmath125@xmath126@xmath127@xmath128@xmath129@xmath130@xmath131utilizing ensemble - averaged reduced state , we obtain : @xmath132 in conclusion , @xmath133 is the fourier transform of @xmath134 for ghz states .
\(3 ) w state in quantum mechanics , w state is the most robust entanglement state @xmath135 . from ref .
@xcite , by proper transformation on equation ( eq25 ) , we can obtain the expression of these three fields as following:@xmath136the formal product state of these three fields can be expressed as:@xmath137 \times \left ( \left\vert 100\right\rangle + \left\vert 010\right\rangle \right .
\nonumber \\ & & \left .
+ \left\vert 001\right\rangle \right ) + \left [ e^{i\left ( \lambda ^{\left ( 1\right ) } -\lambda ^{\left ( 3\right ) } \right ) } + e^{i\left ( \lambda ^{\left ( 1\right ) } -\lambda ^{\left ( 2\right ) } \right ) } \right ] \left ( \left\vert 011\right\rangle + \left\vert 110\right\rangle + \left\vert 101\right\rangle \right ) + e^{i\left ( 2\lambda ^{\left ( 1\right ) } -\lambda ^{\left ( 2\right ) } -\lambda ^{\left ( 3\right ) } \right ) } \left\vert 111\right\rangle \nonumber \\ & & \left .
+ 2\left [ e^{i\left ( 2\lambda ^{\left ( 2\right ) } -\lambda ^{\left ( 1\right ) } -\lambda ^{\left ( 3\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 3\right ) } -\lambda ^{\left ( 2\right ) } -\lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( \lambda ^{\left ( 2\right ) } -\lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( \lambda ^{\left ( 3\right ) } -\lambda ^{\left ( 1\right ) } \right ) } % \right ] \left\vert 000\right\rangle \right\ } \label{eq58}\end{aligned}\]]from ref .
@xcite , utilizing the concept of pseudorandom phase ensemble and the properties of pseudorandom sequence , we obtain the ensemble - averaged reduced state:@xmath138similarly , except for normalization constant and overall phase factor , the state is identical to w state .
using the above algorithm , we can easily obtain the coefficients of the fourier transform of this state respectively:@xmath139
\nonumber \\ & & + e^{3i\lambda ^{\left ( 1\right ) } } + e^{3i\lambda ^{\left ( 2\right ) } } + e^{3i\lambda ^{\left ( 3\right ) } } , \label{eq60}\end{aligned}\]]@xmath140 -\left ( 1-\omega -\omega ^{2}\right ) \nonumber \\ & & \times \left [ e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 1\right ) } \right ) } \right ] + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 2\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 3\right ) } \right ) } \nonumber \\ & & -\omega ^{3}e^{3i\lambda ^{\left ( 1\right ) } } + e^{3i\lambda ^{\left ( 2\right ) } } + e^{3i\lambda ^{\left ( 3\right ) } } , \label{eq61}\end{aligned}\]]@xmath141 + \omega ^{2}\left [ e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 1\right ) } \right ) } \right ] + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 2\right ) } \right ) } \nonumber \\ & & + e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 3\right ) } \right ) } -\omega ^{2}e^{3i\lambda ^{\left ( 1\right ) } } + e^{3i\lambda ^{\left ( 2\right ) } } + e^{3i\lambda ^{\left ( 3\right ) } } , \label{eq62}\end{aligned}\]]@xmath142 -\left ( 1+\omega ^{2}-\omega ^{3}\right ) \nonumber \\ & & \times \left [ e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 1\right ) } \right ) } \right ] + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 2\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 3\right ) } \right ) } \nonumber \\ & & + \omega ^{5}e^{3i\lambda ^{\left ( 1\right ) } } + e^{3i\lambda ^{\left ( 2\right ) } } + e^{3i\lambda ^{\left ( 3\right ) } } , \label{eq63}\end{aligned}\]]@xmath143 + 2\left [ e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 1\right ) } \right ) } \right ] + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 2\right ) } \right ) } \nonumber \\ & & + e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 3\right ) } \right ) } -e^{3i\lambda ^{\left ( 1\right ) } } + e^{3i\lambda ^{\left ( 2\right ) } } + e^{3i\lambda ^{\left ( 3\right ) } } , \label{eq64}\end{aligned}\]]@xmath144 -\left ( 1+\omega -\omega ^{2}\right ) \nonumber \\ & & \times \left [ e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 1\right ) } \right ) } \right ] + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 2\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 3\right ) } \right ) } \nonumber \\ & & + \omega ^{3}e^{3i\lambda ^{\left ( 1\right ) } } + e^{3i\lambda ^{\left ( 2\right ) } } + e^{3i\lambda ^{\left ( 3\right ) } } , \label{eq65}\end{aligned}\]]@xmath145 -\omega ^{2}\left [ e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 1\right ) } \right ) } \right ] + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 2\right ) } \right ) } \nonumber \\ & & + e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 3\right ) } \right ) } + \omega ^{2}e^{3i\lambda ^{\left ( 1\right ) } } + e^{3i\lambda ^{\left ( 2\right ) } } + e^{3i\lambda ^{\left ( 3\right ) } } , \label{eq66}\end{aligned}\]]@xmath146 -\left ( 1+\omega ^{2}+\omega ^{3}\right ) \nonumber \\ & & \times \left [ e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 1\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 1\right ) } \right ) } \right ] + e^{i\left ( 2\lambda ^{\left ( 3\right ) } + \lambda ^{\left ( 2\right ) } \right ) } + e^{i\left ( 2\lambda ^{\left ( 2\right ) } + \lambda ^{\left ( 3\right ) } \right ) } \nonumber \\ & & -\omega ^{5}e^{3i\lambda ^{\left ( 1\right ) } } + e^{3i\lambda ^{\left ( 2\right ) } } + e^{3i\lambda ^{\left ( 3\right ) } } .
\label{eq67}\end{aligned}\]]utilizing ensemble - averaged reduced state , we obtain : @xmath147 in conclusion , @xmath133 is also the fourier transform of @xmath134 for w states .
utilizing pseudorandom phase ensemble model , we propose a transform similar to quantum fourier transform .
the computational resources required for this transform is in @xmath18 similar to quantum fourier transform , which means an exponential speedup compared with classical fourier transform . in the future , we will utilize this method to implement shor s algorithm and other algorithms which need quantum fourier transform , and then conduct computer - based simulation of this algorithms to verify the feasibility of them .
d. deutsch , quantum theory , the church - turing principle and the universal quantum computer , proceedings of the royal society of london a : mathematical , physical and engineering sciences 400 , 97 ( 1985 ) .
j. fu and x. wu , effective simulation of quantum entanglement using classical fields modulated with pseudorandom phase sequences , scienceopen research 2015 ( doi : 10.14293/s2199 - 1006.1.sorphys.anvyqz.v1 ) . | in this paper , we introduce an optical analogy to quantum fourier tanformation based on a pseudorandom phase ensemble .
the optical analogy also brings about exponential speedup over classical fourier tanformation . using the analogy ,
we demonstrate three classcial fields to realize fourier transform similar to three quantum particles . |
thin sheets with repetitive metallic inclusions or exclusions have been used extensively in the past in antenna applications to control the properties of reflected or transmitted power , e.g. its direction , phase , or polarization @xcite .
such surfaces have received increasing attention lately , as part of the intensive research in the field of optical and microwave metamaterials , in an attempt to harness ideas from bulk metamterial explorations to design low - profile components with extraordinary wavefront manipulation capabilities @xcite .
in contrast to bulk metamaterials , where subwavelength elements are combined to form a volumetric entity with prescribed local response to electromagnetic fields , in metamaterial sheets , or metasurfaces , these subwavelength atomic units are confined to a region with subwavelength thickness .
this geometrical difference should decrease significantly fabrication complexity of metasurfaces and also loss - related problems ; however , it requires development of new design methodologies , as the interaction of electromagnetic fields with metasurfaces is naturally described via effective boundary conditions @xcite , as opposed to effective permeabilities and permittivities ( or effective wave equations ) , more suitable for modelling volumetric metamaterials @xcite . in particular , it was recently recognized that as metasurfaces act as sources of tangential field discontinuities , they can be modelled by a distribution of electric and magnetic surface currents , prescribed by the equivalence principle [ , pp .
575 - 579 ] .
hence , in principle , for a given incident field , a desirable electromagnetic field distribution in space can be achieved by engineering the surface to induce currents that would produce the required tangential fields on both of its facets . approximating the required continuous surface currents by a dense distribution of electric and magnetic dipoles
, passive surfaces implementing plane - wave refraction were demonstrated @xcite .
the elementary sources were formed by subwavelength inductive and capacitive elements that produced the suitable magnitudes of the current in response to the exciting incident plane - wave .
simultaneously , it was demonstrated that also active elementary sources may be utilized to introduce desirable field discontinuities , e.g. , to implement a cloaking device based on the same equivalence principle @xcite . as these surfaces were composed of orthogonal electric and magnetic dipoles engineered to induce unidirectional radiation , i.e. acting as huygens sources [ , pp .
653 - 660],@xcite , they were named huygens metasurfaces ( hms ) @xcite .
in addition to plane - wave refraction and cloaking , recent reports proposed designs of huygens metasurfaces which implement beam shaping , transmission or reflection coefficient engineering and polarization manipulation ( using tensor huygens metasurfaces ) @xcite .
although the design methodologies differ between the various authors , they all rely on the fact that if the dimensions of the unit cells and their spatial arrangement obey certain conditions , the metasurface can be modelled by effective electric and magnetic polarizability distributions , which are translated to position - dependent sheet boundary conditions @xcite .
these , in turn , can be equivalently described as surface impedance and surface admittance matrices relating the electric and magnetic field components at the two facets of the metasurface @xcite . following this approach
greatly simplifies device design , as it facilitates the development of simple circuit models to huygens metasurfaces @xcite .
moreover , as the effective surface impedance and admittance matrices are directly related to the _ locally averaged _ polarizabilities of the elementary scatterers , we may assess the local equivalent surface impedance of a unit cell by simulating or measuring the effective impedance of an infinite periodic array of such identical unit cells ( local periodicity approximation ) @xcite .
indeed , this modelling approach was used in recent demonstrations of hmss : given the incident field and the desirable transmitted ( or reflected ) field , the required boundary conditions to support the field discontinuities can be formulated , resulting in the required surface impedance and admittance matrices ( or electric and magnetic polarizability distributions ) . however , for realizing the huygens metasurfaces , it is desirable for the elements implementing the required polarizability distributions to be passive and lossless , i.e. the surface impedance and admittance to possess pure imaginary values . implementing impedance or admittance sheets with nonvanishing real parts requires engineering of gain or loss elements , thereby complicating greatly the design and realization . nonetheless , following the simplistic methodology in which the discontinuities between the desirable and incident fields are directly translated into surface impedances and admittances by no means guarantees the passivity of the resultant metasurface ( see , e.g. , @xcite ) .
in fact , in @xcite the design procedure did not consider the passivity limitation , however the resultant leading to a well - functioning , prototype .
the reasons for that encouraging outcome were not analyzed therein , though .
in addition , almost all huygens metasurfaces presented in the literature to date were designed to be excited by a plane - wave or a beam propagating towards the surface in homogeneous medium @xcite .
nevertheless , to facilitate the development of realizable antenna devices based on huygens metasurfaces it would be necessary to extend the current design techniques to enable excitation of the metasurface by localized ( impulsive ) sources , or waveguided modes .
a step in that direction was made by holloway _
@xcite , where a line source excitation was considered , but only for metasurfaces with _ constant _ polarizability density ; as demonstrated by @xcite , allowing the polarizability densities to vary along the metasurface could be beneficial , providing more degrees of freedom for the design . in this work
, we derive simple rules for designing passive lossless huygens metasurfaces producing directive radiation to a prescribed angle when excited by a given ( arbitrary ) source field . decomposing the fields to their plane - wave spectrum and generalizing the approach presented in @xcite for plane - wave excitation
, we show that satisfying two physical conditions is sufficient to guarantee that the desirable functionality can be achieved by purely reactive surfaces : local power conservation across the surface , and local impedance the fields on both sides of the metasurface . enforcing these conditions locally ,
i.e. at each point on the surface , leads to a complementary set of simple expressions for the surface impedance and surface admittance , as well as facilitates the semi - analytical evaluation of the reflected and transmitted fields .
these results enable design of directive hms radiators with a wide range of source excitations , thus extending significantly the possible applications .
moreover , the derivation provides clear physical interpretation of the conditions required to implement passive lossless hmss , which may also be indicative to equivalent requirements in more generalized scenarios ( e.g. hmss which perform other functionalities ) .
altogether , this forms an efficient and powerful tool for hms engineering , promoting design of novel antenna devices .
we consider a 2d configuration @xmath0 in which a huygens metasurface situated at @xmath1 is excited by an arbitrary current distribution limited to the half - space @xmath2 ( fig .
[ fig : physicalconfiguration ] ) .
the surrounding media is assumed to be homogeneous , with permittivity @xmath3 and permeability @xmath4 , defining the wave impedance @xmath5 .
harmonic time dependency of @xmath6 is assumed ( and suppressed ) , defining the wavenumber @xmath7 . .
the formalism applies without modification also to scenarios in which the region @xmath8 is occupied by an inhomogeneous medium , as long as its cross - section remains uniform with respect to the @xmath9 axis , i.e the permittivity , permeability and conductivity are a function of @xmath9 coordinate only ( e.g. , plane - stratified media ) [ , pp .
183 - 202].,width=321 ] the is characterized by its surface impedance @xmath10 and surface admittance @xmath11 , inducing discontinuities in tangential magnetic and electric field components , respectively , given by the generalized sheet transition conditions ( gstc ) as formulated by kuester _
@xcite @xmath12 = \\ \bigskip = \dfrac{1}{2}\left[\left.\vec{e}\right|_{x\rightarrow0^{+}}+\left.\vec{e}\right|_{x\rightarrow0^{-}}\right ] \\
y_{sm}\left(z\right)\vec{m}_s = -y_{sm}\left(z\right)\hat{x}\times\left[\left.\vec{e}\right|_{x\rightarrow0^{+}}-\left.\vec{e}\right|_{x\rightarrow0^{-}}\right ] = \\ \bigskip = \dfrac{1}{2}\left[\left.\vec{h}\right|_{x\rightarrow0^{+}}+\left.\vec{h}\right|_{x\rightarrow0^{-}}\right ] , \end{array } \right.\!\!\!\!\!\!\ !
\label{equ : general_impedance_admittance_discontinuity}\end{aligned}\ ] ] where @xmath13 and @xmath14 are the surface currents induced by the tangential electric and magnetic field components , respectively , and we assumed the impedance and admittance matrices can be described by scalar quantities .
the half - spaces below and above the metasurface are referred to as region 1 and region 2 , respectively , and we require the sources not to be infinitesimally close to the hms such that a source - free region @xmath15 can be defined just below it ( fig . [
fig : physicalconfiguration ] ) . in region 1 , we distinguish between the incident field and reflected field , while region 2 is populated only by the transmitted fields ( denoted by @xmath16 , @xmath17 , and @xmath18 superscripts , respectively ) @xmath19 and the incident fields are defined as the fields produced by the sources in the absence of the hms includes scattering elements ( e.g. , as we allow in subsection [ subsec : results : line_source_pec ] ) will be given in - . ] . in 2d configurations
maxwell s equations can be separated to two decoupled sets of equations related to the transverse electric ( te ) or transverse magnetic ( tm ) polarized field components . in source - free regions , the te - polarized ( @xmath20 ) nonvanishing field components are @xmath21 , and maxwell s equations can be reduced to a wave equation for @xmath22 and curl equations relating the other components to it , @xmath23 analogously , in source - free regions , the tm - polarized ( @xmath24 ) nonvanishing field components are @xmath25 , and maxwell s equations can be reduced to @xmath26 as this work does not deal with polarization manipulation of the source fields , we may design an hms directive radiator for each polarization independently , namely a te - hms and a tm - hms .
the te - hms would interact only with @xmath22 and @xmath27 , therefore the required surface impedance @xmath10 would be implemented using scatterers sensitive to electric field in the @xmath28 direction ( e.g. loaded wires parallel to the @xmath28-axis ) and the surface admittance @xmath11 would be implemented using scatterers sensitive to magnetic field in the @xmath29 direction ( e.g. loaded loops whose axis is parallel to the @xmath29-axis ) @xcite . on the other hand
, the tm - hms should interact only with @xmath30 and @xmath31 , therefore @xmath10 and @xmath11 would be composed by scatterers sensitive to electric field in the @xmath29 direction and magnetic field in the @xmath28 direction , respectively .
as the elements implementing the te - hms and the tm - hms are orthogonal , there should be no coupling between them ; thus , ideally , the two hmss may be combined to a single metasurface without any changes to the independent designs . in view of this observation
, we formulate the procedure to design an hms assuming the incident field is te - polarized ; the design rules for tm - polarized excitation can be readily derived by duality ( appendix [ app : theory_tm_excitation ] ) .
extension of this work to polarization manipulating hmss @xcite will be addressed in a separate report . as denoted , when the excitation field is te - polarized , the nonvanishing field components are @xmath32 , @xmath33 , and @xmath34 ; the scalar surface impedance only induces _ electric _ currents in the @xmath28 direction ; and the scalar surface admittance only induces _
magnetic _ currents in the @xmath29 direction . in view of a general solution for the fields in the _ source - free region _
@xmath35 can be formulated in the spectral domain as @xcite @xmath36 where @xmath37 is the transverse wavenumber ( associated with the propagation along @xmath29 ) , and @xmath38 is the longitudinal wavenumber ( associated with the propagation along @xmath9 ) ; to satisfy the radiation condition we demand that @xmath39 .
the @xmath40 left superscript denotes te - hms related parameters throughout the paper , and @xmath41 is a unit current magnitude .
it can be readily verified that all three integrals in satisfy the wave equation of , where @xmath42 is the source - related plane - wave spectrum , @xmath43 is the reflection coefficient in the spectral domain , and @xmath44 corresponds to the spectral content of the transmitted fields . as the sources reside at @xmath8 in region 1 , and
the scattering metasurface is situated at @xmath1 , the general solution in @xmath15 must consider both upwards ( incident ) and downwards ( reflected ) propagating waves .
in contrast , as neither sources nor scatterers exist in region 2 , only upwards ( transmitted ) propagating waves are allowed for @xmath45 , as to satisfy the radiation condition .
we emphasize that the formal solution presented in is valid in general only for @xmath35 and that the presence of the sources at @xmath8 introduces a discontinuity in the fields , which should be accounted for in that region .
nonetheless , for designing the hms only the tangential fields at @xmath46 as formulated in are required @xcite , and this formulation is valid _
regardless _ of the nature of the source distribution @xmath8 .
in fact , applies without modification also to scenarios in which the region @xmath8 is occupied by an inhomogeneous medium , as long as its cross - section remains uniform with respect to the @xmath9 axis , i.e the permittivity , permeability and conductivity are a function of @xmath9 coordinate only ( e.g. , plane - stratified media ) [ , pp .
183 - 202 ] . as shall be demonstrated in subsection
[ subsec : results : line_source_pec ] , this allows for even a wider range of excitation schemes to be investigated using our model . from and the curl equations of we derive the corresponding tangential component of the magnetic field at @xmath35 , @xmath47 and can be rewritten as @xcite @xmath48}{h_z^{\mathrm{trans}}\left(0,z\right)-\left[h_z^{\mathrm{inc}}\left(0,z\right)+h_z^{\mathrm{ref}}\left(0,z\right)\right ] } \\ \!\!\!\!\!{{\vphantom{y}}^{e}_{}{y}}_{sm}\left(z\right)=-\dfrac{1}{2 } \dfrac{h_z^{\mathrm{trans}}\left(0,z\right)+\left[h_z^{\mathrm{inc}}\left(0,z\right)+h_z^{\mathrm{ref}}\left(0,z\right)\right]}{e_y^{\mathrm{trans}}\left(0,z\right)-\left[e_y^{\mathrm{inc}}\left(0,z\right)+e_y^{\mathrm{ref}}\left(0,z\right)\right ] } \end{array}\!\!\!\!\!\!\!\!\ ! \right .
\label{equ : te_impedance_admittance_discontinuity}\end{aligned}\ ] ] given an incident field , we would like to determine the required variation of the surface impedance and admittance along the metasurface such that * both the surface impedance and surface admittance are purely reactive ( passive and lossless ) , namely @xmath49 . for that we are willing to allow some reflections from the metasurface . * the transmitted field @xmath50 , @xmath51 , @xmath52 will form a directional radiation to a specified angle @xmath53 with respect to the @xmath9 axis ( fig .
[ fig : physicalconfiguration ] ) . to this end
we would like to form a virtual aperture on the surface with surface current having the suitable linear phase variation and as uniform as possible magnitude . in order to satisfy the second demand , we require that the spectral content of the transmitted fields @xmath44 will be localized around @xmath54 .
@xmath55 and demand that the hms would be designed such that the resulting virtual aperture window would form a slowly - varying as @xmath56 , namely , @xmath57 applying this constraint on enables us to approximate the tangential magnetic field at @xmath56 as @xmath58 substituting and into yields @xmath59 where we have defined dimensionless quantities , proportional to the fields on the lower ( minus - sign superscript ) and upper ( plus - sign superscript ) facets of the metasurface , as follows @xmath60 \\ \vspace{5pt } { { \vphantom{f}}^{e}_{}{\!f}}_{\!\!h}^{-}\left(z\right ) \triangleq \dfrac{1}{i_0k\cos\theta_0 } \left[h_z^{\mathrm{inc}}\left(0,z\right)+h_z^{\mathrm{ref}}\left(0,z\right)\right ] \\ { { \vphantom{f}}^{e}_{}{\!f}}^{+}\left(z\right ) \triangleq { { \vphantom{w\left(0,z\right)}}^{e}_{}{w\left(0,z\right)}}e^{-jkz\sin\theta_0}. \end{array}\!\!\!\!\!\ ! \right .
\label{equ : te_dimensionless_quantities}\end{aligned}\ ] ] as we require the hms to be purely reactive , the real power across the metasurface must be conserved .
however , to obtain sufficient conditions for designing passive lossless hmss we may require a stricter condition to be met , namely , that the real power impinging the metasurface from region 1 is equal _ locally _ to the real power transmitted to region 2 , at _ each point _
@xmath29 on the surface . to assess the consequences of this local power conservation condition we utilize to express the local power densities along the metasurface in region 1 and 2 using the dimensionless field quantities .
these are given , respectively , by the projection of the poynting vector on the @xmath9 axis as @xmath61 @xmath62_{x\rightarrow 0 ^ - } \nonumber \\ & \bigskip = \frac{1}{2}k\eta\left|i_0\right|^2{{\vphantom{f}}^{e}_{}{\!f}}_{\!\!e}^{-}\left(z\right ) { { \vphantom{f}}^{e}_{}{\!f}}_{\!\!h}^{-*}\left(z\right)k\cos\theta_0 , \label{equ : te_poynting_vector_region1}\end{aligned}\ ] ] and as @xmath63 @xmath64_{x\rightarrow 0^+ } \nonumber \\ &
\bigskip = \frac{1}{2}k\eta\left|i_0\right|^2\left|{{\vphantom{f}}^{e}_{}{\!f}}^{+}\left(z\right)\right|^2k\cos\theta_0 , \label{equ : te_poynting_vector_region2}\end{aligned}\ ] ] where the asterisk indicates the complex - conjugate operation .
the condition for local power conservation is @xmath65 which reads , using - , @xmath66 multiplying the rational functions in by the complex conjugate of their respective denominators and plugging in the local power conservation requirement yields @xmath67 in view of , a _ sufficient _ condition for the real part of both @xmath68 and @xmath69 to vanish is given by @xmath70 resulting in a purely imaginary numerator for both fractions .
the physical meaning of the latter is revealed by rewriting using : the condition locally the wave impedance of the fields on the two facets of the metasurface . indeed , substituting into leads to the local impedance condition
@xmath71 when is satisfied , the local power conservation condition takes a simpler form , namely , @xmath72 subsequently , we may define the dimensionless field quantities above and below the metasurface as @xmath73 where @xmath74 . substituting these definitions into leads to a complementary set of compact expressions for the desirable surface impedance and surface admittance @xmath75 \\ \!\!\!\!{{\vphantom{y}}^{e}_{}{y}}_{sm}\left(z\right)=-j\dfrac{{{\vphantom{y}}^{e}_{}{y}}_0}{2 } \cot\left[\dfrac{\varphi_{-}\left(z\right)-\varphi_{+}\left(z\right)}{2}\right ] , \end{array}\!\!\!\!\ ! \right .
\label{equ : te_impedance_admittance_final}\end{aligned}\ ] ] where @xmath76 is the wave impedance of a te - polarized plane - wave propagating in region 2 at an angle of @xmath53 with respect to the @xmath9 axis , given generally by [ ] @xmath77 , \label{equ : transmitted_wave_impedance}\end{aligned}\ ] ] @xmath78 and @xmath79 being the tangential components of the fields . as @xmath80 , @xmath81 , @xmath82 are all real , the hms defined via is indeed purely reactive , as required .
equation prescribes the required variation of the surface reactance and susceptance to implement the desirable hms .
however , in order to use this formula , we should instruct how the phases of the fields at @xmath46 , defined as @xmath82 , are to be evaluated . to that end , we first indicate how to find the reflected and transmitted fields from the given source field and the requirement that the power is locally conserved and the impedance is locally .
local impedance requires @xmath83 , \label{equ : te_local_impedance_matching_explicit_2}\end{aligned}\ ] ] from which , using the spectral representation of the fields - , the reflection coefficient can be evaluated as @xmath84 _ regardless _ of the source field .
equation is merely a private case of the fresnel reflection formula for an incident plane - wave at an angle @xmath85 and a transmitted plane - wave at an angle of @xmath53 , travelling in media with the same permittivity and permeability @xcite .
actually , it is a generalization of the reflection coefficient derived in @xcite for passive lossless hmss excited by a plane - wave .
however , in the general case considered herein , the source field consists of an infinite number of plane - waves , propagating in different directions .
therefore , the reflection coefficient of actually ensures _
individual _ impedance of each of the plane - waves in the source field spectrum - to that of a plane - wave propagating in region 2 with an angle of @xmath53 with respect to the @xmath9 axis .
c |l| l| l| l no . & step & input & output & relevant equations + 1 & l local impedance + & [ cols= " < " , ] & @xmath86 & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 & @xmath94 + to verify our formulation , we have designed and simulated three huygens metasurfaces according to the procedure described in this subsection ( with @xmath95 ) , designated to convert the fields produced by an electric line source at @xmath96 to directive radiation towards @xmath97 , @xmath98 , and @xmath99 ( fig . [ fig : linesourceresults ] ) . for an hms stretching from @xmath100 to @xmath101 ( total length @xmath102 ) the required variation of the surface reactance @xmath103 and surface susceptance @xmath104 with @xmath29
is presented in fig .
[ fig : linesourceresults](a)-(c ) . for clarity ,
only values of @xmath105 and @xmath106 are presented .
recalling the physical meaning of the equivalent angle of incidence we can indeed observe a change of signs in the cotangent argument of when the angles formed between the source and the evaluation point approaches @xmath53 , i.e. in the proximity of @xmath107 .
this would happen around @xmath108 ( fig .
[ fig : linesourceresults](a ) ) , @xmath109 ( fig .
[ fig : linesourceresults](b ) ) , and @xmath110 ( fig .
[ fig : linesourceresults](c ) ) , for @xmath97 , @xmath98 , and @xmath99 , respectively .
moreover , as the difference between the equivalent angles and the transmission angle becomes larger , the effective period of the cotangent should become shorter ; indeed , if we focus on the @xmath111 region of the metasurface , where the equivalent angles of incidence are mostly negative , we observe that the effective period for @xmath97 ( fig .
[ fig : linesourceresults](a ) ) , is longer than that of @xmath98 ( fig .
[ fig : linesourceresults](b ) ) , which is , in turn , longer than the one corresponding to @xmath99 ( fig .
[ fig : linesourceresults](c ) ) .
the practical implication of this observation is that if we desire to harness those rays incident upon the metasurface at angles which differ significantly from the desirable transmission angle , we should anticipate a fast variation of the corresponding surface impedance and admittance , which , in turn , requires smaller distances between the elements implementing the metasurface .
we have implemented the designed hmss in a commercially available finite - element solver ( ansys hfss ) using one hundred @xmath112-long unit cells comprised of loaded wires and loops @xcite , as described in appendix [ app : hfss_implementation ] .
the hms was excited by an electric line source carrying @xmath113 current oscillating at a frequency of @xmath114 ; the simulated electric field variation as a function of position , @xmath115 , is presented in fig .
[ fig : linesourceresults](g)-(i ) for the three hmss considered . to compare these results with the theoretical predictions ,
we have calculated the spectral integrals - in conjunction with , , , and - , to evaluate the fields in the region @xmath116 , as presented in fig .
[ fig : linesourceresults](j)-(l ) .
the theoretical plots rely on semi - analytical approximations ( appendix [ app : asymptotic_evaluation ] ) , which assume the hms is of infinite extent to calculate the fields in region 1 ( @xmath2 ) and on the virtual aperture @xmath63 . then , to account for the finite hms length when evaluating the fields in region 2 , the virtual aperture window function @xmath117 is truncated at @xmath118 before utilizing .
these approximations yield accurate results when most of the excitation power interacts with the hms on its finite extent ( @xmath119 ) .
in addition , the steepest - descent - path method is employed to evaluate the radiated power in the far - field regions @xmath120 using the spectral integrals - and based on the same assumptions @xcite ( appendix [ app : asymptotic_evaluation ] ) .
this allows us to compare in fig .
[ fig : linesourceresults](d)-(f ) the theoretically predicted far - field radiation patterns ( dashed blue line ) with the ones calculated by the hfss simulation ( solid red line ) ; all radiation patterns are normalized to their maximum .
although some discrepancies between the simulation results and the theoretical predictions are observed in fig .
[ fig : linesourceresults](d)-(l ) , it is clear that the designed hms successfully convert the line source fields to directive radiation toward the desirable angle . both the beam - width and the immediate side lobe levels are in a good agreement with the semi - analytical theory . in all three cases ,
the simulated directivity values outside the main beam ( in region 2 ) are at least @xmath121 below the peak directivity .
importantly , the theoretical calculations indicate that the slowly - varying condition is indeed satisfied for all transmission angles considered , except at some points towards the edge of the metasurface , where the exclusion of diffraction effects introduce a discontinuity in the fields along the @xmath29 axis ( not shown ) .
another support for the validity of the theoretical results is provided by the fact that the absolute values of the fields as presented in fig .
[ fig : linesourceresults](g)-(i ) and fig .
[ fig : linesourceresults](j)-(l ) in @xmath122 units , are to scale .
table [ tab : line_source_hms_performance ] concentrates performance parameters calculated from the theoretical and simulated radiation patterns of fig .
[ fig : linesourceresults](d)-(f ) .
these include the transmission efficiency , i.e. the ratio between the power transmitted to region 2 and the total power radiated by the source ; the half - power beamwidth ( hpbw ) , i.e. the angular difference between the half - power points ; the aperture efficiency , i.e. the ratio between the hpbw of the hms radiation and the hpbw of a uniformly excited aperture with the same length @xmath123 @xcite ; and the ( 2d ) peak directivity @xcite
. for each parameter , table [ tab : line_source_hms_performance ] indicates its value as calculated from the simulation results and the theoretical predictions , and the respective relative performance , defined as the ratio between the two . in consistency with the results presented in fig .
[ fig : linesourceresults](d)-(l ) , the performance parameters also indicate that the power radiated into region 2 is successfully funnelled into a directive beam , with performance comparable with the theoretical predictions , peak directivity excepted . in view of the theoretically predicted values themselves ,
two comments are in place .
first , we refer to the predicted transmission efficiencies , which are calculated to be around @xmath124 . due to the finite length @xmath123 of the hms , only of the line source power
interacts with the hms . if we assume minor reflections from the metasurface , and consider @xmath125 as in our case , this rough estimation leads to a transmission efficiency of @xmath126 , very close to the values in table [ tab : line_source_hms_performance ]
hence , these values indicate that if the hms implementation would perfectly match its design , most of the power interacting with the metasurface is expected to be transmitted to region 2 . in other words ,
the reflection coefficient resulting from the enforcement of local impedance , of which we have little control , should not deteriorate significantly the hms performance .
the second comment refers to the aperture efficiencies presented in table [ tab : line_source_hms_performance ] , which do not exceed @xmath127 for all transmission angles considered , even in the ideal theoretical scenario .
the reason for these values is the utilization of local power conservation in the hms design , that coerces the virtual aperture window function to follow the profile of the total ( incident+reflected ) fields at @xmath128 .
as the line source is situated only @xmath129 away from the metasurface in our configuration , most of the incident power is concentrated in a small region near @xmath108 .
thus , effectively , the full length of the hms can not be utilized for radiation , and , in turn , the theoretical limit for the aperture efficiency is below @xmath130 . to conclude this subsection
, we note that both the results presented in table [ tab : line_source_hms_performance ] and in fig .
[ fig : linesourceresults](d)-(l ) indicate that the main deviations from the theoretical predictions are the increased reflections from the metasurface in region 1 and the incomplete elimination of the incident field in region 2 ; both discrepancies contribute to a significant difference in peak directivity ( table [ tab : line_source_hms_performance ] ) .
we believe that these differences originate in the fact that the unit cells implementing the hms have not yet been optimized to exhibit the prescribed surface admittance and impedance accurately over the entire required dynamical range .
however , the optimization of the metasurface unit cells requires specialized treatment , including scattering element selection @xcite , and is outside the scope of this paper . to further illustrate the versatility of our formulation , we consider a third excitation configuration , that of an electric line source positioned in front of a pec
, the latter is separated from the hms by a distance @xmath131 ( fig .
[ fig : physicalconfigurationthreesources](c ) ) ; as in subsection [ subsec : results : line_source ] , the line source current is given by @xmath132 .
we present this configuration herein for two reasons : first , to demonstrate how scenarios including multiple reflections ( more generally , plane - stratified configurations ) can be treated using the design procedure presented in section [ sec : theory ] ; second , to provide an example for virtual aperture engineering via careful selection of the source excitation .
when region 1 contains not only sources but also scatterers ( e.g. , abrupt interfaces ) the field expressions must take into account the boundary conditions induced by these scatterers , on top of the source condition already encountered in subsection [ subsec : results : line_source ] .
the enforcement of these boundary conditions gives rise to multiple - reflection terms in the spectral response of the fields @xcite , which in turn form a dependency between the source - related spectrum @xmath42 of - and the reflection coefficient of the hms . in the configuration considered herein ( fig .
[ fig : physicalconfigurationthreesources](c ) ) the boundary conditions introduced by the scatterers at @xmath133 require that the tangential electric field vanishes on the pec , i.e. @xmath134 for each @xmath29 .
combining this requirement with the source condition at @xmath135 yields the following expressions for the incident and reflected fields - @xcite @xmath136 where @xmath137 , @xmath138 , and the upper and lower signs in should be used when @xmath35 or @xmath139 , respectively . as in
, this change of signs for the regions below and above the source provides the required discontinuity in the magnetic field at @xmath135 due to the electric current ( source condition ) .
we emphasize once more that the formulation of - is valid for any reflection coefficient dependency @xmath43 , retaining this degree of freedom required to employ our design procedure .
the fraction in the braces of the integrands accounts for the reflection from the pec , and its numerator vanishes when @xmath140 as required .
its denominator corresponds to the multiple reflections taking place between the hms and the pec@xcite ; consequently , the poles of the integrands correspond to guided or leaky modes of this structure @xcite .
moreover , although the source - related spectrum is now dependent on the reflection coefficient , it can be readily verified that when @xmath141 the incident field in - is reduced to the field produced by a line source and its image , positioned symmetrically @xmath142 below the pec . as in subsubsection
[ subsubsec : results : line_source : design ] , we compare - with - in the _ source - free region _
@xmath15 ( i.e. , where @xmath143 and @xmath144 ) to extract the source - related spectrum .
this results in @xmath145 a careful examination of the derivation in section [ sec : theory ] reveals that the dependency of the source - related spectrum in the reflection coefficient does not affect the conditions for local impedance ( table [ tab : design_rules ] / step 1 ) , as in the source - free region , this dependency is the same for all fields ( e.g. , - ) .
this means that enforcing local impedance on the fields - in that region will result in the same expression for the reflection coefficient . utilizing this
, the incident and reflected fields can be completely evaluated in region 1 ( @xmath146 ) , and the dimensionless total field at @xmath61 will thus be given by , reading @xmath147 } { j\beta\cos\left(\beta d\right)-k\cos\theta_0\sin\left(\beta d\right)}e^{jk_tz}. \label{equ : line_source_pec_spectral_domain_f_minus}\end{aligned}\ ] ] as the poles of @xmath148 and the integrand of @xmath149 coincide , the pole contributions to the integral indicate the guided and leaky modal fields in region 1 @xcite . however , as these poles are complex ( @xmath150 ) , their presence do not , in general , introduce significant difficulties to numerical evaluation of the integral .
hence , step 1 of table [ tab : design_rules ] may be completed by evaluating @xmath151 and @xmath152 from . as in subsubsection
[ subsubsec : results : line_source : design ] we follow [ table [ tab : design_rules ] / step 2 ] and [ table [ tab : design_rules ] / step 3 ] to establish the shape of the virtual aperture and the phase compensation of the hms , respectively , leading to the formulation of the hms surface impedance and surface admittance @xmath153 \\ \!\!\!\!{{\vphantom{y}}^{e}_{}{y}}_{sm}\left(z\right)=-j\dfrac{{{\vphantom{y}}^{e}_{}{y}}_0}{2 } \cot\left[\dfrac{kz\left(\sin\theta_0-\sin\theta_i\left(z\right)\right)+\xi_0}{2}\right ] , \end{array}\!\!\!\!\ ! \right .
\label{equ : line_source_pec_impedance_admittance_final}\end{aligned}\ ] ] where we have used again the definition of the equivalent angle of incidence . executing these steps
enable the assessment of the transmitted fields as well .
as the total field at @xmath61 is now an outcome of multiply - reflected field interference , it seems that controlling the shape of the virtual aperture window function becomes an even more difficult task .
however , the introduction of the pec reflector actually enhances our ability to control this function .
as discussed in subsubsection [ subsubsec : results : line_source : aperture ] , in many cases the total field on the lower facet of the hms can be approximated by the incident field .
when the pec is present , the incident field is created by an interference between the source and its image .
therefore , by controlling the relative position of the source and the pec , we may affect this interference pattern , and subsequently the variation of the total field magnitude at @xmath61 . as an example
, we may utilize the fact that in the absence of the hms , the distance between the source and the pec , @xmath154 , is related to the internal angle @xmath155 in which constructive interference occurs in region 1 ( fig .
[ fig : physicalconfigurationthreesources](c ) ) via @xcite @xmath156 formation of two lobes travelling towards @xmath157 within region 1 should , in general , broaden the effective interaction length of the source field with the hms on its lower facet , which , in turn , may enhance its aperture efficiency .
as shall be demonstrated in the following subsubsection , can be used as an initial aperture engineering step , by which the suitable hms - pec distance @xmath131 is determined for given source position @xmath158 and desirable constructive interference direction @xmath155 . in front of a pec at @xmath159 to directive radiation towards @xmath97 .
( a ) required surface reactance @xmath160 ( blue dashed line ) and surface susceptance @xmath161 ( red solid line ) calculated from .
( b ) theoretically predicted ( blue dashed line ) and hfss - simulated ( red solid lines ) normalized radiation pattern ( db scale ) .
( c ) real part of the electric field phasor @xmath115 as simulated by hfss .
( d ) theoretical ( semi - analytical ) prediction of @xmath115.,width=207 ] to verify the design procedure for the configuration considered in this subsection , we have designed and simulated a huygens metasurface according to the prescribed procedure ( with @xmath95 ) , designated to convert the fields produced by an electric line source at @xmath162 , positioned in front of a pec at @xmath163 , to directive radiation towards @xmath97 ( fig .
[ fig : linesourcepecresults ] ) . following the discussion in subsubsection [ subsubsec : results : line_source_pec : aperture ] , we aimed at harnessing the pec reflector to improve the theoretical limit for aperture efficiency for a line source @xmath129 below the hms , calculated in subsubsection [ subsubsec : results : line_source : results ] to be @xmath164 for @xmath97 . to that end
we positioned the pec such that constructive interference would take place at @xmath165 in region 1 , yielding , utilizing with @xmath166 , an hms - pec distance of @xmath167 .
although other values of @xmath155 also broaden the effective interaction length of the incident field and the hms , we have found using the semi - analytically estimated fields ( appendix [ app : asymptotic_evaluation ] ) that @xmath165 yields an optimal result for the @xmath168-long hms under consideration .
[ fig : linesourcepecresults](a ) presents the surface reactance and surface susceptance required to implement the desirable hms . as in subsubsection
[ subsubsec : results : line_source : results ] , the design procedure output was used to implement the hms in ansys hfss ( following the same procedure and unit cell structure , _ cf .
_ appendix [ app : hfss_implementation ] ) , as well as to assess the predicted reflected and transmitted fields ( following the semi - analytical approach of appendix [ app : asymptotic_evaluation ] ) .
[ fig : linesourcepecresults](b ) compares between the normalized radiation patterns evaluated using the hfss simulation ( red solid line ) and the semi - analytical theoeretical calculations ( blue dashed line ) , showing a very good agreement between the theory and simulation for the angular range @xmath169 . as implied by the simulated ( fig .
[ fig : linesourcepecresults](c ) ) and theoretical ( fig .
[ fig : linesourcepecresults](d ) ) field plots , most of the discrepancy at large angles ( @xmath170 ) originate from the fact that the semi - analytical approximations neglect the contribution of fields incident at the plane @xmath1 outside the metasurface @xmath171 to the radiation in region 2 ( appendix [ app : asymptotic_evaluation ] ) .
[ fig : linesourcepecresults](c)-(d ) highlight two additional properties of the line - source / pec configuration .
first , these two subfigures indicate in a clear manner that the profile of the virtual aperture @xmath63 indeed follows the total field magnitude at @xmath61 .
the two spots formed on the lower facet of the hms due to the interference between the line source and its image ( subsubsection [ subsubsec : results : line_source_pec : aperture ] ) are clearly translated into two dominant beams originating from the same positions on the upper facet .
second , as discussed briefly in subsubsection [ subsubsec : results : line_source_pec : design ] , the expression for the total field in region 1 contains complex poles , and indeed the field plots indicate that the examined structure supports leaky modes .
this is particularly pronounced in the theoretical prediction fig .
[ fig : linesourcepecresults](d ) , as the semi - analytical formulation used assumes the hms is of infinite extent over the @xmath29 axis when evaluating the fields in region 1 ( appendix [ app : asymptotic_evaluation ] ) , thus facilitating the observed long - range guidance of moderately leaking modes .
l|c c c & hfss & theory & relative performance + + half - power beam width & @xmath172 & @xmath172 & @xmath130 + aperture efficiency & @xmath173 & @xmath173 & @xmath130 + peak directivity & @xmath174 & @xmath175 & @xmath164 + table [ tab : line_source_pec_hms_performance ] summarizes the performance of the els / pec excited hms as a directive radiator . in consistency with fig .
[ fig : linesourcepecresults](b ) , the theoretically predicted and numerically simulated hpbw perfectly agree .
as anticipated in subsubsection [ subsubsec : results : line_source_pec : aperture ] , the aperture efficiency has indeed increased from @xmath164 for the els excited hms ( table [ tab : line_source_hms_performance ] ) to @xmath173 after introduction of the pec , due to the utilization of the interference between the source and its image to broaden the incident field profile .
an enhanced directivity is also recorded , where we note that when the pec is present no power is lost due to radiation to region 1 .
it should be also noted that the theoretically predicted directivity is clearly overestimated , as is does not account for the fraction of the source power reaching region 2 without interacting with the hms ( i.e. , via @xmath171 , _ cf .
_ appendix [ app : asymptotic_evaluation ] ) , which has non - negligible contribution , as the numerical simulations reveal ( fig . [ fig : linesourcepecresults](c ) ) .
we have presented a formulation of a design procedure for scalar huygens metasurface directive radiators , applicable for an arbitrary 2d source excitation .
our derivation reveals that satisfaction of two physical conditions is sufficient to guarantee that the designed hms is passive and lossless : local power conservation and local impedance . by expressing the incident ,
reflected and transmitted fields via their plane - wave spectrum , and utilizing the slowly - varying approximation , we have shown that enforcing local impedance results in a fresnel - like reflection for the various spectral components . furthermore , enforcing local power conservation dictates that the profile of the virtual aperture forming the transmitted directive radiation follows the magnitude of the total ( incident and reflected ) excitation fields . at the end of the design procedure ,
the fields at all regions may be assessed semi - analytically , and the required variation of the surface impedance and surface admittance along the hms is prescribed .
we have verified our formulation using three different source configurations , showing good agreement between the semi - analytical predictions and finite - element simulation results executed by implementing the hms as consecutive unit cells in ansys hfss , yet to be optimized in a dedicated future work
. means to control the virtual aperture via modification of the source excitation were discussed and demonstrated as well .
the proposed design procedure establishes a flexible and robust foundation for the design and verification of novel antennas , allowing exploration of a vast variety of excitation forms via a single uniform formalism .
moreover , the derivation provides insight regarding the physical requirements to achieve passive lossless huygens metasurfaces with desirable functionality , shedding light on previously investigated hmss as well as indicating possible directions for future hms applications .
for completeness , we provide here the final results of the derivation of the surface reactance required to implement the dual hms , converting an arbitrary tm - polarized source field to directive radiation towards @xmath53 ( derivable by duality @xcite ) .
as denoted in subsection [ subsec : theory : formulation ] , when the excitation field is tm - polarized , the nonvanishing field components are @xmath176 , @xmath177 , and @xmath178 ; the scalar surface impedance only induces _ electric _ currents in the @xmath29 direction ; and the scalar surface admittance only induces _ magnetic _ currents in the @xmath28 direction .
analogously to - , the tangential magnetic fields in the _ source - free region _
@xmath35 can be generally formulated in the spectral domain as @xmath179 and the respective tangential electric fields can be derived via ; the @xmath180 left superscript denotes tm - hms related quantities .
the scalar surface impedance and admittance may be written in analogy to as @xmath181}{h_y^{\mathrm{trans}}\left(0,z\right)-\left[h_y^{\mathrm{inc}}\left(0,z\right)+h_y^{\mathrm{ref}}\left(0,z\right)\right ] } \\ \!\!\!\!\!{{\vphantom{y}}^{m}_{}{y}}_{sm}\left(z\right)\!=\!\dfrac{1}{2 } \dfrac{h_y^{\mathrm{trans}}\left(0,z\right)+\left[h_y^{\mathrm{inc}}\left(0,z\right)+h_y^{\mathrm{ref}}\left(0,z\right)\right]}{e_z^{\mathrm{trans}}\left(0,z\right)-\left[e_z^{\mathrm{inc}}\left(0,z\right)+e_z^{\mathrm{ref}}\left(0,z\right)\right]}. \end{array}\!\!\!\!\!\!\ ! \right .
\label{equ : tm_impedance_admittance_discontinuity}\end{aligned}\ ] ] the transmitted tangential fields are defined using a virtual aperture window function @xmath182 , similarly to , @xmath183 which we assume to satisfy the slowly - varying condition as @xmath56 ( with @xmath40 superscript replaced by @xmath180 ) .
the dimensionless field quantities are defined for tm - polarized source field as @xmath184 \\
\vspace{5pt } { { \vphantom{f}}^{m}_{}{\!f}}_{\!\!h}^{-}\left(z\right ) \triangleq -\dfrac{1}{i_0k } \left[h_y^{\mathrm{inc}}\left(0,z\right)+h_y^{\mathrm{ref}}\left(0,z\right)\right ] \\ { { \vphantom{f}}^{m}_{}{\!f}}^{+}\left(z\right ) \triangleq { { \vphantom{w\left(0,z\right)}}^{m}_{}{w\left(0,z\right)}}e^{-jkz\sin\theta_0 } \end{array}\!\!\!\!\!\ !
\label{equ : tm_dimensionless_quantities}\end{aligned}\ ] ] using which local impedance and local power conservation conditions retain the same formulae as in and , with the @xmath40 superscript replaced by @xmath180 , likewise defining the analogue of .
finally , the surface reactance for the tm case is given by @xmath185 \\ \!\!\!\!{{\vphantom{y}}^{m}_{}{y}}_{sm}\left(z\right)=-j\dfrac{{{\vphantom{y}}^{m}_{}{y}}_0}{2 } \cot\left[\dfrac{\varphi_{-}\left(z\right)-\varphi_{+}\left(z\right)}{2}\right ] , \end{array}\!\!\!\!\ !
\label{equ : tm_impedance_admittance_final}\end{aligned}\ ] ] where @xmath186 is the wave impedance of a tm - polarized plane - wave propagating in region 2 at an angle of @xmath53 with respect to the @xmath9 axis ( see ) .
despite the different wave impedance for te and tm polarizations , the reflection coefficient arising from local impedance remains the same .
consequently , explicit evaluation of the scattered fields and the phases @xmath82 may be obtained by using - with the @xmath40 left superscript replaced by @xmath180 .
the huygens metasurface designs investigated in subsubsections [ subsubsec : results : line_source : results ] and [ subsubsec : results : line_source_pec : results ] were simulated using a commerically available finite - element solver ( ansys hfss ) to verify the theoretical predictions ( calculated using the continuous equivalent surface impedance and admittance ) via comparison to a more realistic implementation of the hms .
-axis while the electric dipole formed by the loaded wires is oriented parallel to the @xmath28-axis.,width=245 ] the simulated @xmath168-long hmss were implemented using one - hundred @xmath112-long unit cells , where the frequency of the time - harmonic excitations was @xmath114 .
[ fig : unitcell ] presents the physical configuration of a unit cell , along with its dimensions .
each unit cell is composed of two square pec loops , loaded by identical lumped capacitors , and two pec wires , loaded either by identical lumped capacitors or by identical lumped inductors , positioned symmetrically with respect to the center of the cell .
the magnetic dipole formed by the loaded loops is oriented parallel to the @xmath29-axis while the electric dipole formed by the loaded wires is oriented parallel to the @xmath28-axis , indicating that the unit cell is only sensitive to te - polarized fields ( see subsubsection [ subsubsec : theory : formulation : spectral_decomposition ] ) .
we simulate a 2d environment by placing two pec surfaces at @xmath187 .
the surface impedance and admittance of a unit cell for a given lumped loading is evaluated by simulating the response of an infinite periodic array of identical unit cells to a normally incident ( te - polarized ) plane - wave excitation . utilizing the impedance matrix calculated by hfss for the scattering problem , and the circuit model introduced in @xcite ,
we extract the equivalent surface impedance and admittance of the cell .
variation of the lumped loading of the wires and loops induces , respectively , variation of the equivalent surface impedance and admittance of the cell @xcite .
this facilitates the generation of a lookup table , matching pairs of surface reactance and susceptance to sets of lumped capacitors or inductors . invoking the principle of local periodicity , a required polarizability variation such as the ones in fig .
[ fig : linesourceresults](a)-(c ) and fig .
[ fig : linesourcepecresults](a ) can be sampled at a suitable unit - cell - length period ( @xmath112 in our case ) and translated using the lookup table into a set of unit cells with prescribed lumped loadings @xcite .
although the electric and magnetic dipoles are orthogonal , and ideally should exhibit decoupled responses to orthogonal electric and magnetic fields , we discovered that due to the finite size of the structures , some coupling does exist .
this is especially pronounced when the lumped loading is such that the element is near resonance at the working frequency @xmath188 .
thus , to enhance the accuracy of the lookup table , we have chosen to simulate the unit cell as a whole and to relate pairs of reactance / susceptance to pairs of lumped elements .
to establish an effective engineering methodology based on the design procedure introduced in section [ sec : theory ] , it is desirable to develop computationally efficient tools to estimate the hms performance before resorting to numerical simulation tools for final optimization . in subsubsection [ subsubsec :
results : line_source : results ] and subsubsection [ subsubsec : results : line_source_pec : results ] we have employed such tools , in the form of semi - analytical formulae , to assess the agreement between simulated results of detailed design and the predictions of the more idealized theory . in this appendix
we describe the details regarding the evaluation of these semi - analytical formulae , as well as the key approximations used . formally , the theoretical derivation of section [ sec : theory ] is valid only for infinitely long hmss .
if the hms is finite , the problem is no longer uniform ( the configuration is not separable ) , and the spectral analysis employed herein can not be carried out . nevertheless , in practice , if an efficient conversion of the source power to directive radiation is to be achieved , the hms length @xmath123 should be judiciously chosen such that most of the interaction between the source power and an infinitely long hms takes place over a finite region @xmath119 of the @xmath189 plane . assuming this is indeed the case ( _ assumption 1 _ ) , we may treat the hms as infinite over the @xmath29 axis without introducing significant errors with respect to the realistic finite - length implementation .
this , in turn , facilitates the utilization of to evaluate the fields in region 1 , disregarding any effects arising from the edges of the hms , e.g. diffraction or a discontinuity in the reflection coefficient .
a second assumption essential for the validity of our derivation is that the magnitude of the total field impinging upon the hms at @xmath61 , including the reflection induced by local impedance , varies moderately such that the slowly - varying condition is satisfied . utilizing this assumption ( _ assumption 2 _ )
, we may use - to evaluate the fields on the upper facet of the hms ( @xmath63 ) .
however , when evaluating the fields in region 2 , we would like to account for the fact that the hms is of finite length @xmath123 , which may have significant effects on the radiated fields . hence , we truncate the virtual aperture window function @xmath117 at the edges of the implemented hms ; practically , the integral of is executed with the infinite limits replaced by @xmath190 .
this is equivalent to assuming that the fields at the hms plane @xmath1 vanish in the region where the hms is absent @xmath171 .
_ assumption 1 _ above facilitates this approximation .
these assumptions are used to evaluate the spectral content of the reflected and transmitted fields , and consequently , via the spectral integrals - , the fields everywhere above and below the hms ( fig .
[ fig : linesourceresults](j)-(l ) and fig .
[ fig : linesourceresults](d ) ) . nonetheless , when far - field evaluation is required , such as for the assessment of the radiation pattern , straightforward spectral integration is problematic due to the highly oscillating nature of the radiation integrals as @xmath191 . hence , for the evaluation of far - field radiated power we apply the steepest - descent - path method , yielding closed - form asymptotic approximations for the spectral integrals .
for the typical spectral integrals - @xmath192 the asymptotic evaluation at a point @xmath193 where @xmath194 and @xmath195 $ ] is given by @xcite @xmath196 where @xmath197 for @xmath45 and @xmath198 for @xmath2 , assuming the sign attached to @xmath199 in the exponent of satisfies the radiation condition at the respective regions .
a.e . gratefully acknowledges the support of the lyon sachs postodoctoral fellowship foundation as well as the andrew and erna finci viterbi fellowship foundation of the technion - israel institute of technology , haifa , israel .
d. sievenpiper , l. zhang , r. f. j. broas , n. alexopolous , and e. yablonovitch , `` high - impedance electromagnetic surfaces with a forbidden frequency band , '' _ ieee trans .
theory techn .
_ , vol .
47 , no . 11 , pp . 20592074 , nov .
1999 . c. l. holloway , e. f. kuester , j. a. gordon , j. ohara , j. booth , and d. r. smith , `` an overview of the theory and applications of metasurfaces : the two - dimensional equivalents of metamaterials , '' _ ieee antennas propag . mag .
_ , vol .
54 , no . 2 ,
pp . 1035 , apr .
2012 .
e. kuester , m. mohamed , m. piket - may , and c. holloway , `` averaged transition conditions for electromagnetic fields at a metafilm , '' _ ieee trans .
antennas propag .
_ , vol .
51 , no .
26412651 , oct .
2003 .
e. martini , g. sardi , and s. maci , `` homogenization processes and retrieval of equivalent constitutive parameters for multisurface - metamaterials , '' _ ieee trans .
antennas propag .
_ , vol .
62 , no . 4 , pp . 20812092 , apr . 2014 .
m. selvanayagam and g. v. eleftheriades , `` discontinuous electromagnetic fields using orthogonal electric and magnetic currentsfor wavefront manipulation , '' _ opt . express _
21 , no . 12 ,
pp . 1440914429 , june 2013 .
y. g. rapoport , s. tretyakov , and s. maslovski , `` nonlinear active huygens metasurfaces for reflectionless phase conjugation of electromagnetic waves in electrically thin layers , '' _
j. electromagnet . wave .
_ , vol . 27 , no . 11 , pp .
13091328 , 2013 .
c. holloway , m. mohamed , e. kuester , and a. dienstfrey , `` reflection and transmission properties of a metafilm : with an application to a controllable surface composed of resonant particles , '' _ ieee trans .
_ , vol .
47 , no . 4 , pp . 853865 , 2005 .
c. l. holloway , d. c. love , e. f. kuester , j. a. gordon , and d. a. hill , `` use of generalized sheet transition conditions to model guided waves on metasurfaces / metafilms , '' _ ieee trans .
antennas propag .
_ , vol . 60 , no . 11 , pp .
51735186 , 2012 .
a. epstein , `` rigorous electromagnetic analysis of optical emission of organic light - emitting diodes , '' ph.d .
dissertation , dept .
, technion - israel insititute of technology , haifa , israel , 2013 .
[ online ] .
available : http://arielepstein.webs.com/research/thesis.pdf g. lovat , p. burghignoli , f. capolino , d. jackson , and d. wilton , `` analysis of directive radiation from a line source in a metamaterial slab with low permittivity , '' _ ieee trans .
antennas propag .
_ , vol . 54 , no . 3 , pp . 10171030 ,
ariel epstein ( s12-m14 ) received the b.a .
degree in computer science from the open university of israel , raanana , the b.a .
degree in physics , and the b.sc . and ph.d .
degrees in electrical engineering from the technion - israel institute of technology , haifa , israel , in 2000 , 2003 , and 2013 respectively .
he is currently a lyon sachs postdoctoral fellow at the department of electrical and computer engineering in the university of toronto , ontario , canada .
his current research interests include utilization of electromagnetic theory , with emphasis on analytical techniques , for novel applications , e.g. electromagnetic metamaterials for antennas and microwave components .
epstein was the recipient of the young scientist best paper award in the ursi commission b international symposium on electromagnetic theory ( emts2013 ) , held in hiroshima , japan on may 2013 , as well as the best poster award at the 11th international symposium on functional @xmath200-electron systems ( f@xmath200 - 11 ) , held in arcachon , france on june 2013 .
george v. eleftheriades ( s86-m88-sm02-f06 ) earned the m.s.e.e . and ph.d .
degrees in electrical engineering from the university of michigan , ann arbor , mi , usa , in 1989 and 1993 , respectively , where he was involved in the development of sub - millimeter - wave and thz technology sponsored by nasa . from 1994 to 1997 , he was with the swiss federal institute of technology , lausanne , switzerland , where he was engaged in the design of millimeter and sub - millimeter - wave receivers for the european space agency .
currently , he is a professor in the department of electrical and computer engineering at the university of toronto , on , canada , where he holds the canada research chair in nano- and micro - structured electromagnetic materials and applications and the velma m. rogers graham chair in engineering .
he is a recognized international authority and pioneer in the area of negative - refractive - index metamaterials .
these are man - made materials which have electromagnetic properties not found in nature .
he introduced a method for synthesizing metamaterials using loaded transmission lines .
together with his graduate students , he provided the first experimental evidence of imaging beyond the diffraction limit and pioneered several novel microwave components and antennas using these transmission - line based metamaterials .
his research has impacted the field by demonstrating the unique electromagnetic properties of metamaterials ; used in lenses , antennas , and other microwave components to drive innovation in fields such as defence , medical imaging , microscopy , automotive radar , and wireless telecommunications .
presently , he is leading a group of graduate students and researchers in the areas of electromagnetic negative - refraction metamaterials , transformation optics , metasurfaces , small antennas and components for broadband wireless communications , novel antenna beam - steering techniques , plasmonic and nanoscale optical components , and fundamental electromagnetic theory .
eleftheriades served as an associate editor for the ieee transactions on antennas and propagation ( ap ) .
he also served as a member of the ieee ap - society administrative committee ( adcom ) from 2007 to 2012 and was an ieee ap - s distinguished lecturer from 2004 to 2009 .
he served as the general chair of the 2010 ieee international symposium on antennas and propagation held in toronto , on , canada .
papers that he coauthored have received numerous awards such as the 2009 best paper award from the ieee microwave and wireless propagation letters , twice the r. w. p. king best paper award from the ieee transactions on antennas and propagation ( 2008 and 2012 ) , and the 2014 best paper award from the ieee antennas and wireless propagation letters .
his work has been cited more than 10,000 times , and his h - index is 47 .
he received the ontario premier s research excellence award and the university of toronto s gordon slemon award , both in 2001 . in 2004
he received an e.w.r .
steacie fellowship from the natural sciences and engineering research council of canada .
he received the 2008 ieee kiyo tomiyasu technical field award `` for pioneering contributions to the science and technological applications of negative - refraction electromagnetic materials . '' in 2009 , he was elected a fellow of the royal society of canada . | we present a formulation of the interaction between a given source field and a scalar huygens metasurface ( hms ) , a recently introduced promising concept for wavefront manipulation based on a sheet of orthogonal electric and magnetic dipoles . utilizing the equivalent surface impedance representation of these metasurfaces , we establish that an arbitrary source field can be converted into directive radiation via a passive lossless hms if two physical conditions are met : local power conservation and local impedance . expressing the fields via their plane - wave spectrum and harnessing the slowly - varying approximation
we obtain semi - analytical formulae for the scattered fields , and prescribe the surface reactance required for the metasurface implementation .
the resultant design procedure indicates that the local impedance induces a fresnel - like reflection , while local power conservation forms a radiating virtual aperture which follows the total excitation field magnitude .
the semi - analytical predictions are verified by finite - element simulations of hmss designed for different source configurations . besides serving as a flexible design procedure for hms radiators
, the proposed formulation also provides a mechanism to incorporate a variety of source configurations into general hms models , as well as physical insight on the conditions enabling purely reactive implementation of this novel type of metasurfaces .
epstein and eleftheriades : passive lossless huygens metasurfaces for directive radiation metasurfaces , huygens sources , wavefront manipulation , plane - wave spectrum . |
a nanopore is simply a hole of small size in an impermeable membrane separating two regions containing an electrolytic buffer .
a size range of 1100 nm is fairly typical .
living cells and intracellular organelles are usually bounded by lipid membranes containing nanopores constructed of membrane - bound proteins .
the transport of small molecules and polymers across such nanopores is a very common feature in living cells and is essential to their normal function @xcite .
synthetic nanopores @xcite have been the focus of much interest in recent years following the demonstration of their use as effective single molecule sensors @xcite .
the main distinguishing feature of nanopore systems responsible for many of the novel effects is that their geometric dimensions are small enough so that electrokinetic effects are important .
such effects have been invoked to explain a range of observations relating to experiments involving free as well as hindered translocation of dna across synthetic nanopores @xcite .
nanopores also exhibit other unusual properties , some of which could be potentially exploited to build novel microfluidic devices .
for example , conical nanopores fabricated on plastic films using the ion track - etching technique @xcite have been shown to exhibit ion current rectification similar to that of semiconductor diodes @xcite .
a similar effect has been reported recently for electroosmotic flow out of nanocapillaries @xcite . a nonlinear electrokinetic effect known as induced charge electroosmosis ( iceo )
@xcite produces vortices at the edges of nanopores resembling recirculation vortices in separated flows even though the reynolds number in such applications is essentially zero . mixing due to these flow structures @xcite together with electroconvective instabilities @xcite are thought to be responsible for the `` overlimiting '' behaviour of the current voltage characteristics of perm selective pores and membranes described by @xcite .
similar vortical structures may be generated in cylindrical channels that undergo a sudden constriction when the relevant length scales are on the order of the debye length @xcite .
if an electric field @xmath0 is applied across an uncharged membrane , induced charge electroosmosis leads to velocities @xmath1 with no net flow through the membrane unless symmetry is somehow broken .
molecular dynamics simulations of flow through nanopores in uncharged membranes , such as graphene sheets @xcite , show that differences in mobility between cations and anions can result in asymmetric debye layers and consequent net flow through the membrane .
symmetry is also broken if the membrane is charged , so that the intrinsic field due to the membrane competes with the externally applied field in determining the distribution of ions in the debye layer @xcite .
if a voltage is applied across a charged membrane containing a long narrow pore an electroosmotic flow is generated by the electric field acting on the charge cloud of counter - ions in the fluid adjacent to the fixed charges at the solid / fluid interface .
the strength of this flow is proportional to the applied electric field , and thus , inversely proportional to the membrane thickness if the membrane is sufficiently thick .
however , the flow does not increase indefinitely as the membrane is progressively thinned .
if the membrane is much thinner than the diameter of the hole , the flow is driven mainly by electroosmosis at the membrane surface exterior to the pore , rather than by electric forces within the pore itself . here
we present results for the flow rate through a pore in a charged membrane in the limit of a weak applied field . in section [ sec : zerothick ]
we use the reciprocal theorem to calculate the flow rate through a circular hole in a charged membrane of zero thickness .
the result is obtained in terms of an integral which in general has to be evaluated numerically but can be obtained analytically when the pore size is much smaller than the debye length . in section [ sec : fullsim ] we present computer simulations of the full problem based on numerical solutions of the poisson - nernst - planck - stokes system of equations using a finite volume method .
both thick and thin membranes are considered and compared with analytical results for membranes of zero thickness and of thickness large in comparison to the pore radius .
conclusions are provided in section [ sec : conclusion ] .
we consider a hole ( of arbitrary shape ) in an infinite plane membrane immersed in an incompressible homogeneous electrolyte containing @xmath2 ionic species ( figure [ fig : sketch ] ) .
the number density of the @xmath3th ionic species is @xmath4 , with @xmath5 in the bulk electrolyte far from any charged surfaces .
the electrolyte has viscosity @xmath6 and electrical permittivity @xmath7 . a surface charge of fixed density @xmath8 exists at the membrane / electrolyte interface , and in the electrolyte adjacent to the membrane there is charge cloud of counter ions , with thickness characterized by the debye length @xmath9 where @xmath10 is the boltzmann constant , @xmath11 the absolute temperature , @xmath12 the proton charge , and @xmath13 the valence of the @xmath3th species of ion .
we assume that the surface charge density @xmath8 is sufficiently small so that the poisson - boltzmann equation describing the potential @xmath14 in the equilibrium charge cloud may be linearized .
far from any hole in the membrane , the zeta potential at the surface of the membrane is @xmath15 , with @xmath16 .
an electric potential difference @xmath17 is applied across the membrane with a resulting current @xmath18 .
the hole within the membrane has characteristic size @xmath19 , and we make no assumption concerning @xmath20 , the ratio of the hole size to the debye length @xmath21 . when @xmath22 , the charge clouds from opposite sides of the perimeter of the hole overlap , but this has little effect upon the electrical conductivity of the hole when ( as here ) the surface charge density @xmath8 and the resulting perturbations to the ionic number densities @xmath23 are small .
similarly , any ion exclusion effects of the overlapping charge cloud are negligible .
we also note that the ion exclusion properties of a thin membrane are in general smaller than those of a long cylindrical pore .
when @xmath24 , the potential in the unperturbed electrical double layer within the pore , on the plane of the membrane , is @xmath25 . inside a uniform cylindrical
pore with surface charge density @xmath8 , the potential when @xmath22 is @xmath26 , and the condition @xmath27 required for the perturbation of the ionic number densities to be small within the pore implies a smaller charge density @xmath8 for the pore than for the membrane .
only when @xmath28 will the long cylindrical pore and the hole in a membrane exhibit similar ion exclusion properties . when @xmath29 the charge cloud is thin compared to the lateral dimension of the hole , but it remains thick compared to the membrane of zero thickness @xmath30 considered in [ subsec : roundhole ] .
thus , we are unable to appeal to smoluchowski s analysis for thin charge clouds in
[ subsec : roundhole ] .
ion exclusion effects are negligible in this limit . in [ subsec : charge_perturbation ]
we discuss how the charge cloud is deformed by the applied electric field and by fluid motion .
we then ( [ subsec : formalism ] ) describe a theoretical framework for calculating the flow rate @xmath31 through the hole , exploiting the reciprocal theorem : the analysis is similar to that of @xcite . in [ subsec : roundhole ] we consider the special case of a circular hole in a thin membrane for which the integral for the flow rate @xmath31 can be computed numerically .
.,scaledwidth=80.0% ] when the electrical potential difference @xmath32 is applied across the membrane , the charge cloud adjacent to the surface of the membrane is perturbed , both by the direct electrical field @xmath33 acting on the ions , and by motion of the fluid .
ions are convected with the fluid velocity @xmath34 , and move relative to the fluid under the influence of electric fields and thermal diffusion .
the conservation equation for the number density @xmath4 of the @xmath3th ionic species , in steady state , is therefore @xmath35 where @xmath36 is the mobility of the @xmath3th species of ion .
we follow @xcite and nondimensionalise potentials by @xmath37 , lengths by the typical hole dimension @xmath19 , velocities by @xmath38 and mobilities by a characteristic mobility value @xmath39 .
we assume that the potential @xmath32 which characterises the applied field is small compared to the equilibrium zeta potential @xmath40 , so that @xmath41 , where @xmath42 has already been assumed small in order that we may describe the charge cloud by means of the linearized poisson - boltzmann equation .
we use the dimensionless field strength @xmath43 as the basis for a perturbation expansion : & = & _ 1 + + & = & _ 0 + _ 1 + + n^i & = & n_0^i + n_1^i + , where the subscript 0 refers to the equilibrium cloud , and the caret @xmath44 denotes a nondimensional quantity .
the steady - state ion conservation equation , correct to @xmath45 , becomes @xmath46 where the peclet number @xmath47 characterizes the ratio of ionic convection to diffusion .
since the nondimensional equilibrium potential @xmath48 has been assumed to be small , ( [ convdiff ] ) reduces to @xmath49 the boundary conditions at infinity are n_1^i&&0 , + _ 1&~&/2 [ bc_infinity ] we assume that no ions enter or leave the surface of the membrane . hence @xmath50 where @xmath51 is the normal to the membrane . at @xmath45 , and assuming @xmath52 , this zero flux boundary condition becomes @xmath53 multiplying equation ( [ eqn1 ] ) by @xmath54 and summing over @xmath3 , we obtain , at @xmath45 , @xmath55 where @xmath56 and @xmath57 is the ( non - dimensional ) perturbation to the charge density @xmath58 .
the boundary conditions for ( [ chieqn1 ] ) are similarly obtained from ( [ bc_infinity ] ) and ( [ bc1 ] ) : _ 1&~&1/2 + _ 1&=&0 [ chieqn3 ] and we note that the boundary condition ( [ chie_normal_bc ] ) represents zero flux of ions into the membrane , rather than a zero normal electric field .
the potential @xmath59 is thus obtained by solving the laplace equation for the potential created by applying the potential difference @xmath32 across the insulating membrane containing the pore .
the potential @xmath60 is related to the linearized change @xmath61 in the electrochemical potential of the @xmath3th species of ion , with @xmath62 . as discussed by @xcite , when @xmath63 the perturbation @xmath64 to the charge density is negligibly small , so that @xmath65 , and our analysis is equivalent to that of @xcite .
the stokes equations governing fluid motion are modified by the presence of an electric force @xmath66 acting on the fluid .
expanding in powers of @xmath43 , we find @xmath67 where we have used the relation @xmath68 between the equilibrium charge density @xmath69 and the equilibrium potential @xmath14 given by the linearized poisson - boltzmann equation .
hence the stokes equations become @xmath70 where the term @xmath71 in ( [ rhophi ] ) has been incorporated into the pressure @xmath72 .
note that the fluid motion caused by direct electrical forces acting on the fluid create additional deformation of the charge cloud , but , as seen from equation ( [ convdiff ] ) , this deformation is @xmath73 , and may be neglected when @xmath48 is small .
we now determine the @xmath45 total volumetric flow rate @xmath31 through the pore , created by the electrical force acting on the charge cloud within the fluid .
consider two stokes flows @xmath34 and @xmath74 in a volume @xmath75 with boundary conditions given on the bounding surface @xmath76 with outward normal @xmath51 .
a body force @xmath77 acts on the fluid , in which the pressure is @xmath72 , the viscosity is @xmath6 , the strain rate tensor is @xmath78 , and the stress tensor is @xmath79 .
barred variables represent the corresponding quantities for the second flow @xmath74 .
the reciprocal theorem gives the identity : @xmath80 we suppose that flow 1 ( @xmath34 ) is the flow of interest , namely , the electrokinetic flow through a hole in a charged membrane .
the body force @xmath77 , by ( [ stokes ] ) , is @xmath81 where , by ( [ chieqn1 ] ) , @xmath82 with boundary conditions ( [ chieqn3 ] ) @xmath83 on the membrane surface @xmath84 and @xmath85 on surfaces @xmath86 far from the pore ( figure [ fig : sketch ] ) .
we take flow 2 ( @xmath74 ) to be that due to a pressure difference @xmath87 imposed across a pore in an uncharged membrane ; thus , the body force @xmath88 . since the stokes flow equations are linear , we may write @xmath89 where @xmath90 is a function that depends solely on the pore geometry .
we now substitute the two flows into the reciprocal relation ( [ eq : rectheo ] ) .
the bounding surface @xmath76 is as shown in figure [ fig : sketch ] ; it consists of two hemispheres @xmath91 and @xmath92 of very large radius @xmath93 , together with the membrane surface , @xmath94 . on @xmath94 , the velocities @xmath95 whereas on @xmath86 we have @xmath96 and @xmath97 where @xmath98 are the pressures at great distance @xmath93 from the pore on either side of the membrane . for flow 1 , @xmath99 and for flow 2 , @xmath100 . substituting ( [ eq : forcef ] ) and ( [ eq : defineg ] ) into ( [ eq : rectheo ] ) , and cancelling the pressure difference @xmath87 from both sides of the equation , we obtain @xmath101 where @xmath31 is the volumetric flux of flow 1 from the side @xmath91 to the side @xmath92 of the membrane .
we now consider a circular pore of radius @xmath19 .
we adopt cylindrical co - ordinates @xmath102 , with origin at the centre of the pore and @xmath103 along the axis of symmetry , together with oblate spherical coordinates @xmath104 where @xmath105 , @xmath106 such that @xmath107 the scale factors are @xmath108 the imposed electric field is given by , with potential @xmath109 .
\label{eq : chi}\ ] ] give the stream function @xmath110 for flow 2 .
comparing the resulting velocity , @xmath74 , with ( [ eq : defineg ] ) , @xmath111 substituting ( [ eq : chi ] ) and ( [ eq : u2 ] ) into ( [ eq : q1 ] ) yields the electroosmotic flow rate @xmath112 the equilibrium charge density @xmath69 in the linearized , debye hckel limit may be obtained by excising the solution for a uniformly charged disk @xcite from that for a charged infinite plate .
hence @xmath113 .
\label{eq : rho0}\end{aligned}\ ] ] the integral in ( [ eq : q2 ] ) can not be evaluated in closed form when @xmath69 is given by ( [ eq : rho0 ] ) .
however , in the long debye length limit @xmath114 , the rate of decay of @xmath90 and @xmath115 is such that the major contribution to the integral ( [ eq : q2 ] ) comes from a region ( near the hole ) of volume @xmath116 , within which @xmath117 .
thus , @xmath118 where @xmath119 is a convenient characteristic flow rate . in other cases ,
the integral must be evaluated numerically .
it is convenient to introduce new variables @xmath120 , @xmath121 , @xmath122 , @xmath123 and @xmath124 in terms of which ( [ eq : q2 ] ) becomes @xmath125 , \label{eq : q3}\ ] ] with @xmath126 and @xmath127 defined as @xmath128 dt , \label{eq : i1_define } \\
i_2 & = & \int_0 ^ 1 q^2 \left [ ci(x q)\sin(x
q ) \cos(x q ) \right ] d q , \label{eq : exptermint}\end{aligned}\ ] ] where @xmath129 are the sine and cosine integrals @xmath130 the integrals @xmath126 ( [ eq : i1_define ] ) and @xmath127 ( [ eq : exptermint ] ) were evaluated using the matlab routine * quadgk * @xcite .
@xmath126 represents the potential due to a thin charged disk , and decays exponentially at large distances , as does @xmath131 . the @xmath132 integration in ( [ eq : q3 ] )
could therefore be truncated at a large value , taken to be @xmath133 ( corresponding to a distance @xmath134 from the pore ) .
to evaluate the second term in brackets in ( [ eq : q3 ] ) , the @xmath104 space was divided into sub
regions , with typically 500 intervals for @xmath132 and 200 for @xmath135 .
smaller intervals were used near the pore ( @xmath136 ) and near the membrane ( @xmath137 ) . for every pair @xmath104 , the integral @xmath138
was numerically evaluated and @xmath31 was obtained via trapezoidal summation within matlab .
results are shown in figure [ fig : qakappa ] with the asymptotic regime @xmath114 , described by ( [ eq : qakappasmall ] ) , depicted by the dashed straight line of unit slope .
in order to examine the validity of ( [ eq : q3 ] ) , the full poisson - nernst - planck - stokes system of equations was solved numerically using a finite volume method based on the open source cfd library openfoam @xcite .
our model has been described by @xcite and details of the implementation for the current problem are presented in the appendix .
the cases studied include membranes of thickness @xmath30 and @xmath139 in the regime of weak applied fields and low membrane charge .
these conditions , stated in [ subsec : formalism ] , may be restated more conveniently as @xmath140 where @xmath141 and @xmath142 are dimensionless parameters characterizing the degree of membrane charge and the strength of the applied field respectively . in the simulations presented here , the values of these parameters were @xmath143 and @xmath144 so that ( [ eq : q3 ] ) can be expected to be a reasonable approximation in the range @xmath145
. the computed flow rate @xmath31 normalized by @xmath146 is shown by the symbols in figure [ fig : qakappa ] .
good agreement with ( [ eq : q3 ] ) is obtained , but thicker membranes result in somewhat increased flow rates .
the discrepancy increases at shorter debye lengths .
the analysis of [ sec : zerothick ] assumed that effects due to induced charge electroosmosis ( iceo ) are negligible .
however , @xcite have shown that iceo generates vortices at sharp corners , and such vortices will inevitably be generated in the membrane geometry considered here . in the vicinity of the edge of the nanopore , where @xmath147 , the potential @xmath60 ( [ eq : chi ] ) on the membrane surface @xmath148 may be expanded as @xmath149 if we assume that this potential is little modified when the membrane has a finite thickness @xmath139 , the potential gradient within the solid membrane due to the external potential @xmath60 is @xmath150 and if @xmath151 the induced potential gradient normal to the surface within the liquid is @xmath152 .
this corresponds to an induced surface charge ( with accompanying charge cloud of couter - ions ) @xmath153 the analysis breaks down when @xmath154 where the detailed geometry near the edge of the pore becomes important .
if the membrane has rounded edges , the curvature @xmath155 and the induced charge is at most @xmath156 .
this may be neglected as long as it is small compared to @xmath8 , or equivalently , if @xmath157 . for common membrane materials ( e.g. lipids and silica ) @xmath158 ,
thus , as long as the applied field remains weak , iceo effects are restricted to the neighbourhood of sharp corners .
the results of full numerical solutions of the poisson - nernst - planck - stokes equations reported in figure [ fig : qakappa ] were computed with membrane permittivity @xmath159 .
however , it is shown in the appendix that numerical solutions for appropriate non - zero solid permittivities @xmath160 predict flow rates @xmath31 that differ little from those with @xmath159 in the parameter regime under consideration .
the contribution of such effects to the net fluid flux , @xmath31 , is at best very weak .
this is perhaps not surprising , since even when ( as here ) there are sharp corners , if the membrane is uncharged and the electrolyte is symmetric ( with identical ionic mobilities ) , symmetry dictates that iceo can not generate a net flow @xmath31 through the pore .
so although nonlinear effects such as iceo can generate a net fluid flow through a charged membrane , the applied field @xmath32 must be larger than the fields considered here .
an iceo contribution to the fluid flux is in principle possible , but only for a charged membrane at high applied fields , as , for example , in the numerical results presented by @xcite .
the reciprocal theorem used in [ subsec : formalism ] enabled us to determine the volumetric flow rate @xmath31 through the pore without a full computation of the velocity field .
this has the advantage of leading quickly to a value for @xmath31 .
however , this approach hides other interesting features of the flow , such as eddies , whether generated by iceo @xcite or by the pore throat restricting the flow @xcite .
these features are only revealed by full numerical computations , such as those discussed in the appendix .
since the fluid flux through the pore is generated by the applied potential , @xmath17 , we can define an `` electroosmotic conductance '' @xmath161 in analogy to the electric conductance .
if the membrane thickness @xmath162 , we have a long cylindrical pore with surface charge density @xmath8 at the wall .
the electroosmotic flow velocity is then @xcite @xmath163 \label{eq : axial_velocity_u}\ ] ] where @xmath164 is the equilibrium potential in the double layer , @xmath40 is the equilibrium potential at the wall , and @xmath165 .
the equilibrium potential of a cylindrical pore in the debye - hckel limit is @xmath166 integrating the fluid velocity @xmath167 ( [ eq : axial_velocity_u ] ) over the cross - section we obtain the volumetric flow rate @xmath31 and hence the the electroosmotic conductance @xcite @xmath168 .
\label{eq : hc}\ ] ] on the other hand , when the thickness @xmath169 , we expect the system to be identical to a hole in a zero thickness membrane , and the electroosmotic conductance may be obtained using ( [ eq : q3 ] ) : @xmath170 .
\label{eq : hp}\ ] ] if a membrane of thickness @xmath171 containing a circular hole of radius @xmath19 separates two uniformly conducting regions , then the electrical resistance of the cylindrical hole increases , proportional to @xmath171 .
this might suggest a vanishing resistance for an infinitely thin membrane .
however , in reality , as @xmath172 the electrical resistance is dominated by entrance and exit effects , and can be determined from the electrical potential ( [ eq : chi ] ) .
this is called the `` access resistance '' of the pore and for a circular pore in an infinitely thin membrane it is described by a simple analytical formula @xcite .
an analogous situation applies to the problem of electroosmotic flow through a pore in a membrane .
when the pore length @xmath171 is large compared to the pore radius @xmath19 , the flow conductance @xmath173 from ( [ eq : hc ] ) a consequence of the fact that the electric field in the pore , @xmath174 .
however , @xmath175 does not increase indefinitely as @xmath176 but instead approaches a finite value @xmath177 given by ( [ eq : hp ] ) .
the surface charge @xmath178 within the cylindrical pore goes to zero as @xmath172 .
electroosmotic motion is therefore determined by flow in the fluid on either side of the membrane , as described in
[ sec : zerothick ] , and not by the cylindrical pore .
thus , in analogy to the corresponding electrical problem , @xmath179 may be regarded as an `` access resistance '' of the pore to electroosmotic flow .
figure [ fig : flowconduc ] shows the electroosmotic conductance @xmath175 obtained from the finite volume numerical computations as a function of @xmath180 .
the computed value of @xmath175 is normalized by @xmath177 obtained from ( [ eq : hp ] ) .
it is seen that @xmath181 approaches unity as @xmath182 and approaches @xmath183 for large @xmath180 .
the dashed line representing @xmath184 was obtained from ( [ eq : hc ] ) and ( [ eq : hp ] ) .
the results of the full computations indicate that though @xmath175 does exhibit the expected limiting behaviors , it does not vary monotonically with @xmath171 at short debye lengths .
the origin of the peak at intermediate values of @xmath180 will be investigated further in future work . determined from the full numerical simulation ( symbols ) as a function of the normalized membrane thickness , @xmath180 .
left panel corresponds to @xmath185 and right panel @xmath186 .
the dashed lines correspond to the thin ( @xmath187 ) and thick ( @xmath188 ) membrane limits obtained from ( [ eq : hc ] ) and ( [ eq : hp]).,title="fig:",scaledwidth=45.0% ] determined from the full numerical simulation ( symbols ) as a function of the normalized membrane thickness , @xmath180 .
left panel corresponds to @xmath185 and right panel @xmath186 .
the dashed lines correspond to the thin ( @xmath187 ) and thick ( @xmath188 ) membrane limits obtained from ( [ eq : hc ] ) and ( [ eq : hp]).,title="fig:",scaledwidth=45.0% ]
we have assumed that the surface charge density @xmath8 is sufficiently low that the zeta potential is small , @xmath189 at @xmath190 .
thus , the poisson - boltzmann equation can be linearized .
non - dimensional zeta potentials @xmath42 in colloidal systems , though not always small , are typically at most 5 , and it is found that theories based on small potentials usually give useful qualitative insight into electrokinetic behaviour over this range of potentials ( e.g. @xcite ) . we have also assumed that the applied potential difference @xmath191 .
potential differences applied in experiments are typically of the same order as typical @xmath40-potentials which in silica substrates vary in magnitude between @xmath192 and @xmath193 mv depending mainly on counter - ion concentration @xcite . for example , @xcite describe experiments in which @xmath32 was in the range 30100 mv .
nanopores ( radius @xmath194 510 nm ) in graphene sheets have recently been used in dna translocation experiments @xcite . the applied voltage @xmath195 0200
mv in these experiments .
the computations of @xcite predict that the electroosmotic flow rate through a pore in a membrane varies non - linearly with @xmath32 only at voltages greater than 100 mv .
thus , we again expect the results presented here to give at least a qualitative understanding of electroosmotic flow in such experiments .
electroosmotic flow through nanopores has been shown to control the translocation velocity of charged polymers in resistive pulse experiments @xcite .
when the free translocation of the polymer is hindered by tethering it to a colloid held in an optical trap , the tethering force has been shown to be determined by the electroosmotic flow within the pore @xcite .
furthermore , it has been argued that the flow outside and in the vicinity of the nanopore controls the capture rate of polymers into the pore @xcite , though the experimental evidence for this appears tentative at present .
in addition to the single molecule experiments mentioned above , our results should also be helpful in understanding the properties of nanoporous membranes that are used in batteries , water desalination and numerous other industrial applications .
@xcite have recently studied the electrical conductivity of a model membrane consisting of an array of nanopores .
however , the applied voltage should also result in an electroosmotic flux , the calculation of which may be undertaken as a suitable generalization of the approach presented here .
since membrane bound organelles in cells contain nanopores that control the traffic of biological molecules across the membrane , our results may also be of interest in the biological context ( see e.g. @xcite ) .
mm & sg acknowledge support from the nih through grant 4r01hg004842 - 03 .
sg was hosted by the cavendish laboratory , university of cambridge , as visiting professor with funds provided by the leverhulme trust .
jds thanks the department of applied mathematics and theoretical physics , university of cambridge , and the institut de mcanique des fluides de toulouse , for hospitality .
an electrohydrodynamic solver was developed to solve the poisson nernst planck
stokes ( pnp stokes ) system of equations using the finite volume method .
the solver was based on the openfoam cfd library @xcite , a c++ library designed for computational mechanics , containing a collection of object oriented classes developed to represent mesh , fields , matrices and the necessary operations on fields and tensors .
it also provides functions to handle finite volume discretization and matrix equation solving .
we apply the following scheme to solve the pnp stokes equations .
we start from a zero flow field
. equations ( [ eq : poisson ] ) and ( [ eq : np ] ) are solved sequentially in a loop with under - relaxation until the absolute residual is smaller than @xmath197 . under
relaxation is necessary because the pnp system is non linear .
the electric volume force @xmath198 is obtained from this solution and used explicitly in the next step : the solution of the incompressible stokes flow : ( [ eq : stokes ] ) and ( [ eq : continuity ] ) . the simple algorithm is used with a fixed volume force density . the flow field
is then substituted into ( [ eq : np ] ) .
the pnp equations are then solved again using the updated flow field .
an outer loop is constructed to iterate over the pnp loop and stokes flow module . for the finite volume discretization of the governing equations ,
central differences are used for all diffusive terms in ( [ eq : np ] ) and viscous terms in ( [ eq : stokes ] ) .
a second order upwind scheme is used for the convective terms in ( [ eq : np ] ) .
the discretized linear system is solved using a pre - conditioned conjugate gradient solver if the matrix is symmetric or a pre - conditioned bi
conjugate gradient solver if the matrix is asymmetric .
the details of the numerical algorithm are given by .
a schematic view of the axisymmetric geometry used for the full numerical simulations is provided in figure [ fig : system ] .
it consists of a circular hole of radius @xmath19 in a solid dielectric membrane cdef of arbitrary thickness @xmath199 .
the membrane surfaces cd , de and ef have a uniform surface charge density @xmath8 .
two large cylindrical reservoirs are connected to the pore , one at each end .
the length and radius of both the reservoirs are identical , and are @xmath200 , chosen to be much larger than either the hole radius @xmath19 or the debye length @xmath21 in order to approximate an infinite reservoir .
we adopt the following boundary conditions @xcite .
the ion number densities on ab and gh are constant , and equal to the number density @xmath201 in the bulk solution far from any charged surfaces .
the electrical potentials are uniform on ab and on gh , with a potential difference of @xmath32 between the top ( ab ) and bottom ( gh ) .
the pressure @xmath202 on ab is uniform and equal to that on gh . on
the side walls bc and fg , the radial electric field , ionic flux and radial velocity , which decay away from the pore , are set to zero .
a zero tangential shear stress is imposed on flow parallel to the side walls . at the membrane surfaces cd , de and ef
, a no flux condition is used for ( [ eq : np ] ) , a no slip condition for the flow ; the electric field @xmath203 undergoes a jump across the solid - fluid interface such that @xmath204 where @xmath7 is the electrical permittivity of the fluid and @xmath205 is the permittivity of the membrane , @xmath206 is the electric field at the interface within the membrane and @xmath207 is the unit normal at the surface directed into the fluid .
the potential is continuous across the interface .
the strength of the applied field and the amount of surface charge can be characterized by the dimensionless parameters @xmath208 and @xmath209 respectively . in the simulations presented here ,
the values of these parameters were kept fixed at @xmath144 and @xmath143 .
the flow rate @xmath31 was obtained by numerically integrating the @xmath103component of the velocity over the plane @xmath148 .
if the membrane polarizability is sufficiently small that @xmath210 , then the jump condition of the normal component of the field may be replaced by @xmath211 . in this case , the computational domain may be restricted to include only the fluid phase .
this approximation was adopted for the results presented in figures [ fig : qakappa ] and [ fig : flowconduc ] .
thus , effects due to induced charge electroosmosis ( iceo ) were neglected .
the results of a calculation to test the validity of this assumption in the parameter range of interest are shown in figure [ fig : qakappaepw0 ] .
the data from figure [ fig : qakappa ] for a non - polarizable membrane of thickness @xmath212 are reproduced in figure [ fig : qakappaepw0 ] .
for comparison , the result of a second calculation in which the dielectric constant of the membrane material was set to @xmath213 ( corresponding to silica ) is also shown .
the electrolyte is considered polarizable with a dielectric constant of @xmath214 .
it is seen that the effect of membrane polarizability on the flow rate is negligible . | electroosmotic pumping of fluid through a nanopore that traverses an insulating membrane is considered .
the density of surface charge on the membrane is assumed uniform , and sufficiently low for the poisson - boltzmann equation to be linearized .
the reciprocal theorem gives the flow rate generated by an applied weak electric field , expressed as an integral over the fluid volume . for a circular hole in a membrane of zero thickness , an analytical result is possible up to quadrature . for a membrane of arbitrary thickness , the full poisson
nernst planck stokes system of equations is solved numerically using a finite volume method .
the numerical solution agrees with the standard analytical result for electro - osmotic flux through a long cylindrical pore when the membrane thickness is large compared to the hole diameter .
when the membrane thickness is small , the flow rate agrees with that calculated using the reciprocal theorem . |
quantum chromodynamics ( qcd ) is generally accepted as the underlying theory of the strong interaction . at low energies relevant to nuclear physics ,
however , nucleons and mesons are convenient and efficient degrees of freedom . in particular ,
relativistic field theories of hadrons , called quantum hadrodynamics ( qhd ) , have been quite successful in describing the bulk and single - particle properties of nuclei and nuclear matter in the mean - field and dirac brueckner hartree fock approximations@xcite .
qhd studies based on renormalizable models , however , have encountered difficulties due to large effects from loop integrals that incorporate the dynamics of the quantum vacuum@xcite . on the other hand ,
the `` modern '' approach to renormalization @xcite , which makes sense of effective , `` cutoff '' theories with low - energy , composite degrees of freedom , provides an alternative .
when composite degrees of freedom are used , the structure of the particles is described with increasing detail by including more and more _ nonrenormalizable _ interactions in a derivative expansion@xcite .
we seek to merge the successful phenomenology of the qhd approach to nuclei with the modern viewpoint of effective field theories .
the present work is an extension of refs .
@xcite , which explored some aspects of incorporating qcd symmetries into effective hadronic models . here
we build a more complete effective lagrangian and apply it to nuclei at the one - baryon - loop level .
our two main goals are to include effects of compositeness , such as single - nucleon electromagnetic structure , in a unified , self - consistent framework and to test whether `` naive dimensional analysis '' ( nda ) and naturalness ( see below ) are useful concepts for effective field theories of nuclei . eventually , we will need to demonstrate a systematic expansion scheme that includes corrections beyond one - loop order
nevertheless , ref .
@xcite describes how a general energy functional with classical meson fields can implicitly incorporate higher - order contributions .
we therefore expect that higher - order many - body effects can be absorbed into the coefficients of our one - loop energy functional when we determine parameters by fitting to nuclear observables .
we return to this point later in this section .
important physical constraints on the effective lagrangian are established by maintaining the symmetries of qcd .
these include lorentz invariance , parity invariance , electromagnetic gauge invariance , isospin , and chiral symmetry .
these symmetries constrain the lagrangian most directly by restricting the form of possible interaction terms . in some cases
, the magnitude of ( or relations between ) parameters may also be restricted .
chiral symmetry has been known to play an important role in hadronic physics at low energies since the early days of current algebra @xcite ; it is exploited systematically in terms of effective fields in the framework known as chiral perturbation theory ( chpt)@xcite . there is also a long history of attempts to unite relativistic mean - field phenomenology based on hadrons with manifest chiral symmetry .
in particular , it has been tempting to build upon the linear sigma model @xcite , with a light scalar meson playing a dual role as the chiral partner of the pion _ and _ the mediator of the intermediate - range nucleon
nucleon ( nn ) attraction .
the generic failure of this type of model has been shown in refs .
@xcite . however , a light scalar is completely consistent with a _ nonlinear _ realization of chiral symmetry @xcite .
moreover , in a nonlinear realization , one need not assume that the hadrons are grouped into particular chiral multiplets @xcite .
the phenomenological success of chpt establishes the dominant contributions of the vector resonances to the low - energy constants in the pion sector@xcite .
in particular , a tree - level effective lagrangian with vector mesons is found to be essentially equivalent to chpt at one - meson - loop order . in the baryon sector , a satisfactory description of the nucleon form factors from vector - meson dominance ( vmd ) @xcite alone has not been found@xcite .
however , we _ can _ describe the long - range parts of the single - nucleon form factors ( that is , magnetic moments and mean - square radii ) using an effective lagrangian that includes vector mesons and direct couplings of photons to nucleons .
our approach is analogous to that implemented long ago in ref .
@xcite . in virtually all previous qhd calculations of finite nuclei ,
the single - nucleon electromagnetic structure was incorporated `` by hand . '' after the nuclear structure calculation was completed and `` point - nucleon '' densities were obtained , they were convoluted with empirical single - nucleon form factors to compute the charge density .
the results were typically insensitive to the nature of the form factor , as long as the nucleon charge radius was accurate . in the present work ,
we do not need external form factors .
the large oscillations in the point - nucleon charge density of a nucleus are significantly reduced by virtual mesons inherent in the model , and the low - momentum parts of the charge form factors of various nuclei are well reproduced .
thus we have a _ unified _ framework . to achieve a systematic framework for calculating nuclei , however
, we must develop an expansion scheme .
chpt is an expansion in powers of small external momenta and the pion mass .
weinberg@xcite applies this small - momentum expansion further to a calculation of the @xmath0-nucleon `` effective potential '' , defined as the sum of all connected diagrams for the @xmath1 matrix ( in time - ordered perturbation theory ) without @xmath0-nucleon intermediate states . on the other hand
, an expansion of the nn interaction kernel in terms of intermediate states of successively shorter range provides a successful and economical way to accurately describe nn scattering data @xcite .
here we combine the ideas of the small - momentum expansion and the intermediate - state expansion by including the low - lying _ non - goldstone bosons _ into the low - energy effective lagrangian .
these bosons account for the intermediate - range interactions and conveniently describe the nonvanishing expectation values of nuclear bilinears such as @xmath2 and @xmath3 .
there are well - defined resonances in the vector channels , and thus it is natural to identify them with fields in the effective lagrangian ; in contrast , the dynamics in the scalar channel is more complicated .
numerous calculations@xcite show that correlated two - pion exchange leads to a distribution of strength in the isoscalar - scalar channel at center - of - mass pion energies of roughly 500 to 600 mev , even in the absence of a low - energy resonance .
we therefore introduce an isoscalar - scalar meson as an _ effective _ degree of freedom to simulate this strong interaction , just like the non - goldstone bosons in the vector channels .
such a scalar , with a mass of 500 to 600 mev , is essential in one - boson - exchange models of the nn potential @xcite and in relativistic mean - field models of nuclei@xcite .
we emphasize that at the level of approximation used in this paper , the scalar field can be introduced without loss of generality .
( for example , it can be viewed as a way to remove terms with more than two nucleon fields from a general pion - nucleon lagrangian . )
a deeper understanding of the nn interaction in the scalar channel is an important pursuit , but is not needed for our purposes here .
we shall expand our effective hadronic lagrangian in powers of the fields and their derivatives , and organize the terms using `` naive dimensional analysis '' ( nda)@xcite . the nda framework is used to identify the dimensional factors of any term . after these dimensional factors and some appropriate counting factors are extracted , the remaining _ dimensionless _ coefficients are all assumed to be of order unity .
this is the so - called naturalness assumption .
if naturalness is valid , the effective lagrangian can be truncated at a given order with a reasonable bound on the truncation error for physical observables .
thus we have a controlled expansion , at least at the tree level .
evidently , for an expansion in powers of the fields and their derivatives to be useful , the naturalness assumption must hold .
implicit in this assumption is that all the short - distance physics is incorporated into the coefficients of the effective lagrangian ; otherwise , the truncation error will not be reliably given by nda power counting when loops are included .
thus , just as in chpt or in heavy - baryon theories , the nucleon vacuum loops and non - goldstone - boson loops will be `` integrated out '' , with their effects buried implicitly in the coefficients .
nevertheless , the nucleon and non - goldstone - boson fields are still needed in the lagrangian to account for the valence nucleons , to treat the mean fields conveniently , and to apply the usual many - body and field - theoretic techniques .
any contributions from vacuum nucleon or non - goldstone - boson loops that arise in many - body calculations can formally be cancelled by the appropriate counterterms .
these cancellations can always be achieved , since _ all possible interaction terms consistent with the symmetries are already included in the effective lagrangian .
_ on the other hand , long - distance , finite - density effects should be calculated explicitly in a systematic application of the effective lagrangian . in principle , the naturalness assumption can be verified by deriving the effective lagrangian from qcd . since this is not yet feasible , one must fit the unknown parameters to experimental data . at finite density
, however , the fermi momentum enters as a new scale .
this complicates the situation , because a treatment of the effective lagrangian at the level of classical meson fields and valence nucleons ( the dirac
hartree approximation ) implies that several different physical effects are implicitly contained in the parameters .
not only are short - distance contributions from meson and nucleon vacuum loops absorbed , but long - distance , many - body effects from well - known ladder and ring diagrams are included approximately as well @xcite .
thus the usefulness of nda , the naturalness assumption , and the expansion in powers of the fields and their derivatives requires further examination when these ideas are applied to problems at finite density .
it is not obvious that an effective lagrangian with natural coefficients that include only short - distance physics will remain natural in a one - loop calculation that accounts for higher - order many - body effects by adjustments of the coefficients .
here we rely on fits to the properties of finite nuclei to determine whether the coefficients are natural and our expansion of the lagrangian is useful .
( friar and collaborators have recently investigated naturalness as exhibited by a mean - field point - coupling model@xcite . ) an important topic for future study is to include the many - body dynamics explicitly , so as to disentangle it from the short - range and vacuum dynamics already implicit in the parameters of the low - energy effective lagrangian .
the interaction terms of the nucleons and the non - goldstone bosons can be quite complicated , and different forms have been used in various models in the literature @xcite .
some of these models are actually equivalent if one uses the freedom of field redefinitions@xcite , which leave the on - shell @xmath1 matrix invariant@xcite .
we discuss how to take advantage of this freedom to simplify the interaction terms between the nucleons and the non - goldstone bosons . as a result
, the lagrangian can be conveniently written in a `` standard form '' , in which one mainly has the traditional yukawa interactions .
the rest of the paper is organized as follows . in sec .
ii , we discuss the construction of an effective hadronic lagrangian consistent with chiral symmetry , gauge invariance , and lorentz and parity invariance .
chiral symmetry is realized in the standard nonlinear form of callan , coleman , wess , and zumino ( ccwz)@xcite .
neutral scalar and vector mesons are introduced to efficiently describe the mid- and short - range parts of the nn interaction , and the role of broken scale invariance in qcd is considered .
georgi s naive dimensional analysis and field redefinitions are also discussed , with the new feature of off - shell non - goldstone bosons . in sec .
iii , we show how the electromagnetic structure of the pion and the nucleon can be described in the low - energy regime .
applications to nuclear matter and finite nuclei at the dirac hartree level are presented in sec .
iv , and the results are shown in sec . v. a discussion and summary is given in sec . vi .
we begin this section by showing how we implement nonlinear chiral symmetry and electromagnetic gauge invariance using nucleons , goldstone pions , and non - goldstone isovector mesons .
next , we discuss the introduction of neutral vector and scalar mesons that are used to provide an efficient description of the short- and mid - range nucleon nucleon interaction . to use the resulting lagrangian , which in principle has an infinite number of terms
, we must have some systematic truncation procedure , and we develop this using the ideas of naive dimensional analysis and naturalness , together with some observations on the sizes of meson mean fields in nuclei .
we illustrate a nonlinear realization of chiral symmetry@xcite using a system that consists of rho mesons , pions , and nucleons only .
we construct a general effective lagrangian that respects chiral symmetry , lorentz invariance , and parity conservation .
the role of vector - meson dominance ( vmd ) is also discussed .
we restrict ourselves to the chiral limit with massless pions and postpone the generalization to a finite pion mass for future study .
the goldstone pion fields @xmath4 , with @xmath5 , form an isovector , which can be considered as the phase of a chiral rotation of the identity matrix in isospin space : @xmath6 here @xmath7 is the pion - decay constant and the pion field is compactly written as @xmath8 , with @xmath9 being pauli matrices .
the matrix @xmath10 is the standard exponential representation and the `` square root '' representation in terms of @xmath11 is particularly convenient for including heavy fields in the chiral lagrangian .
the isospinor nucleon field is represented by a column matrix @xmath12 where @xmath13 and @xmath14 are the proton and neutron fields respectively .
the rho fields @xmath15 also form an isovector ; we use the notation @xmath16 .
( see refs .
@xcite for discussions of the formal equivalence between the vector and the tensor formulations for spin-1 fields . ) following ccwz@xcite , we define a nonlinear realization of the chiral group @xmath17 such that , for arbitrary global matrices @xmath18 and @xmath19 , we have the mapping @xmath20 where @xmath21 \rho'_\mu(x ) & = & h(x ) \rho_\mu ( x ) h^{\dagger}(x ) \ , \label{eq : rhotrans } \\[4pt ] n'(x ) & = & h(x)n(x ) \ .
\label{eq : ntrans}\end{aligned}\ ] ] as usual , the matrix @xmath10 transforms as @xmath22 . the second equality in eq .
( [ eq : xitrans ] ) defines @xmath23 implicitly as a function of @xmath24 , @xmath25 , and the local pion fields : @xmath26 . in terms of @xmath24 , @xmath25 , and @xmath10 as @xmath27 .
given the decomposition @xcite @xmath28 , @xmath29 , and @xmath30 , with @xmath31 , @xmath32 , and @xmath33 real , the infinitesimal expansion of @xmath34 is @xmath35 @xcite .
] under parity , which is defined to be @xmath36 the pion field transforms as a pseudoscalar : @xmath37 .
the existence of the parity automorphism ( [ eq : parity ] ) allows the simple decomposition @xmath38 in eq .
( [ eq : u - def ] ) @xcite .
it follows from eq .
( [ eq : xitrans ] ) that @xmath23 is invariant under parity , that is , @xmath39 , with @xmath40 the unbroken vector subgroup of @xmath41
( [ eq : rhotrans ] ) and ( [ eq : ntrans ] ) ensure that the rho and nucleon transform linearly under @xmath42 in accordance with their isospins .
note that the matrix @xmath23 becomes global only for @xmath42 transformations , in which case @xmath43 .
it is straightforward to show that eqs .
( [ eq : nonlr ] ) to ( [ eq : ntrans ] ) indeed yield a realization of the chiral group @xcite .
we emphasize that there is no local gauge freedom involved , since there are no arbitrary local gauge functions in the transformations .
the nonlinear realization implies that chiral rotations involve the absorption and emission of pions , as expected from a spontaneously broken symmetry@xcite .
for example , while vector transformations mix the proton with the neutron , axial transformations mix the proton with states of nucleons plus zero - momentum pions . to complete the building blocks of the chiral effective lagrangian
, we define an axial vector field @xmath44 and a polar vector field @xmath45 by @xmath46 v_{\mu } & \equiv & -{i \over 2}(\xi^{\dagger } \partial_{\mu } \xi + \xi \partial_{\mu}\xi^{\dagger } ) = v_{\mu } ^\dagger \ , \label{eq : vdef}\end{aligned}\ ] ] both of which contain one derivative . under a chiral transformation ,
we find @xmath47 a_\mu & \rightarrow & h a_\mu h^{\dagger } \ .\end{aligned}\ ] ] note that to leading order in derivatives , @xmath48 + \cdots \ .
\label{eq : leading}\ ] ] to maintain chiral invariance , instead of using an ordinary derivative @xmath49 acting on a rho or nucleon field , we should use a covariant derivative @xmath50 .
the chirally covariant derivative of the rho field is @xmath51 \ .
\label{eq : rhocovar}\ ] ] we also define the covariant tensors @xmath52 = - i [ a_\mu , a_\nu ] \ , \\[4pt ] \rho_{\mu\nu}&= & d_\mu\rho_\nu - d_\nu\rho_\mu + i g [ \rho_\mu,\rho_\nu ] \ .
\label{eq : rhomunu}\end{aligned}\ ] ] the term proportional to @xmath53 $ ] on the right - hand side of eq .
( [ eq : rhomunu ] ) vanishes at the mean - field level , so the coupling @xmath54 will be left unspecified . to write a general effective lagrangian
, we need an organizational scheme for the interaction terms .
we organize the lagrangian in increasing powers of the fields and their derivatives .
while we do not consistently include loops in this investigation , we assign to each interaction term a size of the order of @xmath55 , with @xmath56 a generic small momentum , and @xmath57 where @xmath58 is the number of derivatives , @xmath59 the number of nucleon fields , and @xmath60 the number of non - goldstone boson fields in the interaction term .
the first two terms in eq .
( [ eq : order ] ) are suggested by weinberg s work @xcite .
the last term is a generalization that arises because a non - goldstone boson couples to _ two _ nucleon fields .
equation ( [ eq : order ] ) is also consistent with finite - density applications when the density is not too much higher than the nuclear - matter equilibrium density , as we will see in sec .
ii.d . taking into account chiral symmetry , lorentz invariance , and parity conservation , we may write the lagrangian through quartic order ( @xmath61 ) as @xmath62n \nonumber \\ & & -{f_\rho g_\rho\over 4m}\overline n \rho_{\mu\nu } \sigma^{\mu\nu } n - { \kappa_\pi\over m } \overline{n } v_{\mu\nu}\sigma^{\mu\nu } n \nonumber \\ & & -{1\over 2}{\rm tr}(\rho_{\mu\nu}\rho^{\mu\nu } ) + m_\rho^2 { \rm tr}(\rho_\mu \rho^\mu ) -g_{\rho \pi\pi}{2\fpi^2 \over m_\rho^2 } { \rm tr}\,(\rho_{\mu\nu}v^{\mu\nu } ) \nonumber \\ & & + \fpi^2{\rm tr}(a_\mu a^\mu ) + { 4\beta_\pi\over m } \overline{n } n { \rm tr}\ , ( a_\mu a^\mu ) + l_1[{\rm tr}(a_\mu a^\mu)]^2 + l_2 { \rm tr}(v_{\mu\nu } v^{\mu\nu } ) \ , \label{eq : lag1}\end{aligned}\ ] ]
where @xmath63/2 $ ] , @xmath64 is the axial coupling constant , @xmath65 is the nucleon mass , @xmath66 is the @xmath67 coupling , @xmath68 is the so - called tensor coupling , @xmath69 is the @xmath70 coupling , and @xmath71 , @xmath72 , @xmath73 , and @xmath74 are couplings for higher - order @xmath75 and @xmath76 interactions .
the pion couplings can be determined from @xmath75 and @xmath76 scattering@xcite .
we have omitted terms with higher powers of the rho fields , which have small expectation values in finite nuclei , and four- or more - nucleon `` contact '' terms , which can be represented by appropriate powers of other meson fields , as discussed shortly . a third - order ( @xmath77 )
term of the form @xmath78 will be included by coupling a scalar to two rho fields in sec .
the antisymmetry of rho meson field tensor @xmath79 guarantees three independent degrees of freedom , because there is no momentum conjugate to the time component @xmath80 , and @xmath80 satisfies an equation of constraint .
chiral symmetry permits isoscalar - scalar and isoscalar - vector terms like @xmath81 and @xmath82 .
however , analyses of nn scattering with one - boson exchange potentials @xcite shows that the low - lying mesons are sufficient for describing low - energy many - nucleon systems and that they dominate the nn interaction .
one may then expect that contact interactions of four ( or more ) fermions can be removed in favor of couplings to massive non - goldstone bosons .
the following example shows that this should be a good approximation .
consider an additional isovector - vector contact term @xmath83 ^ 2 $ ] .
the contribution of this term can be absorbed completely into a renormalization of @xmath66 as follows .
if we had constructed @xmath84 by retaining several isovector - vector non - goldstone bosons , we could eliminate them using their field equations to obtain @xmath85 ^ 2 - { 1\over 2}\big ( { { g'_\rho}^2 \over m_\rho^4 } + { g_1 ^ 2 \over m_1 ^ 4 } + \cdots \big ) \{\partial_\mu[\overline{n}(\bbox{\tau}/2)\gamma_\nu n]\}^2 + \cdots \ , \label{eq : rhoexp}\ ] ] where @xmath86 is the @xmath87 coupling when the heavier mesons are retained , with couplings to nucleons given by @xmath88 , and masses given by @xmath89 , .
evidently , the effects of the heavier particles can be absorbed into a redefinition of the @xmath67 coupling : @xmath90 if we approximate the second term in eq .
( [ eq : rhoexp ] ) as @xmath91\}^2 \ , \ ] ] the leading error is given by the following term : @xmath92\}^2 \ .\ ] ] for a typical resonance in the isovector - vector channel @xcite , we have @xmath93 and @xmath94 .
if @xmath95 is roughly @xmath96 , the error is about @xmath97 of the already small second term .
one should be able to test the sensitivity to these errors by allowing the non - goldstone boson masses to vary in the fit to nuclear properties . if we were to assume universal vmd in eq .
( [ eq : lag1 ] ) , then @xmath98 and @xmath99 in eq .
( [ eq : rhomunu ] ) . if we further assume the kawarabayashi
suzuki riazuddin
fayyazuddin ( ksrf ) @xcite relation @xmath100 then we may redefine the rho field to be @xmath101 and the covariant field tensor to be @xmath102 , and recover the lagrangian of weinberg@xcite or bando _
et al_.@xcite . in finite nuclei
, electromagnetism also plays an important role
. we may include electromagnetic gauge invariance in our formalism in a straightforward manner by noticing how the pion , rho , and nucleon fields transform under the local @xmath104 symmetry . here the electric charge @xmath105 , the third component of the isospin @xmath106 , and the hypercharge @xmath107 are related by @xmath108 . under @xmath104 , the electromagnetic field
@xmath109 transforms in the familiar way @xmath110 the pion , rho , and nucleon fields transform under the local @xmath104 rotation just as in eqs .
( [ eq : xitrans ] ) to ( [ eq : ntrans ] ) , with both @xmath24 and @xmath25 set equal to @xmath111 \ , \ ] ] where @xmath112 for the pion and the rho , and @xmath113 for the nucleon
. indeed , the resulting transformations for these hadrons are just as expected from their electric charges .
that is , @xmath114 ( \rho_\mu^0,\ \rho_\mu^\pm ) & \rightarrow & ( \rho_\mu^0,\ e^{\pm i\alpha}\rho_\mu^\pm ) \ , \\[5pt ] ( n,\ p ) & \rightarrow & ( n,\ e^{+i\alpha}p ) \ .\end{aligned}\ ] ]
electromagnetism breaks chiral symmetry .
the lagrangian in eq .
( [ eq : lag1 ] ) becomes gauge invariant if the ordinary derivatives are replaced with the covariant derivatives @xmath115 \ , \label{eq : xider } \\
\partial_{\mu } n & \rightarrow & \bigl(\partial _ { \mu } + { i\over 2}ea_{\mu}(1+\tau_3)\bigr)n\ , \\ \partial_{\mu } \rho_\nu & \rightarrow & \partial _
{ \mu}\rho_\nu + ie a_{\mu}\bigl[{\tau_3 \over 2},\rho_\nu\bigr ] \ .
\label{eq : rhoder}\end{aligned}\ ] ] as a result , the axial and vector pion fields become @xmath116 -\xi^{\dagger}\,[{\tau_3\over 2},\xi ] \big ) \ , \\ \widetilde{v}_{\mu } & = & v_{\mu } + { 1\over 2}ea_{\mu}\big ( \xi\,[{\tau_3\over 2},\xi^{\dagger } ] + \xi^{\dagger}\,[{\tau_3\over 2},\xi ] \big ) \ , \end{aligned}\ ] ] respectively , and the pion `` kinetic '' term is @xmath117 in addition to terms arising from the preceding minimal substitutions , there can be other non - minimal terms in the general effective lagrangian .
for example , it is possible to have a direct rho - photon coupling term @xcite @xmath118 where @xmath119 is given by eq .
( [ eq : rhomunu ] ) with the replacements in eqs .
( [ eq : xider ] ) and ( [ eq : rhoder ] ) , and @xmath120 is the usual electromagnetic field tensor . in principle , a similar term proportional to @xmath121 is also possible , although vmd applied to the pion form factor suggests that the coefficient of this term is small @xcite .
furthermore , the photon can couple to the nucleon as @xmath122 where @xmath123 with @xmath124 and @xmath125 the anomalous magnetic moments of the proton and the neutron , respectively . the first term in eq .
( [ newterms ] ) is needed to reproduce the magnetic moments of the nucleons , and the second one contributes to the rms charge radii .
note that these terms have @xmath77 and @xmath126 , respectively .
we may now write down the resulting manifestly gauge - invariant lagrangian , which contains the preceding `` tilde '' fields . instead of showing this lagrangian in its entirety , however , we omit terms that are irrelevant to our applications here .
the remaining lagrangian is manifestly chiral invariant , except for some terms involving the photon field that are proportional to the electric charge .
we find @xmath127 it is well known that the isoscalar - vector @xmath103 meson is needed to describe the short - range repulsion of the nn interaction . to have a realistic description of nuclear systems , the @xmath103 meson should be included .
note that once the @xmath103 meson is introduced , an isoscalar - vector four - fermion term @xmath128 can be completely accounted for by adjusting @xmath129 to fit experimental data , as discussed above for the rho meson .
with respect to chiral @xmath130 symmetry , the @xmath103 meson can be treated as a chiral singlet represented by a vector field @xmath131 .
the relevant lagrangian can be written as @xmath132 where @xmath133 .
the kinetic terms again reflect the spin-1 nature of the particle .
analyses of nn scattering show that the @xmath103 tensor coupling is not important , since @xmath134 is small , and this is indeed also the case from our fits ( see below ) .
the factor of @xmath135 in the coupling to the photon arises naturally from @xmath136 symmetry and the assumption of ideal mixing of the vector @xmath137 and @xmath103 mesons .
the ellipsis represents higher - order terms involving additional powers of the fields and their derivatives , some of which will be considered explicitly in sec .
ii.e . before ending this section
, we note that it is consistent to integrate out the vector @xmath138 field because it has a small coupling to the nucleon .
the effects of the vector @xmath137 in the isoscalar - vector channel can be absorbed into @xmath129 , as discussed near the end of sec .
for example , if we integrate out the @xmath137 , we obtain a contribution to @xmath139 of @xmath140 and a four - fermion term @xmath141 with a coefficient proportional to @xmath142 , where @xmath143 is its coupling constant to the nucleons and @xmath144 is its mass .
the contribution of this latter term is absorbed into the renormalization of @xmath129 . in general , one
need only keep the lowest - lying meson in a given channel at low energy .
unlike the vector channels , the absence of well - defined , low - lying resonances in the isoscalar - scalar channel makes the dynamics more difficult to identify and to model .
it is well known from the empirical nn scattering amplitude@xcite , from one - boson - exchange models of the nn interaction@xcite , and from fits to the properties of finite nuclei @xcite that there is a strong , mid - range nn attraction arising from the isoscalar - scalar channel , with a range corresponding to a mass of roughly 500 mev .
it has also been shown in calculations based on either phenomenological scattering amplitudes @xcite or explicit chirally symmetric models @xcite that an attraction of the appropriate strength can be generated dynamically by correlated two - pion exchange between nucleons .
no nearby underlying resonance at the relevant mass is needed . nevertheless , this dynamics is efficiently , conveniently , and adequately represented by the exchange of a light isoscalar - scalar degree of freedom .
therefore , in analogy to what we did with the vector mesons , we will introduce an effective scalar meson with a yukawa coupling to the nucleon to incorporate the observed mid - range nn attraction , and thereby avoid the explicit computation of multi - loop pion contributions . based on the results of the two - pion - exchange calculations and the success of one - boson - exchange models ,
it is sufficient to give the scalar a well - defined mass , which we will choose here by fitting to properties of finite nuclei .
moreover , the light scalar is a chiral singlet , so it does not spoil the chiral symmetry of the model .
in addition to the evidence from the nn interaction , one can ask what other information exists on the dynamics in the isoscalar - scalar channel . a natural place
to look is the physics of broken scale invariance in qcd .
this leads to an anomaly in the trace of the qcd energy - momentum tensor and also to `` low - energy theorems '' on connected green s functions constructed from this trace @xcite . in pure - glue qcd
, these low - energy theorems can be satisfied by introducing a scalar glueball field ( `` gluonium '' ) with a mass of roughly 1.6 gev .
when light quarks are added and the chiral symmetry is dynamically broken , the trace anomaly can acquire a contribution in addition to that from gluonium . in ref .
@xcite , we considered this additional contribution to arise from the self - interactions of the light scalar mentioned above , and used the low - energy theorems to constrain these interactions .
the lagrangian , which contained the light scalar , heavy scalar ( glueball ) , and other degrees of freedom , was scale invariant , except for some minimal scale - dependent self - interactions among the glueballs and the light scalar .
since the glueballs are heavy , they can be integrated out along with other heavy particles and high - momentum modes .
if one assumes no mixing between the scalar and the glueball , the scalar potential is constrained to satisfy the low - energy theorems ( at tree level ) , and the potential is given by@xcite @xmath145 here @xmath146 is the vacuum expectation value of the light scalar field @xmath1 , @xmath147 is its mass , and @xmath1 is assumed to have scale dimension @xmath58 ; that is , when @xmath148 , @xmath149 .
if we introduce the fluctuation field @xmath150 and expand @xmath151 as a power series in @xmath137 , the coefficients are all determined by the three parameters @xmath146 , @xmath147 , and @xmath58 .
the results of ref .
@xcite showed , however , that finite - nucleus observables are primarily sensitive only to the first three terms in this expansion ; thus , even in this extremely simple model of the scale breaking , the low - energy theorems do not place significant constraints on the self - interactions of the light scalar field .
if we allow for the possibility of scalar glueball mixing and employ field redefinitions to simplify the interaction terms between the light scalar and the nucleons , the form of the potential @xmath151 can be modified and additional parameters need to be introduced .
therefore , to take advantage of a simple yukawa coupling to the nucleon , we adopt here a general potential for the scalar meson , which can be expanded in a taylor series : @xmath152 here we have anticipated the discussion of naive dimensional analysis in the next subsection by including a factor of @xmath153 for each power of @xmath137 ; these factors are then eliminated in favor of @xmath154 .
various counting factors are also included , as discussed in the next subsection .
if the naturalness assumption is valid , the parameters @xmath155 , @xmath156 , should be of order unity .
this is indeed supported by our fits to finite nuclei , as shown in sec .
v. as with the vector mesons discussed earlier , only one low - mass scalar field is needed to a good approximation .
from a different point of view , chiral symmetry allows for fermion contact terms , such as @xmath157 , @xmath158 ^ 2 $ ] , @xmath159 , and so on .
these terms are relevant if there are strong interactions between nucleons in the isoscalar - scalar channel . by introducing a scalar field with mass @xmath147 , yukawa coupling @xmath160 , and nonlinear couplings @xmath155 ,
@xmath156 , adjusted to fit experimental data , we can account for all of the contact terms .
the only relevant questions are which implementation allows for the most efficient truncation and how many terms are needed in practical calculations .
we shall not include an isovector - scalar field in our effective lagrangian since the nn interaction in that channel is weak @xcite .
there is no meson with these quantum numbers with a mass below 1 gev , and two ( identical ) pions in a @xmath161 state can not have @xmath162 .
the lagrangian constructed thus far is based on symmetry considerations and the desire to reproduce realistic electromagnetic and nn interactions while keeping pion - loop calculations to a minimum .
we have illustrated the lowest - order terms in powers of the fields and their derivatives , but in principle , an infinite number of terms is possible , and we must have a meaningful way to truncate the effective lagrangian for the theory to have any predictive power . a naive dimensional analysis ( nda ) for assigning a coefficient of the appropriate size to any term in an effective lagrangian has been proposed by georgi and manohar@xcite .
the basic assumption of `` naturalness '' is that once the appropriate dimensional scales have been extracted using nda , the remaining overall dimensionless coefficients should all be of order unity .
for the strong interaction , there are two relevant scales : the pion - decay constant @xmath163 and a larger scale @xmath164 , which characterizes the mass scale of physics beyond goldstone bosons .
the nda rules for a given term in the lagrangian density are 1 . : : include a factor of @xmath153 for each strongly interacting field .
2 . : : assign an overall factor of @xmath165 .
3 . : : multiply by factors of @xmath166 to achieve dimension ( mass)@xmath167 . the appropriate mass for @xmath168 may be the nucleon mass @xmath65 or a non - goldstone boson mass ;
the difference is not important for verifying naturalness but can be relevant for @xmath169 counting arguments @xcite .
as noted by georgi@xcite , rule 1 states that the amplitude for generating any strongly interacting particle is proportional to the amplitude @xmath170 for emitting a goldstone boson .
rule 2 can be understood as an overall normalization factor that arises from the standard way of writing the mass terms of non - goldstone bosons ( or , in general , the quadratic terms in the lagrangian ) .
for example , one may write the mass term of an isoscalar - scalar field @xmath171 as @xmath172 where the scalar mass @xmath147 is treated as roughly the same size as @xmath168 . by applying rule 1 and extracting the overall factor of @xmath165 ,
the dimensionless coefficient is of @xmath173 .
terms with gradients ( or external fields ) will be associated with one power of @xmath166 for each derivative ( or field ) as a result of integrating out physics above the scale @xmath168 .
( a simple example is the expansion at low momentum of a tree - level propagator for a heavy meson of mass @xmath174 , which leads to terms with powers of @xmath175 . )
it is because of these @xmath166 suppression factors and dimensional analysis that one arrives at rule 3 .
after the dimensional nda factors and some appropriate counting factors are extracted ( such as @xmath176 for the @xmath177 power @xmath178 ; see below ) , the naturalness assumption implies that any overall dimensionless coefficients should be of order unity .
without such an assumption , an effective lagrangian will not be predictive .
until one can derive the effective lagrangian from qcd , the naturalness assumption must be checked by fitting to experimental data . as an example of nda and naturalness
, we can estimate the sizes of the coupling constants @xmath179 , @xmath160 , and @xmath129 by applying the rules to the following interaction terms : @xmath180 , @xmath181 , and @xmath182 .
note that the pion field in @xmath183 is already associated with an overall factor of @xmath153 [ see eq .
( [ eq : leading ] ) ] so @xmath184 is simply counted as a derivative .
it is straightforward to find that @xmath185 that is , eq .
( [ eq : ggg ] ) gives the natural sizes of these couplings , where we use the symbol @xmath186 to denote order of magnitude .
was originally proposed by georgi and manohar @xcite and subsequently refined to @xmath187 , where @xmath188 is the number of light quark flavors @xcite .
qcd appears to essentially saturate this bound .
] it is reassuring that the preceding values are indeed consistent with phenomenology . according to nda , a generic term in the effective lagrangian involving the isoscalar fields and
the nucleon field can be written as ( generalizing to include the pion , rho , and photon is straightforward ) @xmath189 which includes @xmath190 nucleon bilinears , @xmath191 scalar fields , @xmath59 vector fields , and @xmath192 derivatives .
gamma matrices and lorentz indices have been suppressed .
the coupling constant @xmath54 is dimensionless [ and of @xmath173 if naturalness holds ] .
numerical counting factors for terms with multiple powers of meson fields are included because the nda rules are actually meant to apply to the tree - level amplitude generated by the corresponding vertex .
thus , a term containing @xmath193 , for example , should have a factor @xmath194 associated with it .
these additional factors are certainly relevant when , say , @xmath195 . an explicit illustration of the origin of the counting factors can be found in the pion mass term in chpt : @xmath196 an important feature of the effective lagrangian is that each derivative on a field is associated with a factor of @xmath166 . thus organizing
the lagrangian in powers of derivatives can be a good expansion for describing processes where the dominant energies and momenta are not too large .
nevertheless , this organization is intimately connected with the important issue of how to remove redundant terms in the effective lagrangian . by employing the freedom to redefine the fields , one can organize the lagrangian such that there are no terms with @xmath197 acting on a boson field or @xmath198 acting on a fermion field , _ except _ in the kinetic - energy terms .
these field redefinitions do not change the on - shell @xmath1-matrix elements@xcite . in finite - density applications
, however , the non - goldstone bosons exchanged by the nucleons are far off shell , which motivates a different definition of the corresponding field variables .
short - distance effects due to heavier particles , high - momentum modes , and dynamical vacuum contributions are assumed to be integrated out , so that their consequences appear only indirectly in the coefficients of the lagrangian .
derivatives on the non - goldstone fields are already small compared to their masses in our applications because the momenta are limited by the fermi momentum .
thus , we do not need to eliminate these derivatives . instead
, we can use nonlinear field redefinitions to simplify the form of the interaction terms between the nucleons and the non - goldstone bosons so that they take the conventional yukawa form .
this ensures the consistency of the power counting rule ( [ eq : order ] ) .
recall that one of the main purposes for introducing the non - goldstone bosons is to describe the intermediate - range nn interactions . since the nonlinear field redefinitions do not change the quadratic kinetic and mass terms
, no changes occur to the ranges of the one - boson - exchange interactions in free space .
all other heavy boson - exchange contributions are short - ranged , so we can efficiently parametrize them using meson self - couplings .
we can easily show the invariance of the one - loop energy functional under field redefinitions , although a general proof to all orders seems difficult .
it is still convenient to use the field redefinitions to remove all derivatives on the nucleon fields except in the kinetic term .
these derivatives can be large because valence - nucleon energies are of the order of the nucleon mass @xmath65 . at finite density
, the fermi momentum appears as a new scale and meson fields develop expectation values .
we must then also decide how to organize the nonderivative terms .
the key observation is that at normal nuclear - matter density ( or in the center of a heavy nucleus ) , the scalar and vector mean - fields are typically @xmath199 and the gradients of the mean fields in the nuclear surface , where they are largest , are @xmath200 ( note that the _ magnitudes _ of the mean fields are reduced in the surface . ) equations ( [ eq : fdest ] ) and ( [ eq : derest ] ) are our basis for estimating the leading contributions to the energy of any term in our lagrangian . to summarize , the assumption of naturalness and the observation that the mean fields and their derivatives are small compared to @xmath65 up to moderate densities allows us to organize the lagrangian in powers of the fields and their derivatives .
we will take advantage of the freedom to redefine the fields to simplify the interaction terms between the nucleons and the non - goldstone bosons .
for the nucleons , pions , and photons , we use field redefinitions to remove high - order derivatives , whereas the scalar and vector meson parts of the lagrangian can contain powers of @xmath197 in addition to the kinetic terms , since these terms are small . by truncating the lagrangian at a given order in @xmath201 , we have a finite number of parameters to determine through fitting to nuclear observables .
based on the preceding discussion , we illustrate here how nonlinear field redefinitions allow us to simplify the terms involving interactions between the non - goldstone bosons and the nucleons .
derivatives on the nucleon field other than the kinetic term are removed by redefining the field because these derivatives on the nearly on - shell nucleons are of @xmath202 .
for the photons and the goldstone pions , field redefinitions are used to remove higher derivatives @xcite , so that their interactions with the nucleons are unconstrained except by gauge invariance and chiral symmetry . through @xmath126 ,
the part of the effective lagrangian involving nucleons can then be written as @xmath203 & & -{f_\rho g_\rho \over 4m}\overline n \rho_{\mu\nu } \sigma^{\mu\nu } n -{\fomega \gomega \over 4m}\overline n\omega_{\mu\nu } \sigma^{\mu\nu }
n - { \kappa_\pi\over m } \overline{n } v_{\mu\nu}\sigma^{\mu\nu } n \nonumber \\[3pt ] & & -{e \over 4m}f_{\mu\nu } \overline n \lambda \sigma^{\mu\nu } n -{e\over 2m^2 } \overline{n}\gamma_{\mu}(\beta_{\rm s } + \beta_{\rm v}\tau_3)n \partial_\nu f^{\mu\nu } \ , \label{eq : lnucl}\end{aligned}\ ] ] where the covariant derivative is @xmath204 one might expect more general isoscalar - scalar couplings to the nucleons .
however , all possible isoscalar - scalar field combinations of the non - goldstone boson fields that can couple to @xmath2 can be redefined in terms of a single isoscalar - scalar field with a yukawa coupling @xmath205 .
for example , the following couplings are redundant : @xmath206 the redefinition of the field induces changes in the coefficients of the mesonic part of the lagrangian , but since all terms to a given order in @xmath201 are already included , there is no loss of generality .
a similar observation holds for the isoscalar - vector and the isovector - vector couplings .
a term of the form @xmath207 is irrelevant because the baryon current is conserved .
moreover , a coupling of the form @xmath208 can be removed from the hamiltonian by using the constraint equation for the vector field . the identity @xmath209 and the equations of motion allow us to reduce a term of the form @xmath210 in the lagrangian to a sum of yukawa and tensor terms and the term in ( [ eq : dv ] ) in the hamiltonian .
as noted , bilinears in the nucleon fields and derivatives acting on such bilinears can be eliminated in favor of low - mass non - goldstone bosons .
furthermore , we can use eq .
( [ eq : relation ] ) and partial integrations to reduce mixed - derivative terms such as @xmath211 to the form @xmath212 , which we assume is small compared to terms that are retained based on the minor role of tensor mesons in meson - exchange phenomenology . in eq .
( [ eq : lnucl ] ) , we have omitted the following @xmath126 terms : @xmath213 the leading effect of the terms in ( [ eq : p5terms ] ) is a slight modification of the tensor couplings @xmath68 and @xmath134 at finite density due to the nonvanishing expectation value of the scalar field . however , the effects of the tensor terms are themselves already small , so the further modification is negligible .
thus , while we may keep terms through @xmath126 in the lagrangian , not all @xmath126 terms contribute equally in a given application .
in fact , the pion terms do not contribute to the energy of nuclei at tree - level .
the mesonic part of the lagrangian is also organized in powers of the fields and their derivatives . keeping terms through @xmath126 , we find @xmath214 & & \null -{1\over 2}{\rm tr}\ , ( \rho_{\mu\nu}\rho^{\mu\nu } ) -{1 \over 4}\big ( 1 + \alpha_2 { g_{\rm s}\phi\over m } \big ) \omegav_{\mu\nu}\omegav^{\mu\nu } -g_{\rho \pi\pi}{2\fpi^2 \over m_\rho^2 } { \rm tr}\,(\rho_{\mu\nu}v^{\mu\nu } ) \nonumber \\[3pt ] & & \null -2ef_{\pi}^2a^{\mu}{\rm tr}\,(v_{\mu}\tau_3 ) -{e \over 2g_{\gamma}}f_{\mu\nu}\big[\ , { \rm tr}\,(\tau_3\rho^{\mu\nu } ) + { 1\over 3}\,\omegav^{\mu\nu } \big ] \nonumber \\[3pt ] & & \null + { 1\over 2}\bigg ( 1 + \eta_1 { g_{\rm s}\phi\over m } + { \eta_2\over 2 } { g_{\rm s}^2\phi^2\over m^2 } \bigg ) \momega^2 \omegav_{\mu}\omegav^{\mu } + { 1\over 4!}\zeta_0 \gomega^2 ( \omegav_{\mu}\omegav^{\mu})^2 \nonumber \\[3pt ] & & \null + \bigg ( 1 + \eta_\rho { g_{\rm s}\phi\over m } \bigg ) m_\rho^2 { \rm tr}\,(\rho_{\mu}\rho^\mu ) - m_{\rm s}^2\phi^2\bigg ( { 1\over 2}+{\kappa_3\over 3!}\ , { g_{\rm s}\phi \over m } + { \kappa_4 \over 4!}\ , { g_{\rm s}^2\phi^2\over
m^2}\bigg ) \ ,
\label{eq : nlag}\end{aligned}\ ] ] where , apart from conventional definitions of some couplings ( @xmath215 , @xmath129 , and @xmath216 ) and the masses , we have defined the parameters so that they are of order unity according to naive dimensional analysis .
( again , nuclei will test this hypothesis for us ! ) also , since the expectation value of the @xmath217 field is typically an order of magnitude smaller than that of the @xmath218 field , we have retained nonlinear @xmath217 couplings only through @xmath77 . note that the @xmath219 and @xmath220 terms have @xmath221 . however , from eqs .
( [ eq : fdest ] ) and ( [ eq : derest ] ) we can estimate that these two @xmath221 terms are numerically of the same magnitude as the quartic scalar term in the nuclear surface energy , so we have retained them . thus , numerical factors such as @xmath222 , which are cancelled in scattering amplitudes , are relevant in deciding the importance of a term in the energy .
our full lagrangian through @xmath126 is then @xmath223 for applications to normal nuclei , keeping these terms should be sufficient if the parameters are natural .
this may be checked by including the following @xmath221 terms : @xmath224 as discussed below , the inclusion of @xmath225 improves our fits only marginally .
for an effective lagrangian to be useful for calculations of nuclear matter and finite nuclei , it should incorporate the low - energy structure of the pion and the nucleon .
this physics has been extensively studied ; there is a long history of vector - meson dominance ( vmd ) descriptions of electromagnetic form factors @xcite .
we illustrate here how the electromagnetic form factors of the pion and the nucleon at low momenta arise from our effective lagrangian through vmd and also through direct couplings of photons to pion and nucleon fields @xcite . the electromagnetic ( em ) current can be obtained from the lagrangian ( [ eq : lnucl ] ) , ( [ eq : nlag ] ) , and ( [ eq : fulll ] ) by taking @xmath226 after some partial integration : @xmath227 \nonumber \\[4pt ] & & + { 1\over g_\gamma}(\partial_\nu \rho_3^{\mu\nu } + { 1\over 3 } \partial_\nu\omegav^{\mu\nu } ) + 2 \fpi^2{\rm tr}(v^\mu\tau_3 ) \ .
\label{eq : emcrnt}\end{aligned}\ ] ] note that the photon can couple to two or more pions either directly or through the exchange of a neutral rho meson at non - zero momentum transfers .
the coupling of the photon to the nucleons is also direct or through the exchange of neutral vector mesons ( rho or omega ) .
we can determine the tree - level em form factor of the pion @xmath228 from the current ( [ eq : emcrnt ] ) and the lagrangian ( [ eq : nlag ] ) .
for spacelike momentum transfers @xmath229 , we have @xmath230 thus the mean - square charge radius of the pion is @xmath231 where we have used the experimental values@xcite @xmath232 which reproduce the rho widths @xmath233mev and @xmath234kev , respectively .
the observed charge radius is @xmath235 .
similarly , we can determine the em isoscalar and isovector charge form factors of the nucleon , @xmath236 f_1^{\rm v}(q^2 ) & = & { 1\over 2 } - { \beta_{\rm v}\over 2}\,{q^2 \over m^2 } -{g_\rho\over 2g_\gamma}\ , { q^2 \over q^2+m_\rho^2}+\cdots \ , \end{aligned}\ ] ] and the anomalous form factors @xmath237 f_2^{\rm v}(q^2 ) & = & { \lambda_p-\lambda_n\over 2 } -{f_\rho g_\rho\over 2g_\gamma}\ , { q^2 \over q^2+m_\rho^2}+\cdots \ .\end{aligned}\ ] ] the corresponding mean - square charge radii are @xmath238 \langle r^2 \rangle_{{\rm v}1 } & = & 6\biggl ( { \beta_{\rm v}\over m^2 } + { g_\rho\over g_\gamma m_\rho^2}\biggr)\ , \end{aligned}\ ] ] and @xmath239 \langle r^2 \rangle_{{\rm v}2 } & = & { 6\over \lambda_p-\lambda_n}\ , { f_\rho g_\rho\over g_\gamma m_\rho^2 } \ .
\label{eq : size2}\end{aligned}\ ] ] we will fit @xmath129 and @xmath66 to the properties of finite nuclei and then choose @xmath139 , @xmath240 , and @xmath68 to reproduce the experimental charge radii of the nucleon@xcite : @xmath241 we do not use eq .
( [ eq : sizze ] ) to fix @xmath242 because the remaining radius @xmath243 is not sufficiently well determined @xcite .
as in refs .
@xcite and @xcite , we have a contribution from vector dominance and a correction to the intrinic structure to second order in a derivative expansion , that is , to order @xmath244 .
this correction is adequate for our applications to nuclear structure ; applications involving higher momenta would require us to include additional nonrenormalizable interaction terms @xcite and loop effects . thus we do not expect our nucleon form factors to be accurate at large momentum transfers .
we work at one - baryon - loop order in this paper , which is equivalent to the dirac hartree approximation @xcite
. this should be sufficient to test the consistency of nda and the truncation of our lagrangian . as discussed in ref .
@xcite , a hartree calculation can be viewed as equivalent to a density - functional approach , in which higher - order many - body corrections are treated approximately .
based on dirac brueckner hartree fock calculations of nuclear matter , exchange and correlation corrections to the nucleon self - energies are expected to be small @xcite .
the stability of the hartree results will be tested in a two - loop calculation , which will be presented elsewhere .
the derivation of the hartree equations and energy functional for nuclei is similar to that described in ref . @xcite . because of the large masses of the nucleons and non - goldstone bosons , loop integrals involving these particles include dynamics from distance scales that are much shorter than the scale set by the valence - nucleon momenta , which are limited by the fermi momentum @xmath245 .
as discussed above , these short - range effects are included implicitly in the coefficients of the lagrangian .
formally , one can include counterterms to remove these loop effects to all orders , which is always possible , since _ all such terms are already contained in the effective lagrangian_. our present approach differs from ref .
@xcite in that we take a more general point of view and employ field redefinitions to simplify the couplings between the nucleon and the non - goldstone bosons .
the structure of the nucleon is now described _ within the theory _ , so that no external electromagnetic form factors are needed , as in earlier calculations @xcite . by taking @xmath155 , @xmath156 , etc . in the scalar potential as free parameters ,
we also allow for a more general description of the dynamics in the isoscalar - scalar channel , as well as for modifications from field redefinitions .
the single - particle dirac hamiltonian can be written as @xcite @xmath246 & & -{i\over 2m}\beta \bbox{\alpha\,\cdot\ , } ( f_\rho{\textstyle{1\over 2}}\tau_3\bbox{\nabla}r + \fomega\bbox{\nabla}w ) + { 1\over 2m^2 } ( \beta_{\rm s } + \beta_{\rm v}\tau_3 ) \bbox{\nabla}^2a -{i\over 2m}\lambda\beta \bbox{\alpha\,\cdot\ , } \bbox{\nabla}a \ , \end{aligned}\ ] ] where @xmath247 , @xmath248 , and we have defined scaled mean fields by including the couplings @xcite : @xmath249 \phi({\bf x } ) & = & g_{\rm s}\phi_0({\bf x})\ , \label{eq : phi } \\[4pt ] r({\bf x } ) & = & g_\rho b_0({\bf x})\ , \label{eq : r } \\[4pt ] a({\bf x } ) & = & e a_0({\bf x})\ .
\label{eq : a}\end{aligned}\ ] ] the dirac equation with eigenvalues @xmath250 and eigenfunctions @xmath251
is@xcite @xmath252 we follow the notation and conventions of ref .
@xcite and write the eigenfunctions for spherically symmetric nuclei as @xmath253 where @xmath254 is a spin spherical harmonic , @xmath255 for protons , and @xmath256 for neutrons .
the radial equations for @xmath257 and @xmath258 become @xmath259 f_a(r ) - u_3 g_a(r ) & = & 0 \ , \\
\biggl({d\over dr } - { \kappa\over r}\biggr ) f_a(r ) + [ e_a - u_1(r ) - u_2(r ) ] g_a(r ) + u_3 f_a(r ) & = & 0 \ , \end{aligned}\ ] ] where we have defined the single - particle potentials @xmath260 \bigr\}\ , \end{aligned}\ ] ] and a prime denotes a radial derivative .
the mean - field equations for @xmath261 , @xmath262 , @xmath25 , and @xmath263 are @xmath264 & & \null + { g_{\rm s}^2 \over 2 m } \big(\eta_1+\eta_2{\phi\over m } + { \eta_3\over 2}{\phi^2\over m^2}\big ) { \momega^2\over \gomega^2 } w^2 + { \zeta_1 g_{\rm s}^2\over 4 !
\gomega^2}{w^4\over m } \nonumber \\[4pt ] & & \null + { g_{\rm s}^2 \eta_\rho \over 2 m } { m_\rho^2 \over g_\rho^2 } r^2 + { \alpha_1\over 2 m } [ ( \bbox{\nabla}\phi)^2 + 2\phi{\bbox{\nabla}}^2 \phi ] + { \alpha_2g_{\rm s}^2\over 2m\gomega^2}(\bbox{\nabla}w)^2 \ , \\[4pt ] -{\bbox{\nabla}}^2 w + \momega^2 w & = & \gomega^2 [ \rho_{\rm b}({\bf x } ) + { \fomega\over 2m}\bbox{\nabla}\cdot ( \rho_{\rm b}^{\rm t}({\bf x})\widehat r ) ] -\big(\eta_1+{\eta_2\over 2}{\phi\over m } + { \eta_3\over 3!}{\phi^2\over m^2}\big){\phi\over m } \momega^2 w \nonumber \\[4pt ] & & \null -{1\over 3!}\big ( \zeta_0 + \zeta_1{\phi\over m}\big)w^3 + { \alpha_2 \over m } ( \bbox{\nabla}\phi\bbox{\cdot\nabla}w + \phi\bbox{\nabla}^2w ) -{e^2\gomega \over 3 g_\gamma}\rho_{\rm chg}({\bf x } ) \ , \\[4pt ] -{\bbox{\nabla}}^2 r + m_\rho^2 r & = & { 1\over 2}g_\rho^2[\rho_3({\bf x } ) + { f_\rho\over 2m}\bbox{\nabla}\cdot ( \rho_3^{\rm t}({\bf x})\widehat r ) ] -\eta_\rho { \phi\over m } m_\rho^2 r -{e^2g_\rho \over g_\gamma } \rho_{\rm chg}({\bf x } ) \ , \\[4pt ] -{\bbox{\nabla}}^2 a & = & e^2 \rho_{\rm chg}({\bf x } ) \ .\end{aligned}\ ] ] here the various densities are defined as @xmath265 \rho_{\rm b}({\bf x } ) & = & \sum_{\alpha}^{\text{occ } } \psi^\dagger_{\alpha}({\bf x})\psi_{\alpha}({\bf x } ) = \sum_a^{\text{occ } } { 2j_a+1 \over 4\pi r^2 } \bigl ( g_a^2(r ) + f_a^2(r ) \bigr ) \ , \\[4pt ] \rho_{\rm b}^{\rm t}({\bf x } ) & = & \sum_{\alpha}^{\text{occ } } \psi^\dagger_{\alpha}({\bf x } ) i\beta \bbox{\alpha\cdot\widehat r } \psi_{\alpha}({\bf x } ) = \sum_a^{\text{occ } } { 2j_a+1 \over 4\pi r^2 } 2 g_a(r ) f_a(r ) \ , \\[4pt ] \rho_3({\bf x } ) & = & \sum_{\alpha}^{\text{occ } } \psi^\dagger_{\alpha}({\bf x})\tau_3\psi_{\alpha}({\bf x } ) = \sum_a^{\text{occ } } { 2j_a+1 \over 4\pi r^2 } ( 2 t_a ) \bigl ( g_a^2(r ) + f_a^2(r ) \bigr ) \ , \\[4pt ] \rho_3^{\rm t}({\bf x } ) & = & \sum_{\alpha}^{\text{occ } } \psi^\dagger_{\alpha}({\bf x } ) i\tau_3\beta \bbox{\alpha\cdot\widehat r } \psi_{\alpha}({\bf x } ) = \sum_a^{\text{occ } } { 2j_a+1 \over 4\pi r^2 } ( 2 t_a ) 2 g_a(r ) f_a(r ) \ , \end{aligned}\ ] ] where the summation superscript `` occ '' means that the sum runs only over occupied ( valence ) states in the fermi sea . evidently , these densities depend only on the radial coordinate .
the charge density is given by @xmath266 where the `` direct '' nucleon charge density is @xmath267 \ , \label{eq : direct}\ ] ] and the vector meson contribution is @xmath268 here the `` point '' proton density and nucleon tensor density are determined ( as always ) from the solutions to the dirac equation in the mean fields : @xmath269 with the anomalous magnetic moment @xmath270 from eq .
( [ eq : lambda ] ) . as seen in the next section , the vector meson contribution and the final term in eq .
( [ eq : direct ] ) generate the nucleon form factor that reduces the fluctuations of the point - nucleon charge densities .
note that the mean - field equations can be used to eliminate the second derivatives in the charge density in favor of first derivatives , which are more accurately calculated numerically .
the energy functional is given by @xmath271 where @xmath272 & & + { m_{\rm s}^2\over g_{\rm s}^2}{\phi^3\over m } \big({\kappa_3\over 12}+{\kappa_4\over 24}{\phi\over m } + { \kappa_5\over 80}{\phi^3\over m^3}\big ) -{\eta_\rho \over 4}{\phi\over m } { m_\rho^2\over g_\rho^2 } r^2 \nonumber \\[4pt ] & & -{\phi \over 4 m } \big(\eta_1+\eta_2{\phi\over m } + { \eta_3\over 2}{\phi^2\over m^2}\big ) { \momega^2 \over \gomega^2 } w^2 \nonumber \\[4pt ] & & -{1\over 4 !
\gomega^2 } \big(\zeta_0 + { 3\zeta_1\over 2}{\phi\over m}\big ) w^4 + { \alpha_1\over 4g_{\rm s}^2}{\phi\over m } ( \bbox{\nabla}\phi)^2 - { \alpha_2\over 4\gomega^2}{\phi\over m } ( \bbox{\nabla}w)^2 \ , \end{aligned}\ ] ] and we have applied the mean - field equations to arrive at this form . the center - of - mass ( c.m . )
correction to the nuclear binding energy can be estimated nonrelativistically using @xmath273 where @xmath274 is the baryon number of the nucleus , and @xcite @xmath275 with @xmath276 for simplicity , we use an empirical estimate given by reinhard@xcite : @xmath277 which determines @xmath278 as well from eq .
( [ eq : ecm ] ) .
the binding energy @xmath279 is then @xmath280 note that we can not replace @xmath281 by @xmath282 , the sum of the masses of the protons and neutrons , unless we use different masses for the proton and neutron in the dirac equation .
there are also c.m .
corrections to the nuclear charge density and charge radius .
we estimate these in analogy to the well - known nonrelativistic results for harmonic - oscillator wavefunctions , which are exact . the charge density in momentum space ,
_ i.e. _ , the charge form factor , becomes @xmath283 where @xmath284 is the fourier transform of the charge density @xmath285 .
we expect this correction to be accurate for momenta @xmath286 . with the c.m .
correction , the predicted charge density in coordinate space is , to a good approximation , @xmath287 the mean - square charge radius is given by @xmath288 where @xmath289 the energy density for uniform , symmetric nuclear matter in the one - loop approximation can be obtained from the preceding results by observing that the single - particle energy eigenvalue becomes @xmath290 where @xmath291 , and @xmath261 and @xmath262 are now constant mean fields [ see eqs .
( [ eq : phi ] ) and ( [ eq : w ] ) ] .
the energy density @xmath292 becomes @xmath293 & = & { m_{\rm s}^2\over g_{\rm
s}^2}\phi^2 \big({1\over 2}+ { \kappa_3\over 3!}{\phi\over m } + { \kappa_4\over 4!}{\phi^2\over m^2 } + { \kappa_5\over 5!}{\phi^3\over m^3}\big ) \nonumber \\[4pt ] & & \null -{1\over 2\gomega^2 } \big(1+\eta_1{\phi\over m}+{\eta_2\over 2}{\phi^2\over m^2 } + { \eta_3\over 3!}{\phi^3\over m^3}\big ) \momega^2 w^2 \nonumber \\[4pt ] & & \null - { 1\over 4 ! \gomega^2 } \big(\zeta_0 + \zeta_1 { \phi\over m}\big ) w^4 + w\rho_{\scriptscriptstyle\rm b } + { 4\over ( 2\pi)^3}\ !
\int^{k_{\scriptscriptstyle\rm f } } { \kern-.1em}{\rm d}^3{\kern-.1em}{k } \ , \sqrt{{\bf k}^2+{m^*}^2 } \ , \label{eq : nuclmatt}\end{aligned}\ ] ] where @xmath294 is the fermi momentum and @xmath295 is the baryon density .
an extensive analysis of energy densities of this form is given in ref .
one can also compute the bulk symmetry - energy coefficient @xcite @xmath296 where the effective rho mass @xmath297 is defined here by @xmath298
to test the physical ideas discussed in the preceding sections , we fit our one - loop results to the properties of finite nuclei .
we choose a set of spherical nuclei and solve the dirac and mean - field equations for each nucleus by iteration , until self - consistency is reached @xcite .
we then calculate a set of observables \{@xmath299 } for each nucleus and tune the parameters to minimize the generalized @xmath300 @xcite defined by @xmath301 ^ 2 \ , \label{eq : chisq}\ ] ] where @xmath302 runs over the set of nuclei , @xmath303 runs over the set of observables , the subscript `` exp '' indicates the experimental value of the observable , and @xmath304 are the relative weights .
the nuclei we choose are @xmath305o , @xmath306ca , @xmath307ca , @xmath308sr , and @xmath309pb .
similar fitting procedures have been standard for determining parameter sets for relativistic mean - field models @xcite . in the previous studies , however , the principal goal was to determine minimal parameter sets that provided a good description of bulk nuclear properties , so as to maximize predictability .
the set of observables used to make the fit was also kept at a minimum . in these fits ,
the parameters included only the scalar mass @xmath147 , the nucleon - meson couplings @xmath160 , @xmath310 , and @xmath311 , and two scalar self - interaction parameters , @xmath155 and @xmath156 .
( the exception is the point coupling model of ref .
@xcite , which has a corresponding set of nucleon - only point couplings . )
the nucleon and vector meson masses were generally fixed at experimental values .
the resulting parameter sets provide good fits to nuclear binding energies , charge distributions , and single - particle structure , for a wide variety of nuclei .
our effective lagrangian , truncated at @xmath126 , contains many additional parameters .
thus we expect to find good fits to nuclear properties . however , our goals are quite different from the previous studies .
we are not concerned with maximizing predictive power ; instead , we want to use experimental data to test whether the parameters of the effective lagrangian are , in fact , natural . in so
doing we want to gain insight into an appropriate level of truncation .
therefore we constrain the parameters by using every observable we believe should be well reproduced . the @xmath300 function in eq .
( [ eq : chisq ] ) serves as a figure of merit for our fit .
we do not expect our model to reproduce the experimental observables at the level of the actual experimental uncertainties in each case ; this means that a direct statistical interpretation of @xmath300 will not be meaningful .
rather we choose the weights to be the relative accuracy we expect in a good fit .
a `` perfect fit '' corresponds to a contribution of unity to @xmath300 from each observable . in choosing weights
we follow closely the prescription in ref .
@xcite , so that we will have a basis of comparision . different choices of weights will emphasize different aspects of the physics and lead to different `` optimal '' parametrizations .
however , we have tried a range of reasonable weights and the qualitative conclusions given below are independent .
the choice of observables starts with the basic physics of the semi - empirical mass formula , which should be reproduced by any reasonable model of nuclei . here
this is achieved directly by fitting to the binding energies of individual nuclei .
we adopt a relative weight of 0.15% for each binding energy .
this implies that binding energies per nucleon should be reproduced to two decimal places , with an uncertainty of roughly unity in the second place .
but it is also important that the _ systematics _ of the energies are followed ; otherwise , the minimization procedure may find reasonable fits to energies but with poor systematics ( particularly when fewer parameters are used , as with @xmath312 or @xmath77 truncations ) .
therefore we also use the deviation of experimental and calculated binding energies @xmath313 to determine @xmath314 , @xmath315 , and @xmath316 according to @xmath317 here @xmath318 and @xmath319 are the number of neutrons and protons in the @xmath320 nucleus and @xmath321 .
thus an exact fit to the energies would also have @xmath322 .
the deviations @xmath315 and @xmath316 are included as separate terms in eq .
( [ eq : chisq ] ) in the form @xmath323 ^ 2 $ ] , with @xmath324 .
( this weight was selected based on the results for low - order truncations ; it has no intrinsic significance . )
we also expect the effective theory to accurately reproduce low - momentum observables , which leads us to focus on the charge form factor .
we choose the rms charge radii of the nuclei and the so - called diffraction - minimum - sharp ( d.m.s . )
radii as observables ( another choice would have been the surface thickness @xcite ) .
the d.m.s .
radius of a nucleus is defined to be@xcite @xmath325 where @xmath326 is the three - momentum transfer at the first zero of the nuclear charge form factor @xmath327 with @xmath328 .
the d.m.s .
radius is expected to be a clean low - momentum quantity for a nucleus and should be well described .
the weights chosen for the radii correspond to an accuracy of roughly two decimal places .
finally , we expect to reproduce various features of the single - particle structure in nuclei .
we focus in particular on spin - orbit splittings near the fermi surface , which are reasonably well known for both proton and neutron levels .
these splittings are closely related to the size of the mean fields at the one - loop level . in a previous study , a good reproduction of these splittings
was found even when they were not included as observables @xcite .
however , our expanded parameter space means that this correlation may not follow without explicitly including them .
in addition , we use the proton @xmath329 energy level in @xmath309pb to fix the overall scale of the energy levels ; the positioning of this level helps to determine parameters that influence the symmetry energy .
finally , the proton splitting @xmath330 is chosen as a characteristic splitting between states with large and small @xmath190 , which constrains the shape of the effective single - particle potential @xcite .
the additional weight on observables for lead also leads to a more sensitive determination of the isovector parameters . to summarize , the observables ( a total of 29 ) and their relative weights are taken as follows : * the binding energies per nucleon @xmath331 , with a relative weight of 0.15% * the rms charge radii @xmath332 , with a relative weight of @xmath333 * the d.m.s .
radii @xmath334 , with a relative weight of @xmath335 * the spin - orbit splittings @xmath336 of the least - bound protons and neutrons , with a relative weight of @xmath97 for @xmath305o , @xmath337 for @xmath309pb , @xmath338 for @xmath306ca and @xmath307ca , and @xmath339 for @xmath308sr * the proton energy @xmath340 and the proton level splitting @xmath341 in @xmath309pb , with relative weights of @xmath97 and @xmath338 , respectively * the surface - energy and symmetry - energy deviation coefficients @xmath315 and @xmath316 [ see eq .
( [ eq : semf ] ) ] , each with a weight of @xmath342
. for further discussion of the relative weights for @xmath331 , @xmath343 , and @xmath336 , see ref.@xcite .
additional observables that might be added in future investigations include neutron radii , quadrupole deformations , monopole resonance energies , and an enlarged set of test nuclei .
we take the nucleon , @xmath103 , and @xmath217 masses to be given by their experimental values : @xmath344mev , @xmath345mev , and @xmath346mev . as a free parameter starting from the parameter sets in table [ tab : one ] leads to only a tiny improvement in @xmath300 ( less than unity ) and a small change in @xmath347 ( about 10 mev ) .
therefore we fix the @xmath103 and @xmath217 masses , since the latter also has little effect . ]
the anomalous magnetic moments of the nucleon are fixed at @xmath124 and @xmath125 .
we also fix @xmath348 as given in eq .
( [ eq : ggamma ] ) .
the free - space charge radii of the nucleon , as given in eqs .
( [ eq : nucsize1 ] ) and ( [ eq : nucsize2 ] ) , are used to fix @xmath139 , @xmath240 , and @xmath68 by solving eqs .
( [ eq : size1 ] ) to ( [ eq : size2 ] ) .
the remaining thirteen parameters @xmath160 , @xmath129 , @xmath311 , @xmath349 , @xmath350 , @xmath351 , @xmath155 , @xmath156 , @xmath352 , @xmath147 , @xmath134 , @xmath219 , and @xmath220 for the @xmath126 parametrization are then obtained by the optimization procedures described above .
this optimization program faces serious difficulties .
with such a large parameter set and because the observables themselves are highly correlated , there is a definite problem of underdetermination .
the minimization routines used here are reliable for determining local minima but are not guaranteed to find a global minimum , so we can not be sure to have sampled all relevant values of the parameters .
there are further complications in obtaining a more global fit because the required stability of the hartree iterations precludes large steps through parameter space .
thus the minimization routines must be efficient in navigating narrow valleys in parameter space . to make our results as robust as possible , we used two different codes and made many runs with different starting parameters .
# 1 .parameter sets from fits to finite nuclei .
[ cols="^ , > , > , > , > , > , > " , ] [ tab : five ] experience with a broad class of relativistic mean - field models shows that models that successfully reproduce bulk properties of finite nuclei share characteristic properties in infinite nuclear matter @xcite .
these properties are the equilibrium binding energy and density , the compression modulus @xmath353 , the value of @xmath354 at equilibrium , and the symmetry energy @xmath355 .
one further condition is that the light scalar mass satisfies @xmath356 .
this condition ensures reasonably smooth charge densities , good surface - energy systematics , and an appropriate shape for the single - particle potentials @xcite .
this is confirmed from the properties of nuclear matter predicted from our fits to finite nuclei , as shown in table [ tab : five ] for the parameter sets in table [ tab : one ] .
it is instructive to examine the contributions to the energy of nuclear matter from various terms in the energy density .
this is shown in fig .
[ fig : twelve ] , where the contributions to the energy per nucleon at equilibrium from the quadratic , cubic , and quartic powers of the fields @xmath261 and @xmath262 in eq .
( [ eq : nuclmatt ] ) are indicated separately .
we set the scale for the natural size of contributions expected from an @xmath177-order term using the power counting estimate [ eq .
( [ eq : generic ] ) ] for an @xmath177-order contact term and the approximation @xmath357 , with @xmath358 being the baryon density and the bracket indicating an expectation value .
these estimates are marked with crosses , and the error bars reflect a range of @xmath168 from 500 mev to 1 gev .
we observe that the estimates are consistent with the field energies found from the fits .
furthermore , these estimates also set the scales for the mean meson fields , which are consistent with eq .
( [ eq : fdest ] ) .
this is an indication that the naturalness assumption is valid ( see below ) .
several observations can be made based on the figure : * the contributions decrease steadily with increasing powers of the fields , so that a truncation is justified .
note that this decrease will become more gradual as the density is increased above equilibrium density .
* the nuclear matter binding energy ( indicated by the arrow ) is an order of magnitude smaller than the individual @xmath312 contributions .
this implies that the @xmath312 scalar and vector contributions cancel almost completely and also that @xmath126 contributions are still quite significant ( at least for set g1 ) .
* the @xmath221 contributions [ see eq . ( [ eq : fifth ] ) ] are representative results obtained when the @xmath126 fit is extended .
( they correspond to the lowest values of @xmath300 ) .
they are not well determined ; almost identical values of @xmath300 can be found for very different @xmath221 parameters .
nevertheless , even when unnatural @xmath221 parameters are introduced , their _ net _ contribution to the energy is typically a few tenths of an mev .
thus a truncation at @xmath126 is consistent and adequate .
* the systematics of the two parameter sets at @xmath126 are quite different
. set g1 parallels the systematics of the standard parameter sets from refs .
@xcite , while the @xmath126 terms for g2 are quite small .
this is possible because of the contributions from the @xmath219 and @xmath220 terms , which involve gradients of the meson fields and thus do not contribute in nuclear matter .
the contributions of these gradients [ note the signs of the @xmath359 and the signs in eq .
( [ eq : nlag ] ) ] to the spin - orbit splittings in finite nuclei allow set g2 to accurately reproduce the empirical spectra , even though @xmath360 in nuclear matter is larger than in conventional models .
if we return now to the parameters in table [ tab : one ] , we see that while the yukawa couplings and the scalar mass are close for the two sets , the meson self - couplings are very different .
this is consistent with our earlier hartree calculations @xcite , which showed that the bulk and single - particle observables of interest are determined to a large extent by a relatively small number of nuclear properties .
as noted above , by reproducing five important properties of nuclear matter and by choosing an appropriate scalar mass , one essentially fixes the predicted properties of doubly magic nuclei , and thus the thirteen parameters in the present model are underdetermined .
the parameters are therefore highly correlated and are sensitive to the relative weights chosen for the observables , including the spin - orbit splittings . to determine the parameters better , one must include additional observables that are sensitive to different aspects of the dynamics .
to explore these issues further and to put our @xmath126 fit in perspective , it is useful to find parameter sets for more constrained truncations .
if we optimize with the `` standard '' mean - field model , which includes the yukawa couplings , the scalar mass , and @xmath155 and @xmath156 , we find @xmath361 , and the parameter set ( which we label q1 ) is qualitatively close to the `` best - fit '' sets of the past and also to the corresponding parameters of set g1 ( see table [ tab : one ] ) . adding the @xmath126 vector self - interaction , with parameter @xmath352 ,
improves the @xmath300 to about 90 ( set q2 ) .
the @xmath300 values of sets g1 and g2 are slightly less than 50 , so it is clear that a better fit is obtained with more parameters , but not dramatically so .
if we truncate our model at @xmath312 ( set w1 ) , keeping only the yukawa couplings and the scalar mass and setting all other parameters to zero , the best @xmath300 is over 1700 .
the `` optimal '' set in this case depends strongly on the relative weights , because only a subset of the observables can be reproduced accurately .
the standard choice @xcite has been to reproduce the charge radii accurately at the expense of significant underbinding .
the present set of weights leads to much better binding energies ( despite a large compressibility ) but the systematics of the charge radii ( oxygen in particular ) is badly reproduced .
the spin - orbit splittings are too large in all cases , corresponding to a low value for @xmath360 ( see table [ tab : five ] ) . at @xmath77 ( set c1 ) , we add two parameters , @xmath155 and @xmath349 .
( we hold @xmath351 at zero for simplicity . )
if the number of parameters was the sole key to a good fit , we would expect @xmath300 to be comparable to that from the standard models .
however , the best fit has @xmath362 . in practice
the @xmath349 contribution plays the same role as @xmath156 if one looks at energy contributions in nuclear matter , but it is insufficient to accurately reproduce nuclear sizes .
it is likely that the density dependence of the two @xmath77 terms is too similar to provide enough flexibility for a good fit .
thus we find that we need at least @xmath126 terms to find a good fit to nuclear properties .
examples are sets q1 and q2 .
once we allow all @xmath126 terms , however , the fit is underdetermined , as is evident by the differences between our sets g1 and g2 .
one of the important new features are the @xmath219 and @xmath220 parameters , which accompany derivative corrections . without these
included in the fit , set g2 can not achieve a @xmath300 better than 100 . if we include @xmath221 parameters from eq .
( [ eq : fifth ] ) in the fit , only small improvements in @xmath300 ( @xmath363 units ) are found .
the net contribution to the energy from these terms is very small ( roughly a few tenths of an mev ) in these fits , even if the coefficients are artificially made large so that individual contributions are significant .
furthermore , the changes in the lower - order coefficients from the @xmath126 fits are small .
thus we find @xmath126 contributions are necessary and sufficient for this set of observables .
clearly the @xmath221 results , when included , do not drive the physics , and their effects can be absorbed into small changes in the lower - order couplings . the insensitivity to the @xmath221 terms validates our truncation of the lagrangian .
the parameters in table [ tab : one ] have been displayed in such a way that they should all be of order unity according to nda and the naturalness assumption .
this is seen to be the case . from set g1 .
its contribution to the energy is still compatible with natural estimates despite its size because of the @xmath364 counting factor .
] it is important to emphasize this nontrivial result : without the guidance from nda , these coefficients could be any size at all !
we conclude that nda and the naturalness assumption are indeed compatible with and implied by the observed properties of finite nuclei , even though we are absorbing many - body effects in the coefficients . however , the detailed mechanism by which the nuclear observables constrain the parameters to be natural is not transparent .
second - order ( @xmath312 ) coefficients vary only slightly , regardless of the number of parameters used .
the other parameters , while natural , are not as well determined .
we searched for unnatural parameter sets by starting from unnatural values , but the optimization always produced natural results .
the @xmath221 parameters can be made large , but they are tuned by the optimization so that they cancel almost completely in the hartree equations . despite the narrow range of densities present in ordinary nuclei , the different density dependence of various terms in the energy is apparently sufficient to enforce naturalness .
our goals in this paper are to construct an effective field theory based on hadrons that is appropriate for calculations of finite - density nuclear systems and to propose methods for making these calculations tractable .
the lagrangian is consistent with the underlying symmetries of qcd , and heavy , non - goldstone bosons are introduced to avoid the calculation of ( some ) virtual pion loops . to organize the lagrangian
, we rely on naive dimensional analysis ( nda ) , which allows us to extract the dimensional scales of any term , on the assumption of naturalness , which says that the remaining dimensionless coefficient for each term should be of order unity , and on the observation that the ratios of mean meson fields ( and their gradients ) to the nucleon mass ( which generically represents the `` heavy '' mass scale ) are good expansion parameters .
since the meson fields are roughly proportional to the nuclear density , and since the spatial variations in nuclei are determined by the momentum distributions of the valence nucleon wave functions , this organizational scheme is essentially an expansion in @xmath365 , where @xmath245 is a fermi wavenumber corresponding to normal nuclear densities . if nda and the naturalness assumption are valid , one can expand the lagrangian in powers of the fields and their derivatives , truncate at some finite order , and thus have some predictive power for the properties of nuclei .
our fits to bulk and single - particle nuclear properties show that this is indeed the case . as in all effective field theories ,
the coefficients in our nonrenormalizable lagrangian implicitly include contributions from energy scales heavier than our generic mass @xmath168 .
thus contributions from baryon vacuum loops and non - goldstone boson loops ( including , in principle , particles more massive than those considered here ) are parametrized in this way .
moreover , since we presently compute the nuclear observables at the one - baryon - loop or hartree level , fitting the parameters to nuclei means that they also implicitly include the effects of many - body corrections .
these consist of nucleon exchange and correlation contributions , which may involve virtual states with large excitation energies , modifications of the vacuum dynamics at finite density , like the nuclear lamb shift , and virtual pion loops , since the mean pion field vanishes in spherical nuclei .
these many - body effects involve both short- and long - range dynamics , so it is notable that our fitted parameters are still natural
. we can understand this result with the following arguments .
first , it is unlikely that natural parameters arise from a sensitive cancellation between unnatural parameters in the lagrangian and compensating large many - body effects .
it is much more likely that the parameters in the lagrangian are natural , and the modifications from many - body effects are small .
this is consistent with explicit calculations of relativistic exchange and correlation effects , which show that these contributions do not significantly modify the nucleon self - energies nor introduce large state dependence , at least for states in the fermi sea .
this `` hartree dominance '' of the mean fields @xcite and of the parameters needed to produce them is a crucial element in the expansion and truncation scheme described above .
an important topic for future work is the explicit evaluation of correlation effects in this model , both to disentangle these contributions from the fit parameters and to verify the extent to which hartree dominance is valid
. we can also understand why earlier relativistic hartree approaches to nuclear structure were successful .
many of these calculations used lagrangians that were truncated arbitrarily at quartic interactions in the meson fields , but we have seen that quintic and higher - order terms are unimportant ( and undetermined ) for the nuclear observables of interest .
moreover , even though most of these investigations arbitrarily set some of the cubic and quartic meson parameters to zero , we have also seen that the full complement of parameters through fourth order ( @xmath61 ) are underdetermined by the data .
therefore , keeping only a subset of the parameters does not preclude the possibility of a realistic fit to nuclei . in summary
, we have constructed an effective lagrangian that maintains the symmetries of qcd , such as lorentz invariance , parity conservation , chiral symmetry , and electromagnetic gauge invariance .
the bulk and single - particle properties of nuclei can be described realistically by this effective hadronic theory at the one - baryon - loop level .
naive dimensional analysis and the naturalness assumption are found to be compatible with the observed nuclear properties .
the fitted parameters are natural , and the results are not driven by the last terms kept .
the description of the spin - orbit splittings of the least bound nucleons and the shell structure are generally good , although the errors are relatively large compared with the other observables .
the long - range electromagnetic structure of the nucleon is included explicitly using vector - meson dominance and derivative couplings to the photon ; it is not necessary to introduce _ ad hoc _ electromagnetic form factors . the mesonic and gradient contributions are important for reducing fluctuations in the point - nucleon charge density , and the resulting nuclear charge form factors agree well with experiment .
the extension of these calculations beyond one - baryon - loop order is currently in progress .
we thank b. clark , p. ellis , s. jeon , l. mclerran , and j. rusnak for useful comments and stimulating discussions . this work was supported in part by the department of energy under contract no .
de fg0287er40365 , the national science foundation under grants no .
phy9511923 and phy9258270 , and the a. p. sloan foundation .
999 b. d. serot and j. d. walecka , adv .
* 16 * ( 1986 ) 1 .
b. d. serot , rep .
* 55 * ( 1992 ) 1855 .
r. j. perry , phys .
b * 199 * ( 1987 ) 489 .
r. j. furnstahl and c. j. horowitz , nucl .
* a485 * ( 1988 ) 632 .
r. j. furnstahl , r. j. perry , and b. d. serot , phys .
c * 40 * ( 1989 ) 321 .
lepage , `` what is renormalization ? '' , in _ from actions to answers _ ( tasi-89 ) , edited by t. degrand and d. toussaint ( world scientific , singapore , 1989 ) , p. 483 .
j. polchinski , in _ recent directions in particle theory : from superstrings and black holes to the standard model _
( tasi-92 ) , edited by j. harvey and j. polchinski ( world scientific , singapore , 1993 ) , p.235 .
h. georgi , ann .
. nuc . part .
sci . * 43 * ( 1993 ) 209 .
r. d. ball and r. s. thorne , ann .
( ny ) * 236 * ( 1994 ) 117 .
s. weinberg , _ the quantum theory of fields , vol .
i : foundations _ ( cambridge university press , new york , 1995 ) .
r. j. furnstahl and b. d. serot , phys .
c * 47 * ( 1993 ) 2338 ; phys .
b * 316 * ( 1993 ) 12 .
r. j. furnstahl , h .- b .
tang , and b. d. serot , phys .
c * 52 * ( 1995 ) 1368 .
r. j. furnstahl , b. d. serot , and h .- b .
tang , nucl .
* a598 * ( 1996 ) 539 .
s. adler and r. dashen , _ current algebras _
( benjamin , new york , 1968 ) . v. de alfaro , s. fubini , g. furlan , and c. rossetti , _ currents in hadron physics _ ( north - holland , amsterdam , 1973 ) .
s. weinberg , physica * a96 * ( 1979 ) 327 .
r. dashen and m. weinstein , phys .
* 183 * ( 1969 ) 1261 .
j. gasser and h. leutwyler , ann .
( ny ) * 158 * ( 1984 ) 142 ; nucl
. phys .
* b250 * ( 1985 ) 465 , 517 , 539 .
m. gell - mann and m. lvy , nuovo cim . * 16 * ( 1960 ) 705 . b. w. lee , _ chiral dynamics _ ,
( gordon and breach , new york , 1972 ) . t. d. lee and g. c. wick , phys .
d * 9 * ( 1974 ) 2291 .
e. k. heide , s. rudaz , and p. j. ellis , nucl .
phys . ( 1994 ) 713 .
g. carter , p. j. ellis , and s. rudaz , nucl .
* a603 * ( 1996 ) 367 .
s. weinberg , phys .
* 166 * ( 1968 ) 1568 .
g. ecker , j. gasser , a. pich , and e. de rafael , nucl .
* b321 * ( 1989 ) 311 . g. ecker , j. gasser , h. leutwyler , and e. de rafael , phys .
b * 223 * ( 1989 ) 425 .
j. f. donoghue , c. ramirez , and g. valencia , phys .
d * 39 * ( 1989 ) 1947 .
m. gell - mann and f. zachariasen , phys .
* 124 * ( 1961 ) 953 .
j. j. sakurai , _ currents and mesons _
( univ . of chicago , chicago , 1969 ) .
t. h. bauer , r. d. spital , d. r. yennie , and f. m. pipkin , rev .
( 1978 ) 261 .
g. e. brown , m. rho , and w. weise , nucl .
* a454 * ( 1986 ) 669 .
f. iachello , a. d. jackson , and a. lande , phys .
b * 43 * ( 1973 ) 191 . s. weinberg , phys .
b * 251 * ( 1990 ) 288 ; nucl .
* b363 * ( 1991 ) 3 ; phys .
b * 295 * ( 1992 ) 114 .
r. machleidt , adv .
* 19 * ( 1989 ) 189 .
a. d. jackson , d. o. riska , and b. verwest , nucl .
( 1975 ) 397 .
j. w. durso , m. saarela , g. e. brown , and a. d. jackson , nucl .
a278 * ( 1977 ) 445 .
j. w. durso , a. d. jackson , and b. j. verwest , nucl .
* a345 * ( 1980 ) 471 .
w. lin and b. d. serot , phys .
b * 233 * ( 1989 ) 23 .
w. lin and b. d. serot , nucl .
* a512 * ( 1990 ) 637 .
h. georgi , phys .
b * 298 * ( 1993 ) 187 .
r. j. furnstahl , b. d. serot , and h .- b .
tang , in preparation .
j. l. friar , d. g. madland , and b. w. lynn , phys .
c * 53 * ( 1996 ) 3085 .
j. zimanyi and s. a. moskowski , phys .
c * 42 * ( 1990 ) 1416 .
a. delfino , c. t. coelho , and m. malheiro , phys .
c * 51 * ( 1994 ) 2188 .
h. lenske and c. fuchs , phys .
b * 345 * ( 1995 ) 355 .
h. georgi , nucl .
* b363 * ( 1991 ) 339 .
s. coleman , j. wess , and b. zumino , phys .
* 177 * ( 1969 ) 2239 . c. callan , s. coleman , j. wess , and b. zumino , phys .
* 177 * ( 1969 ) 2247 .
bando , t. kugo , and k. yamawaki , phys .
rep . * 164 * ( 1988 ) 217 .
j. bijnens and e. pallante , mod .
( 1996 ) 1069 .
b. borasoy and u. g. meissner , preprint hep - ph/9511320 , nov .
v. bernard , n. kaiser , u. g. meissner , int .
j. mod .
e4 * ( 1995 ) , 193 .
h. georgi , _ weak interaction and modern particle theory _ ( benjamin / cummings , menlo park , ca , 1984 ) .
r. m. barnett et al .
d * 54 * ( 1996 ) 1 .
k. kawarabayashi and m. suzuki , phys .
* 16 * ( 1966 ) 255 .
riazuddin and fayyazuddin , phys . rev .
( 1966 ) 1071 .
m. bando , t. kugo , and k. yamawaki , nucl .
phys . * b259 * ( 1985 ) 493 . n. kroll , t. d. lee , and b. zumino , phys . rev . * 157 * ( 1967 ) 1376 .
j. a. mcneil , j. r. shepard , and s. j. wallace , phys .
* 50 * ( 1983 ) 1439 .
j. r. shepard , j. a. mcneil , and s. j. wallace , phys .
* 50 * ( 1983 ) 1443 .
b. c. clark , s. hama , r. l. mercer , l. ray , and b. d. serot , phys .
* 50 * ( 1983 ) 1644 . c. j. horowitz and b. d. serot , nucl .
* a368 * ( 1981 ) 503 .
m. svec , a. de lesquen , l. van rossum , phys .
d * 46 * ( 1992 ) 949 .
a. i. vainshtein , v. i. zakharov , v. a. novikov , and m. a. shifman , sov .
j. part .
* 13 * ( 1982 ) 224 . v. a. miransky and v. p. gusynin , prog .
* 81 * ( 1989 ) 426 .
h. georgi and a. manohar , nucl .
* b234 * ( 1984 ) 189 . r.s .
chivukula , m.j .
dugan , and m. golden , phys .
d * 47 * ( 1993 ) 2930 ; + m.j .
dugan , and m. golden , phys .
d * 48 * ( 1993 ) 4375 .
t. hakioglu and m. d. scadron , phys .
d * 43 * ( 1991 ) 2439 . c. j. horowitz and b. d. serot , nucl .
* a464 * ( 1987 ) 613 ; * a473 * ( 1987 ) 760 ( e ) .
reinhard , rep .
* 52 * ( 1989 ) 439 .
b. a. nikolaus , t. hoch , and d. g. madland , phys .
c * 46 * ( 1992 ) 1757 .
a. r. bodmer , nucl .
* a526 * ( 1991 ) 703 .
m. rufa , p .-
reinhard , j. a. maruhn , w. greiner , and m. r. strayer , phys .
c * 38 * ( 1988 ) 390 .
m. m. sharma , g. a. lalazissis , and p. ring , phys .
b * 317 * ( 1993 ) 9 .
j. friedrich and n. voegler , nucl .
* a373 * ( 1982 ) 192 .
a. l. fetter and j. d. walecka , _ quantum theory of many - particle systems _ ( mcgraw hill , new york , 1971 ) . h. de vries , c. w. de jager , and c. de vries , atomic data and nuclear data tables * 36 * ( 1987 ) 495 .
a. bohr and b. r. mottelson , _ nuclear structure , vol .
( benjamin , new york , 1969 ) .
x. campi and d. w. sprung , nucl .
phys . * a194 * ( 1972 ) 401 . c. j. horowitz and b. d. serot , nucl .
* a399 * ( 1983 ) 529 . | an effective hadronic lagrangian consistent with the symmetries of quantum chromodynamics and intended for applications to finite - density systems is constructed .
the degrees of freedom are ( valence ) nucleons , pions , and the low - lying non - goldstone bosons , which account for the intermediate - range nucleon nucleon interactions and conveniently describe the nonvanishing expectation values of nucleon bilinears .
chiral symmetry is realized nonlinearly , with a light scalar meson included as a chiral singlet to describe the mid - range nucleon
nucleon attraction .
the low - energy electromagnetic structure of the nucleon is described within the theory using vector - meson dominance , so that external form factors are not needed .
the effective lagrangian is expanded in powers of the fields and their derivatives , with the terms organized using georgi s `` naive dimensional analysis '' .
results are presented for finite nuclei and nuclear matter at one - baryon - loop order , using the single - nucleon structure determined within the model .
parameters obtained from fits to nuclear properties show that naive dimensional analysis is a useful principle and that a truncation of the effective lagrangian at the first few powers of the fields and their derivatives is justified . |
we denote by @xmath25 the poset of all the set partitions of @xmath18 $ ] ordered by refinement , that is , define @xmath26 if every block of @xmath27 is contained in a block of @xmath28 .
let @xmath29 be the minimal element of @xmath25 , i.e. , @xmath30 .
consider an interval @xmath31 $ ] in the poset @xmath25 and suppose @xmath32 and @xmath33 is partitioned into @xmath34 blocks in @xmath27
. then we have @xmath31\cong\pi_{\lambda_1}\times\pi_{\lambda_2}\times\cdots\times \pi_{\lambda_k}$ ] . for the particular case @xmath35
, we have @xmath36\cong\pi_{\vert b_1\vert}\times\pi_{\vert b_2\vert}\times\cdots\times \pi_{\vert b_k\vert}$ ] .
if we set @xmath37 for every @xmath38 , then we can rewrite @xmath39\cong \pi_1^{a_1}\times\pi_2^{a_2}\times\cdots\times\pi_n^{a_n}.\end{aligned}\ ] ] the poset @xmath25 of set partitions is the archetype of exponential structures .
the concept of exponential structure was introduced by stanley as a generalization of compositional and exponential formulas @xcite .
an _ exponential structure _ is a sequence @xmath0 of posets such that : 1 . for each @xmath40 ,
the poset @xmath2 is finite , has a unique maximal element @xmath8 and every maximal chain of @xmath2 has @xmath41 elements . 2 . for @xmath42 ,
the interval @xmath43 $ ] is isomorphic to the poset @xmath44 of set partitions for some @xmath45 .
the subposet @xmath46 of @xmath2 is isomorphic to @xmath47 for unique @xmath48 .
suppose @xmath42 and @xmath49 is a minimal element of @xmath2 satisfying @xmath50 .
by @xmath51 and @xmath52 , we obtain that @xmath53\cong\pi_n$ ] .
it follows from ( [ e : sigmapi ] ) that @xmath36\cong \pi_1^{a_1}\times \pi_2^{a_2}\times\cdots\times \pi_n^{a_n}$ ] for unique @xmath48 satisfying @xmath54 and @xmath55 .
in particular , if @xmath56 is another minimal element of @xmath2 satisfying @xmath57 , then we have @xmath58\cong [ \rho,\pi]$ ] .
we will define the numbers @xmath11 associated with an exponential structure @xmath0 in the following way .
let @xmath1 be the number of minimal elements of @xmath2 for @xmath59 and set @xmath3 .
then a sequence of numbers @xmath4 is defined by the equation @xmath60 let @xmath6 denote the poset @xmath2 with a @xmath7 adjoined and let @xmath8 denote the unique maximal element in the poset @xmath2 .
furthermore , let @xmath9 be the mbius function on the poset @xmath6 . then from chapter 5.5 of @xcite ,
we know @xmath61 and thus , @xmath10 .
this implies that the numbers @xmath11 are integers for any exponential structure @xmath0 . in the case @xmath62 , the number @xmath1 of minimal elements in the poset @xmath25 is @xmath63 .
it follows immediately from ( [ e : qexp ] ) and ( [ e : defr ] ) that @xmath64 for @xmath65 and @xmath66 .
there are three other examples of exponential structures @xmath0 in @xcite . 1 .
@xmath67 which is the poset of direct sum decompositions of the @xmath41-dimensional vector space @xmath68 over the finite field @xmath69 .
let @xmath68 be an @xmath41-dimensional vector space over the finite field @xmath70 .
let @xmath71 consist of all the collections @xmath72 of subspaces of @xmath68 such that @xmath73 for all @xmath74 , and such that @xmath75 ( direct sum ) .
an element of @xmath71 is called a _ direct sum decomposition _ of @xmath68 .
we order @xmath71 by refinement , i.e. , @xmath76 if each @xmath77 is contained in some @xmath78 .
@xmath79 which is the poset of set partitions of @xmath16 $ ] whose block sizes are divisible by @xmath17 .
@xmath13 which is the poset of @xmath17-partitions of @xmath18 $ ] .
the definition of @xmath17-partition will be given in section [ s : rpar ] .
the poset @xmath71 had been studied by welker @xcite who used the theory of free monoid to give an expression of the mbius function @xmath80 ; see theorem 4.4 in @xcite . here
we prove that the number @xmath19 is the generalized euler number @xmath21 and @xmath22 is the number of complete non - ambiguous trees of size @xmath23 .
first , we show that @xmath19 and @xmath20 enumerate the pyramids by applying the cartier - foata monoid identity on the respective posets @xmath14 and @xmath15 .
secondly , we build a bijection between the set of the pyramids counted by @xmath19 and the set of permutations of @xmath81 $ ] with descent set @xmath82 .
this gives a new proof of welker s theorem that @xmath24 .
welker @xcite proved that @xmath83 by counting the number of descending chains in the chain lexicographic shellable poset @xmath14 .
thirdly , we establish a bijection between the set of the pyramids counted by @xmath22 and the set of complete non - ambiguous trees of size @xmath23 .
this implies the construction of @xmath17-dimensional complete non - ambiguous trees . as a bonus of applying the theory of heaps , we provide a bijection between the set of complete non - ambiguous forests and the set of pairs of permutations with no common rise
this answers an open question raised by aval _
et al._. this paper is organized as follows . in section [ s :
hmfoata ] , we introduce the terms heap , pyramid and the cartier - foata identity . in sections [ s : parde ] and [ s : rpar ] , we prove our main results .
that is , that @xmath19 is the generalized euler numbers @xmath21 and @xmath22 is the number of complete non - ambiguous trees of size @xmath23 . in subsection
[ s : foraval ] , we establish a bijection between the set of complete non - ambiguous forests and the set of pairs of permutations with no common rise .
the theory of heaps was introduced by viennot to interpret the elements of the cartier - foata monoid in a geometric manner ; see @xcite . here
we will adopt the notations from @xcite .
let @xmath84 be a set of pieces with a symmetric and reflexive binary relation @xmath85 , i.e. , we have @xmath86 and @xmath87 for every @xmath88 .
we will define a _
heap _ as follows : [ d : heap ] a _ heap _ is a triple @xmath89 where @xmath90 is a finite poset and @xmath91 is a map @xmath92 satisfying : 1 . for every @xmath93 , if @xmath94 , then @xmath95 and @xmath96 are comparable , i.e. , @xmath97 or @xmath98 .
2 . for every @xmath93 , if @xmath96 covers @xmath95 ( i.e. , @xmath97 and for any @xmath99 such that @xmath100 , @xmath101 or @xmath102 ) , then @xmath94 .
we represent the heap @xmath89 by the hasse diagram of the poset @xmath90 where every element @xmath95 is labelled by @xmath103 , and we will give an example of heap from @xcite .
+ * example @xmath63 .
* let @xmath104 be a set of integers with a binary relation @xmath105 defined by : @xmath106 if and only if @xmath107 for every @xmath108 .
the poset @xmath109 is defined by its hasse diagram in figure [ f:1 ] and the map @xmath110 is defined as such : if @xmath111 lies on the vertical line @xmath112 , then @xmath113 .
see figure [ f:1 ] .
it is clear that @xmath114 satisfies conditions @xmath51 and @xmath52 in definition [ d : heap ] . +
* remark @xmath63 . *
a heap @xmath89 is equivalent to the directed graph given as we will now state .
let @xmath115 be the graph endowed with an orientation @xmath116 whose vertex set is @xmath117 and such that there is an oriented edge from @xmath95 to @xmath96 if and only if @xmath118 in the poset @xmath117 .
since @xmath119 is an order relation in the poset @xmath90 , @xmath116 is acyclic .
then the heap @xmath120 is equivalent to the directed graph @xmath121 .
( left ) and the heap @xmath122 of integers ( right ) .
[ f:1 ] ] we denote @xmath123 if the heap @xmath124 is empty . to clarify our subsequent discussion on the set of heaps , we need to first review some definitions .
[ d : monoid ] a _ monoid _
@xmath125 is a set that is closed under an associative binary operation @xmath126 and has an identity element @xmath127 such that for all @xmath128 , we have @xmath129 .
[ d : tranclosure ] the _ transitive closure _ of a binary relation @xmath85 on a set @xmath130 is the transitive relation @xmath131 on the set @xmath130 such that @xmath132 is minimal and contains @xmath85 .
let @xmath133 be the set of heaps @xmath89 given in definition [ d : heap ] , in particular , include @xmath134 .
then the set @xmath133 of heaps can be equipped with a monoid structure as follows : [ d : circ ] suppose @xmath135 , @xmath136 , and the composition @xmath137 is the heap defined by the following : 1 .
@xmath138 is the disjoint union of @xmath139 and @xmath140 .
@xmath141 is the unique map @xmath142 such that @xmath143 if we restrict @xmath138 to @xmath144 for @xmath145 .
the partial order @xmath141 is the transitive closure of the following relation @xmath131 . for @xmath146 , @xmath147 if and only if one of @xmath148 is satisfied .
1 . @xmath149 and @xmath150 2 .
@xmath151 and @xmath152 3 .
@xmath153 and @xmath154
. simply put , the heap @xmath155 is obtained by putting the pieces in @xmath156 on top of @xmath157 in its hasse diagram representation .
see figure [ f:2 ] . of integers ( left ) and the composition @xmath155 ( right ) where @xmath157 is given in figure [ f:1 ] .
[ f:2 ] ] by definition [ d : circ ] , the set @xmath133 is closed under the associative composition @xmath126 and the identity element @xmath158 is the empty heap @xmath159 from which it follows that the set @xmath133 is a monoid .
the size of a heap @xmath124 , denoted by @xmath160 , is the number of pieces in the heap @xmath124 .
[ d : triheap ] a _ trivial heap _
@xmath161 is a heap consisting of pieces that are pairwise unrelated , i.e. , @xmath162 for every distinct @xmath163 in @xmath161 .
let @xmath164 be the set of trivial heaps contained in the set @xmath133 .
let @xmath165 $ ] be the ring of formal power series in @xmath133 with coefficients in the commutative ring @xmath166 .
then the cartier - foata identity in the ring @xmath165 $ ] is @xmath167 [ d : cfmonoid ] consider the set of letters @xmath168 and the set @xmath169 of relations .
then the _ cartier - foata monoid _ is the set @xmath170 of equivalent classes of words of @xmath125 subject to the relations in @xmath171 . in other words , two words
are equivalent if we can transform one into another by applying successive relations of @xmath171 .
the cartier - foata monoid @xmath170 is isomorphic to the monoid @xmath133 .
here we only show the isomorphism @xmath172 without the proof .
we refer the readers to the proposition 3.4 in @xcite for a detailed proof that @xmath173 is an isomorphism .
given a heap @xmath174 of size @xmath41 , let @xmath175 $ ] be a bijection such that @xmath176 for every @xmath177 where @xmath178 , and define a word @xmath179 where @xmath180 .
here we call the bijection @xmath27 a _ natural labeling _
( sometimes also called a _ linear extension _ ) of the poset @xmath90 .
for any two different natural labelings @xmath181 of the poset @xmath90 , it was proven in @xcite that @xmath182 in the monoid @xmath170 . in view of this , we set @xmath183 and , thus , @xmath184 is well - defined .
in particular , we have @xmath185 and , for any two heaps @xmath186 , we have @xmath187 .
namely , under the isomorphism @xmath173 , the empty heap @xmath159 corresponds to the identity word @xmath63 and the composition of two heaps @xmath155 corresponds to the product of two words @xmath188 and @xmath189 by juxtaposition .
+ * example @xmath190 . *
we continue using the heap @xmath122 in example @xmath63 to show the isomorphism @xmath173 .
we start labeling the minimal elements of @xmath109 with integers @xmath191 from left to right , then we remove these minimal elements from the poset @xmath109 and label the minimal elements of the remaining poset by increasing integers from left to right .
we continue this process until all the elements in the poset @xmath109 are labeled .
it is clear this labeling process is a natural labeling of the poset @xmath109 which gives us the corresponding word @xmath192 in the monoid @xmath193 .
similarly , we get the corresponding words for the heaps @xmath156 and @xmath155 in figure [ f:2 ] , which are @xmath194 and @xmath195 .
it is easy to verify that @xmath196 because @xmath197 commutes with every @xmath198 for @xmath199 .
+ let @xmath200 $ ] be the ring of formal power series in @xmath170 with coefficients in the commutative ring @xmath166
. then we can express ( [ e : cf1 ] ) in the ring @xmath200 $ ] as follows : @xmath201 a _ pyramid _ is a heap with exactly one maximal element .
let @xmath202 denote the set of pyramids consisting of pieces in @xmath84 . then , according to the exponential formula for the unlabeled combinatorial objects @xcite , @xmath203 where @xmath204 means the identity holds in the commutative extension of @xmath133 , that is , in the commutative monoid which arises from @xmath133 by letting all pieces in @xmath84 commute . in combination of ( [ e : cf2 ] )
, it follows that @xmath205 we will mainly use ( [ e : logpy ] ) to connect the numbers @xmath19 and @xmath20 with the pyramids in section [ s : parde ] and section [ s : rpar ] .
it is worthwhile to mention that via ( [ e : logpy ] ) josuat - vergs had proved that a new sequence enumerates the pyramids which are related to the perfect matchings in @xcite . with a slight abuse of notation , we simply use the labels @xmath103 in the heap @xmath89 to represent the elements @xmath206 in the heap and we use @xmath207 to denote the heap @xmath89 .
the poset of the set partitions of @xmath16 $ ] whose block sizes are divisible by @xmath17 is @xmath14 and its elements are ordered by refinement . in particular ,
let @xmath209 be the set of all the set partitions of @xmath16 $ ] whose block sizes are exactly @xmath17 .
then @xmath209 is the set of minimal elements of the poset @xmath14 and , consequently , @xmath210 . for @xmath211 ,
the sequence @xmath212 defined by ( [ e : defr ] ) satisfies @xmath213 let @xmath214 be the set of permutations of @xmath18 $ ] . for @xmath215 , we write @xmath216 if @xmath217
. then the descent set of permutation @xmath218 is @xmath219 .
the generalized euler number @xmath21 counts the number of permutations @xmath220 such that @xmath221 .
one of our main results is a new proof of the following result due to welker @xcite : [ t : eulercm ] @xmath24
. we will prove theorem [ t : eulercm ] by first showing that @xmath19 enumerates the pyramids of the monoid @xmath222 in lemma [ l : cirn ] and then building a bijection between the set of the pyramids counted by @xmath19 and the set of permutations of @xmath81 $ ] with descent set @xmath82 in lemma [ l : bijrn ] .
before we proceed , we introduce some definitions and notations .
we say @xmath223 is an _ alternating permutation _ if @xmath224 , namely , des @xmath225 .
it is well - known that the number of alternating permutations of @xmath226 $ ] is counted by the euler number @xmath227 ( also called the tangent number ) , whose exponential generating function is @xmath228 recall that @xmath209 is the set of set partitions of @xmath16 $ ] where each block has size @xmath17 .
we say @xmath229 is a block of partition @xmath230 , denoted by @xmath231 , if @xmath232 .
the diagram representation of @xmath230 is given as follows : we draw @xmath233 dots in a line labeled with @xmath234 .
for each block @xmath235 , we connect @xmath236 and @xmath237 by an edge for all @xmath238 .
we say two blocks @xmath229 and @xmath239 of @xmath28 are not crossing if @xmath240 or @xmath241 .
otherwise two blocks @xmath229 and @xmath239 are crossing .
see figure [ f:41 ] for an example . where @xmath242 and @xmath28 has blocks @xmath243 .
[ f:41 ] ] stanley @xcite proved @xmath244 by using the exponential generating function of @xmath227 given in ( [ e : euler2 ] ) .
more precisely , the hyperbolic tangent function @xmath245 given by @xmath246 satisfies @xmath247 . by setting @xmath248 to ( [ e : rq2 ] ) and replacing @xmath249 by @xmath250 , we get @xmath251 by differentiating both sides of the above equation with respect to @xmath99 , we have @xmath252 in view of ( [ e : euler2 ] ) and @xmath247 , we obtain @xmath253 which yields @xmath244 . for general @xmath17 , we shall show that @xmath19 counts the number of pyramids related to the set @xmath209 .
let @xmath254 be the set of blocks with a binary relation @xmath255 defined by @xmath256 if and only if two blocks @xmath257 and @xmath258 are crossing .
let @xmath90 be a poset where every element is labeled by an element @xmath259 such that : 1 .
@xmath257 and @xmath258 are comparable if they are crossing
if @xmath258 covers @xmath257 in the poset @xmath90 , then they are crossing . by definition [ d : heap ] , the poset @xmath90 is a heap @xmath260 . in a geometric way , we can represent each block @xmath257 by a line @xmath261 and put @xmath261 above @xmath262 if @xmath263 in the heap @xmath207 and the line @xmath261 can not move downwards without touching the line @xmath262 .
see figure [ f:3 ] .
( left ) in the monoid @xmath222 and its geometric representation ( right ) .
the corresponding word in the monoid @xmath264 is @xmath265 .
[ f:3 ] ] now we are in a position to establish a connection between @xmath19 and the set of pyramids @xmath266 via ( [ e : logpy ] ) .
[ l : cirn ] @xmath19 counts the number of pyramids @xmath267 in @xmath266 such that @xmath267 has @xmath41 elements which are exactly the blocks of a partition from @xmath209 and the unique maximal element of @xmath267 is the block containing @xmath233 . from ( [ e : logpy ] ) we know @xmath268 we now consider the cartier - foata monoid @xmath264 .
let @xmath269 be a sequence of variables such that @xmath270 for every @xmath74 and @xmath271 for every @xmath272 .
furthermore let @xmath273\rightarrow \mathbb{z}[[a_1,a_2,\ldots]]$ ] be a ring homomorphism such that @xmath274 and @xmath275 .
in other words , the commutative ring @xmath276 $ ] is a commutative extension of the ring @xmath277 $ ] .
we apply @xmath278 on both sides of ( [ e:1 ] ) and obtain @xmath279 let @xmath280 be the set of trivial heaps of size @xmath41 contained in the set @xmath281 and suppose that @xmath282 is a trivial heap having @xmath41 pieces @xmath229 , @xmath283 , @xmath284 such that @xmath285
. then @xmath286 must be a strictly increasing sequence .
we use @xmath287 to denote this strictly increasing sequence .
thus , the map @xmath288 is a bijection between @xmath280 and the set of strictly increasing sequences of length @xmath233 .
this yields @xmath289 where the second summation runs over all the strictly increasing sequences @xmath290 of length @xmath233 and the last equation holds because @xmath270 for every @xmath74 .
it follows immediately that the right hand side of ( [ e : logpy1 ] ) is @xmath291 on the other hand , let @xmath292 be the set of pyramids of size @xmath41 contained in the set @xmath266 .
suppose that @xmath293 is a pyramid of size @xmath41 such that @xmath294 , then the elements of the poset @xmath295 are exactly the blocks of a set partition of @xmath233 integers where every block has size @xmath17 , and every block is crossing with at least one other block of this set partition .
we continue using @xmath290 to represent any strictly increasing sequence @xmath286 .
a strictly increasing sequence @xmath290 uniquely corresponds with a set @xmath296 . for a given
set @xmath296 , let @xmath297 count the number of pyramids @xmath295 such that the elements of @xmath295 are exactly the blocks of a set partition of @xmath296 where each block has size @xmath17 .
then we have @xmath298 note that the number @xmath297 is independent of the choice of the set @xmath296 . without loss of generality ,
we choose the set @xmath299 .
it turns out that the left hand side of ( [ e : logpy1 ] ) is @xmath300 in view of ( [ e : rq2 ] ) , we get @xmath301 . since every pyramid in @xmath266 has @xmath41 elements , it follows that @xmath19 counts the number of pyramids @xmath302 in @xmath266 such that the unique maximal element of @xmath267 is the block containing @xmath233 and @xmath267 has @xmath41 elements which are exactly the blocks of a partition in @xmath209 .
the proof is complete .
for example , @xmath303 counts the pyramid @xmath304 such that @xmath305 and the pyramid @xmath306 such that @xmath307 .
[ l : bijrn ] there is a bijection between the set of permutations of @xmath81 $ ] with descent set @xmath82 and the set of pyramids @xmath267 in @xmath266 such that @xmath267 has @xmath41 elements which are exactly the blocks of a partition in @xmath209 and the unique maximal element of @xmath267 is the block containing @xmath233 .
let @xmath308 be a permutation of @xmath81 $ ] such that @xmath221 .
we consider a partition @xmath309 whose blocks are exactly @xmath310 and we now prove that we can choose them in such a way that they are the elements of a pyramid @xmath267 .
it remains to define an order @xmath119 of the elements in the pyramid @xmath267 .
the correspondence @xmath311 is defined inductively .
let @xmath312 be the minimal integer such that @xmath313 are in the same block .
we notice that @xmath312 and @xmath233 are in two crossing blocks of @xmath27 and we consider the block that contains @xmath312 .
suppose that @xmath314 is the block of @xmath27 such that @xmath315 and let @xmath316 be a subset of the blocks of @xmath27 .
let @xmath317 be the set of remaining blocks .
then we can write @xmath318 and we use @xmath319 to denote the number of blocks in the partition @xmath320 , for @xmath145 . by induction
, @xmath321 is a pyramid such that the unique maximal element is the block containing @xmath312 if @xmath322 , or the block containing @xmath233 if @xmath323 , and @xmath321 has @xmath319 elements which are exactly the blocks of @xmath320 , for @xmath145 . for every element @xmath95 in the pyramid @xmath324 , suppose that @xmath325 is one of the minimal elements in the pyramid @xmath326 that is crossing with @xmath95 but not crossing with any @xmath99 such that @xmath327 in the pyramid @xmath324 , then the pyramid @xmath311 is obtained by letting @xmath325 cover @xmath95 for every @xmath95 and @xmath325 .
in particular , @xmath328 in the pyramid @xmath311 , which implies that the unique maximal element of @xmath311 is the block containing @xmath233 .
in fact , this provides an inductive process to successively construct the pyramid @xmath311 whose unique maximal element is the block of @xmath27 containing @xmath233 and @xmath311 has @xmath41 elements that are exactly the blocks of @xmath27 .
conversely , consider a pyramid @xmath267 in @xmath266 such that @xmath267 has @xmath41 elements which are exactly the blocks of a partition @xmath329 and the unique maximal element is @xmath330 .
we consider the pyramid @xmath139 that is induced by the block that contains @xmath312 , i.e. , @xmath139 contains all the elements @xmath95 such that @xmath331 in the pyramid @xmath267 where @xmath315 .
let @xmath140 be the pyramid of remaining elements from @xmath267 .
then @xmath332 for @xmath145 which implies the above inductive process is bijective and therefore the proof is complete .
see figure [ f:4 ] for an example of this bijection @xmath333 . where @xmath334 and @xmath27 is a partition with blocks @xmath335 , @xmath336 , @xmath337 , @xmath338 , @xmath339 .
an @xmath17-partition of @xmath18 $ ] is a set @xmath340 satisfying the following two conditions : 1 . for each @xmath341 $ ] , the set @xmath342 forms a partition of @xmath125 ( into @xmath45 blocks ) , 2 . for fixed @xmath74 , @xmath343 .
the set of all the @xmath17-partitions of set @xmath18 $ ] , denoted by @xmath15 , has a partial ordering by refinement .
namely , let @xmath344 be two @xmath17-partitions of @xmath18 $ ] where we set for @xmath345 , @xmath346 then @xmath347 if for every @xmath345 , we have @xmath348 , i.e. , every block of partition @xmath349 is contained in a block of partition @xmath350 .
for instance , @xmath351 are two @xmath190-partitions of @xmath352 $ ]
. then we have @xmath353 , @xmath354 , @xmath355 and @xmath356 , @xmath357 by refinement .
that implies that @xmath347 in the poset @xmath358 . by definition ,
every minimal element of @xmath15 can be identified as an @xmath359-tuple @xmath360 of permutations @xmath361 by @xmath362 as an immediate consequence , the number of minimal elements of @xmath15 is equal to the number of ordered permutations of @xmath18 $ ] , i.e. , @xmath363 where @xmath361 .
let @xmath364 be the set of @xmath359-tuple @xmath363 of permutations @xmath361 , it follows that @xmath365 and that the sequence @xmath366 where @xmath211 defined by ( [ e : defr ] ) satisfies @xmath367 for any @xmath368 , the diagram representation of @xmath27 is described as follows : we draw @xmath233 dots in @xmath17 rows with each row having @xmath41 dots labeled by @xmath369 , and , for every @xmath370 , we connect @xmath312 from the @xmath74-th row with @xmath371 from the @xmath372-th row . for every @xmath373
$ ] , we call the sequence @xmath374 a _ path _ of @xmath27 that starts from @xmath312
. see figure [ f:5 ] .
clearly every @xmath375 has @xmath41 paths .
two paths @xmath376 and @xmath377 are not crossing if and only if @xmath378 for all @xmath45 .
otherwise two paths @xmath376 and @xmath377 are crossing . and @xmath27 has paths @xmath379 , @xmath380 , @xmath381 , @xmath382 and @xmath383 .
[ f:5 ] ] let @xmath384 be the set of paths of length @xmath385 with a symmetric and reflexive binary relation @xmath386 defined by : @xmath387 if and only if two paths @xmath388 and @xmath389 are crossing .
let @xmath390 be a poset where each element is labeled by an element @xmath391 such that : 1 .
@xmath388 and @xmath389 are comparable if @xmath388 and @xmath389 are crossing
if @xmath389 covers @xmath388 in the poset @xmath390 , then they are crossing . by definition [ d : heap ] ,
the poset @xmath390 is a heap @xmath392 . in a geometric way
, we put the path @xmath389 on top of the path @xmath388 if @xmath393 in the heap @xmath394 .
see figure [ f:6 ] .
( left ) in the monoid @xmath395 and its geometric representation ( right ) .
the corresponding word in the monoid @xmath396 is @xmath397 .
[ f:6 ] ] furthermore , _ non - ambiguous trees _ were introduced by aval _
_ @xcite because of their connection to the tree - like tableaux .
the non - ambiguous trees are embedded in a @xmath190-dimensional @xmath398 grid .
let each vertex @xmath399 have coordinates @xmath400 .
then a @xmath190-dimensional non - ambiguous tree of size @xmath41 is a set @xmath401 of @xmath41 points @xmath402 such that : 1 . @xmath403 ; we call this point the root of @xmath401 .
2 . for a given non - root point @xmath404
, there exists one point @xmath405 such that @xmath406 and @xmath407 or one point @xmath408 such that @xmath409 and @xmath410 ( but not both ) .
3 . there is no empty line between two given points : if there exists a point @xmath404 such that @xmath411 ( resp .
@xmath412 ) , then for every @xmath413 ( resp .
@xmath414 ) there exists @xmath415 such that @xmath416 ( resp .
@xmath417 ) . a _ complete non - ambiguous tree _ is a non - ambiguous tree whose vertices have either @xmath418 or @xmath190 children .
the non - ambiguous tree @xmath401 has a unique tree structure since , except for the root , every point @xmath419 has a unique parent , which is the nearest point @xmath420 or @xmath421 in condition @xmath52 .
in other words , the set @xmath401 of points determines the tree structure .
let @xmath422 be the unique underlying tree associated to the vertex set @xmath401 .
for instance , there are four complete non - ambiguous trees of size @xmath423 whose underlying trees are given here .
( 30,18)(32,-14 ) ( 10,0)(10,0)(-1,-1)8 ( 10,0)(1,-1)8(2,-8 ) ( 14,-4)(18,-8 ) ( 14,-4)(-1,-1)4(10,-8 ) ( 7,1)(-2,-5 ) ( 14,-3)(6,-11 ) ( 18,-7 ) ( 35,0)(35,0)(-1,-1)8 ( 35,0)(1,-1)8(27,-8 ) ( 31,-4)(43,-8 ) ( 31,-4)(1,-1)4(35,-8 ) ( 32,1)(26,-11 ) ( 25,-3)(33,-11 ) ( 43,-7 ) ( 60,0)(60,0)(-1,-1)8 ( 60,0)(1,-1)4(64,-4 ) ( 56,-4)(52,-8 ) ( 56,-4)(1,-1)8(64,-12 ) ( 57,1)(50,-3 ) ( 64,-3)(50,-12 ) ( 57,-12 ) ( 85,0)(85,0)(-1,-1)4 ( 85,0)(1,-1)8(89,-4 ) ( 81,-4)(93,-8 ) ( 89,-4)(-1,-1)8(81,-12 ) ( 82,1)(75,-3 ) ( 89,-3)(93,-7 ) ( 83,-12 ) one of our main results is stated as follows : [ t : rnqn ] @xmath20 counts the number of pyramids @xmath424 in @xmath425 such that @xmath424 has @xmath41 elements which are exactly the paths of an @xmath359-tuple of permutations in @xmath426 and the unique maximal element of @xmath424 is the path starting from @xmath63 . in particular , @xmath22 counts the number of complete non - ambiguous trees of size @xmath23 .
we divide theorem [ t : rnqn ] into lemma [ l : bcm ] and lemma [ l : bicm ] .
we shall prove lemma [ l : bcm ] again by using the connection between @xmath20 and heaps and prove lemma [ l : bicm ] by a bijection .
now we first establish a connection between @xmath20 and the set of pyramids @xmath425 via ( [ e : logpy ] ) . [
l : bcm ] @xmath20 counts the number of pyramids @xmath424 in @xmath425 such that @xmath424 has @xmath41 elements which are exactly the paths of an @xmath359-tuple of permutations in @xmath426 and the unique maximal element of @xmath424 is the path starting from @xmath63 . from ( [ e : logpy ] ) we know @xmath427 we now consider the cartier - foata monoid @xmath396 .
let @xmath428 be a sequence of variables such that @xmath429 for every @xmath272 and @xmath430 for every @xmath431 .
furthermore let @xmath432\rightarrow \mathbb{z}[[a_{1,1},a_{1,2},\ldots]]$ ] be a ring homomorphism such that @xmath433 and @xmath434 .
we apply @xmath435 on both sides of ( [ e : c1 ] ) and obtain @xmath436 let @xmath437 be the set of trivial heaps of size @xmath41 contained in the set @xmath438 and suppose that @xmath439 is a trivial heap having @xmath41 paths @xmath440 , @xmath441 , @xmath442 such that @xmath285 . then the sequence @xmath443 is a strictly increasing sequence for every @xmath38 , @xmath345 .
it follows that @xmath444 where @xmath445 and @xmath443 is a bijection between the set of trivial heaps of size @xmath41 and the set of @xmath17-tuples of strictly increasing sequences of length @xmath41 .
this gives @xmath446 where the second summation runs over all the @xmath17-tuples @xmath447 of strictly increasing sequences of length @xmath41 and the last two equations hold because @xmath429 for every @xmath272 .
it follows immediately that the right hand side of ( [ e : c2 ] ) is @xmath448 on the other hand , let @xmath449 be the set of pyramids of size @xmath41 contained in the set @xmath425 .
suppose that @xmath450 is a pyramid of size @xmath41 such that @xmath451 , then the elements of the poset @xmath295 are exactly the paths of an @xmath359-tuple @xmath452 of bijections such that @xmath350 is a bijection between two sets of @xmath41 integers for every @xmath38 , and every path of @xmath27 is crossing with at least one other path of @xmath27 .
we continue using @xmath445 to represent any @xmath17-tuple of strictly increasing sequences of length @xmath41 .
a strictly increasing sequence @xmath453 uniquely corresponds with a set @xmath454 . for a given @xmath17-tuple @xmath447 of strictly increasing sequences , let @xmath455 count the number of pyramids @xmath295 such that the elements of @xmath295 are exactly the paths of an @xmath359-tuple @xmath452 of bijections such that @xmath350 is a bijection between the set @xmath456 and @xmath457 for every @xmath38 .
then we have @xmath458 note that the number @xmath455 is independent of the choice of the @xmath17-tuple @xmath447 of strictly increasing sequences and we choose the set @xmath459 for every @xmath38 .
it turns out the left hand side of ( [ e : c2 ] ) is @xmath460 in view of ( [ e : rq3 ] ) , we conclude that @xmath461 .
since every pyramid in @xmath425 has @xmath41 elements , it follows that @xmath20 counts the number of pyramids @xmath424 in @xmath425 such that @xmath424 has @xmath41 elements which are exactly the paths of an @xmath359-tuple of permutations in @xmath426 and the maximal element of @xmath424 is the path starting at @xmath63 .
the proof is complete .
for example , @xmath462 counts the pyramid @xmath463 such that @xmath464 , the pyramid @xmath465 such that @xmath466 , and the pyramid @xmath467 such that @xmath468 .
let @xmath469 be the number of complete non - ambiguous trees of size @xmath23 , aval _ et al .
_ proved the integers @xmath469 satisfy ( [ e : rq3 ] ) when @xmath248 .
this implies that @xmath470 .
we will prove that @xmath470 by a bijection in lemma [ l : bicm ] .
[ l : bicm ] there is a bijection between the set of complete non - ambiguous trees of size @xmath23 and the set of pyramids @xmath424 in @xmath425 such that @xmath424 has @xmath41 elements which are exactly the paths of a permutation @xmath471 and the unique maximal element of @xmath424 is the path @xmath472 . given a complete non - ambiguous tree @xmath401 of size @xmath23 , we first do a coordinate translation .
let @xmath473 be a bijection such that for any @xmath474 having coordinate @xmath400 , we set @xmath475 .
recall that @xmath422 is the unique underlying tree associated to @xmath401 .
thus via bijection @xmath476 , we shift the tree @xmath422 rooted at @xmath477 to the tree @xmath478 rooted at @xmath479 .
next we consider the coordinates of the leaves of @xmath478 . from the definition of complete non - ambiguous tree , we know the coordinates of the leaves of @xmath478 must be @xmath480 where @xmath481 and we now prove that we can choose them in such a way that they are the elements of a pyramid @xmath424 .
it remains to define an order @xmath119 of the elements in the pyramid @xmath424 .
let @xmath482 be any subset of @xmath401 such that the underlying tree @xmath483 is a subtree of @xmath478 .
the correspondence @xmath484 is defined inductively .
suppose @xmath482 is rooted at @xmath485 and let @xmath486 be two subsets of @xmath482 such that @xmath487 is the left subtree and @xmath488 is the right subtree of @xmath483 .
then for @xmath145 , the leaves of @xmath489 can be identified as a permutation @xmath490 which is @xmath27 restricted to a subset of @xmath18 $ ] .
clearly @xmath491 , @xmath492 and the two paths @xmath493 are crossing .
by induction , @xmath494 is a pyramid such that the unique maximal element is @xmath495 , @xmath496 is a pyramid such that the unique maximal element is @xmath497 and all the elements of @xmath498 are exactly the paths of @xmath490 for @xmath145 .
we use @xmath499 to denote the number of paths in @xmath490 for @xmath145 .
if @xmath483 is the left subtree of its parent in @xmath478 or @xmath500 , then , for every element @xmath95 in the pyramid @xmath496 , suppose that @xmath325 is one of the minimal elements in the pyramid @xmath494 that is crossing with @xmath95 but not crossing with any @xmath99 such that @xmath327 in the pyramid @xmath496 .
thus the pyramid @xmath484 is obtained by letting @xmath325 cover @xmath95 for every @xmath95 and @xmath325 . in particular
, @xmath501 holds in the pyramid @xmath484 which implies that the unique maximal element of @xmath484 is @xmath495 .
if @xmath483 is the right subtree of its parent in @xmath478 , then , for every path @xmath96 in the pyramid @xmath494 , suppose @xmath502 is one of the minimal elements in the pyramid @xmath496 that is crossing with @xmath96 but not crossing with any @xmath99 such that @xmath503 in the pyramid @xmath494 .
thus , the pyramid @xmath484 is obtained by letting @xmath502 cover @xmath96 for every @xmath96 and @xmath502 . in particular
, @xmath504 holds in the pyramid @xmath484 which implies that the unique maximal element of @xmath484 is @xmath497 .
in fact , this gives us an inductive process to successively construct the pyramid @xmath505 whose unique maximal element is @xmath472 and @xmath505 has @xmath41 elements which are exactly the paths of @xmath471 .
conversely , without loss of generality , consider a pyramid @xmath424 such that the elements of @xmath424 are exactly the paths of the permutations @xmath506 and the unique maximal element is @xmath495 .
let @xmath507 be the minimal integer such that @xmath508 for all the elements @xmath509 , then @xmath510 and we consider the pyramid @xmath511 that is induced by the element @xmath497 , i.e. , @xmath511 contains all the elements @xmath95 such that @xmath512 in the pyramid @xmath424 .
let @xmath513 be the pyramid of remaining elements from @xmath424 .
then @xmath514 , for @xmath145 , which implies the above inductive process is bijective and therefore the proof is complete .
see figure [ f:7 ] for an example .
( left ) and the corresponding pyramid @xmath515 whose unique maximal element is the path @xmath516 ( right ) .
the pyramid @xmath515 has @xmath423 elements which are exactly the paths of the permutation @xmath517 .
[ f:7 ] ] a non - ambiguous forest introduced by aval _ et al . _
@xcite is a set of points @xmath518 satisfying the following conditions : 1 . for a given non - root point @xmath404 ,
the set @xmath519 has at most @xmath63 element .
2 . there is no empty line between two given points : if there exists a point @xmath404 such that @xmath411 ( resp .
@xmath412 ) , then for every @xmath413 ( resp .
@xmath414 ) there exists @xmath415 such that @xmath416 ( resp .
@xmath417 ) . in the same way
, a non - ambiguous forest has an underlying binary forest structure .
a complete non - ambiguous forest is a non - ambiguous forest such that all trees in the underlying binary forest are complete .
for instance , there are @xmath520 complete non - ambiguous forests that contain leaves @xmath521 , which are ( 30,18)(32,-14 ) ( 10,0)(10,0)(-1,-1)4 ( 10,0)(1,-1)8(14,-4 ) ( 6,-4)(18,-8 ) ( 14,-4)(-1,-1)8(6,-12 ) ( 7,1)(0,-3 ) ( 14,-3)(18,-7 ) ( 8,-12 ) ( 31,-4)(43,-8 ) ( 31,-12 ) ( 27,-2 ) ( 40,-6 ) ( 33,-12 ) ( 64,-4)(1,-1)4 ( 64,-4)(1,-1)4 ( 64,-4 ) ( 56,-4)(68,-8 ) ( 64,-4)(-1,-1)8(56,-12 ) ( 53,-2 ) ( 61,-2)(68,-7 ) ( 58,-12 ) ( 85,0)(85,0)(-1,-1)4 ( 85,0)(1,-1)8 ( 81,-4)(93,-8 ) ( 81,-12 ) ( 82,1)(75,-3 ) ( 93,-7 ) ( 83,-12 ) we denote by @xmath522 the number of complete non - ambiguous forests with @xmath41 leaves
. then aval _ et al . _
@xcite proved that @xmath523 in addition , carlitz _ et al . _
@xcite proved the same identity by enumerating the pairs of permutations with no common rise .
let @xmath524 count the number of pairs @xmath525 of permutations of @xmath18 $ ] such that any two consecutive entries can not be rising both in @xmath28 and @xmath526 , i.e. , there is no integer @xmath74 such that @xmath527 and @xmath528 .
@xcite showed that @xmath529 this implies that @xmath530 and aval _ et al .
_ asked for a bijective proof of @xmath530 . here
we will give a bijection between the set of complete non - ambiguous forests with @xmath41 leaves and the set of pairs @xmath525 of permutations @xmath531 with no common rise by using heaps as intermediate objects .
there is a bijection between the set of complete non - ambiguous forests with @xmath41 leaves and the set of pairs @xmath525 of permutations of @xmath18 $ ] with no common rise .
we first define the number @xmath532 by the equation @xmath533 similar to the proof of lemma [ l : bcm ] , via ( [ e : cf2 ] ) we can prove that @xmath532 counts the number of heaps @xmath534 in @xmath395 such that @xmath534 has @xmath41 elements which are exactly the paths of an @xmath359-tuple of permutations in @xmath214 .
we will first show that @xmath535 by a bijection . given a complete non - ambiguous forest @xmath401 with @xmath41 leaves ,
let @xmath474 have coordinates @xmath536 and we again obtain the set @xmath475 via the bijection @xmath476 given in lemma [ l : bicm ] . recall that @xmath478 is the underlying forest associated with the vertex set @xmath537 . from the definition we know the coordinates of the leaves of @xmath478 must be @xmath538 where @xmath481 and we now prove that we can choose them in such a way that they are the elements of a heap @xmath534 .
it remains to define the order @xmath119 of the elements of the heap @xmath534 .
the correspondence @xmath539 is defined as follows .
suppose the forest @xmath478 contains exactly @xmath45 complete non - ambiguous trees @xmath540 that are associated to the vertex sets @xmath541 , respectively , and the leaves of every @xmath489 can be identified as a permutation @xmath490 which is @xmath27 restricted to a subset of @xmath18 $ ] .
furthermore , suppose the leftmost leaf of @xmath489 is @xmath542 for @xmath543 and @xmath544 is strictly increasing .
then by applying the bijection @xmath545 given in lemma [ l : bicm ] on every @xmath33 , we have the corresponding pyramid @xmath498 such that the unique maximal element is @xmath542 and all the elements of @xmath498 are exactly the paths of a permutation @xmath490 . for every @xmath272 such that @xmath546 and @xmath547 , and every element @xmath95 in the pyramid @xmath498 , suppose @xmath548 is one of the minimal elements in the pyramid @xmath549 that is crossing with @xmath95 but not crossing with any @xmath99 such that @xmath327 in the pyramid @xmath498
. then the heap @xmath539 is obtained by letting @xmath548 cover @xmath95 for every @xmath95 and @xmath325 .
we next prove that the map @xmath550 is a bijection by showing its inverse map . for a given heap @xmath534 in @xmath395
such that @xmath534 has @xmath41 elements which are exactly the paths of a permutation @xmath27 in @xmath214 , namely the elements of @xmath534 are @xmath551 , we consider the pyramid @xmath139 induced by the element @xmath472 , i.e. , @xmath139 contains all the elements @xmath96 in @xmath534 such that @xmath552 . from the remaining heap @xmath553
, we choose an element @xmath554 such that @xmath555 is minimal of all the elements contained in @xmath553 .
let @xmath140 be a pyramid induced by the element @xmath554 , i.e. , @xmath140 contains all the elements @xmath96 in @xmath553 such that @xmath556 .
we continue this process until no element is left and we decompose @xmath534 into a sequence of pyramids @xmath557 whose unique maximal elements are respectively @xmath558 such that @xmath559 is strictly increasing .
in fact , @xmath560 .
for any two crossing paths @xmath561 such that @xmath562 and @xmath563 , we have @xmath564 if @xmath565 .
by applying the inverse bijection @xmath566 given in lemma [ l : bicm ] on the pyramids @xmath557 , we retrieve a forest of binary trees @xmath540 whose vertex set is a complete non - ambiguous forest , from which it follows that @xmath550 is a bijection .
see figure [ f:8 ] for an example of the bijection @xmath567 .
leaves ( left ) and its corresponding heap @xmath568 whose elements are exactly the paths of the permutation @xmath517 ( right ) .
furthermore , the heap @xmath568 uniquely corresponds with the pair @xmath569 of permutations with no common rise .
[ f:8 ] ] in the next step , we will show that @xmath570 by constructing a bijection between the set of pairs @xmath525 of permutations of @xmath18 $ ] with no common rise and the set of heaps @xmath534 in @xmath395 such that @xmath534 has @xmath41 elements which are exactly the paths of a permutation @xmath481 . for a given pair @xmath525 of permutations @xmath531 without common rise , suppose @xmath571 and @xmath572 .
then , we choose a permutation @xmath481 satisfying @xmath573 for every @xmath74 .
let @xmath574 be the elements of the corresponding heap @xmath534 , then it remains to define the order @xmath575 in the heap @xmath534 .
for every two crossing paths @xmath576 , we set @xmath577 if @xmath578 .
thus , we have given the map @xmath579 .
we next prove this map is a bijection by showing its inverse map . given a heap @xmath534 in @xmath395 such that @xmath534 has @xmath41 elements which are exactly the paths of a permutation @xmath471 , for any two paths @xmath576
, we define @xmath580 if @xmath577 in the heap @xmath534 .
otherwise , we define @xmath581 if @xmath582 and @xmath542 , @xmath583 are incomparable in the heap @xmath534 . in this way
, we give a total order @xmath584 of the paths @xmath542 for all @xmath74 which allows us to rewrite the elements @xmath585 by this linear order @xmath584 .
that is , suppose the sequence @xmath586 satisfies @xmath587 , then we choose @xmath588 and @xmath589 .
there is no common rise of the pair @xmath525 .
if there exists @xmath236 such that @xmath590 and @xmath591 , then the paths @xmath592 and @xmath593 are not crossing , so they have no cover relation in the heap @xmath534 .
it follows that @xmath592 and @xmath593 are incomparable in the heap @xmath534 and , in view of @xmath590 , we have @xmath594 which contradicts the assumption that @xmath595 .
consequently , the pair @xmath525 has no common rise and the map @xmath579 is a bijection .
see figure [ f:8 ] for an example of this bijection .
the author thanks two anonymous reviewers for their very helpful suggestions and comments .
the author also thanks jiang zeng , zhicong lin , ren ciak and laura silverstein for related discussions and would like to give special thanks to the joint seminar arbeitsgemeinschaft diskrete mathematik for their valuable feedback .
this work was done during my stay in the ag algorithm and complexity , technische universitt kaiserslautern , germany .
the author would like to thank markus nebel , sebastian wild and raphael reitzig for their kind help and support . | take @xmath0 to be an exponential structure and @xmath1 to be the number of minimal elements of @xmath2 where @xmath3 .
then a sequence of numbers @xmath4 is defined by the equation @xmath5 let @xmath6 denote the poset @xmath2 with a @xmath7 adjoined and let @xmath8 denote the unique maximal element in the poset @xmath2 .
furthermore , let @xmath9 be the mbius function on the poset @xmath6 .
stanley proved that @xmath10 .
this implies that the numbers @xmath11 are integers . in this paper
, we study the cases @xmath12 and @xmath13 where @xmath14 and @xmath15 are posets , respectively , of set partitions of @xmath16 $ ] whose block sizes are divisible by @xmath17 and of @xmath17-partitions of @xmath18 $ ] . in both cases
we prove that @xmath19 and @xmath20 enumerate the pyramids by applying the cartier - foata monoid identity and further prove that @xmath19 is the generalized euler number @xmath21 and that @xmath22 is the number of complete non - ambiguous trees of size @xmath23 by bijections .
this gives a new proof of welker s theorem that @xmath24 and implies the construction of @xmath17-dimensional complete non - ambiguous trees . as a bonus of applying the theory of heaps , we establish a bijection between the set of complete non - ambiguous forests and the set of pairs of permutations with no common rise .
this answers an open question raised by aval _
et al._. corresponding author email : [email protected] ; [email protected] . the author is supported by the german research foundation dfg , ji 207/1 - 1 and the austrian research fund fwf , project sfb f50 algorithmic and enumerative combinatorics . ] |
cosmic objects emit in diverse electromagnetic wavelengths from radio to @xmath0-rays . some of them ( e.g. , stars ) are luminous primarily in a narrow wavelength range , while others ( e.g. , active galactic nuclei ) emit in a broad energy spectrum .
it is , therefore , essential to observe a celestial source in various wavelengths in order to fully understand its nature . up to the middle of the twentieth century ,
visible light was the primary window for observation .
but in the last few decades , wavelength - based branches of astronomy have not only been developed , but also got matured . moreover
, many new types of sources have been discovered by the observations in wavelengths other than visible lights .
one such discovery was made in x - rays using rocket - borne instruments in 1962 @xcite .
these instruments detected x - rays from sources outside the solar system for the first time , and specifically discovered a new type of source scorpius x-1 .
this discovery brought a new era in astronomy , and formally gave birth to a new branch called x - ray astronomy " .
the new source scorpius x-1 was the first observed x - ray binary , and more specifically it was the first detected low - mass x - ray binary system containing a neutron star and a low - mass stellar companion . unlike the optical ( i.e. , visible light ) and radio detection instruments , the x - ray instruments must be sent above the atmosphere for observations .
this is because x - rays from space can not reach the surface of the earth . in 1960
s , balloons and sub - orbital rockets were extensively used to lift the x - ray instruments . however , as these carriers are short - lived , satellites were thought to be ideal for x - ray observations .
the first astronomy satellite _ uhuru _ was launched by nasa in 1970 .
it identified over 300 discrete x - ray sources .
nasa s _ einstein _ satellite ( launched in 1978 ) was the first fully imaging x - ray telescope put into space .
its few arcsecond angular resolution increased the sensitivity enormously , and made the imaging of extended objects , and the detection of faint sources possible .
the _ rosat _ satellite launched in 1990 by the germany / us / uk collaboration did the first x - ray all - sky survey using a highly sensitive imaging telescope , and detected more than 150000 objects .
the scientific outcome of these key x - ray space missions , as well as the other past x - ray satellites , have revolutionized our knowledge of the x - ray sky .
the current primary x - ray space missions are ( 1 ) nasa s _ rossi x - ray timing explorer _
( _ rxte _ ) ; ( 2 ) european space agency s _ xmm - newton _ ; ( 3 ) nasa s _ chandra _ ; and ( 4 ) japan s _ suzaku_.
they primarily operate in the @xmath1 kev range ( i.e. , soft x - rays ) , and contain a variety of instruments .
the primary instrument of _ rxte _ is a set of five proportional counters ( called proportional counter array " or pca ) with large collecting area and very good time resolution ( @xmath2 microsecond ) .
therefore , pca is an ideal instrument to study the fast timing phenomena , as well as the continuum energy spectrum .
however , it is not an imaging instrument , and due to poor energy resolution , pca is not very suitable to study spectral lines .
this instrument operates in @xmath3 kev band , although the effective collecting area becomes small above @xmath4 kev .
another very useful instrument of _ rxte _ is all - sky monitor " or asm .
it consists of three relatively low - sensitivity proportional counters with large field of view , and hence monitors the entire sky to detect transient x - ray sources .
_ xmm - newton _ contains x - ray telescopes , charge - coupled devices and high resolution reflection grating spectrometers ( rgs ) " .
therefore , this satellite , that operates in @xmath5 kev range , is a good imager , and is ideal for spectral analysis .
observatory contains an x - ray telescope , charge - coupled devices , high resolution camera and transmission gratings .
its angular resolution ( @xmath6 subarcsecond ) is the best among all the past and current x - ray instruments , which makes it a unique imager .
_ chandra _ is also ideal to detect and study narrow spectral lines , and it operates in @xmath7 kev . finally , the _ suzaku _ satellite also has x - ray telescopes and charge - coupled devices .
since it contains an additional hard x - ray detector " ( hxd ) , _ suzaku _ can study the broadband energy spectrum .
all these x - ray space missions , especially the current ones , have gathered a wealth of information about the low - mass x - ray binary systems , and hence our understanding of these sources has grown enormously . in this review , we will give a brief overview of the x - ray observational aspects of neutron star low - mass x - ray binaries .
we note that the reference list will not be exhaustive in this short review .
neutron stars are extremely compact objects with a mass of about one solar mass ( @xmath8 ) , but a radius of @xmath9 km .
the core density of neutron stars are @xmath10 times the nuclear density , and their magnetic field can be anywhere between @xmath11 gauss .
they can spin with a wide range of frequency values : very slow to several hundred hertz .
study of these stars provides a unique opportunity to probe some extreme aspects of the universe , that can not be achieved by other means .
although the neutron stars were discovered in 1967 , their plausible existence was predicted in 1930 s . around this time ,
subrahmanyan chandrasekhar calculated the upper limit of the mass of a collapsed star ( white dwarf ) that is supported by the electron degeneracy pressure against the gravity .
this limit is the celebrated chandrasekhar mass limit " , with a value of about @xmath12 .
if the mass of the collapsed star is greater than this limit , then the star is expected to collapse more .
but if the mass is less than a certain value , the neutron degeneracy pressure will eventually balance the gravity to form a stable star , known as neutron star @xcite .
this remained as a theoretical calculation almost for three decades , after which hewish et al .
@xcite reported the discovery of a 1.377 sec period radio pulsar at 81.5 mhz , and gold @xcite showed that the pulsars are spinning neutron stars with large surface magnetic fields .
the density of a neutron star increases by @xmath13 orders of magnitude from the surface to the centre . as a result ,
the constituents and their properties change dramatically with the radial distance ( see , for example , shapiro & teukolsky @xcite ) .
the dense core of a neutron star may contain pion condensates , deconfined quarks , or other exotic ( and possibly unknown ) forms of matter , which makes the study of this part of the star extremely interesting . moreover according to a model , some of the neutron stars are actually strange stars " made entirely of u , d and s quarks @xcite .
unfortunately , in spite of many theoretical studies , the properties of matter in the core of a neutron star is not yet known , because such a dense degenerate matter can not be created in the laboratory . perhaps the only way to understand
the nature of this high density matter is to constrain the theoretically proposed equation of state models of the stellar core .
this can be done by measuring three independent structural parameters ( e.g. , mass , radius , spin frequency ) of the same neutron star , because the stable stellar structure gives a single mass vs. radius curve for a known spin frequency and a given equation of state model @xcite .
measurement of the stellar parameters is , therefore , a primary goal of the neutron star astronomy .
such a measurement , especially of the radius , is an extremely formidable task .
this can be appreciated from the fact that we need to measure the @xmath9 km radius of a neutron star with @xmath14% error , while the typical distance of the star is @xmath15 km .
neutron stars can emit in various wavelengths depending on their properties and the systems they are in .
they are observed as radio , x - ray and @xmath0-ray pulsars ( that may create pulsar wind nebulae ) , millisecond pulsars in binary systems , magnetars , and so on , and can be found in supernova remnants , x - ray binaries , etc . in this review we will concentrate on neutron stars in x - ray binaries .
point ) are called roche lobes , and the inner lagrange point @xmath16 , having a saddle - like potential , allows the matter to pass from one lobe to the other ( figure courtesy : dipankar bhattacharya ) .
, width=288 ] a low - mass x - ray binary ( lmxb ) is an old binary stellar system ( typical age @xmath17 years ) with a low - mass companion star ( @xmath18 ; main sequence star , evolved star or white dwarf ) and a primary compact star ( neutron star or black hole ) rotating around each other . in such a system , the companion star fills its roche lobe , i.e. , the critical equipotential surface ( see fig .
1 ) , and matter from it flows towards the compact star . because of the initial angular momentum , this matter can not move radially towards the primary star , and follows slow spiral paths that form an accretion disk . in this review ,
we consider the systems with neutron stars as the primary stars . in such neutron star lmxbs
, the accreted matter eventually hits the stellar surface , and generates electromagnetic radiation .
the accretion disk also emits such radiation by viscous dissipation , and both these emissions are powered by the gravitational potential energy release .
the inner part of the accretion disk , as well as the neutron star surface , emit x - rays , as their typical temperature is @xmath19 k. fig .
2 gives an artist s impression of such a system . in the next section
, we will give the basic theory of a simple geometrically thin accretion disk in newtonian gravity .
note that a more realistic approach may include complicated fluid dyanamical processes ( e.g. , mukhopadhyay 2006 @xcite ; mukhopadhyay et al .
@xcite ) and general relativistic effects ( e.g. , mukhopadhyay & misra 2003 @xcite ) . moreover , the magnetosphere of the neutron star can affect the accretion process depending on the accretion rate and the magnetic field strength .
ghosh et al .
@xcite and ghosh & lamb @xcite provided important theoretical understanding of this process . due to accretion induced angular momentum transfer , the neutron stars in lmxbs are expected to be spun up .
this is consistent with the observed stellar spin frequencies ( typically @xmath20 hz ) .
the magnetic fields of these neutron stars are relatively low ( @xmath21 gauss ) compared to the expected fields of these stars when they are born , and several groups have worked on this problem of stellar magnetic field evolution @xcite . note that the low magnetic field allows the accretion disk to approach very close to the neutron star .
this , in combination with the fact that x - rays can be detected both from the star and the inner part of the disk for many lmxbs , provides an excellent opportunity to study the extreme physics of and around neutron stars
. however , the modeling of any observation of the neutron star vicinity should consider the special and general relativistic effects including light bending ( e.g. , datta & kapoor @xcite ) .
lmxbs are observed mostly in the disk , bulge and globular clusters of our galaxy .
some of them have also been discovered from the nearby galaxies , and the locations much above our galactic disk plane . according to the catalogue of liu et al .
@xcite , 187 lmxbs ( containing black hole or neutron star ) were known from our galaxy , large magellanic cloud " and small magellanic cloud " up to that time .
lmxbs emit primarily in x - ray wavelengths , and the optical to x - ray luminosity ratio is normally less than 0.1 for them . the central x - ray emission from these sources can not be spatially resolved using the current x - ray imaging instruments , because the distances of even our galactic lmxbs are very large ( typically @xmath15 km ) . therefore , in order to understand the neutron star lmxbs , we have to rely on their spectral and timing properties .
in this review , we will describe some of these properties , as well as the information they provide . in a neutron star lmxb
, the companion mass is transferred to the neutron star typically via a geometrically thin accretion disk .
therefore , in this section we will briefly discuss the basic theory of such a disk . here
we note that the standard model of steady thin accretion disks in newtonian gravity was worked out by shakura & sunyaev @xcite . in a thin accretion disk using cylindrical coordinates ( @xmath22 )
, we expect that the @xmath23 component @xmath24 should be the main component of velocity , and the @xmath25 component @xmath26 .
the accreted matter should move around the neutron star with a nearly keplerian speed ( i.e. , the angular speed @xmath27 ; @xmath28 mass of the neutron star ) , and a small radial inflow ( speed @xmath29 ) caused by the effect of viscosity should exist .
we consider @xmath30 because of the @xmath23-symmetry of the system .
then the equation of continuity is @xmath31 where @xmath32 is the surface density of the accretion disk . using this and the @xmath23-component of the navier - stokes equation
, we get @xmath33 where @xmath34 and @xmath35 is the term involving viscosity .
we note that @xmath36 is the angular momentum of an annulus between @xmath37 and d@xmath37 .
hence the equation ( 2 ) multiplied with @xmath38 gives how the angular momentum of the annulus changes by the viscous torque .
therefore , @xmath39 where @xmath40 is the viscous torque at the radial distance @xmath37 .
it is normally assumed that the term @xmath41 gives the velocity shear .
however for the accretion disk , @xmath42 in which @xmath43 is associated with pure rotation , and @xmath44 gives the shear .
therefore , the viscous stress is @xmath45 , where @xmath46 is the coefficient of viscosity .
then the viscous torque is @xmath47 here , @xmath48 is the kinematic viscosity and @xmath49 is the volume density . therefore using the equations ( 2 ) , ( 3 ) and ( 5 )
, we get @xmath50 the equations ( 1 ) and ( 6 ) govern the accretion disk dynamics .
if the disk is nearly keplerian , then these two equations can be combined to get @xmath51.\end{aligned}\ ] ] this equation gives the time evolution of the accretion disk .
let us now consider a special case , viz .
, a steady ( @xmath52 ) geometrically thin accretion disk . from the equations ( 1 ) and ( 6 )
, we get @xmath53 @xmath54 where @xmath55 and @xmath56 are constants . for a steady disk ,
the mass accretion rate is @xmath57 .
therefore from equation ( 8) , @xmath58 in order to calculate @xmath56 , we consider a boundary condition @xmath59 let us justify this condition for the following two cases : ( 1 ) when the disk is extended up to the stellar surface , and ( 2 ) when the star is well inside the disk , and the matter falls from the disk inner edge towards the star quasi - radially . for a near - keplerian disk ,
the angular speed @xmath43 increases with the decrease of @xmath37 . for the first case ,
since the stellar angular speed is less than keplerian ( otherwise the star would break up ) , the disk @xmath43 must decrease with the decrease of @xmath37 close to the star .
therefore , @xmath60 must be zero at some point ( assumed to be @xmath61 ) near the stellar surface .
on the other hand , for the second case , the torque at the disk inner edge is expected to vanish ( note : it may not be exactly true in reality ) making @xmath62 .
the region ( between the star and the disk ) in which the flow deviates considerably from keplerian is called the boundary layer . from the equations ( 8) , ( 9 ) and ( 11 ) ,
we get @xmath63 where @xmath64 is the stellar mass . substituting this in the equation ( 9 ) , and using the equations ( 8) and ( 10 ) ,
we get , @xmath65.\end{aligned}\ ] ] this shows that the accretion rate is greater for larger viscosity .
the viscous dissipation in the accretion disk causes the emission from the disk , and the source of this emitted energy is the gravitational potential energy release .
this viscous dissipation rate per unit volume of the accretion disk is @xmath66 . integrating this over the disk thickness and assuming keplerian angular speed
, we get the rate of energy emitted per unit area of the disk from equation ( 13 ) : @xmath67.\end{aligned}\ ] ] therefore , the total power emitted from a disk extending from @xmath68 to infinity is @xmath69 this shows that , for a steady geometrically thin accretion disk in newtonian gravity , half of the gravitational potential energy is emitted by viscous dissipation in the disk , and the rest half remains stored in the accreted matter as the kinetic energy , and gets dissipated in the boundary layer .
it is expected that a geometrically thin keplerian disk should emit like a blackbody .
therefore , the temperature @xmath70 of the disk surface can be calculated from equation ( 14 ) : @xmath71\}^{1/4},\end{aligned}\ ] ] where @xmath72 is the stefan - boltzmann constant . as the blackbody temperature varies with the radial distance ,
the spectrum from a disk is known as multicolor blackbody . for luminous lmxbs
, the x - rays from the neutron star boundary layer and the inner part of the accretion disk can substantially irradiate the disk surface .
this irradiation temperature @xmath73 ( which is blackbody temperature ) is proportional to @xmath74 @xcite .
therefore , the net effective disk temperature is @xmath75^{1/4}$ ] , in which @xmath70 dominates for the inner portion ( that emits x - rays ) of the disk , and @xmath73 dominates far away from the neutron star .
note that the latter can have significant consequences for the disk instability @xcite .
a class of neutron star lmxbs are called transients .
these sources have two distinctly different intensity states : ( 1 ) the quiescent state in which the source remains for months to years ; and ( 2 ) the outburst state which continues for weeks to months ( sometimes for years ; e.g. , for the source ks 1731 - 260 ) .
the luminosity of the source typically increases by several orders of magnitude as it evolves from the quiescent state into an outburst .
3 shows the long term intensity profile of such a transient 1a 1744 - 361 , exhibiting two clear outbursts .
the transient nature of these lmxbs is normally attributed to an accretion disk instability , that causes high accretion rates for certain periods of time , and almost no accretion during other times ( see king 2001 @xcite for a review ) .
these sources are often not detectable in their quiescent states , and in some cases when they are detectable , the poor statistics due to the small number of observed photons hinders a detailed study . for most of the scientific purposes ,
it is therefore preferable to observe these transients during their outbursts .
new transients are also typically discovered during the outburst states .
however , it can not normally be predicted when a transient will come out of its quiescent state .
therefore , instruments with large field of view that can monitor the entire sky almost continuously are required to find the transients in their outbursts .
the _ rxte _
asm is such a very useful instrument , that has discovered many new transient lmxbs , as well as detected subsequent outbursts of them .
however , there are scientific reasons that motivate the observations of transient neutron star lmxbs in their quiescent states . for example , during the quiescence , the neutron star surface should be the primary x - ray emitting component , which can be modelled to constrain the stellar radius .
besides , a long outburst heats up the neutron star crust taking it out of thermal equilibrium with the rest of the star .
therefore , tracking the lmxb system going into the quiescence , and the analysis of the quiescent x - ray data can provide a unique opportunity to probe some aspects of the neutron star extreme physics ( e.g. , cackett et al .
@xcite ) . for example
, such a study allows one to compare the observed neutron star cooling rate with the theoretical cooling curve , that involves the physics of stellar constituents .
note that the satellites ( e.g. , _ chandra _ , _ xmm - newton _ ) with high sensitivity imaging instruments are used to observe the sources in quiescence .
many neutron star lmxb systems show a spectacular eruption in x - rays every few hours to days . during these bursts ,
the observed x - ray intensity goes up sharply in @xmath76 seconds , and decays relatively slowly in @xmath77 seconds ( see fig .
4 ; strohmayer & bildsten @xcite ) . the typical energy emitted in a few seconds during such
a burst is @xmath78 ergs ( note that the sun takes more than a week to release this energy ) .
these bursts are known as type - i or thermonuclear bursts , and were first discovered in 1970 s @xcite . soon after their discovery it was realized that they originate from recurrent unstable nuclear burning of accreted matter on the neutron star surfaces @xcite .
the neutron star surface origin was supported by the observational fact that the burst emission area ( estimated from the energy spectrum of the bursts ) matched with the expected surface area of a neutron star @xcite .
although , a competetive emission area could be the inner portion of the accretion disk , nuclear ignition in order to produce the observed bursts can not occur in such a rare medium .
but what shows that the nuclear burning causes the bursts ? before coming to that point , we would like to convince the readers that such a nuclear burning has to be unstable .
this is because , while the nuclear energy released per nucleon by fusion is only a few mev , the gravitational energy release by a nucleon falling on the neutron star surface from a large distance is @xmath79 ( @xmath64 and @xmath80 are neutron star mass and radius respectively ) , which is typically @xmath81 mev .
therefore , the energy released from a stable burning of the accreted matter would be entirely lost in the observed gravitational energy .
so the nuclear burning can power the bursts , only if the accreted matter accumulates on the neutron star surface for a long time , and then burns quickly in an unstable manner .
this instability appears if the nuclear energy generation rate is more temperature sensitive than the radiative cooling rate . for an observed data set from a given source , the ratio of the total energy content in all the bursts to that in the non - burst emission roughly tallies with the expected nuclear energy release ( per nucleon ) to the expected gravitational energy release ( per nucleon ) ratio @xcite
this is one of the strongest evidences of the nuclear energy origin of the bursts .
here we note that , only a year before these bursts were discovered , hansen and van horn @xcite predicted that thin hydrogen and helium layers on neutron star surfaces could be susceptible to the above mentioned instability @xcite .
the accreted material consists of mostly hydrogen , some helium and a small amount of heavier elements , except when the donor companion is a white dwarf .
this matter , that accumulates on the neutron star surface , undergoes hydrostatic compression as more and more material piles on , and the ignition density and temperature are reached in a few hours to days .
the burning typically happens at a depth of @xmath82 m at a column density of @xmath83 gm @xmath84 . the compression rate and the ignition depends on the accretion rate , which sets four distinct regimes of burning @xcite . _
regime 1 _ : in most cases , the temperature of the accumulated material exceeds @xmath85 k , and hence the hydrogen burns via the cno cycle instead of the proton - proton ( pp ) cycle .
this causes mixed hydrogen and helium bursts triggered by the unstable hydrogen ignition , which happens for the accretion rate per unit neutron star surface area m @xmath86 gm @xmath84 sec@xmath87 ( @xmath88)@xmath89 @xcite . here
@xmath90 is the mass fraction of cno . _
regime 2 _ : at higher temperatures ( @xmath91 k ) , the proton capture timescale becomes shorter than the subsequent @xmath92 decay lifetimes .
hence , the hydrogen burns via the hot " cno cycle of fowler & hoyle @xcite : @xmath93 in this case , the hydrogen burning happens in a thermally stable manner ( i.e. , without triggering a thermonuclear burst ) simultaneously with the accumulation of matter .
this allows a helium layer to build up below the hydrogen layer . for the m range @xmath94 gm @xmath84
sec@xmath87 ( @xmath88)@xmath95 m @xmath96 gm @xmath84 sec@xmath87 ( @xmath88)@xmath97 , the hydrogen becomes entirely burned before this helium is ignited @xcite .
so when the helium ignition happens , a short ( @xmath9 sec ) but very intense burst occurs by the unstable triple - alpha reaction of the pure helium ( @xmath98 @xmath99 @xmath100 ) .
such a burst is called a helium burst " . many of these bursts are so intense , that the local x - ray luminosity in the neutron star atmosphere may exceed the eddington limit ( for which the radiative pressure force balances the gravitational force ) , and the photospheric layers may be lifted off the neutron star surface .
such bursts are called photospheric radius expansion ( pre ) bursts .
_ regime 3 _ : for m @xmath101 gm @xmath84 sec@xmath87 ( @xmath88)@xmath97 , hydrogen burns via the hot " cno cycle in a stable manner ( as for the regime 2 ) , but enough unburnt hydrogen remains present at the time of helium ignition .
this is because at this higher accretion rate , the helium ignition conditions are satisfied much sooner .
therefore in this regime , mixed hydrogen and helium bursts are triggered by the helium ignition .
during such a burst , the thermal instability can produce elements beyond the iron group @xcite via the rapid - proton ( rp ) process of wallace & woosley @xcite .
this process burns hydrogen via successive proton capture and @xmath92 decays .
the long series of @xmath92 decays makes these bursts typically much longer ( @xmath102 sec ) than the helium bursts . _
regime 4 _ : finally at a very high accretion rate ( comparable to the eddington limit ) , the helium burning temperature sensitivity becomes weaker than the cooling rate s sensitivity @xcite .
therefore in this regime , the stable burning sets in , and the thermonuclear bursts do not occur . apart from type - i x - ray bursts , another type of thermonuclear x - ray bursts with much larger recurrence time ( @xmath6 years ) has been observed . these are called superbursts , and they differ from the type - i bursts also in the decay time ( @xmath103 hours ) and the amount of released energy ( @xmath104 ergs ) .
the first superburst was discovered by cornelisse et al .
@xcite from the known type - i burster 4u 1735 - 44 .
so far about 15 superbursts have been detected from 10 sources @xcite .
it is believed that these bursts of large fluence happen by the @xmath105 fusion reaction at a column depth of @xmath106 gm @xmath84 @xcite .
however , currently our understanding of superbursts is not as clear as that of type - i bursts .
thermonuclear bursts provide a unique opportunity to study some aspects of extreme physics .
moreover , such a study brings together several branches of physics : nuclear physics , astrophysics , general theory of relativity , magnetohydrodynamics , etc .
these bursts can also be useful to constrain the neutron star parameters , and hence to understand the nature of the stellar core matter at supranuclear densities ( see the spectral and timing properties of these bursts in later sections ) .
therefore , the thermonuclear bursts belong to a truly multidisciplinary field , which is not yet well understood .
one such poorly explored aspect of these bursts is the thermonuclear flame spreading on the neutron star surfaces .
burst is ignited at a certain location on the stellar surface , and then the burning region expands to cover the entire surface @xcite .
the theoretical modeling of thermonuclear flame spreading including all the main physical effects is extremely difficult , and only recently spitkovsky et al .
@xcite have provided some insights in this field . although these authors have ignored several physical effects ( magnetic field , strong gravity , etc . ) in their model , they have considered the effects of the coriolis force , which is important as the bursting neutron stars are rapidly spinning ( spin frequency @xmath20 hz ) .
the observational study of flame spreading is also no less difficult than the theoretical study .
this is because , this spreading mostly happens within @xmath2 sec during the burst rise , and in order to study it observationally , one needs to detect and measure the evolution of the burst spectral and timing properties during this short time .
only recently this has been possible for a few bursts by analyzing the _ rxte _ pca data @xcite .
moreover , an indication of the effects of the coriolis force on the flame spreading ( as predicted by spitkovsky et al .
@xcite ) has also been found @xcite .
a series of bursts , known as type - ii x - ray bursts , are observed from neutron star lmxbs in quick succession . to the best of our knowledge , only two such systems ( rapid burster and gro j1744 - 28 ) exhibit these bursts @xcite .
the intervals between the bursts vary between seconds to @xmath107 hour , and the burst duration is in the range of seconds to minutes .
the typical emitted energy range during a burst from the rapid burster is @xmath108 ergs to @xmath109 ergs @xcite .
these bursts often show successive peaks and flat tops .
note that a source is not always active for type - ii x - ray bursts , and a series of such bursts are observed in certain periods of time .
type - ii x - ray bursts are distinctly different from the type - i bursts in many ways .
for example , the latter bursts show a spectral softening during burst decay , while the former bursts lack this signature .
type - ii bursts are believed to be caused by spasmodic accretion , possibly due to an accretion instability @xcite .
therefore , these bursts are powered by the gravitational energy , while the type - i bursts are powered by the nuclear energy .
the recent models of the type - ii bursts include the effects of magnetosphere @xcite and intermittent advection - dominated accretion flow @xcite . however , the origin of the accretion instability is not yet well understood .
some neutron star lmxbs show periodic temporary decrease in x - ray intensity .
this timing feature can be divided into two distinct categories : eclipses and dips .
both of them occur with the binary orbital period , and hence provide an excellent way to measure this period @xcite .
partial and total eclipses are believed to be caused by the obscuration of the x - ray source by the companion star @xcite .
such an obscuration requires an observer s line - of - sight close to the plane of the binary system ( i.e. , close to the accretion disk plane ) .
therefore , the observation of eclipses from a source implies that its inclination angle is large ( @xmath110 ; frank et al .
this inclination angle @xmath111 is measured from the direction perpendicular to the accretion disk plane .
the dips are periodic but irregular , and typically are more pronounced at lower energies @xcite .
the shape , depth and duration of each dipping interval widely vary from cycle to cycle , and within such an interval the observed intensity varies dramatically on a timescale of a few seconds ( see fig . 5 ) .
these dips are believed to be caused by a bulge on the edge of the accretion disk at the region where the gas stream from the companion star hits the disk ( see fig . 2 for a view of this impact region ; white & swank @xcite ) .
therefore , dips imply @xmath112 @xcite .
the detailed nature of this bulge is still a matter of debate , and spectral analysis can shed light on this . )
spectral line from the inner accretion disk of a neutron star lmxb serpens x-1 @xcite .
the x - axis shows the x - ray energy , and the y - axis gives the observed intensity in excess to the best - fit continuum spectral model .
the data points ( shown with error bars ) clearly shows an asymmetric spectral line , while the dotted profile is a relativistic model of spectral line that fits the observed line well .
, width=288 ]
the study of the continuum spectrum of the persistent emission ( i.e. , non - burst , non - dipping , non - eclipsing ) from the neutron star low - mass x - ray binary systems can be very useful to understand the various components of the accretion flow and their structures , including the flow in the strong gravity region near the neutron star .
such a spectrum can be typically fitted with two spectral components .
mitsuda et al .
@xcite used a single - temperature blackbody plus a multicolor blackbody model .
the former component is expected to be produced from a boundary layer between the accretion disk and the neutron star , while the latter may be produced from the accretion disk . in an alternative two - component model , white et al .
@xcite used a comptonization component instead of a multicolor blackbody .
they assumed that the comptonization component was originally the disk component which was dominated by the comptonization processes . in the later years , it has been found that a comptonization component is often required to fit the observed persistent emission continuum spectrum , and an additional blackbody is needed when the source is more luminous and spectrally softer ( see , for example , bhattacharyya et al .
@xcite ) . apart from these
, there is evidence of an extended accretion disk corona ( adc ) " for some sources , that can be modelled with a power law @xcite .
however , we note that much work needs to be done in order to build a self - consistent spectral model involving all the x - ray emitting and absorbing components , and to fit the observed spectrum with such a single model . spectrum of a thermonuclear x - ray burst is obtained usually by subtracting the preburst emission spectrum ( assumed to be powered by the gravitational energy release ) from the observed spectrum during the burst @xcite .
the continuum burst spectra are thermal , and in most cases they are quite well described with a blackbody model @xcite .
the temperature of this blackbody naturally increases rapidly during the burst intensity rise , and then decreases relatively slowly during the burst decay .
if the entire neutron star surface emits like a blackbody of the same temperature at a given time , then the neutron star radius ( @xmath113 ) can , in principle , be measured from the following formula : @xmath114^{1/2}$ ] . here ,
@xmath115 and @xmath116 are the observed flux and temperature respectively , and @xmath117 is the source distance , which may be known by an independent technique .
note that the measurement of the neutron star radius can be useful to understand the nature of the dense matter of the stellar core .
however , in order to do such measurement , one needs to include the following two effects .
( 1 ) the electron ( compton ) scattering in the neutron star atmosphere shifts the blackbody spectrum ( originated from the burning layer ) towards the higher energies , and changes its shape slightly @xcite .
such a blackbody is called a diluted blackbody " , and its temperature ( @xmath118 , i.e. , the color temperature ) is related with the blackbody temperature ( @xmath119 ) by the color factor @xmath120 .
( 2 ) the surface gravitational redshift shifts the temperature towards a lower value .
therefore , the observed temperature @xmath116 is related to @xmath119 by @xmath121 .
correspondingly , the actual neutron star radius @xmath122 is then related to the observed radius @xmath113 by @xmath123 .
therefore , in order to measure the neutron star radius from the continuum burst spectra , one needs to independently estimate or measure the color factor @xmath120 and the surface gravitational redshift factor @xmath124 .
the former can be estimated from the temperature , surface gravity and the atmospheric chemical composition @xcite , while the latter can be calculated from the neutron star radius - to - mass ratio .
this ratio can be independently measured from the surface atomic spectral lines or the burst oscillations ( see later ) .
the evolution of the continuum spectrum of a thermonuclear burst can be best measured using the pca instrument of the _ rxte _ satellite .
such a measurement during the burst rise will be extremely useful to study the thermonuclear flame spreading .
the chemical composition of the neutron star atmosphere may change during the burst . therefore , and since the burst temperature also evolves , the observed radius @xmath113 changes substantially , and apparently irregularly during the burst decay @xcite .
an understanding of these changes will be essential to ( 1 ) measure the neutron star radius using the burst continuum spectrum , ( 2 ) understand the nuclear physics of bursts , and ( 3 ) probe the magnetohydrodynamics of the burning layer and the atmosphere . in order to probe the nature of the bulge that causes dips ,
it is essential to understand how the continuum spectrum changes as the source goes into the dipping interval .
a simple increase in the neutral absorber column density from its persistent emission value can not explain the dip spectra . currently , there are two competing spectral models to explain dips .
( 1 ) the progressive covering " model ( e.g. , church et al .
@xcite ) : in this approach , the x - ray emission is considered to originate from a point - like blackbody and a power - law from an extended component ( e.g. , a corona ) .
the dipping is assumed to be caused by the partial and progressive covering of the extended component by a neutral absorber .
( 2 ) the ionized absorber model ( e.g. , boirin et al .
@xcite ) : in this approach , the dips are caused by the increase in column density and the decrease in ionization state of an ionized absorber , along with a neutral absorber column density increase .
this model can naturally explain some of the observed narrow spectral features . many of the high inclination neutron star lmxbs ( for example , dipping sources ) exhibit narrow spectral absorption features ( e.g. , boirin et al .
@xcite ; parmar et al .
@xcite ; daz trigo et al .
@xcite ) , as well as narrow recombination emission lines @xcite .
the most prominent of these lines are fe xxv and xxvi absorption lines .
this spectral features are believed to originate from the material above the accretion disk , and can be observed only when the observer s line - of - sight passes through this material .
this is possible only for the high inclination sources .
the narrow spectral features , therefore , can be very useful to constrain the structures and properties ( e.g. , ionization states , relative abundances of various elements ) of the photoionized plasma above the accretion disks in these systems @xcite .
note that , although the spectral features are mostly seen from high inclination lmxbs , this plasma must be common to all lmxbs , and hence these features will be useful in understanding lmxbs in general .
broad asymmetric spectral iron emission line has been observed from many accreting supermassive and stellar - mass black hole systems @xcite .
it is believed to originate from the inner portion of the accretion disk , when the disk is irradiated by a hard x - ray source ( e.g. , an accretion disk corona ) .
such an irradiation creates a reflection " component of the x - ray spectrum .
when the hard x - ray photon enters the colder disk , it is either scattered out of the disk , or destroyed by auger de - excitation , or reprocessed into a fluorescent line photon that escapes the disk @xcite .
the observed iron k@xmath125 line near 6.4 kev is the strongest fluorescent line from such a system . as the matter in the inner accretion disk rotates very fast , and is close to the black hole , doppler effect broadens the line , special relativistic beaming makes it asymmetric , and the gravitational redshift shifts it towards the lower energies .
therefore , these lines can be useful to study the accretion flow close to the black hole , to test the theory of general relativity , and to measure the black hole spin . for long time
it has been known that the persistent spectra of many neutron star lmxbs contain broad iron emission line @xcite .
it was suspected that such a line might have an origin similar to that of the broad iron line from black hole systems
. however , the asymmetry of these lines was not established , and hence their origin was not well understood .
recently , bhattacharyya & strohmayer @xcite have , for the first time , established the inner disk origin of such a line from the neutron star lmxb serpens x-1 ( see fig .
this was later supported for several other sources ( e.g. , cackett et al .
@xcite ; pandel et al .
@xcite ; di salvo et al .
@xcite ) . therefore
, the high precision observations of these broad iron k@xmath125 line from neutron star lmxbs could be useful to constrain the stellar parameters ( radius , angular momentum ) and to probe the inner accretion disk .
however , the evidence of the reflection " component of the x - ray spectrum is still marginal @xcite , which the theory and the future observations should address . observing atomic spectral lines from the surface of a neutron star
has been one of the long - standing goals of the neutron star astrophysics .
this is because , such an observation and the identification of the line provide the cleanest way to measure the neutron star radius - to - mass ratio using the following formula ( for the schwartzschild spacetime appropriate for a non - spinning neutron star ) : @xmath126^{-1/2}$ ] @xcite . here
, @xmath127 is the emitted energy of the line photons , @xmath128 is the observed energy of these photons ( different from @xmath127 because of the surface gravitational redshift ) , and @xmath64 and @xmath80 are the mass and the radius of the neutron star respectively . note that the measurement of the stellar parameter @xmath129 can be useful to understand the nature of the dense core matter of neutron stars .
as the neutron stars in lmxb systems typically spin with high frequencies ( @xmath20 hz ) , doppler effect and special relativistic beaming should make the lines broad and asymmetric ( similar to the lines from disks mentioned in the previous section ) .
measuring @xmath129 accurately from such a broad line ( for which the definition of @xmath128 is not certain ) using the above formula is difficult .
bhattacharyya et al .
@xcite have recently suggested a way to measure the stellar @xmath129 using the above formula from such lines .
these broad asymmetric lines can have low - energy and high - energy peaks ( or dips for absorption lines ) .
bhattacharyya et al .
@xcite have found that the ratio of energy contents of these peaks ( or , dips ) contains information about the general relativistic frame - dragging .
therefore , surface atomic lines can be used to test the theory of general relativity in the strong gravity regime .
surface lines are likely to be detected from thermonuclear x - ray bursts and hence from neutron stars in lmxb systems .
this is because : ( 1 ) the continuous accretion and the radiative pressure may keep the line - forming heavy elements in the stellar atmosphere for the time - duration required for the line detection ; ( 2 ) the relatively low magnetic fields ( @xmath130 gauss ) of the neutron stars in lmxbs should make the line identification easier by keeping the magnetic splitting negligible ; ( 3 ) the lower magnetic field does not complicate the modeling of the atmosphere ; ( 4 ) the luminous bursts give good signal - to - noise ratio ; and ( 5 ) during the bursts , typically 90% of the total emission originate from the neutron star surface , and hence the surface spectral line may be largely free from the uncertainty due to the other x - ray emission components , such as the accretion disk .
indeed , the only plausible observation of the neutron star surface atomic spectral lines was from an lmxb exo 0748 - 676 @xcite .
these authors detected two significant absorption features in the _ xmm - newton _ rgs energy spectra of the thermonuclear x - ray bursts , which they identified as fe xxvi and fe xxv @xmath131 lines with a redshift of @xmath132 ( see fig .
7 ) . in order to detect these lines
, they had to combine the spectra of 28 bursts .
the identification of these two lines as the surface atomic spectral lines is reasonable for the following reasons : ( 1 ) the early burst phases ( when the temperature was higher ) showed the fe xxvi line , while the late burst phases ( when the temperature was lower ) showed the fe xxv line .
this is consistent with qualitative temperature dependence of the iron ionization states .
( 2 ) the narrow lines were consistent with the relatively slow spin rate of the neutron star ( 45 hz ; villarreal & strohmayer @xcite ) .
( 3 ) both the lines showed the same surface gravitational redshift , as expected .
this redshift is reasonable for the surface of a typical neutron star .
the absorption of some of the bursts photons ( originated in deep burning layers ) by the iron ions in the upper atmosphere might give rise to these redshifted lines .
chang et al .
@xcite showed that the observed strength of the iron lines could be produced by a neutron star photospheric metallicity , which was @xmath133 times larger than the solar metallicity .
however , these lines were not significantly detected from the new data from the later observations of the same source .
although , this casts doubts in the reality of these lines , it may be caused by the change in the photospheric conditions that might weaken the lines .
therefore , the evidence of the existence of surface atomic lines currently remains uncertain , which , given the importance of these lines , provides the motivation for the future generation x - ray instruments .
a convenient way to study how the persistent spectrum of a neutron star lmxb changes with time , and how it is correlated with various timing properties and the occurrence of other features ( e.g. , thermonuclear x - ray bursts ) , is to compute the color - color diagram . in order to do this ,
typically five photon energies with increasing values are considered : @xmath134 , @xmath135 , @xmath136 , @xmath137 and @xmath138 .
let us assume that the observed background - subtracted photon count numbers in the ranges @xmath139 , @xmath140 , @xmath141 and @xmath142 are @xmath143 , @xmath144 , @xmath145 and @xmath146 respectively .
then we define soft color @xmath147 and hard color @xmath148 .
the color - color diagram is a plot of hard color vs. soft color , which shows the change of spectral hardness of the source .
neutron star lmxbs can be divided into two classes based on their color - color diagrams : the so called z sources " , that show distinct z - like curves ( fig .
8) , and the atoll sources " , that exhibit c - like curves ( fig .
these two types of curves are not only different by shape , but also they are correlated with distinctly different timing properties @xcite . apart from these differences , the z sources are very luminous ( luminosities are close to eddington ) , while the luminosities of the atoll sources typically remain between @xmath149 of the eddington luminosity @xcite .
the three branches of a z curve , from upper to lower , are called horizontal branch ( hb ) , normal branch ( nb ) and flaring branch ( fb ) .
the lower part of a typical atoll curve is called the banana state , while the upper part is known as the island state .
the x - ray luminosity typically increases from hb to fb for a z source , and from island state to banana state for an atoll source .
we note that , although the various characteristics of the color - color diagrams are observationally very robust , most of them are not yet theoretically well understood . accretion rate is known to be a parameter that changes along the z curves and the atoll curves .
but , given the complex properties of these curves , more parameters must be involved in the fundamental level .
it is also not understood what , apart from the accretion rate , distinguishes between the z and atoll properties .
the recent discovery of the first transient z source , that shows substantially changed properties at lower luminosities , holds the promise to take our understanding of these sources to a higher level @xcite .
most of the radio pulsars ( that are spinning neutron stars emitting periodic radio pulses ) have spin period @xmath150 milliseconds , and high magnetic field ( typically @xmath151 gauss ) .
these are relatively young neutron stars .
this is because , as a pulsar spins down ( due to magnetic dipole radiation ) and its magnetic field decays ( in timescales of several million years ) , it eventually ceases to be a pulsar @xcite .
however , a small fraction of the radio pulsar population is rapidly spinning ( period @xmath152 ms ) , old ( age @xmath6 billions of years ) , and having weak surface magnetic fields ( @xmath153 gauss ) . since many of these pulsars are in binary systems , it has been believed for long time that these neutron stars were spun up by the accretion - induced angular momentum transfer in lmxb systems @xcite . this model would be supported by the observation of millisecond - period x - ray pulsations from neutron star lmxbs , as such pulsations would be a missing link between the millisecond radio pulsars and lmxbs .
but years of searching did not yield the detection of any such coherent millisecond - period x - ray pulsations . finally in 1998 , the first millisecond x - ray pulsations ( with the frequency @xmath154 hz ) were discovered from the transient lmxb sax j1808.4 - 3658 with the _ rxte _ satellite @xcite . as x - ray pulsations ( like the radio pulsations ) measure the neutron star spin frequency
, this showed that the neutron star in this source spins with @xmath154 hz , and hence , indeed the neutron stars in the lmxbs can be spun up .
several such sources have been discovered after sax j1808.4 - 3658 ( see , for example , wijnands 2006 @xcite ; watts et al .
2009 @xcite .
10 shows the power spectrum of one such lmxb .
the narrow peak at the frequency of 442.36 hz shows the pulsations .
note that the computation of a power spectrum is a convenient way to detect timing features .
such a spectrum ( power vs. frequency ) is a fourier transform of the intensity profile ( photon counts vs. time ) .
for example , if @xmath155 is the number of photons detected in bin @xmath156 , then the discrete fourier transform @xmath157 decomposes the intensity into @xmath158 sine waves .
the direct and the inverse fourier transforms are then given by : @xmath159 in the lmxbs containing these pulsars , the accretion flow is channeled by the neutron star magnetic field to the magnetic polar cap .
the x - ray hot spot on the neutron star created by this channeled flow rotates around the stellar spin axis ( if the magnetic axis and the spin axis are misaligned ) and gives rise to the observed x - ray pulses .
since the gravitational energy release from the accreted matter powers the x - ray pulsations , these sources are called accretion - powered millisecond pulsars ( amps ) " . since the first discovery of these pulsars
, it intrigues the astrophysicists how these sources are different from other lmxbs , and why all the lmxbs do not exhibit such regular pulsations .
the recent discovery of several intermittent amps ( that show sporadic regular pulsations ) holds the promise to resolve these problems @xcite .
better understanding of the amps will be very important to probe the evolution of lmxbs , as well as to model the magnetically channeled flows in the strong gravity regions near the neutron stars .
modelling of this flow , its energy spectrum and the timing properties will be useful to constrain the neutron star parameters , and hence the nature of the stellar core matter @xcite .
high frequency timing features are observed from many ( so far about 20 ) neutron star lmxbs during thermonuclear x - ray bursts @xcite .
they are called burst oscillations " , and were first discovered with the _ rxte _ pca instrument in bursts from the lmxb 4u 1728 - 34 @xcite .
it was soon realized that this timing feature might be originated from an aymmetric brightness pattern on the surface of a spinning neutron star .
this was confirmed when burst oscillations were detected from the amp sax j1808.4 - 3658 close to the known neutron star spin frequency @xcite .
later , these oscillations were observed from other amps near the known stellar spin rates @xcite . these established a new technique to measure the neutron star spin frequency .
in fact , regular pulsations and burst oscillations are the only methods to measure the spin rates of neutron stars in lmxbs , and together they have provided spin frequencies of about 25 sources . apart from measuring the stellar spin
, the modelling of shapes and amplitudes of the phase - folded burst oscillation intensity profiles can be useful to constrain the other neutron star parameters ( including the radius - to - mass ratio ; e.g. , bhattacharyya et al .
@xcite ) . fig .
11 shows the burst oscillation power contours and the power spectrum for a neutron star lmxb .
the power contours indicate that the oscillation frequency changes with time .
indeed such small evolution of frequencies are observed for many bursts , which suggests that the aymmetric brightness pattern slowly moves on the spinning neutron star surfaces .
it is not yet understood what causes such motion , although several models exist in the literature @xcite .
the oscillations during burst rise are normally believed to be caused by the expansion of the burning region on the stellar surface immediately after the ignition .
therefore , the study of the evolution of the burst oscillation properties during burst rise can be useful to probe the thermonuclear flame spreading .
the nature of the aymmetric brightness pattern that causes the oscillations during burst decay is not yet well understood .
this is because the entire stellar surface is believed to be ignited at this time .
the existing models involve non - radial global oscillations in the surface layers @xcite , shear oscillations @xcite , etc . , each of which has shortcomings . hz frequency .
note that the shape of the power contours indicates a slight change in frequency with time , which is common for burst oscillation .
lower panel : the corresponding power spectrum .
the peak at @xmath160 hz implies the burst oscillation with that frequency .
, width=288 ] neutron star lmxbs show a variety of observationally robust timing features , many of which are not well understood yet .
continuum power , in excess of the mean random power , is often observed at frequencies less than 100 hz .
the nature of this power depends on the source state determined from the position on the color - color plot .
for example , a power - law - like continuum ( @xmath161 ; @xmath162 is the frequency in the power spectrum ; and @xmath125 is a constant ) below 1 hz is typically dominant for the atoll sources in the end part of the banana state @xcite .
apart from the broad continua , neutron star lmxbs often exhibit peaked features with a wide range of frequencies ( millihertz to kilohertz ) .
since they are somewhat broad , and hence not quite periodic , these features are known as quasi - periodic oscillations ( qpos ) .
similar to the continuum features , qpos are correlated with the color - color diagram .
for example , a type of @xmath163 hz qpo appears in the horizontal branch of the color - color diagram of z sources , and hence known as the horizontal branch oscillation ( hbo ) " .
the qpos , when modelled correctly , are expected to significantly enhance our understanding of neutron star lmxbs .
the observed correlations among different types of qpos can be helpful for their modeling .
however , in this short review we will not discuss the qpos and their correlations in detail , and we refer to an extensive review by van der klis @xcite that gives a large amount of information about the fast x - ray timing properties of neutron star lmxbs .
the only qpos , that we would like to briefly discuss here , are called kilohertz ( khz ) qpos .
these high frequency qpos are observed from both z and atoll sources .
the frequencies of khz qpos vary between @xmath164 hz , and often they appear in a pair ( see fig . 12 ) . for a given source , the lower and upper khz qpos
are seen to move up and down in frequency together without changing the frequency separation much .
this frequency separation is normally found to be within 20% of the neutron star spin frequency , or half of that , depending on source .
these khz qpos are believed to be originated from a region close to the neutron star , and hence , when their origin is fully understood , will be very useful to measure the neutron star parameters , and to test the theory of general relativity ( e.g. , miller et al .
1998 @xcite ; stella & vietri 1998 @xcite ; barret et al . 2005 @xcite ) .
neutron star lmxbs provide a unique opportunity to study several aspects of extreme physics that we have briefly discussed in this short review .
these include ( 1 ) constraining the equation of state models of the supranuclear density matter of neutron star cores from the measurements of the stellar parameters ; ( 2 ) testing the theory of general relativity ; ( 3 ) understanding the physics of neutron star atmospheres ; and ( 4 ) probing the accretion flow in the strong gravity region near the neutron stars .
in addition , the x - ray study of neutron star lmxbs allows us ( 1 ) to understand the evolution of these binary systems ; ( 2 ) to probe the long term accretion process and the structures of the accreted matter ; and ( 3 ) to find out how and why these systems are different from black hole lmxbs .
since the observed x - ray properties mentioned in this review can be used as tools to achieve these scientific goals , we need to understand these properties better , and explore ( 1 ) the correlations among them , and ( 2 ) their correlations with the observations in other wavelengths .
these can be achieved with more sensitive x - ray instruments ( typically with larger photon collecting areas ) , and simultaneous observations in multiple wavelengths .
it will also be required to monitor the entire sky frequently with high sensitivity in order to catch the transient neutron star lmxbs during their outbursts .
here we give a short description of some of the proposed future space missions with these capabilities . the proposed indian multiwavelength astronomy space mission
_ astrosat _ will observe simultaneously in a wide energy range ( optical to hard x - ray of 100 kev ) .
apart from this unprecedented capability , its laxpc instrument should be able to detect and measure the high frequency timing features , such as khz qpos and regular pulsations , for the first time in hard x - rays ( say , up to @xmath165 kev ) .
note that in its own time , laxpc will probably be the only instrument capable of detecting the high frequency features in the x - ray band .
the proposed next generation joint american , european and japanese satellite _ international x - ray observatory _
( _ ixo _ ) will have several highly sensitive instruments that will take the x - ray study of neutron star lmxbs to a higher level . in fig .
13 , we demonstrate how significant an improvement the high timing resolution spectrometer ( htrs ) instrument of _ ixo _ will make over the _ rxte _
pca for the high frequency timing observations .
moreover , ray et al .
@xcite mention that the proposed _ advanced x - ray timing array _
( _ axtar _ ) satellite will have better capabilities than _ ixo _ for these timing observations . finally
, the japanese mission _ monitor of all - sky x - ray image _
( _ maxi _ ) will monitor the x - ray sky in the @xmath166 kev range with high sensitivity .
therefore , the study of the neutron star lmxbs is likely to remain an active and promising field in order to explore some aspects of extreme physics .
ghosh , p. , pethick , c. j. , & lamb , f. k. , accretion by rotating magnetic neutron stars .
i - flow of matter inside the magnetosphere and its implications for spin - up and spin - down of the star , apj , 1977 , 217 , 578 - 596 .
strohmayer , t. e. , & bildsten , l. , new views of thermonuclear bursts , in _ compact stellar x - ray sources _ ( eds .
lewin , w. h. g. , and van der klis , m. ) .
cambridge university press , cambridge , 2006 , pp .
113 - 156 .
cornelisse , r. , heise , j. , kuulkers , e. , verbunt , f. , & int zand , j. j. m. , the longest thermonuclear x - ray burst ever observed ? . a bepposax wide field camera observation of 4u 1735 - 44 , a&a , 2000 , 357 , l21-l24 .
lewin , w. h. g. , van paradijs , j. , & taam , r. e. , x - ray bursts , in _ x - ray bimaries _ ( eds .
lewin , w. h. g. , van paradijs , j. , and van den heuvel , e. p. j. ) , cambridge university press , cambridge , 1995 , pp .
175 - 232 .
boirin , l. , mndez , m. , daz trigo , m. , parmar , a. n. , & kaastra , j. s. , a highly - ionized absorber in the x - ray binary 4u 1323 - 62 : a new explanation for the dipping phenomenon , a&a , 2005 , 436 , 195 - 208 .
boirin , l. , parmar , a. n. , barret , d. , paltani , s. , & grindlay , j. e. , discovery of x - ray absorption features from the dipping low - mass x - ray binary xb 1916 - 053 with xmm - newton , a&a , 2004 , 418 , 1061 - 1072 .
cackett , e. m. , miller , j. m. , bhattacharyya , s. , grindlay , j. e. , homan , j. , van der klis , m. , miller , m. c. , strohmayer , t. e. , & wijnands , r. , relativistic iron emission lines in neutron star low - mass x - ray binaries as probes of neutron star radii , apj , 2008 , 674 , 415 - 420 .
di salvo , t. , da , a. ; iaria , r. , et al . , a relativistically smeared spectrum in the neutron star x - ray binary 4u 1705 - 44 : looking at the inner accretion disc with x - ray spectroscopy , mnras , 2009 , in press .
homan , j. , van der klis , m. , wijnands , r. , et al . , rossi x - ray timing explorer observations of the first transient z source xte j1701 - 462 : shedding new light on mass accretion in luminous neutron star x - ray binaries , apj , 2007 , 656 , 420 - 430 .
bhattacharyya , s. , strohmayer , t. e. , miller , m. c. , & markwardt , c. b. , constraints on neutron star parameters from burst oscillation light curves of the accreting millisecond pulsar xte j1814 - 338 , apj , 2005 , 619 , 483 - 491 .
barret , d. , olive , j .- f . , & miller , m. c. , drop of coherence of the lower kilo - hz qpo in neutron stars : is there a link with the innermost stable circular orbit ?
, astronomische nachrichten , 2005 , 326 , 808 - 811 .
altamirano , d. , van der klis , m. , mndez , m. , wijnands , r. , markwardt , c. , & swank , j. , discovery of kilohertz quasi - periodic oscillations and state transitions in the low - mass x - ray binary 1e 1724 - 3045 ( terzan 2 ) , apj , 2008b , 687 , 488 - 504 .
altamirano , d. , casella , p. , patruno , a. , wijnands , r. , & van der klis , m. , intermittent millisecond x - ray pulsations from the neutron star x - ray transient sax j1748.9 - 2021 in the globular cluster ngc 6440 , apj , 2008a , 674 , l45-l48 . | a neutron star low - mass x - ray binary is a binary stellar system with a neutron star and a low - mass companion star rotating around each other . in this system
the neutron star accretes mass from the companion , and as this matter falls into the deep potential well of the neutron star , the gravitational potential energy is released primarily in the x - ray wavelengths .
such a source was first discovered in x - rays in 1962 , and this discovery formally gave birth to the x - ray astronomy " . in the subsequent decades
, our knowledge of these sources has increased enormously by the observations with several x - ray space missions . here
we give a brief overview of our current understanding of the x - ray observational aspects of these systems . |
magnetic tunnel junctions ( mtj s ) consisting of two ferromagnetic layers separated by the insulating spacer exhibit the tunneling magnetoresistance effect ( tmr ) when they are switched by applying a magnetic field from the antiparallel to parallel alignment of magnetizations in magnetic layers .
the tmr effect was first observed by m. julliere in 1975 @xcite .
julliere found that the tunneling conductance of the trilayer structure fe / ge / co depended on the angle between magnetizations in the fe and co layers .
the measured amplitude of the tmr in these experiments was 14% at 4.2k .
only 20 years later , the large values of tmr at room temperature were obtained in magnetic junctions comprising the al@xmath2o@xmath3 barrier @xcite . since that time
, there has been a renewed technological and fundamental interest to the tunneling phenomenon and during the last decade a lot of experimental and theoretical papers were published on this topic ( see reviews [ 5,6,7,8 ] ) .
experimentally it was observed @xcite that the tmr depends critically on the material of the insulating barrier and on the conditions of its preparation , in particular on the imperfections of the interfaces between the metal and the insulating layer @xcite . on the other hand ,
the first theory suggested by m. julliere expresses the tmr ratio in terms of the effective spin polarizations @xmath4 and @xmath5 of two magnetic layers via the expression @xmath6 , and thus predicts no dependence of the tmr on the parameters of the barrier . in spite of julliere s formula
is widely used for the interpretation of the experimental data @xcite , it does not have rigorous theoretical foundation @xcite .
the dependence of the tmr magnitude on the parameters of the tunnel barrier and the metal / insulator interface was taken into account in the subsequent theories .
slonczewski @xcite considered a quantum mechanical problem of tunneling of a free electron through a simple spatially uniform barrier and showed that the tmr ratio depends on the height of the potential barrier and on the effective mass of the tunneling electron inside the insulator .
later on , the influence of spin - flip scattering at the interfaces on temperature and bias - voltage dependences of the tmr was investigated by zhang _
et al . _
it was shown that mixing of spin - up and spin - down tunnel channels leads to a decrease of the tmr . due to a permanent progress in the development of different methods of ab - inito calculations of the electronic properties of solids , in the last four years
the transport properties of tunnel junctions were investigated using the realistic band structure of the ferromagnetic layers and the insulator ( refs .
[ 1420 ] ) .
the systems where the conductance was calculated were ideal epitaxial fe / znse / fe(001 ) @xcite and fe / mgo / fe(001 ) tunnel junctions @xcite .
for these structures the electron transport is assumed to be ballistic , i.e. the electron momentum @xmath7 parallel to the ferromagnet / insulator interface is conserved .
experimentally it was also demonstrated recently by heinrich _
et al . _
@xcite that fe / mgo / fe(001 ) junctions can be indeed grown by depositing the mgo epitaxially onto a fe whisker and then depositing another fe electrode epitaxially on a top of the mgo .
the results of ab - initio calculations showed that tunneling has more complicated behavior than the predictions of the simple barrier model proposed by slonczewski @xcite .
the main conclusions are as follows @xcite .
( i ) tunneling conductance depends strongly on the symmetry of the electron states in the ferromagnetic electrodes and in the insulating layer .
( ii ) the decay rates of evanescent states in the barrier are different for the states with different symmetry .
the slowest decay rates have the evanescent states which are compatible with @xmath0 symmetry .
the bloch states in the metal couple more efficiently through the interface with the decaying states of the same symmetry in the barrier .
therefore , mostly free - like @xmath8-electrons from the bands with @xmath0 character in the ferromagnet give the essential contribution to the tunneling conductance .
( iii ) for thin insulating layers the tunneling current in the minority channel is dominated by the interfacial resonance states that gives rise to `` hot spots '' for the @xmath7-resolved conductance in the surface brillouin zone .
the contribution from the resonance states is substantially suppressed for the thicker barriers since the bloch states at `` hot spots '' have no @xmath0 character @xcite . moreover , as it follows from recent discussions @xcite , the surface resonance states are very sensitive to the details of the interface .
the asymmetry of the potential barrier and the interfacial roughness considerably reduce the contribution from the surface states to a total conductance . nowadays the ab - initio calculations of the tmr are possible only for ballistic junctions like above mentioned fe / znse / fe and fe / mgo / fe structures which are characterized by a small lattice mismatch between the metal and the insulator , and by the well - defined band structure of the oxide or the semiconductor . in realistic junctions
the electron transport has a diffuse character , i.e. when the electron crosses the ferromagnet / insulator interface its @xmath7 momentum is not conserved . for f / o / f tunnel junctions the most successful material until now has been alumina , al@xmath2o@xmath3@xcite .
the al@xmath2o@xmath3 tunnel barriers are usually fabricated by the natural or plasma oxidation of the al layer @xcite . the subsequent structural analysis , e.g. with the use of the x - ray photoelectron spectroscopy , shows that alumina is amorphous and the obtained alo@xmath9 tunnel barrier deviates from the ideal al@xmath2o@xmath3 structure @xcite . for a uniform coverage the al film thickness
is usually ranged from about 7 to 18 , depending on a type of the ferromagnetic electrode @xcite .
there is a small range of al thicknesses that yields to the best tmr ratio for a given oxidation condition . when the al layer is thin , the ferromagnet surface becomes oxidized leading to the formation of coo and co@xmath3o@xmath10 oxides @xcite or to the fe@xmath3o@xmath10 oxide @xcite . on the other hand , with too thick al film
, an excess of al metal is left unoxidized .
the amorphous barriers , the roughness of the interface and its structural inhomogeneity make the rigorous ab - initio calculations of the tmr virtually impossible and therefore more simplified models are required to treat the diffuse electron transport in mtj s . in our previous paper @xcite
, we attempted to investigate the influence of scattering processes of the electrons ( with and without spin - flip ) at the interfaces on the tmr using a simple two - band ( spin - up and spin - down ) free - electron model .
it was shown that for this simplified model spin - conserving scattering may or may not lead to a decrease of the tmr depending on the amplitude of the scattering potential . in this paper , we proceed to study the diffuse electron transport in magnetic tunnel junctions of the from f / o / f with f s being 3@xmath1 transition metal electrodes and o being the insulating barrier ( al@xmath2o@xmath3 ) .
we use the results of ab - initio calculations to model the band structure of ferromagnetic electrodes , namely , the most important feature of these calculations that at least two groups of electrons form the total band structure : the almost free - like spin - up and spin - down bands and the narrow strongly exchange split bands .
we will call the first group @xmath0 and the second group @xmath1 electrons and will consider that a periodic part of the @xmath0-@xmath1 hybridization between bands is taken into account that results in the nonequivalence of spin - up and spin - down @xmath0-bands .
the parameters of the adopted model can be adjusted to reproduce a value of the tmr observed in the experiments .
we assume that for the case of al@xmath2o@xmath3 barrier the exchange split @xmath0-like quasi - free electrons give the main contribution to the tmr effect . in the framework of this simplified model
we will be able to investigate in a proper way the influence of electron scattering at the interfaces on the tunneling conductance . to describe the nonideal tunnel junction we assume that defects and impurity centers ( the al or o ions , or other artificially embedded ions ) are randomly distributed within few monolayers near the f / o interface .
within the interface , an electron undergoes scattering when it comes to the defect or impurity center .
we take into account these processes assuming that the hybridization between @xmath0 and @xmath1 bands changes randomly on the interface because of this parameter is the most important in the adopted model among the other ones , characterizing the scattering potential .
it yields to the possibility of @xmath0-like electrons to scatter into the @xmath1 sub - band ( and vice versa ) and thus strongly affects tunneling . to treat the electron scattering on the random potential we use the coherent - potential approximation ( cpa ) @xcite that allows to consider the case of strong scattering .
we apply the generalized kubo formalism and the green s functions method to calculate the conductance of the system , and find the vertex corrections to the conductivity with the use of `` ladder '' approximation @xcite combined with the cpa .
it is shown that the so - called ward identity is satisfied in the framework of this approximation that provides the necessary condition for a conservation of a tunneling current .
note , that if it is not the case , the conclusions may be completely misleading . as a result ,
we show that , due to substantial difference in majority and minority @xmath1 density of states at the fermi energy for 3@xmath1 ferromagnetic metals , the inter - band @xmath0-@xmath1 scattering on the interface can strongly reduce the tmr even down to zero value . in accordance with results of zhang _ et al .
@xcite , it is also shown that the spin - flip scattering of electrons on the surface magnons within the interface leads to a further decrease of the tmr ratio at finite temperature .
the paper is organized as follows . in sec .
ii we describe the model hamiltonian , the calculation of the tunneling conductance and vertex corrections .
the discussion of the obtained results is presented in sec .
conclusions are in sec .
the proof of ward identity is given in appendix a. the details of the derivation of the cpa equations are described in appendix b.
we will consider a trilayer tunnel junction of the form f@xmath11/o
/ f@xmath3 , where f@xmath11 and f@xmath3 are two semi - infinite ferromagnetic layers and o is a dielectric oxide spacer ( al@xmath2o@xmath3 ) .
our arguments in behalf of the two - band @xmath0-@xmath1 model which was briefly described in the introduction are as follows .
one of the conclusions of ab - initio calculations of the tmr for ballistic tunnel junctions @xcite is that the expected spin - dependence of the tunneling current can be deduced from the symmetry of the bloch states in the ferromagnet at the fermi energy .
spin - polarized band structure for bcc fe , fcc ni , and fcc co can be found in ref . [ 26 ] .
the type and symmetry of the bloch states for different crystal faces with @xmath12 for fe , co and ni are presented in table i ( in accordance with ref . [ 18 ] ) .
for example , in case of fe electrodes , the examination of the band structure shows that both the majority and minority bands with @xmath0 character in [ 110 ] and [ 111 ] directions ( @xmath13 and @xmath14 ) cross the fermi energy . for [ 100 ]
direction the band with @xmath0 symmetry ( @xmath15 ) crosses the fermi energy in case of the majority channel only .
the similar analysis can also be performed in case of ni and co. thus one can assume that in the polycrystalline fe , co or ni - based films the states with @xmath0 character present for both spin directions and the bloch states will couple efficiently through the f / o interface with @xmath0 states in the insulator and will decay with the equal rates in the barrier region .
we will call the electrons from these bands as @xmath0-like electrons and will describe them as free electrons with the effective mass @xmath16 ( where @xmath17 is a bare electron mass ) and with different fermi momenta @xmath18 and @xmath19 ( @xmath20 ) for up and down spins . the idea about a dominant contribution of the mostly itinerant electrons to tunneling
was originally proposed by m. stearns @xcite and explained the positive polarization of the spin - dependent current in the experiments on tunneling with the superconductors @xcite . according to estimations of stearns @xcite , in the case of fe , @xmath21 @xmath22 , @xmath23 @xmath22 . .
type and symmetry of the bloch states with * k*@xmath24 for fe , co and ni for three different crystal faces ( in accordance with ref .
the symmetry of these bands is as follows : @xmath15 , @xmath14 , and @xmath13 ( @xmath0,@xmath25,@xmath1 ) ; @xmath26 , and @xmath27 ( @xmath25 and @xmath1 ) ; and @xmath28 , @xmath29 , @xmath30 , and @xmath31 ( @xmath1 ) . [
cols="^,^,^,^",options="header " , ] other more localized bands ( compatible with @xmath1 symmetry ) crossing the fermi energy also will be described by two exchange split bands with the isotropic quadratic dispersion law but with larger effective mass @xmath32 .
the fermi momenta @xmath33 and @xmath34 of @xmath1-like electrons can be chosen to reproduce the typical for 3@xmath1 transition metals ratio of the values of spin - up and spin - down @xmath1 density of states at the fermi energy @xmath35 , @xmath36 @xcite . in accordance with the band structure of fe , co and ni @xcite the narrow majority and minority
@xmath1-bands are practically filled .
therefore , the particles with a large effective mass , @xmath37 , must be regarded as holes .
the values of the fermi momenta define the positions of band bottoms ( fig .
1 ) @xmath38 , @xmath39 ( @xmath40 ; @xmath41 ) with respect to the fermi energy @xmath35 .
we note , that the aim of this work is to calculate the relative change of the tmr due to scattering .
therefore the proposed model is rather adequate for this purpose since the scattering rate depends mostly on the density of states and on the matrix elements of the scattering potential .
the al@xmath2o@xmath3 tunnel barriers obtained by oxidation of the al film are amorphous @xcite . concerning the @xmath42-al@xmath2o@xmath3 crystals , it is known from the band structure calculations @xcite that the gap ( which is not direct ) between the upper valence band and the conduction band is of the width of @xmath43 ev .
the dispersion law of the lowest conduction band is not isotropic , and the effective electron masses along the different directions in the brillouin zone vary from @xmath44 to @xmath45 with an average value of about @xmath46 @xcite .
this lowest conduction band is formed by a mixture of al 3@xmath0 , o 2@xmath0 and o 3@xmath0 orbitals . in the top of the upper valence band
the dispersion curves are very flat , i.e. the effective masses of holes are large as compared with the mass of the conduction electrons . at the @xmath47 point ,
the top of the valence band consists of hybridized o 2@xmath25 and al 3@xmath25 orbitals . in a view of that
, one can describe the conduction and valence bands of the amorphous al@xmath2o@xmath3 by isotropic quadratic laws with effective masses
@xmath48 and @xmath49 , @xmath50 . next
, the @xmath0-like electrons from the ferromagnet can easily penetrate into the oxide since the lead bands with @xmath0 character can couple efficiently with @xmath0 states in the barrier , and the tunneling conductance , caused by the specular transmission of the bloch waves through the interface , will decay for both spin - up and spin - down @xmath0-channels with the same law as @xmath51 $ ] , where @xmath52 is the width of a barrier , @xmath53 is a bottom of the conduction band and @xmath54 is the electron momentum parallel to the f / o interface .
following maclaren _
et al._@xcite we can suppose that @xmath1-like electrons from those bands in the ferromagnet without @xmath0 character can not couple efficiently with @xmath0 states in the oxide . nevertheless , these @xmath1-bands have an admixture of @xmath25 symmetry and therefore can couple with the valence bands of the al@xmath2o@xmath3 .
however , in the @xmath1 channel the tunneling conductance due to specular transmission will decay very rapidly with the thickness @xmath52 of the barrier , as@xmath55 $ ] , because of a large mass @xmath49 of holes in the alumina ( here @xmath56 is the top of the upper valence band ) .
the model potentials describing the propagation of the electron through a tunnel junction are presented in fig . 1 .
we suppose that the ferromagnet / insulator interfaces are rough and contaminated by impurity ions .
the contamination of a few metal monolayers close to the f / o interface always presents after the oxidation of the al film .
other impurities ( e.g. cr , ru @xcite or cu ions @xcite ) can be artificially inserted into the interface .
we will characterize these structural defects by the random potentials , which may be divided into the spin - conserving and the spin - flip parts .
let @xmath57 and @xmath58 be the positions of interfaces , the @xmath59-plane be the plane parallel to the interface , and the @xmath60-axis be the axis perpendicular to the barrier ( see fig . 1 ) .
the hamiltonian of the system is written as : @xmath61 with @xmath62 \psi^s_{\mu}({\bf r})\phantom { , } \nonumber \\ { } & \displaystyle + \sum_{\mu}\int d{\bf r}\ , \psi^{d+}_{\mu}({\bf r } ) \left [ -\frac{\hbar^2}{2m_d(z)}\nabla^2 + u^d_{\mu}(z ) \right ] \psi^d_{\mu}({\bf r } ) , \nonumber\end{aligned}\ ] ] @xmath63 , \nonumber\end{aligned}\ ] ] @xmath64 .
\nonumber\end{aligned}\ ] ] here @xmath65 is a spin suffix ; @xmath66 , @xmath67 and @xmath68 , @xmath69 are field operators of the creation and annihilation of @xmath0 and @xmath1-type electrons with the spin @xmath70 at the point @xmath71 ; @xmath72 and @xmath73 are the effective masses of electrons and holes in the ferromagnetic layers ( @xmath74 and @xmath37 ) or in the barrier ( @xmath48 and @xmath49 ) .
@xmath75 , @xmath76 are the spin - dependent step - like potentials shown in fig . 1 : @xmath77 @xmath78 where @xmath38 , @xmath39 ( @xmath41 ) are the majority ( @xmath79 ) and minority ( @xmath80 ) band bottoms in the ferromagnet , @xmath53 and @xmath56 are the bottom of the conduction band and the top of the upper valence band in the insulator .
the positions of `` steps '' for the potential @xmath76 are shifted by a value @xmath81 with respect to the points @xmath82 and @xmath83 in order to describe the finite thickness ( @xmath84 ) of the interface .
the explanation , why it is done in this way , is given below in the text .
@xmath85 is the spin - conserving part of the hamiltonian , @xmath86 are the interface numbers . to describe the defect structure of the nonideal f / o interface we consider that the impurity ions and the ions of a ferromagnetic metal constitute the more or less random alloy of a type a@xmath9b@xmath87 where a denotes the ions of the ferromagnet ( fe , co or ni ) and b denotes the impurities .
therefore , we suppose that each site @xmath88 on the interface @xmath42 is characterized by the random parameter @xmath89 of @xmath0-@xmath1 hybridization taking two different values @xmath90 and @xmath91 with the probabilities @xmath92 and @xmath93 , respectively . we also suppose that a periodic part of @xmath0-@xmath1 hybridization is taken into account leading to the nonequivalence of majority and minority @xmath0-bands in the ferromagnet . for the simple two - band tight - binding model of the binary alloys @xcite
one assumes that : 1 ) parameters @xmath94 and @xmath95 describing the positions of @xmath0 and @xmath1 bands on the energy scale are different for the alloy s components ; 2 ) also the parameter @xmath96 of the hybridization between @xmath0 and @xmath1 bands depends on the type of an ion ( a or b ) . in our particular case , one can not take into account the former effect since the well - defined two types of the electrons will as before exist in the vicinity of the interface and the adopted model of tunneling will not change significantly .
however , the random hybridization makes possible the processes of scattering of quasi - free @xmath0-electrons into the localized @xmath1 sub - band and vice versa and , therefore , can strongly influence on tunneling .
@xmath97 is a part of the hamiltonian describing the spin - flip scattering .
we take into account only the spin - flip processes for @xmath0-like electrons since these electrons are itinerant and give the most essential contribution to the tunneling current .
operators @xmath98 , @xmath99 are defined as @xmath100 , \nonumber \\
\hat{s}_{-}(\rho_n^{\alpha } ) = \frac{1}{\sqrt{2sn } } \sum_{\bf q}e^{iq\rho_n^{\alpha } } [ b^{+}_{\bf q } + \dots\ ] .
\nonumber\end{aligned}\ ] ] here @xmath101 and @xmath102 denote the creation and annihilation operators of the surface magnons , @xmath103 is a number of the lattice sites on the interface , @xmath104 is a spin value .
we used the well - known representation of the spin operators in terms of @xmath102 and @xmath101 and left the first terms of the series .
@xmath105 is a random exchange integral which also takes the values @xmath106 and @xmath107 with probabilities @xmath92 and @xmath93 .
let us now turn back to the step - like potential for @xmath1-holes .
as it will be clear from the following consideration , the amplitude of the effective scattering potential on the interface ( @xmath82 ) for @xmath0-like electrons ( i.e. @xmath108 , where @xmath109 is the self - energy ) is determined by the value @xmath110 , here @xmath111 is the imaginary part of the @xmath1-electron green s function at the point @xmath112 ( density of states ) . if one put @xmath113 in eq .
( [ u_d ] ) , then the vertex contribution to the conductivity ( that is the contribution due to tunneling assisted by interfacial scattering ) will be rather small as compared with the contribution due to direct tunneling .
this result is not enough accurate and is the sequence of a continual type of the model when one neglects the existence of the atomic lattice and for this reason the self - energy has a @xmath114like behavior on the interface .
however , the width @xmath81 of the interfacial layer is about the distance between atomic planes or even larger . for the case of bcc lattice , @xmath115 for [ 100 ]
direction ( here @xmath116 is a lattice constant ) .
one can show that for the present model with @xmath113 the imaginary part @xmath117 of the unperturbed @xmath1-electron green s function ( with @xmath118 ) , i.e. density of states of @xmath1-electrons ( holes ) , has small value at the point of interface @xmath112 and increases inside the f - layer up to the distance of the order of @xmath119 , and then it oscillates near the average value which is approximately ten times larger than @xmath120 .
the period of oscillations is determined by @xmath33 or @xmath34 , depending on the electron spin and on the orientation of magnetization in the f - layer .
such a behavior is easily understood , if to take into account that @xmath1-electrons ( holes ) are almost completely reflected on the metal / insulator interface .
thus , if one shifts the positions of `` steps '' for @xmath76 with respect to @xmath75 as it is given by ( [ u_d ] ) , one can expect the more effective mechanism of scattering due to the larger value of @xmath121 , and therefore the model becomes more realistic . to calculate the non - local conductivity
we apply the kubo formula of the linear response @xcite ( it is valid under the small applied voltage which is much less than the value of the energy gap in the insulator ) : @xmath122 where a matrix operator @xmath123 is defined as @xmath124 and @xmath125 is the asymmetric gradient operator , @xmath126 and @xmath127 are the retarded and advanced @xmath128matrix green s functions ( with components @xmath129 , @xmath130 , @xmath131 and @xmath132 ) , @xmath133 are the spin suffixes , brackets @xmath134 denote the averaging over the configurations and magnon degrees of freedom , the trace ( sp ) goes over @xmath0 and @xmath1 indices of the bands .
below , for convenience , it is assumed that @xmath135 .
we will recall about @xmath136 in the final expressions for the conductance . to calculate the conductivity ( [ cond ] ) of the system one has to find the green s function of the hamiltonian ( [ hamilt ] ) , which can be found by solving the following system of differential equations in the mixed @xmath137 representation @xcite : @xmath138g_{\mu\kappa}^{\gamma\beta}(z , z ' ) = \delta_{\alpha\beta}\delta(z - z ' ) , \label{green_eq}\ ] ] @xmath139 where @xmath54 is a projection of the electron momentum on the @xmath59plane ( parallel to the f / o interface ) , the greek indices @xmath42 , @xmath140 and @xmath141 assume values @xmath0 and @xmath1
, @xmath70 denotes the electron spin .
@xmath142 is the @xmath128matrix linear differential operator where @xmath143 and @xmath144 ( @xmath145 ; @xmath65 ) denote the coherent potentials for spin - up and spin - down electrons which take into account the scattering of the electron by random spin - conserving and spin - flip potentials on the interfaces .
they were found with the use of the coherent potential approximation ( cpa ) @xcite , the details of these calculations are presented in the subsequent section iic and in appendix b. the operator @xmath146 represents the effective single - particle hamiltonian of the system which , however , is non - hermitian since coherent potentials are imaginary quantities . in order to solve eq .
( [ green_eq ] ) for the green s functions , we will follow the procedure described below .
first , let s solve the shrdinger equation with the hamiltonian @xmath146 : @xmath147 \psi_{\beta}(z ) = 0 .
\label{schrod}\ ] ] the solutions of this equation can be easily found since the potentials @xmath148 have a step - like form .
let us put @xmath149 , where @xmath35 is the fermi energy , and introduce the following notations : @xmath150 @xmath151 are fermi momenta in f@xmath11 and f@xmath3 ferromagnetic layers for electrons with the spin @xmath70 , and @xmath152 let also @xmath153 and @xmath154 be the components of electron momentum with spin @xmath70 along @xmath60-axis in f@xmath11 and f@xmath3 layers , respectively ( @xmath155 is the in - plane component of the momentum , @xmath42 is a band index ) , and let @xmath156 be the imaginary electron momentum in the insulating layer .
further , for convenience , we will omit indices @xmath70 and @xmath155 in the notation of some functions .
equation ( [ schrod ] ) has four linear - independent solutions which we denote as @xmath157 and @xmath158 we choose these independent solutions so that two functions @xmath159 and @xmath160 would describe two waves corresponding to the @xmath0-like electrons , and functions @xmath161 and @xmath162 would correspond to the @xmath1-like electrons .
namely , in a layer f@xmath11 @xmath163 the solutions @xmath164 have the form @xmath165 \\ 0 \end{array } \right ) \quad z < a , \label{phi}\ ] ] @xmath166 \end{array } \right ) \quad z < a,\ ] ] and in a layer f@xmath3 @xmath167 the solutions @xmath168 are @xmath169 \\ 0 \end{array } \right ) \quad z > b , \label{psi}\ ] ] @xmath170 \end{array } \right ) \quad z > b.\ ] ] since @xmath149 , then @xmath171 ( @xmath172 ; @xmath173 ) .
thus these solutions satisfy the following boundary conditions : @xmath174 starting from expressions ( [ phi ] ) and ( [ psi ] ) , the solutions @xmath164 , @xmath168 can be easily extended in two other layers .
let us introduce the matrices @xmath175 @xmath176 the wronskian of the system ( [ schrod ] ) is @xmath177 where the matrix operator @xmath178 is defined similar to eq .
( [ d_r ] ) , and the subscript @xmath179 denotes the transposition operation .
it is known from the theory of differential equations that the matrix @xmath180 is a constant since it satisfies the equation @xmath181 .
taking into account that @xmath149 and @xmath182 , the solution of eq .
( [ green_eq ] ) for the retarded green s function can be written in the matrix form as @xmath183^{-1}\psi^t(z ' ) , \quad \mbox{if}\quad z < z ' , \label{green_r } \\
g^r(z , z ' ) & = & \psi(z)\delta^{-1}\phi^t(z ' ) , \qquad \mbox{if}\quad z > z ' . \nonumber\end{aligned}\ ] ] to find the advanced green s function , one has to put @xmath184 in eq .
( [ green_eq ] ) and assume that @xmath185 .
then we obtain @xmath186^{*}. \label{green_a}\ ] ] in the expressions ( [ green_r ] ) and ( [ green_a ] ) the green s functions depend on the in - plane momentum @xmath155 and on the spin @xmath70 of the electron because solutions @xmath168 and @xmath164 also depend on @xmath70 and @xmath155 .
next , one has to find the two - point conductivity ( [ cond ] ) of the system using the green s functions ( [ green_r ] ) and ( [ green_a ] ) . for that
, we introduce the current matrices @xmath187 and @xmath188 ( @xmath65 ) constructed with the use of solutions @xmath168 and @xmath164 , respectively : @xmath189 the total conductance of the system @xmath190 may be presented in the usual form as a sum of the `` bubble '' part and the vertex corrections ( see fig .
2 ) @xcite : @xmath191 the `` bubble '' contribution to the conductance describes direct electron tunneling from the electrode to another one through the barrier when electron momentum parallel to the f / o interface @xmath54 is conserved ( the specular transmission ) .
the scattering on the interfaces affects direct tunneling so that the effective height of the potential barrier increases because the electron also has to pass through @xmath114-like potentials at the points @xmath112 and @xmath192 formed by the self - energies @xmath193 and @xmath194 . the vertex corrections to the conductance describe the tunneling assisted by interfacial roughness that is the processes when the electron with momentum @xmath155 comes to the impurity center on the interface , undergoes scattering ( @xmath195 ) in another channel with @xmath196 , and then goes away to the electrode or to the barrier .
it also can be shown that the contribution to the tunneling conductance is negligibly small ( @xmath197 ) from the diagram containing both vertex parts @xmath198 and @xmath199 compared with other contributions that are of the order of @xmath200 . the diagram with vertices @xmath198 and
@xmath199 corresponds to the interference of waves scattered from both interfaces .
therefore , we can neglect this interference term . as a sum of `` bubble '' and `` vertex '' parts . here
the full lines correspond to the green s functions @xmath201 and @xmath202 , and wavy lines denote the asymmetric gradient operator @xmath203 of velocity at the points @xmath60 and @xmath204 .
the shaded square designates the vertex part @xmath205 at the interface ; _ h.c .
_ denotes the complex conjugate terms . ]
substituting obtained expressions ( [ green_r ] ) , ( [ green_a ] ) for the green s functions and using the definition ( [ j_current ] ) for the current matrices , we come to the following results . the analytical expression for
the `` bubble '' conductance is given by @xmath206^t\frac{1}{\delta_{\mu}^{\dagger } } j_{\mu}^{\psi}(z')\right\ } , \label{s_0}\ ] ] here it is assumed that @xmath207 , @xmath208 is the wronskian ( [ wronsc ] ) , @xmath209 denotes the junction area . for @xmath210 , @xmath211
, the vertex corrections from the left and from the right interfaces can be written as @xmath212 where summation is also performed over repeating indices @xmath213 . here
@xmath214 and @xmath215 are the vertex parts on the interfaces @xmath216 and @xmath217 .
@xmath218 are @xmath128matrices , their components are defined by @xmath219^{\beta_1\beta_2 } \right\ } , \label{lambda } \\
\lambda^{\gamma_1\gamma_2}_{\mu\kappa}(a , z ' ) = { \rm sp } \left\ { \frac{1}{\delta_{\mu } } \left[\rho^{\varphi t}_{\mu}(a)\right]^{\gamma_1\gamma_2 } \frac{1}{\delta_{\mu}^{\dagger } } j^{\psi}(z ' ) \right\ } , \nonumber\end{aligned}\ ] ] where @xmath220^{\beta_1\beta_2}$ ] and @xmath221^{\gamma_1\gamma_2}$ ] are density matrices with the components : @xmath222^{\beta_1\beta_2}_{ik } & = & \psi^{\beta_1*}_i(a)\psi^{\beta_2}_k(a)\quad ( i , k = s , d ) , \label{rho } \\
\left[\rho^{\varphi}_{\mu}(a)\right]^{\gamma_1\gamma_2}_{ik } & = & \varphi^{\gamma_1*}_i(a)\varphi^{\gamma_2}_k(a)\quad ( i , k = s , d ) .
\nonumber\end{aligned}\ ] ] the expressions similar to expressions ( [ lambda ] ) may be also written for the matrices @xmath223 , @xmath224 and for other position of the points @xmath60 , @xmath204 with respect to @xmath216 and @xmath217 .
the vertex parts @xmath225 and @xmath226 are calculated in the `` ladder '' approximation @xcite .
the derivation of the equation for @xmath205 is presented below in the subsection iid .
we have to note , that coherent potentials @xmath193 , @xmath194 calculated in the framework of the cpa and vertices @xmath225 , @xmath226 calculated in the `` ladder '' approximation satisfy the so - called ward identity which in our case , for example , for the interface @xmath82 can be written as follows ( for the details , see appendix a ) : @xmath227 here the summation is also performed over repeating indices @xmath228 and @xmath229 .
the fulfillment of ( [ ward ] ) provides the necessary condition of the nondivergence of the current through the tunnel junction : @xmath230 = 0 . \label{current0}\ ] ] according to eq .
( [ current0 ] ) , the total conductance of the system is a constant value . in view of that
, we will derive the exact expression for @xmath190 evaluating the conductance at points @xmath231 and @xmath232 , i.e. at the left and at the right sides from the interface .
the wronskian matrix @xmath208 and the current matrices @xmath233 and @xmath234 are expressed in terms of matrices @xmath235 and @xmath236 these matrices , as follows from eq .
( [ green_r ] ) , determine the green s function .
the straightforward evaluation of the conductance according to formulae ( [ s_0][lambda ] ) leads to the result that @xmath237 is expressed in terms of the retarded green s functions at the points of interfaces @xmath238 and @xmath239 constructed according to eq .
( [ green_r ] ) .
the explicit form of @xmath240 in the @xmath137 representation is given by the expression : @xmath241 - \sigma_{\mu}^{dd}(a ) \end{array}\right)^{-1 } , \label{g_int}\ ] ] where @xmath242 the similar expression can be written for the green s function @xmath243 on the right interface .
for that , one has to make the substitutions @xmath244 , @xmath245 , @xmath246 in the expression for @xmath240 . here
@xmath247 ( @xmath173 ) and @xmath248 are the effective electron ( hole ) masses in the ferromagnetic and insulating layers ; @xmath249 and @xmath250 ( @xmath251 , @xmath173 ) are the functions on @xmath155 introduced above in the text after eq .
( [ schrod ] ) .
let us define the `` transport '' density of states as follows : @xmath252 where @xmath253 expression ( [ s_0 ] ) for the `` bubble '' conductance then reads @xmath254 where @xmath255 @xmath256 @xmath257 and @xmath258 is the `` width '' of @xmath1barrier . for the vertex correction from the left interface ( @xmath82 ) we obtain @xmath259 in the similar way
, the vertex correction from the right interface ( @xmath83 ) reads @xmath260 expressions ( [ s_0_ab][s_vertex_b ] ) determine the total conductance of the system .
the scattering of the electron by random potentials within the interface ( terms @xmath261 and @xmath262 in the hamiltonian ) is taken into account by introducing the self - energy operators @xmath143 and @xmath144 into the effective single - particle hamiltionian ( [ h_eff ] ) . to calculate the self - energies we apply the coherent - potential approximation ( cpa ) @xcite .
let s denote kets
latexmath:[$|\gamma,\mu,\rho_n^{\alpha}\rangle = the wannier states of the electron on the interface @xmath42 ( @xmath264 ) at the given site @xmath88 in the @xmath265-plane , where @xmath141 refers to the state @xmath0 or @xmath1 and @xmath70 is the spin .
the symbol @xmath266 denotes the direct product .
the problem of finding of the single - particle green s function @xmath267 of the many - body hamiltonian ( [ hamilt ] ) is reduced to the related single electron problem of the propagation of the electron in a random interfacial potential @xmath268 where @xmath269 and the summation goes over the interface number @xmath42 and over the sites @xmath270 . here
@xmath271 is the random potential of @xmath272 hybridization associated with the site @xmath270 and @xmath273 is the exchange - like interaction with the surface magnons .
the random quantities @xmath89 and @xmath274 used here were introduced in sec .
now one can formulate the cpa by the ordinary way and the only difference with respect to the usual situation of the bulk scattering is that the initial green s functions have to be calculated for the trilayer system .
we assume that the averaged propagator of the system @xmath275 differs from the initial green s function , corresponding to the hamiltonian @xmath276 [ see eq .
( [ hamilt ] ) ] , by the self - energy correction in eqs .
( [ green_eq ] ) , ( [ h_eff ] ) .
this means that the system behaves as if coherent potentials @xmath277 had been assigned to each site of the interface @xmath216 and @xmath217 . after the introducing of the effective medium @xmath278 ,
each site @xmath279 becomes a source of the random potential @xmath280 .
the single - site @xmath281-matrix associated with potential @xmath282 is given by @xmath283 where @xmath284 or @xmath217 , @xmath285 is the averaged green s function at the interface @xmath286 which is expressed via @xmath278 in accordance with eq .
( [ g_int ] ) .
the upper limit @xmath287 is a cut - off of the in - plane momentum which originates from the finite size of the brillouin zone .
for that , one has to substitute the brillouin zone s projection onto @xmath288-plane by a circle of the same square with a radius @xmath287 . for the bcc lattice , @xmath289 , where @xmath116 is a lattice constant .
the single - site @xmath281-matrix ( [ t_matrix ] ) is obviously different for different sites . at the same time
it is supposed that scattering by the random potential is taken into account in the averaged propagator @xmath290 by the self - energy operator @xmath291 .
therefore , we require that the ensemble average of the single - site @xmath281-matrix vanishes , i.e. , @xmath292 here @xmath293 and @xmath294 are the values of the single - site @xmath281-matrix associated with a given site @xmath270 which is occupied by the a - type ion ( ferromagnet s ion ) or by the b - type ion ( impurity ) , respectively .
brackets @xmath295 denote the averaging over magnon degrees of freedom .
the equation ( [ t_av ] ) is the well - known self - consistent coherent potential approximation ( cpa ) @xcite that implicitly determines the self - energy operator @xmath278 .
the cpa equations ( [ t_av ] ) are formulated for the particular case of electron scattering within the f / o interface .
the straightforward calculations show that the single - site @xmath281-matrix @xmath296 can be represented in the form : @xmath297 with respect to spin - up and spin - down subspaces . here
@xmath298 , @xmath299 , and the blocks @xmath300 , @xmath301 are @xmath302-matrices , functionally depending on @xmath303 and @xmath304 , with the components designated by indices of bands @xmath0 and @xmath1 . to satisfy the condition ( [ t_av ] )
, one has to consider only the spin - conserving part of this equation as long as @xmath305 since the expression to be averaged contains unequal number of the creation and annihilation operators of the magnons .
in order to calculate @xmath306 we adopted the further approximation and assumed that @xmath307 here @xmath308 is the average number of magnons at the given temperature .
in other words , we substituted the operators @xmath309 by its averaged values .
the function @xmath310 is given by the familiar expression @xmath311 where @xmath312 is the spectrum of surface magnons .
the approximation we made to take into account the spin - flip processes is the static approximation for magnons .
the inelasticity of the electron - magnon processes may be taken into account in the energy conservation rule .
however , the characteristic magnon energy @xmath313 is much less than the fermi energy @xmath35 , and in the first approximation one can neglect this energy @xmath313 it becomes important in the case of finite voltages when the process of emitting of a magnon by a hot electron influences on the form of @xmath314 dependence .
we , therefore , restricted our calculation to the case of small voltage bias with the voltage less than @xmath313 . within this approximation , for the system of cpa equations we get @xmath315 as far as matrices @xmath316 are @xmath128blocks , thus for the general case eq .
( [ cpa_system ] ) represents the system of two matrix equations for 8 unknown quantities @xmath317 .
the system ( [ cpa_system ] ) is one of the possible forms of the cpa equations .
but actually we used another representation of the cpa which was more convenient for the numerical implementation .
for that we exploited the augmented - space formalism by mookerjee @xcite , and the details are presented in appendix b. to find the vertex corrections we used the `` ladder '' approximation combined with the cpa , that was originally proposed by velick @xcite . the diagrammatic representation of the `` ladder '' approximation is given in fig . 3 .
since the scattering potentials on the different interfaces ( at the points @xmath216 and @xmath217 ) do not correlate with each other , the vertex parts @xmath318 can be found independently for each interface and we will omit interface suffix in the subsequent expressions .
let @xmath319 be the average of the product of two random @xmath281-matrices over configurations and magnon distribution , where as before @xmath320 are spin indices and @xmath321 are orbital indices . on the diagram in fig.3 ,
it is presented by the dashed line . the analytical expression for the correlator @xmath322
can be found in accordance with the adopted approximate scheme of averaging over the magnon degrees of freedom .
taking into account eq .
( [ t_form ] ) we obtain : @xmath323 where @xmath324 and @xmath325 are the components of the single - site @xmath281-matrix , eq .
( [ t_form ] ) .
we also define the operator @xmath326 \nonumber\\ & & -\ g_{\mu}^{*\beta_1\gamma_1}(z_{\alpha})g_{\mu}^{\beta_2\gamma_2}(z_{\alpha } ) \label{d}\end{aligned}\ ] ] denoting the propagator of the pair of electrons in the `` ladder '' equation at the interface @xmath327 .
its definition follows from the fact that in the diagram representation of the cpa the multiple scattering on the given site is assumed to be incorporated into the single - site @xmath281-matrix @xmath328 , corresponding to the single vertex of any diagram . due to that
, the subsequent sites in the `` ladder '' diagrammatic equation ( fig.3 ) must not reproduce each other .
therefore , the necessary correction is subtracted in eq .
( [ d ] ) .
the summation over @xmath155 goes up to @xmath287 similar to eq .
( [ g_kappa ] ) .
after that definitions the analytical equation for the vertex part reads as @xmath329 eq .
( [ gamma_eq ] ) is the ordinary system of linear equations .
it can be proved that the vertex parts found with the use of the `` ladder '' approximation and self - energies found with the use of the cpa satisfy the ward identity ( [ ward ] ) , and hence the total tunneling current is a constant value @xmath330 , i.e. it does not depend on @xmath60 .
the proof of the ward identity is presented in appendix a.
we considered the case of fe / al@xmath2o@xmath3/fe tunnel junction .
the following parameters were chosen to describe the system . according to estimations of stearns @xcite , for the itinerant @xmath0like electrons in fe we set @xmath21 @xmath22 , @xmath23 @xmath22 , and @xmath331 (
here @xmath17 is bare electron mass ) . for the more localized @xmath1electrons ( holes )
, we put @xmath332 .
the @xmath1 density of states at the fermi energy is larger for the minority spin band @xcite , therefore , @xmath333 , and we put @xmath334 @xmath22 , @xmath335 @xmath22 .
the values of the fermi momenta define the positions of band bottoms @xmath336 , @xmath337 ( @xmath173 , @xmath338 ) with respect to @xmath35 . for @xmath1-electrons
the fermi momenta were chosen such a way that the interface density of states @xmath339 ( where @xmath340 are the positions of interfaces ) comply with the following ratio @xmath341 , which is the typical situation for the case of @xmath342 transition metals ( see , for a example , the calculations of tsymbal and pettifor @xcite ) .
we also put the width of the barrier @xmath343 , and @xmath115 where @xmath344 is a lattice constant for bcc fe ( @xmath81 is a parameter describing the thickness of the interface ) .
the main features of the band structure of @xmath42-al@xmath2o@xmath3 crystals were briefly presented in sec .
iia . in a view of that discussion
, the following parameters of the model were taken to describe the amorphous al@xmath2o@xmath3 barrier : the effective masses of electrons and holes are @xmath345 , @xmath346 , the positions of the conduction band bottom ( @xmath53 ) and the top of a valence band ( @xmath56 ) are @xmath347 ev , i.e. the width of the band gap is @xmath348 ev , and @xmath35 is assumed to be a zero of energy . in order to illustrate the general formalism presented in the previous sections ,
let us first consider the case of only @xmath272 impurity scattering ( i.e. @xmath349 and there are no spin - flip processes ) when concentration of impurity ions on the interface is @xmath350 .
in this case all formulae have a simple analytical form . as it was mentioned previously in sec .
iia , in the two - band tight - binding model description of a binary alloy the random variables are diagonal matrix elements of the hamiltonian @xmath351 , @xmath352 and @xmath353 depending on whether the site @xmath270 is occupied by an a or b ion . to be more precise , one must consider the concentration dependence of these matrix elements .
it reflects the fact that their values are modified by the existence of a charge transfer in the alloy . since in the framework of our model we are focusing on the hybridization effects ( see sec .
iia for justification of this approach ) , we have @xmath354 , @xmath355 where @xmath356 and @xmath357 are hybridizations of the pure components .
let us assume for simplicity that in eq .
( [ operators ] ) @xmath358 . then for @xmath359 we get @xmath360 , and thus @xmath361 is the single parameter characterizing the amplitude of @xmath0-@xmath1 scattering on the interface .
the nonzero value of @xmath362 will lead to a slight modification of the @xmath0 and @xmath1-like eigenstates describing @xmath0 and @xmath1-like electrons in the vicinity of the interface , and thus will not affect the qualitative results presented below .
we will omit spin suffixes since we consider the spin - conserving scattering now . at @xmath359 only diagonal elements of the self - energy matrix @xmath363 in the @xmath130-space have non - zero values .
then the green s functions , e.g. at the point @xmath216 , become [ see eq .
( [ g_int ] ) ] @xmath364 \nonumber \\ & & -\
\sigma_{\mu}^{dd}(a ) \biggr\}^{-1}\frac{\kappa d\kappa}{2\pi } , \nonumber\end{aligned}\ ] ] and cpa equations are written as follows @xmath365 which must be solved self - consistently by means of converging iterative procedure .
as far as green functions are diagonal , one can denote @xmath366 and can introduce the similar notations for quantities @xmath179 and @xmath367 , defined in section iid .
then one gets @xmath368
@xmath369 @xmath370 @xmath371 since the mass of holes @xmath49 in the insulator is much larger than the electron mass @xmath48 , the exponential factor @xmath372 in formulae ( [ s_0_ab ] [ s_vertex_b ] ) for tunneling conductance is negligibly small as compared with one for @xmath0like electrons .
therefore , one may neglect the contribution from @xmath1holes to the tunneling current . then for the tunneling conductance we obtain @xmath373 where @xmath374 is the conductance corresponding to the `` bubble '' diagram ( here @xmath375 ) , @xmath376 \nonumber \\ & + & \frac{e^2}{2\pi\hbar}\left(\int_0^{\kappa_{max } } a^{ss}_{\kappa\mu}(a ) \frac{\kappa d\kappa}{2\pi}\right)\,\gamma^{(a)\mu\mu}_{\,\,ss } \nonumber \\ & \times & \left [ \int_0^{\kappa_{max } } { \left| g_{\kappa'\mu}^{ss}(a ) \right|}^2 { \left(\frac{q_2^s}{m^0_s}\right)}^2 a^{ss}_{\kappa'\mu}(b)e^{-2 q_2^s w}\frac{\kappa ' d\kappa'}{2\pi}\right ] \nonumber\end{aligned}\ ] ] is the `` vertex '' contribution to the conductance on the left interface , and @xmath377 \nonumber \\ & + & \frac{e^2}{2\pi\hbar}\left(\int_0^{\kappa_{max } } a^{ss}_{\kappa\mu}(b ) \frac{\kappa d\kappa}{2\pi}\right)\,\gamma^{(b)\mu\mu}_{\,\,ss } \nonumber \\ & \times & \left[\int_0^{\kappa_{max } } { \left| g_{\kappa'\mu}^{ss}(b ) \right|}^2 { \left(\frac{q_2^s}{m^0_s}\right)}^2 a^{ss}_{\kappa'\mu}(a)e^{-2 q_2^s w}\frac{\kappa ' d\kappa'}{2\pi}\right ] \nonumber\end{aligned}\ ] ] is the `` vertex '' contribution to the conductance on the right interface . here
@xmath378 are the transport densities of states which for the present case on the left interface are given by the expressions @xmath379/2m_d - \sigma^{dd}_{\mu}(a)\right|}^2}. \quad \nonumber\end{aligned}\ ] ] and by the analogous expressions in the case of right interface .
we remind , that physically the `` bubble '' term in the total conductance is a contribution to the current due to direct tunneling when the electron momentum @xmath54 in the plane of the interface is conserved .
scattering processes renormalize the `` bubble '' term with respect to the case of ballistic transport such that the effective `` height '' of the potential barrier seen by electron increases due to self - energy corrections @xmath380 arising on the interfaces .
the vertex contributions to the conductance describe the tunneling assisted by interfacial scattering that is the processes of tunneling with scattering on the f / o interface when the in - plane momentum is not conserved , i.e. @xmath381 for the scattered electron .
thus , both momenta @xmath155 and @xmath382 in eq .
( [ s_vertex ] ) determine the vertex corrections .
note also , that both vertex corrections @xmath383 and @xmath384 ( [ sa_diff ] ) , ( [ sb_diff ] ) consist of two terms with vertices @xmath385 and @xmath386 .
the terms with vertex part @xmath386 are contribution to the tunneling conductance from @xmath0-like electrons only . the scattering in the @xmath0 channel described by @xmath109 ( [ cpa_50 ] ) is caused initially by @xmath0-@xmath1 scattering .
the terms with @xmath385 describe either the process of diffuse scattering of @xmath1-like electron to the @xmath0 state in the f / o interface and then tunneling of @xmath0-like electron in the barrier , or the process of tunneling of @xmath0-electron and then its scattering to @xmath1 state in the o / f interface and leaving into the electrode .
we have calculated the tunneling conductance and the tmr ratio defined as @xmath387 , where @xmath388 and @xmath389 are the total conductances for the parallel ( p ) and antiparallel ( ap ) alignment of magnetic moments in the f - layers .
the cpa equations defining the self - energies were solved numerically .
the validity of the ward identity , eq .
( [ ward ] ) , was checked after the vertex parts @xmath47 had been computed at every step for a given value of the parameter @xmath141 . per unit square 1@xmath390 ) for the parallel ( p ) alignment of magnetic moments in the ferromagnetic layers as a function of the scattering parameter @xmath141 on the interface in the absence of spin - flip processes .
the parameters of the model are : @xmath0-like electrons @xmath391@xmath22 , @xmath392@xmath22 , @xmath393 , @xmath394 ; @xmath1-like holes @xmath395@xmath22 , @xmath396@xmath22 , @xmath397 , the height of the potential barrier @xmath347 ev , the width of the barrier @xmath343 , the concentration of fe ions on the interface @xmath398 . ]
the results are presented in figs .
47 . in fig .
4 the `` bubble '' ( [ s_ballistic ] ) and the `` vertex '' ( [ sa_diff ] ) , ( [ sb_diff ] ) contributions to the conductance for spin - up and spin - down channels are shown for the parallel ( p ) alignment of magnetic moments as a function of the scattering parameter @xmath141 on the interface .
( we remind that up to now only the spin - conserving scattering is considered . ) as it was mentioned previously , scattering suppresses the `` bubble '' conductance , and hence @xmath399 and @xmath400 are decreasing functions on @xmath141 . the contribution from the `` bubble ''
part is larger for the majority spin @xmath401 channel since @xmath402 . on the contrary
, the contribution from the vertex corrections dominates for the minority spin @xmath403 channel that behavior can be understood as follows .
first , at small @xmath404 the imaginary part of the self - energy @xmath109 , which describes the scattering of @xmath0-electrons , behaves as @xmath405 [ see eq .
( [ cpa_50 ] ) ] , where @xmath406 are spin - up ( down ) @xmath1 density of states on the interface . as far as @xmath407 is of the order of magnitude greater than @xmath408 , the scattering for @xmath0-like itinerant electrons from minority spin band ( @xmath80 ) is more effective than for the majority spin ( @xmath79 ) electrons .
the more effective scattering of spin - down @xmath0-electons leads to the predominance of the minority spin contribution with vertex part @xmath409 over the majority spin contribution with vertex @xmath410 , since at small @xmath141 , @xmath411 and @xmath412 , therefore @xmath413 .
second , the transport density of states @xmath414 [ eq .
( [ a_transport ] ) ] which determines the partial conductance due to mixing of @xmath0 and @xmath1 channels in the interface ( terms with @xmath385 ) is also larger for minority spin channel .
thus , the combination of these factors results in domination of the vertex contribution to the conductance from the minority spin channel . the results of calculations also showed that the @xmath129 contribution to the vertex corrections @xmath415 ( terms with vertices @xmath386 ) was much less than the @xmath130 contribution @xmath416 ( terms with @xmath385 ) . per unit square 1@xmath390 ) for the antiparallel ( ap ) alignment of magnetic moments as a function of the scattering parameter @xmath141 on the interface in the absence of spin - flip processes .
parameters of the model are the same as in fig . 4 .
] the `` bubble '' and vertex contributions to the conductance ( which are the same for spin - up and spin - down channels ) for the antiparallel ( ap ) alignment of moments as a function of @xmath141 are presented in fig .
the total conductances for p and ap configurations are shown in fig .
6 . finally , in fig .
7 the tmr ratio @xmath417 as a function of scattering parameter @xmath141 is shown .
when the amplitude of scattering is negligibly small ( @xmath418 ) , we have positive value of the tmr @xmath419 that is the result of slonczewski s theory @xcite under the chosen parameters for @xmath0like electrons . with increasing of @xmath141 the tmr amplitude is monotonically decreasing and can become even negative if @xmath420 ev . to understand qualitatively
the obtained results , conductance and magnetoresistance can be rewritten in the general approximate form : @xmath421 where we took into account that @xmath422 , and for high and thick enough barrier the main contribution to the tunneling conductance is due to electrons with momentum almost perpendicular to the barrier ( factors @xmath423 ) .
consequently , @xmath424 are the quasi - one - dimensional @xmath0 density of states near the interface , and @xmath425 are the corresponding three - dimensional @xmath1 density of states .
the @xmath426 are renormalized vertex corrections . in expression ( [ dsigma ] ) for @xmath427 , the first term due to direct tunneling is always positive . concerning the second term , @xmath428 since @xmath429 , but @xmath430 and ,
therefore , this term is negative .
this contribution decreases the magnetoresistance when scattering parameter @xmath141 becomes larger . at @xmath431
the @xmath130 contribution overlaps the contribution to @xmath427 from the @xmath129 channel , and we have @xmath432 .
thus , the inverse tmr ratio arisen in our model is caused by extremely strong scattering of negatively polarized @xmath1-like electrons ( which give the indirect contribution to the tunneling current ) to the @xmath0-band on the interface . per unit square 1@xmath390 ) for the parallel ( p ) and antiparallel ( ap ) alignment of magnetic moments as the function of the scattering parameter @xmath141 on the interface in the absence of spin - flip processes .
the parameters are the same as in fig . 4 .
] the parameter @xmath141 determining the amplitude of @xmath0-@xmath1 scattering on the interface is defined as @xmath361 .
one can regard the difference between hybridizations , @xmath433 , as approximately a constant value for given constituents .
however , @xmath93 is a concentration of impurity centers , and thus @xmath141 is a measure of the imperfection of the f / o interface .
the proposed model , therefore , explains qualitatively the strong interface sensitivity of the tunneling magnetoresistance effect @xcite . according to heine s discussion of the hybridization in transition metals @xcite , the hybridization constants are from about 1.0 to 3.0 ev for different elements .
one can assume that the difference @xmath434 is of the order of magnitude smaller and , therefore , the critical value of scattering parameter @xmath435 ev that we obtained is more or less realistic .
the strong reduction of the tmr due to nonideal structure of the metal / insulator interfaces is a well - known observation .
the oxidation of a thin al layer leads to the undesirable oxidation of few metal monolayers close to the f / o interface and thus to the formation of f - o oxides ( fe@xmath3o@xmath10 @xcite , coo and co@xmath3o@xmath10 @xcite ) that reduces the tmr @xcite . on the other hand ,
if the too thick al layer is not oxidized completely , the contamination of the interface by al ions also reduces the tmr @xcite . the dependence of the tmr _
vs._the thickness of the al overlayer has , therefore , a maximum and the best tmr values achieved by moodera s group lie in the range @xmath436 @xcite . in the case of only elastic scattering on the interfaces .
the parameters are the same as in fig . 4 .
] the contamination of the f / o interface by oh ions in the early experiments by merservey and tedrow on tunneling with the superconductors @xcite led to the small measured values of a spin polarization @xmath437 for ni and gd .
the contamination was due to oxidizing of the al films in the laboratory air .
the improved technique of samples preparation in a pure oxygen increased the values of @xmath437 for ni and gd , and for some rare - earth metals @xcite . in a recent work by leclair _
et al . _
@xcite the strong suppression of magnetoresistance was observed in co / al@xmath2o@xmath3/co tunnel junctions with a very thin cr interfacial layer .
it was found that the tmr decayed exponentially on the cr interlayer thickness with a length scale @xmath438 ( approximately 0.5 monolayers ) . with the addition of only 3 cr ( @xmath439 monolayers )
the reduced tmr was only 10% of the initial value for a control junction .
et al . _
@xcite presented some qualitative arguments that the suppression of a spin polarization ( and , hence , the reduction of tmr ) was due to more strongly suppression of majority @xmath0-@xmath25 density of states compared with minority spins caused by the resonant scattering of majority spin @xmath8-electrons with the cr @xmath1 states . from the point of view of our model , we can explain the strong degradation of the tmr by the strong electron scattering within the interfacial co - cr alloy that is formed under the preparation of extremely thin ( @xmath438 monolayer ) cr interlayer . .
] we have also calculated the temperature dependence of the tmr taking into account spin - flip scattering in addition to @xmath272 impurity scattering as it is described in details in sections iic and iid .
for that , the average magnon number @xmath310 ( [ n_t ] ) as a function of the temperature was found in analogy with debye s treatment of phonons using the similar approach that was proposed by s. zhang _ et al _ @xcite . the magnon dispersion relation in eq .
( [ n_t ] ) was replaced by simple isotropic parabolic spectrum @xmath440 where @xmath441 is the equivalent radius of the two - dimensional brillouin zone ( see eq .
[ g_kappa ] ) , and @xmath442 is related to curie temperature @xmath443 and in the mean - field approximation is given by @xmath444 . for the chosen model of the dispersion relation @xmath445 ,
one has to overcome the divergence on the lower limit of the integral in eq .
( [ n_t ] ) .
therefore , one must introduce a lower wave - length cutoff @xmath446 @xcite .
physically , it may represent a finite coherence length due to interfacial roughness . in our calculation
we have taken the same parameters that were used by s. zhang _ et al . _
@xcite for the analysis of the zero - bias anomaly : @xmath447 , @xmath448 mev , and @xmath449 mev .
then , for the temperature range within the room value , we have @xmath450 we have put @xmath451 , @xmath452 ev ( for fe atoms ) and @xmath453 ev , i.e. it is supposed that spin - flip process is possible , if an electron scatters on the fe ion .
the tmr _ vs. _ the scattering parameter @xmath141 at different temperatures @xmath454 and @xmath455 k are plotted in fig
. 8 . in fig .
9 the temperature dependence of the resistance for the p and ap configuration of magnetizations is presented for the same parameters , but @xmath418 .
the results show that the tmr ratio decreases with increasing of the temperature .
moreover , the resistances @xmath456 and @xmath457 of the structure for both configurations , p and ap , are also decreasing when temperature increases that is in the qualitative agreement with experimental data ( e.g. , see fig .
4b in ref .
the physical mechanism of this effect is related with the excitation of spin - flip processes in the system . due to these processes ,
the new channels of electron scattering appear which are frozen at zero temperature . as the result ,
the conductance of the system increases for both p and ap configurations and , therefore , the resistance drops .
the spin - flip processes mix the spin - up and spin - down channels .
therefore , the relative difference of the resistances decreases at different configurations and the tmr decreases with increasing of temperature .
in conclusion , we would like to outline the main results obtained in the present work . based on the analysis of the band structure of 3@xmath1 ferromagnetic metals and al@xmath2o@xmath3 crystals and on the results of ab - initio calculations of the magnetoresistance for epitaxial tunnel junctions , we built on an adaptation of the simplified two - band @xmath0-@xmath1 model to treat the diffuse electron transport in the non - ideal f / o / f magnetic tunnel junctions .
we had modeled the rough f / o interface by the random binary alloy that is formed from the ions of the ferromagnet and impurities ( e.g. the al or o ions ) , and assumed that the main mechanism of electron scattering on the interface which substantially affects tunneling is the @xmath0-@xmath1 scattering .
we used the kubo formalism to calculate the tunneling conductance and found the vertex corrections to conductivity with the use of the `` ladder '' approximation combined with the cpa .
the obtained results show that in the case of strong electron scattering within the interfacial alloy the vertex corrections give the essential contribution to the tunneling conductance .
we proved that adopted approximations lead to the physically correct results , namely , the non - local conductivity tensor is a constant function and , therefore , a tunneling current is conserved .
we showed that interfacial inter - band scattering substantially reduces a value of the tmr , which can become even negative in the case of extremely strong scattering .
the reason of the suppression of the magnetoresistance is the indirect contribution of negatively polarized @xmath1-like electrons to the tunneling current due to strong scattering to the @xmath0 band on the interface .
it is also shown that spin - flip electron scattering on the surface magnons within the interface leads to a further decrease of the tmr at finite temperature .
thus , in the framework of the proposed model , we are able to explain qualitatively the strong interface sensitivity of the tunneling properties that is observed in the experiments .
a. vedyayev and d. bagrets are grateful to cea / grenoble / drfmc / sp2m / nm for fellowships , a. bagrets is grateful to laboratoire louis nel / cnrs ( grenoble , france ) for hospitality .
this work was partially supported by russian foundation for basic research ( grant no .
98 - 02 - 16806 ) .
in this appendix we briefly describe how to obtain eq .
( [ ward ] ) ( ward identity ) , and using a simple example we will show how this identity can be proved . starting from eq .
( [ current0 ] ) , one first has to compute the derivation of the `` bubble '' conductivity . assuming , that the self - energy @xmath277 is symmetric with respect to the rearrangement of band indices , i.e.@xmath458 , and using the fact that functions @xmath459 are the solutions of the schrdinger equation ( [ schrod ] ) , for the derivation of the current matrix we obtain : @xmath460^{\alpha\beta},\ ] ] where @xmath461 and the matrix @xmath462^{\alpha\beta}$ ] is defined by eq .
( [ rho ] ) .
then , from eq .
( [ s_0 ] ) , ( [ lambda ] ) we have @xmath463 .
\nonumber\end{aligned}\ ] ] to proceed further , let us compute the derivation of the vertex correction to conductivity . the initial expression for the matrix @xmath218
( see diagram in fig .
2 ) is written as @xmath464 taking into account that the green s functions are the solution of the eq .
( [ green_eq ] ) , we obtain @xmath465 substituting the obtained expression to formula ( [ s_vertex ] ) for the vertex correction , we get @xmath466 .
\nonumber\end{aligned}\ ] ] the summation here is also performed over indices @xmath229 , @xmath467 , and @xmath228 . a similar expression can be written for the derivation of the vertex correction @xmath468 at interface @xmath83 . from ( [ app_s0 ] ) and ( [ app_s_vertex ] ) one can obtain the final expression ( [ ward ] ) for ward identity .
let s prove identity for the simple case of only @xmath272 scattering , when there are no spin - flip processes , and @xmath359 .
this situation was considered in sec .
iii , and there were introduced the notations @xmath469 , @xmath470 , and @xmath471 ( @xmath472 ) for the components of the @xmath128matricies @xmath47 , @xmath179 , and @xmath367 . within these notations , the identity ( [ ward ] ) , which has to be proved , can be written as @xmath473.\ ] ] here we omitted spin suffixes , all values associated either with the interface @xmath216 or @xmath217 , @xmath474 and @xmath469 are defined by exps .
( [ g_a ] , [ cpa_50 ] ) . from ( [ d ] , [ gamma_eq ] ) it follows that @xmath475 using these expressions , ward identity can be written in the form : @xmath476_{\alpha\beta } + \delta_{\alpha\beta}|g^{\alpha\alpha}|^2 \bigr\ } { \rm im}\,\sigma^{\beta\beta } = 0.\ ] ] using ( [ t_sd ] ) , we then can find that @xmath477 where @xmath478 the same expessions for @xmath479 and @xmath480 follow directly from the cpa equations ( [ cpa_50 ] ) that finishes the proof .
in this appendix we will derive the cpa equations for our particular case using the augmented - space formalism ( asf ) @xcite . as was described previously
, we assume the f / o interface to be a random binary alloy of the type a@xmath9b@xmath87 , where a are ions of the ferromagnet and b are impurities . following the asf , we associate each random variable @xmath89 and @xmath274 with the self - conjugate operators @xmath481 and @xmath482 , respectively , which are determined in the auxiliary 2-dimensional vector space @xmath483 such a way , that the spectrum of these operators coincides with the set of possible values of random variables . for the sake of clarity , hereafter , the sign `` tilda '' is ascribed to any operator acting on the auxiliary space .
we also define the orthonormalized basis @xmath484 , where @xmath485 or @xmath486 , which are eigenvectors of @xmath481 and @xmath482 , so that @xmath487 according to that definition , @xmath481 and @xmath482 commutate with each other .
let now @xmath488 be a function or an operator of random variables @xmath89 and @xmath274 .
then , the operator in the auxiliary space @xmath483 , associated with function @xmath489 , is defined as @xmath490 and according to ( [ phi ] ) , e.g. @xmath491 is a value of @xmath489 , if the site @xmath270 is occupied by an ion a. one can introduce another orthonormal basis in @xmath483 @xmath492 so that the operators @xmath481 and @xmath482 in this representation are written as @xmath493 here @xmath92 and @xmath494 are the concentration of a - type ions ( ferromagnet s ions ) and b - type ions ( impurities ) on the interface , respectively , and @xmath495 then one can prove that the average value of @xmath489 is given by @xmath496 . together with eq .
( [ operators ] ) this property is the way to evaluate the average of any given operator , depending on the random variables , and we can apply this method to average the @xmath281-matrix ( [ t_matrix ] ) . following the general scheme of the asf , outlined above ,
the random effective potential @xmath280 is associated with the operator @xmath497 acting in the augmented vector space @xmath498 , where @xmath499 denotes the 4-dimensional space of orbital ( @xmath500 ) and spin ( @xmath501 ) electron degrees of freedom . in accordance with ( [ v_gamma ] ) , ( [ v_spin ] ) and ( [ operators ] )
this operator has the form @xmath502 where @xmath503 is defined by ( [ self_e ] ) , operators @xmath504 are defined similar to ( [ v_gamma ] ) , and other operators are given by @xmath505 let us also introduce the nonrandom averaged propagator acting in the augmented space @xmath506 and associated with potential @xmath507 ( [ u ] ) the augmented scattering @xmath281-matrix @xmath508 its projection onto the zero - level @xmath509 of the augmented space @xmath510 coincides with the average from the `` physical '' random @xmath281-matrix ( [ t_matrix ] ) .
the subsequent averaging over magnon degrees of freedom @xmath511 must vanish due to condition ( [ cpa_system ] ) . to proceed further ,
let s introduce the electron propagator @xmath512 @xmath513 which is associated with the `` propagation '' of an electron on the first level of the augmented space in the potential @xmath514 .
taking into account the explicit form of @xmath515 with respect to spin - up and spin - down subspaces , @xmath516 where operators @xmath517 and @xmath518 are @xmath519 one gets @xmath520^{-1}\hat g^{\uparrow } , \nonumber \\
\label{g_1 } \hat g_1^{\downarrow } & = & \left [ 1 - \hat g^{\downarrow } \left(\hat\gamma_1 - \sigma^{\downarrow } + \hat
j_1\hat n_{-}\hat g_1^{\uparrow}\hat j_1 \right ) \right]^{-1}\hat g^{\downarrow } , \\
\hat g_1^{+ } & = & \hat g_1^{\uparrow}\hat j_1 \hat s_{- } \hat g_1^{\downarrow } , \quad \hat g_1^{- } = \hat g_1^{\downarrow}\hat j_1 \hat s_{+ } \hat g_1^{\uparrow } , \nonumber\end{aligned}\ ] ] where @xmath521^{-1 } \hat g^{\uparrow(\downarrow)},\ ] ] and @xmath522 , @xmath523 .
the physical meaning of these formulae is rather transparent .
the green s function @xmath524 corresponds to the propagation of the electron in a spin conserving part of the potential @xmath514 which is @xmath525 , while @xmath526 corresponds to scattering on the potential @xmath527 , renormalized with respect to the spin - conserving potential due to the interaction with surface magnons on the interface .
coming back to the evaluation of scattering matrix element @xmath528 , let introduce the `` denominator '' @xmath529 , corresponding to the whole augmented potential @xmath497 @xmath530 again , using the technique of the inversion of a matrix in the block form and taking into account the elements @xmath507 ( [ u ] ) with respect to the auxiliary space @xmath483 , the blocks of @xmath529 can be expressed in terms of the propagator @xmath512 as follows @xmath531^{-1 } , \nonumber \\
\hat d_{10 } & = & \hat g_1 \hat\delta \hat d_{00 } , \quad \hat d_{01 } = \hat g_0 \hat\delta \hat d_1 , \\
\hat d_{11 } & = & \left(1 + \hat g_0
\hat\delta \hat g_1 \hat\delta \right)\hat d_{00 } , \nonumber\end{aligned}\ ] ] where we define the propagator @xmath532 , @xmath533^{-1 } \hat g = \left ( \begin{array}{cc } \hat g_0^{\uparrow } & \hat g_0^{+ } \\
\hat g_0^{- } & \hat g_0^{\downarrow } \end{array } \right),\ ] ] corresponding to the propagation of the electron in the effective potential @xmath534 , where @xmath535 the potential @xmath536 can be regarded as a renormalization of the `` virtual '' crystal potential @xmath537 of the zero level of the augmented space , representing the average of the random potential on a site .
the renormalization comes from the `` interaction '' @xmath538 in the auxiliary space with the first level , being described by the propagator @xmath512 .
using the explicit form of @xmath538 , @xmath539 one can write down the elements @xmath540 and @xmath541 : @xmath542 @xmath543 one can regard @xmath540 as the effective spin - conserving potential , taking into account the effects of a disorder , and @xmath541 as the renormalized electron - magnon interaction . finally , the simple algebra using ( [ u ] ) ( [ den ] ) , yields @xmath544^{-1 } \nonumber \\ & = & \left ( \begin{array}{cc } \hat t_0^{\uparrow } & \hat t_0^{+}\hat
s_{- } \\ \hat t_0^{-}\hat s_{+ } & \hat t_0^{\downarrow } \end{array } \right ) . \nonumber\end{aligned}\ ] ] thus , we have obtained the logical result , that the @xmath281-matrix , averaged over configurations , corresponds to scattering on the effective potential @xmath534 .
to evaluate the spin - conserving @xmath545 and spin - flip @xmath546 parts of scattering matrix , we introduce the propagators @xmath547^{-1}\hat g^{\uparrow(\downarrow ) } , \label{g0_up_down}\ ] ] corresponding to the potentials @xmath548 and define the `` denominator '' @xmath549^{-1 } = \left ( \begin{array}{cc } \hat d_0^{\uparrow } & \hat d_0^{+ } \\
\hat d_0^{- } & \hat d_0^{\downarrow } \end{array } \right).\ ] ] related with the effective potential @xmath550 .
the elements of @xmath551 and propagator @xmath532 can be expressed via @xmath552 using the form of potential @xmath536 ( [ w_eff ] ) , namely @xmath553^{-1}\hat g^{\uparrow } , \nonumber \\
\nonumber \hat g_0^{\downarrow } & = & \left [ 1 - \hat g^{\downarrow } \left(\hat w^{\downarrow } - \sigma^{\downarrow } + \hat j_{-}\hat n_{-}\hat g_0^{\uparrow}\hat j_{+}\right ) \right]^{-1}\hat g^{\downarrow } , \\
\hat g_0^{+ } & = & \hat g_0^{\uparrow}\hat j_{+}\hat s_{- } g_0^{\downarrow } , \quad \hat g_0^{- } = \hat g_0^{\downarrow}\hat j_{-}\hat s_{+ } g_0^{\uparrow } , \label{g_0}\end{aligned}\ ] ] and @xmath554^{-1 } , \nonumber\\ \hat d_0^{\downarrow } & = & \left [ 1 - \hat g^{\downarrow } \left(\hat w^{\downarrow } - \hat \sigma^{\downarrow } + \hat j_{-}\hat n_{-}\hat g_0^{\uparrow}\hat j_{+ } \right ) \right]^{-1 } , \nonumber \\ \hat d_0^{+ } & = & \hat g_0^{\uparrow}\hat j_{+ } \hat s_{- } d_0^{\downarrow } , \quad \hat d_0^{- } = \hat g_0^{\downarrow}\hat j_{- } \hat s_{+ } d_0^{\uparrow } , \label{d_0}\end{aligned}\ ] ] where `` denominators '' @xmath555 are given by @xmath556^{-1}.\ ] ] now , using ( [ g_0 ] ) and ( [ d_0 ] ) and taking into account the obvious relation @xmath557 , one finds @xmath558 according to the asf , @xmath559 and @xmath545 represent the configurationally averaged quantities , and after the averaging over magnon degrees of freedom one must get @xmath560 and @xmath561 .
it means that due to the cpa self - consistency conditions the averaged scattering @xmath281-matrix must vanish and averaged electron propagator has to be equal to the effective green s function ( [ g_int ] ) , determined by the self - energy operator @xmath503 .
carrying out the averaging procedure over magnon degrees of freedom the similar way as it was proposed earlier in sec .
iic , it is possible to satisfy both of these condition if one assumes that @xmath562 we have pointed out the explicit dependence of matrices @xmath563 , @xmath564 and @xmath565 on the average magnon number @xmath270 ( eq . [ n_t ] ) .
this dependence is assumed to be the same as it comes from the initial definition of operators @xmath566 , @xmath541 , and @xmath552 as functions on @xmath309 [ eqs .
( [ w_up_down])([g0_up_down ] ) ] , that is the consequence of the adopted approximate averaging procedure .
the matrices @xmath566 , @xmath541 and @xmath552 functionally depend on @xmath567 . due to that , the system ( [ cpa_iter ] ) represents the alternative to eq .
( [ cpa_system ] ) form of the cpa conditions and it can be simply solved by means of successive numerical iterations .
j. s. moodera , l. r. kinder , t. m. wong , and r. meservey , phys .
* 74 * , 3273 ( 1995 ) ; j. s. moodera , l. r. kinder , j. appl .
phys . * 79 * , 4724 ( 1996 ) ; j. s. moodera , j. nowak , and r. j. m. van de veerdonk , phys . rev .
80 * , 2941 ( 1998 ) t. l. monchesky , a. enders , r. urban , j. f. cochran , b. heinrich , w. wulfhekel , m. klaua , f. zavaliche , and j. kirschner , in _ the physics of low dimensions _ , edited by jose luis moran lopez ( plenum , new york , 2000 ) o. wunnicke , n. papanikolaou , r. zeller , p. h. dederichs , v. drchal , and j. kudrnovsky , communication on the _
european graduate school on condensed matter ( egscm prague01 ) _ ( unpublished ) ; m. zwierzycki , k. xia , p. h. kelly , g. e. bauer , i. turek , ibid.(unpublished ) | we propose the two - band @xmath0-@xmath1 model to describe theoretically a diffuse regime of the spin - dependent electron transport in magnetic tunnel junctions ( mtj s ) of the form f / o / f where f s are 3@xmath1 transition metal ferromagnetic layers and o is the insulating spacer .
we aim to explain the strong interface sensitivity of the tunneling properties of mtj s and investigate the influence of electron scattering at the nonideal interfaces on the degradation of the tmr magnitude .
the generalized kubo formalism and the green s functions method were used to calculate the conductance of the system .
the vertex corrections to the conductivity were found with the use of `` ladder '' approximation combined with the coherent - potential approximation ( cpa ) that allowed to consider the case of strong electron scattering .
it is shown that the ward identity is satisfied in the framework of this approximation that provides the necessary condition for a conservation of a tunneling current .
based on the known results of ab - initio calculations of the tmr for ballistic junctions , we assume that exchange split quasi - free @xmath0-like electrons with the density of states being greater for the majority spin sub - band give the main contribution to the tmr effect .
we show that , due to interfacial inter - band scattering , the tmr can be substantially reduced even down to zero value .
this is related to the fact that delocalized quasi - free electrons can scatter into the strongly localized @xmath1 sub - band with the density of states at the fermi energy being larger for minority spins compared to majority spins .
it is also shown that spin - flip electron scattering on the surface magnons within the interface leads to a further decrease of the tmr at finite temperature . |
simplification leading to understanding of essential features of physical systems is one of the leading principles in theoretical physics .
finding simple solutions to seemingly complicated models is one of the ways to approach this goal . in quantum physics
the complexity of a system grows exponentially with the number of particles . even if an exact solution is possible representing it in a compact way is nontrivial for many - body systems .
one approach to represent a given state is to express its expansion coefficients in a suitable basis in terms of product of matrices , the so - called matrix product states @xcite , used before in a wholly different context of 2d classical lattice model @xcite .
if quantum state is only weakly correlated the resulting matrices can be small . in condensed matter
such an ansatz has been used to exactly describe ground states of many low dimensional spin systems , among the first one for instance the aklt chain @xcite or ladder system @xcite .
in addition to ground states , description in terms of products of matrices is also successfully used in algorithms for simulation of quantum systems @xcite .
crucial for the efficiency of such simulation is the necessary matrix dimension @xmath2 .
unfortunately , generic coherent quantum evolution will cause @xmath2 to grow exponentially with the simulation time @xcite , rendering simulation inefficient .
however , in some special cases of heisenberg evolution of certain operators in integrable systems , like transverse ising chain , the dimension @xmath2 is small and does not grow with time @xcite .
efficient matrix product description with time - independent @xmath2 is possible also for certain open versions of the same integrable models @xcite . besides quantum systems ,
matrix product states are also widely used in classical stochastic models .
there a matrix product formulation is used to study nonequilibrium stochastic lattice models , in particular their nonequilibrium stationary state ( ness ) . since its first use @xcite for the exact solution of a one - dimensional asymmetric exclusion process @xcite
a matrix product formulation has been utilized in a plethora of models , for a review see @xcite .
expectations in the ness can sometimes be calculated with the help of algebra only , sometimes though an explicit representation of matrices is required .
it is not known in general when a finite dimensional matrix representation is possible . in the present work we provide an explicit finite dimensional matrix product solution for a stationary state of quantum system in nonequilibrium situation .
this extends the applicability of matrix product states in quantum domain from ground states and time evolutions to the ness .
the model we consider is a one - dimensional xx spin chain coupled to reservoirs at chain ends . using the jordan - wigner transformation one can reformulate the system in terms of spinless fermions . in the fermionic language
the system is composed of free fermions .
hamiltonian part is therefore trivially integrable .
in fact , for our choice of bath lindblad operators the whole superoperator @xmath3 is quadratic in fermionic operators and can therefore be diagonalized in the operator space @xcite .
an explicit solution for the first two orders in the driving can be also obtained as a particular case of an analytic solution for the xx chain with dephasing @xcite .
slightly different model resulting in the same solution , in which xx chain repeatedly interacts with independent spins of the bath , has been explicitly solved in @xcite . for studies of nonequilibrium states in a doubly infinite chain
see @xcite . our primary goal here is therefore not to solve the system but instead to provide a compact form for the exact ness in terms of a matrix product ansatz with the matrices of low dimension @xmath4 that is independent of the chain length .
this in turn enables one to explicitly evaluate the expectation value in the ness of an arbitrary observable .
hamiltonian of the xx spin chain is given by @xmath5 with standard pauli matrices ; lower indices running from @xmath6 denote a site position .
dynamics of the spin chain coupled to the environment will be described in an effective way using the lindblad master equation @xcite @xmath7 + { \cal l}^{\rm bath}_{\rm l}(\rho)+{\cal l}^{\rm bath}_{\rm r}(\rho)={\cal l}(\rho ) .
\label{eq : lin}\ ] ] to induce a nonequilibrium situation we couple the chain at the first and the last site to a `` bath '' , modelled here by linear operator @xmath8 which is expressed in terms of lindblad operators @xmath9 acting on the first site , and @xmath10 acting on the last site , @xmath11 + [ l_k^{\rm l , r},\rho l_k^{{\rm l , r}\dagger } ] \right).\ ] ]
we take the simplest lindblad operators of the form @xmath12 on the @xmath13st site , while on the @xmath14th site we have @xmath15 @xmath16 .
such bath operators induce an imbalance in magnetization , causing a flow of magnetization from one to the other end . for bath operators only , the stationary state , for which @xmath17 holds , is diagonal in the eigenbasis of @xmath18 , with the expectation value @xmath19 , and correspondingly for @xmath20 .
therefore , parameter @xmath21 plays the role of average magnetization , @xmath22 its difference between the right and left ends , while @xmath23 are coupling strengths .
matrix representation of superoperators @xmath24 can be found in the [ sec : appendix ] .
we are interested in a stationary solution of the whole master equation , @xmath25 , which we call a nonequilibrium stationary state , ness for short , and denote simply by @xmath26 .
we shall write the ness @xmath26 with the matrix product operator ( mpo ) ansatz , @xmath27 indices @xmath28 run over labels of pauli matrices forming an operator basis , @xmath29 , with the convention @xmath30 , matrices @xmath31 are of dimension @xmath32 , while @xmath33 is a @xmath2-dimensional unit vector .
we arbitrarily choose its components to be @xmath34 .
the goal is to write the ness in terms of as small matrices as possible .
our method of solution shall be the following . in ref .
@xcite , an exact solution for an xx model with dephasing has been provided of which our current xx chain is a special limit ( the limit of zero dephasing ) .
it has been observed that in a model without dephasing there are no long - range correlations , that is , all connected correlations are zero .
this leads us to think that one could perhaps construct an exact matrix product operator solution just by observing nontrivial one and two - point observables ; in our case these are magnetization , current and @xmath1 correlations . therefore , based on the solution from @xcite
, we are first going to construct an mpo that accounts only for few - point observables in the ness .
then we are going to show that such mpo in fact provides an exact soluton , that is , all observables and not just few - point are reproduced correctly . to show this we are going to use the algebraic properties of the matrices found . in
@xcite it has been found that ( for @xmath35 , @xmath36 ) the solution is ( upto normalization ) @xmath37 in the above expression we write only linear terms in @xmath22 and @xmath1 term of order @xmath38 .
all other are inessential for the following discussion ) are @xmath39 and @xmath40 . ] .
how can we write such state in terms of products of matrices ? as an easy overture ,
we start with a simpler operator @xmath26 obtained by keeping in the ness ( [ eq : rho ] ) only terms with @xmath41 , i.e. , dropping the current term in eq.([eq : rho ] ) . one can easily convince oneself that the matrices @xmath42 and @xmath43 , where @xmath44 , @xmath45 , @xmath46 , while trivially @xmath47 , gives the wanted state .
therefore , for this simple operator mpo ansatz with dimension @xmath48 would suffice . to describe also the current though
, we have to allow for at least two additional basis states in matrices @xmath49 .
we found that matrices of size @xmath4 are sufficient . to correctly describe all terms of order @xmath22 and @xmath38 ( also those of order @xmath38 that are not explicitly written out in eq.([eq : rho ] ) ) the following set of matrices gives the correct operator , @xmath50 where we use the notation @xmath51 .
matrices @xmath52 and @xmath53 are periodic with period 4 , @xmath54 , and can be concisely written as @xmath55 , while @xmath56 .
we shall show that the mpo with the above matrices ( [ eq : mpo ] ) and appropriately chosen three parameters @xmath57 , @xmath58 , @xmath59 and @xmath60 is an exact ness solution for the lindblad equation ( [ eq : lin ] ) with arbitrary @xmath61 .
to show this we are going to use an algebraic approach similar to the one used in solutions of classical stochastic processes .
let us denote by @xmath62 the superoperator from the commutator part of master equation corresponding to the nearest - neighbor 2-spin term in the hamiltonian , @xmath63 in our case .
writing four mpo matrices in a vector of matrices as @xmath64 , we can form all @xmath65 different products of two matrices at consecutive sites through @xmath66 ( a @xmath67 matrix , each element being a product of two @xmath32 matrices ) . depending on the action of @xmath62 on products of operators , i.e. , on @xmath68 ,
see also the [ sec : appendix ] , solving for ness is relatively simple in two cases : ( i ) if the ness is separable or has only two - particle entanglement , like for instance in valence bond states , then we can have @xmath69 ; ( ii ) other relatively simple situation is when @xmath68 results in a divergence - like term , that is @xmath70 with some matrices @xmath71 .
note that this ansatz is a trivial inhomogeneous extension of a standard procedure used in classical nonequilibrium systems @xcite .
the reason to allow for spatially dependent matrices is that we want to find the mpo solution with the smallest @xmath2 .
we find that representation with @xmath4 is possible irrespective of the chain length @xmath14 .
any inhomogeneous solution can be written as a site - independent one but with larger matrices .
one possibility is to make block site - independent matrix @xmath72 of size @xmath73 out of site - dependent matrices @xmath49 as @xmath74 if equation ( [ eq : bulk ] ) holds terms from consecutive @xmath62 will pairwise cancel , except at the boundaries . to ensure stationarity we have to enforce an additional condition at the boundaries . for our case of bath acting on a single boundary spin we get two equations , @xmath75=0,\quad \left [ { \cal l}_{\rm r}^{\rm bath}(a_n)+m_n \right]|1 \rangle = 0 .
\label{eq : rob}\ ] ] if one manages to find such @xmath49 and the associated @xmath76 that eqs.([eq : bulk ] ) and ( [ eq : rob ] ) are satisfied one has found the ness solution of the master equation ( [ eq : lin ] ) . the condition ( [ eq : bulk ] ) can in our case of the pauli basis of the operator space be written as a set of @xmath77 matrix equations of size @xmath32 .
as we have only @xmath78 matrices @xmath49 and @xmath76 of size @xmath32 , with @xmath79 unknown parameters , it is not guaranteed that the solution exists .
we are now going to show that for the xx model it actually does exist .
we are going to find an explicit representation of matrices @xmath76 , showing that they , together with @xmath49 ( [ eq : mpo ] ) , satisfy equations ( [ eq : bulk ] ) and ( [ eq : rob ] ) .
similarly as for @xmath80 , matrices @xmath81 also have the periodicity 4
. their explicit ( @xmath4)-dimensional representation in the bulk , that is for the sites @xmath82 , is @xmath83 one can convince oneself by direct calculation that the above @xmath76 together with @xmath49 ( [ eq : mpo ] ) satisfy condition ( [ eq : bulk ] ) , written out in full the [ sec : appendix ] , eq .
( [ eq : eqs ] ) . to also fulfill the two boundary conditions ( [ eq : rob ] ) , @xmath84 and @xmath85 must have additional matrix elements and parameters @xmath86 and @xmath87 must take specific values .
we have @xmath88 , @xmath89 and @xmath90 .
on the right end we have @xmath91 , @xmath92 , @xmath93 , where @xmath94 and @xmath95 are the signs in front of last @xmath96 or @xmath97 in eq.([eq : mbar ] ) and depend on the site @xmath14 .
in addition , for the boundary terms to be zero the values of parameters must be @xmath98 note that we have @xmath99 , @xmath100 and @xmath101 .
parameters ( [ eq : a ] ) together with matrices ( [ eq : mpo ] ) form an exact mpo solution of the ness for the xx chain .
if we add a homogeneous magnetic field in the @xmath102-direction to our hamiltonian , that is the term of the form @xmath103 , the model can again be solved exactly .
in fact , the single - site superoperator due to the magnetic field acts as @xmath104 , @xmath105 , while @xmath106 . due to the symmetry of the ness without the field
( see also the explicit form of all nonzero terms given in @xcite ) and the minus sign in the action of @xmath107 , we have @xmath108 .
this means that the solution presented ( [ eq : mpo ] ) is also the exact ness in the presence of an arbitrary homogeneous field of strength @xmath109 .
with an explicit representation of matrix product solution at hand we can evaluate expectations of various operators in the ness .
let us first evaluate expectation of a series of @xmath110 operators , like @xmath111 .
one or two @xmath112s are simple to evaluate by direct calculation , we obtain @xmath113 and @xmath114 .
connected correlation function of @xmath14 operators , @xmath115 , is obtained from an ordinary expectation value by subtracting product of all connected correlations where each involves less than @xmath14 operators .
for instance , @xmath116 , @xmath117 , @xmath118 , and so on .
connected correlation function of two @xmath112s is therefore nonzero only on neighboring sites , @xmath119 .
we are now going to show by induction that all higher order connected correlation functions involving more than two @xmath112s are identically zero .
assume that the statement holds for products of upto @xmath14 @xmath112s .
denoting by @xmath120 a product of @xmath14 not - necessarily neighboring @xmath121 , we would like to show that the connected correlation @xmath122 is zero for @xmath123 . assuming connected correlations of more than two @xmath112s are zero , we can write @xmath124 where we denoted by @xmath125 a sum of all products of connected correlations , each involving @xmath126 operators @xmath112 , for instance , @xmath127 .
we used the fact that to have a nonzero connected correlation @xmath128 must be paired with at most one other @xmath129 and that it must be its neighbor ( non nearest - neighbor connected correlations are zero ) . using our explicit mpo representation ( [ eq : mpo ] ) of matrices for @xmath130 and @xmath131 we will now show that the right - hand side of ( [ eq : highz ] ) is zero .
first , observe that the matrix corresponding to the product of matrices occurring in the operator @xmath132 has an upper - left @xmath133 block equal to @xmath134 upper left element is @xmath135 by definition while the upper right follows through a simple multiplication of matrix corresponding to @xmath136 by @xmath137 . with the explicit form of @xmath132 we have @xmath138 where the kronecker delta takes into account multiplication of the matrix for @xmath132 by @xmath139 in the case of non - neighboring sites @xmath140 and @xmath141 .
plugging this into eq.([eq : highz ] ) we see that the @xmath142-point connected correlation of @xmath112 is indeed zero .
this completes the proof .
single point expectations @xmath143 and @xmath144 are zero . among two - point expectations on neighboring sites , besides @xmath1 , the only nonzero terms are two from the spin current , that is @xmath145 . therefore , the expectation value of the spin current operator @xmath146 is @xmath147 .
if we have a product of non - overlapping current operators at different sites , @xmath148 ( with @xmath149 ) , the expectation value is simple .
observe that @xmath150 .
because of the form of @xmath131 , if we have a product of only @xmath151s and @xmath152 terms @xmath153 at various sites @xmath154 , their expectation value is simply equal to @xmath155 .
this means that the connected correlation function of non - overlapping operators @xmath156 is zero , @xmath157 , apart from @xmath158 .
if two current operators overlap , i.e. @xmath159 , with the hermitean @xmath160 , the corresponding product of matrices is @xmath161 . from this form ,
similarly as before , we can see that for products of @xmath156 the term @xmath162 is not important , resulting again in all connected correlations being zero .
similar argument holds for products of more than two overlapping current operators .
finally , let us discuss expectations of products of @xmath121 and @xmath163 , the only remaining nonzero terms in the ness . for overlapping sites , @xmath164 ( the same for @xmath165 ) , and we have to consider only non - overlapping operators .
two - point expectation is @xmath166 , connected correlation being therefore @xmath167 .
again , taking into account that if we consider only products of @xmath151 , @xmath137 and @xmath153 , the term @xmath168 in @xmath153 is irrelevant because there are no @xmath169 , @xmath170 or their hermitean conjugates in any of the three matrices .
the expectations are therefore trivial products of scalar quantities in front of @xmath171 terms , resulting in all connected correlations being zero . if we have a product of neighboring currents , like for instance in @xmath172 , similar argument holds .
incidentally , we also see that the expectation of the energy current @xmath173 , such that @xmath174=j_i^{\rm e}-j_{i+1}^{\rm e}$ ] , is zero . in the ness with our choice of baths no energy current flows .
another way to write the ness is to rewrite it in terms of an exponential function as @xmath175 .
doing the calculation we observe that the operator @xmath176 contains only few nonzero terms .
these are @xmath110 at all @xmath14 sites ( with different prefactors on different sites ) , spin current @xmath177 at all @xmath178 sites ( the same prefactor on all sites ) as well as @xmath179 , with @xmath180 , and @xmath181 with @xmath182 , where @xmath183 and @xmath184 .
this shows that the ness in the open xx chain can not be exactly written as a quasi - equilibrium generalized gibbs state @xmath185 with locally varying fields @xmath186 , and @xmath187 conserved quantities of the corresponding hamiltonian system @xcite .
namely , conserved quantities in integrable systems can depend on small changes in the system , for instance on boundary conditions . for 1d xx chain with periodic boundary conditions two infinite sequences of conserved quantities exist , one set are @xmath188 and @xmath189 , another @xmath190 and @xmath191 .
if we change boundary conditions to open , only half of conserved quantities survive , that is , for open boundary conditions conserved quantities are only @xmath192 and @xmath193 , where @xmath194 are additional boundary terms @xcite .
first few conserved quantities for open xx chain are @xmath195 , @xmath196 , @xmath197 , @xmath198 .
we can see that @xmath176 can not be written as a sum of conserved quantities @xmath199 of an open chain neither that of a periodic one .
a change in the boundary condition can therefore globally influence constants of motion of hamiltonian system .
this sensitivity also translates to open systems described by master equation : a change in boundary operators of the bath can have a large influence on the ness . in short , for open integrable systems there is an ambiguity which `` conserved '' quantities @xmath187 should one use in the generalized gibbs ensemble , or , in other words , the @xmath187 depend on the bath .
such non - universality , where the functional form of ness does not depend only on the hamiltonian but also on reservoirs , is probably generic situation for open versions of integrable systems . to summarize
, we have shown that all connected correlations , apart from neighboring @xmath200 , are identically zero .
this should not come as a surprise .
in fact , working in fermionic language the system is quadratic , therefore all expectations of products of fermionic operators can be evaluated in terms of two - point expectations using wick s theorem .
we have found an exact solution of an open xx chain in terms of the matrix product ansatz .
an explicit ( @xmath4)-dimensional representation of matrices is found , enabling to evaluate arbitrary expectations .
all connected correlation functions , apart from the correlation function of magnetization at nearest - neighbor sites , are identically zero .
the results presented extend the applicability of matrix product states to quantum nonequilibrium systems .
furthermore , the method of solution is an algebraic one borrowed from the field of classical stochastic processes where it has been used very successfully .
it is hopped that this will lead to new exactly solvable nonequilibrium quantum systems .
one such instance is an open xx chain with dephasing , a solvable diffusive model @xcite where , based on our experience , a compact matrix product solution is also possible .
unfortunately though , in an even more interesting xxz model the algebra seems to be more complicated .
support by the program p1 - 0044 and the grant j1 - 2208 of the slovenian research agency is acknowledged .
[ [ sec : appendix ] ] using a basis of pauli matrices , and tensor products thereof , a single - site superoperator for the bath is @xmath201 superoperator @xmath20 for the right bath is obtained by replacing @xmath202 with @xmath203 and @xmath22 by @xmath204 .
basis elements are ordered as @xmath205 .
matrix representation gives us the operation of @xmath8 on the expansion coefficients of operators in the pauli basis , e.g. , writing @xmath206 , we have @xmath207 , with @xmath208 .
alternatively , the same matrices can be thought of as transforming matrices in the mpo ansatz , e.g. , @xmath209_i=\sum_{j=1}^{4}[{\cal l}^{\rm bath}_{\rm l}]_{i , j } [ a_1]_j$ ] .
superoperator @xmath62 transforms operators according to their commutator with the hamiltonian .
it is therefore fully specified by its operation on a @xmath65 dimensional basis of 2-site operators .
we choose products of pauli matrices as a basis . as an example , for the xx chain we have for instance , @xmath210 .
acting on mpo ansatz , this for instance gives @xmath211 . in the transformed state ,
the term that comes in the product of matrices at the position of @xmath212 is therefore equal to @xmath213 .
because we are interested in how @xmath62 transforms mpo , it is handy to represent its action in terms of a @xmath67 matrix whose element @xmath214_{\alpha,\beta}$ ] gives us the term in front of the @xmath215 . for the xx chain
we have @xmath216 where @xmath217 and @xmath218 , explicitly giving the transformation @xmath219_{\alpha_i,\alpha_{i+1 } } ) \cdots \rangle\ , \sigma_i^{\alpha_i } \sigma^{\alpha_{i+1}}_{i+1}.\ ] ] written explicitly , the algebra ( [ eq : bulk ] ) of matrices @xmath220 and @xmath221 in the bulk that is induced by @xmath62 is for the xx model @xmath222 representation given in eq.([eq : mpo ] ) and ( [ eq : mbar ] ) satisfies this algebra for the arbitrary values of four parameters @xmath223 and @xmath57 . | a one dimensional xx spin chain of finite length coupled to reservoirs at both ends is solved exactly in terms of a matrix product state ansatz .
an explicit reprsentation of matrices of fixed dimension @xmath0 independent of the chain length is found .
expectations of all observables are evaluated , showing that all connected correlations , apart from the nearest neighbor @xmath1 , are zero . |
if @xmath8 is a measure on @xmath9 ( or , more generally , on a metric space ) , the _ local dimension _ of @xmath8 at @xmath10 is defined as @xmath11 provided the limit exists .
( here , and throughout the paper , @xmath12 denotes the open ball of center @xmath10 and radius @xmath13 . ) for many natural measures @xmath8 , such as self - similar measures and measures invariant and ergodic under hyperbolic diffeomorphisms , the local dimension exists and is constant @xmath8-almost everywhere .
however , even in this case there may be many points for which the local dimension takes exceptional values ( or fails to exist ) .
the multifractal analysis for local dimensions is broadly concerned with the study of the level sets @xmath14 the goal is to compute or estimate the _ dimension spectrum _ of @xmath8 , that is , the function @xmath15 where @xmath16 denotes hausdorff dimension . loosely speaking ,
a measure is termed _ multifractal _ if @xmath17 for a range of values of @xmath7 . for many natural measures , such as self - similar measures under the open set condition
, it turns out that @xmath18 is the legendre transform of the so - called @xmath19-spectrum @xmath20 , which is another quantity that reflects the global oscillations of @xmath8 , and is often easier to compute .
see e.g. ( * ? ? ?
* chapter 12 ) .
a large literature is devoted to the study of the dimension spectrum of self - similar measures . in particular , in the case of self - similar measures satisfying the strong separation condition the dimension spectrum was computed in @xcite and this
was extended to the self - similar measures satisfying the open set condition in @xcite . however , despite substantial recent progress @xcite , the overlapping case remains rather mysterious .
in particular , nearly all attention has been focused on _ singular _ self - similar measures @xmath8 .
very recently , feng ( * ? ? ?
* proposition 5.1 ) has shown that certain absolutely continuous self - similar measures may also possess a rich multifractal structure .
a main theme of this paper is to show that this is also the case for what is , perhaps , the simplest and most studied family of overlapping self - similar measures : ( biased ) bernoulli convolutions .
we review their definition and main properties . for @xmath21 and @xmath22
let @xmath0 be the @xmath23-biased bernoulli convolution measure on the real line with contraction rate @xmath4 , i.e. , the self - similar measure satisfying the equation @xmath24 equivalently , @xmath0 is the distribution of the random series @xmath1 where @xmath2 is a sequence of i.i.d .
random variables taking the values 0,1 with probabilities @xmath3 .
( in many papers the series starts with @xmath25 ; here it is more convenient for us to start with @xmath26 . )
yet another useful characterization is the following : @xmath27 , where @xmath28 is the bernoulli measure @xmath29 on @xmath30 , and @xmath31 is the `` natural projection '' from @xmath32 to @xmath33 . in the unbiased , or symmetric , case @xmath34 , we will write @xmath35 . we review some of the key facts about bernoulli convolutions , with a view towards our investigations , while referring the reader to @xcite and references therein for further background .
let @xmath36 where @xmath37 is the @xmath23-entropy .
notice that @xmath38 .
as we will see , @xmath39 is the `` typical '' or `` expected '' dimension of @xmath40 . when @xmath41 , typically @xmath40 is absolutely continuous ; when @xmath42 , on the other hand , @xmath40 is always singular : [ thm : absolute - continuity - bcs]_(peres and solomyak @xcite ) _ * ( i ) * for @xmath43 , the measure @xmath40 is absolutely continuous for almost every @xmath44 . in particular
, @xmath45 is absolutely continuous for almost every @xmath46 . *
( ii ) * on the other hand , @xmath47 for all @xmath48 , whence @xmath40 is singular for @xmath49 .
( here @xmath50 denotes the hausdorff dimension of a measure , defined as the infimum of the hausdorff dimensions of sets of full measure . )
we make some further remarks : finer information on the regularity of the densities of @xmath40 is also available , see ( * ? ? ? * theorem 1.3 ) , as well as the sharpening of these results obtained in @xcite . in the unbiased case
@xmath51 one can say rather more .
for example , @xmath45 has a continuous density for almost every @xmath52 ; this follows from theorem [ thm : absolute - continuity - bcs ] and a convolution argument , see @xcite .
h. toth @xcite showed that for all @xmath53 , @xmath0 is absolutely continuous for almost every @xmath4 in a non - trivial interval @xmath54 .
however , for @xmath23 outside @xmath55 it is not known whether one can take @xmath56 .
( the interval @xmath55 can be expanded somewhat , but existing techniques break down for @xmath23 close to @xmath57 or @xmath58 . )
when @xmath0 is absolutely continuous , it is in fact equivalent to lebesgue measure on its support : [ thm : mutual ] _
@xcite _ let @xmath59 $ ] . if @xmath60 then @xmath61 and @xmath0 are mutually absolutely continuous ( @xmath62 denotes lebesgue measure on the line ) . if @xmath63 is a pisot number , i.e. an algebraic integer larger than @xmath58 all of whose algebraic conjugates are smaller than @xmath58 in modulus , then @xmath40 is singular for all @xmath23 .
it is a major open question whether there are other values of @xmath46 for which @xmath64 is singular .
the only other concrete class of numbers which are believed to , possibly , yield singular bernoulli convolutions are reciprocal salem numbers , see e.g. @xcite for their definition and main properties . in particular , in @xcite feng
proves that bernoulli convolutions associated with ( reciprocal ) salem numbers have a rich multifractal spectrum , a fact which ( in the unbiased case ) had been only established in the pisot case . in this paper
we focus on results which are valid for all or typical @xmath65 , and hence do not deal with these matters . in this section
we consider the symmetric case @xmath66 .
recall that @xmath67 ; we also write @xmath68 .
when @xmath69 , one may ask about further properties of the density @xmath70 ( i.e. the radon - nikodym derivative ) for a typical @xmath4 .
it is not hard to see that @xmath70 becomes @xmath57 at the two endpoints of the support of @xmath71 ( in fact , the local dimension at these two points is @xmath69 ) . in (
* question 8.3.1 ) it was asked whether there can be any other zeros of the density .
it is easy to deduce from theorem [ thm : mutual ] and the ( typical ) existence of continuous density that for a.e .
@xmath72 the density @xmath73 does not vanish in @xmath74 . on the other hand
we show that for _ all _ @xmath75 , where @xmath76 is the ( reciprocal ) golden ratio , there exist infinitely many @xmath77 such that @xmath78 in fact , much more is true , as we will see below .
recall the definition of the local dimension @xmath79 given in .
one may consider also the lower and upper local dimensions @xmath80 , @xmath81 , by taking the lower and upper limits respectively .
these quantities are always defined . in the next theorem ,
@xmath82 is the komornik - loreti constant defined as the unique positive solution of the equation @xmath83 , where @xmath84 is the thue - morse sequence @xmath85 , see @xcite .
we have @xmath86 recall that @xmath87 [ thm : class1 ] for all @xmath88 there exists a set @xmath89 such that @xmath90 moreover , * ( i ) * @xmath91 is countably infinite for @xmath92 ; * ( ii ) * @xmath93 for @xmath94 ; * ( iii ) * @xmath95 as @xmath96 .
the above fails beyond the golden ratio : feng and sidorov ( * ? ? ?
* corollary 1.6 ) have shown that @xmath97 for all @xmath98 and @xmath10 in the interior of @xmath99 .
it is known that for all @xmath100 there exists a constant @xmath101 such that @xmath102 for @xmath45 almost all @xmath10 .
this was stated in @xcite with an outline of a proof , and formally proved in @xcite .
thus by combining this result with theorem [ thm : class1 ] , we can deduce that for all @xmath103 there are at least two distinct values of @xmath7 with @xmath104 .
the definition of the sets @xmath91 is given below , in
[ subsec : expansions ] .
theorem [ thm : class1 ] provides information about a single exceptional value of the local dimension : the similarity dimension @xmath106 .
recall that @xmath0 is a non - linear projection ( under the coding map ) of @xmath28 , the @xmath23-bernoulli measure on the symbolic space @xmath30 .
this measure is multifractal for @xmath107 , which suggests that , in the asymmetric case , @xmath0 may inherit a degree of multifractality .
the results of this section show that this is , indeed , the case .
we start with an extension of theorem [ thm : class1 ] .
we need another number here , @xmath108 , which is the positive solution of the equation @xmath109 , we have @xmath110 assume that @xmath111 without loss of generality .
[ thm : bias1 ]
let @xmath91 be the same sets as in theorem [ thm : class1 ] .
then * ( i ) * for all @xmath112 , @xmath113 * ( ii ) * for all @xmath114 there exists @xmath115 such that @xmath116 & & \mbox{for all}\ \ { \gamma}\in \left[\frac{r_{\lambda}\log p + ( 1-r_{\lambda } ) \log(1-p)}{\log{\lambda } } , \frac{(1-r_{\lambda } ) \log p + r_{\lambda}\log(1-p)}{\log{\lambda } } \right].\end{aligned}\ ] ] moreover , @xmath117 as @xmath118 ; * ( iii ) * for all @xmath114 , @xmath119 the proof will in fact yield quantitative estimates on @xmath120 and the hausdorff dimensions of the sets involved . the second part of the theorem shows that , for @xmath114 and @xmath121 , the spectrum @xmath122 is strictly positive for @xmath7 in some interval which extends ( and possibly starts ) beyond @xmath58 .
our next theorem shows that also many _ small _ local dimensions ( @xmath123 ) arise . unlike theorem [ thm : bias1 ] ,
this is an almost - everywhere result : it holds not only for typical @xmath4 , but also for typical @xmath7 . moreover , we are only able to obtain results for @xmath4 in a so - called _ interval of transversality _ ,
i.e. for @xmath124 , where @xmath125 has the property that proposition [ prop : transversality ] below holds
. in particular , we can take @xmath126 .
see e.g. @xcite for a discussion of transversality in this context and how to find intervals of transversality .
notice that @xmath127 .
nevertheless , we are able to see that biased bernoulli convolutions can have a rich spectrum of small dimensions , even in the parameter region on which they are typically absolutely continuous .
we underline that it is often harder to obtain estimates for @xmath128 when @xmath7 is smaller than the hausdorff dimension of @xmath8 ; see e.g. @xcite for some instances of this .
again assume , without loss of generality , that @xmath129 is fixed .
[ thm : small - dim - bcs ] for almost every @xmath130 , the following holds : the set @xmath131 has positive hausdorff dimension for almost every @xmath7 in @xmath132.\ ] ] more precisely , @xmath133 where @xmath7 and @xmath134 are related by @xmath135 notice that , for any @xmath136 , there is an open set of @xmath4 for which @xmath40 is ( typically ) absolutely continuous and the interval @xmath137 in the above theorem is nontrivial . indeed , by theorem [ thm : absolute - continuity - bcs ]
, @xmath0 is absolutely continuous for a.e .
@xmath138 , while @xmath137 is nontrivial if @xmath139 .
since @xmath140 precisely for @xmath141 , we see that , even for @xmath23 very close to @xmath142 , there are many values of @xmath4 ( more precisely , almost every value in some interval ) for which @xmath0 is absolutely continuous , yet has a positive measure set of local dimensions smaller than @xmath58 .
our last result is of a slightly different kind : it concerns the behavior of the multifractal spectrum as @xmath143 .
we note that , a priori , there is no reason why , for a fixed @xmath10 , @xmath144 should be continuous in @xmath4 .
one may ask whether , in spite of this , the _ global _ spectrum @xmath122 behaves continuously with the parameters @xmath65 and @xmath23 .
in general this appears unlikely , since e.g. reciprocals of pisot numbers are exceptional .
however , the next theorem shows that , as the size of the overlaps tends to @xmath57 , there is continuity : [ thm : continuity - mf - spectrum ] let @xmath145 then , for all @xmath146 and all @xmath147 , @xmath148 although we do not make them explicit , the proof of the theorem provides computable lower and upper bounds : see figure [ fig : mfbounds ] . , for @xmath149 ( left ) and @xmath150 ( right ) .
the dashed curve is @xmath151 ( which can be computed explicitly ) in both cases .
the upper and lower bounds are obtained from the proof of theorem [ thm : continuity - mf - spectrum].,scaledwidth=90.0% ] the following table summarizes the main notation to be used throughout the paper . [ cols="^,^",options="header " , ] we would like to thank mark pollicott and nikita sidorov for useful discussions .
recall that @xmath152 $ ] . for @xmath153 define @xmath154 to be the set of all expansions of @xmath10 in base @xmath4 with digits 0 and 1 .
let @xmath155 be the set of @xmath10 having a unique expansion in base @xmath4 .
of course , 0 and @xmath156 always have a unique expansion .
the following results are known about @xmath157 : * for all @xmath158 and all @xmath77 , the set @xmath159 is uncountable , of positive hausdorff dimension . on the other hand , for all @xmath75 ,
the set @xmath157 is infinite @xcite .
* for all @xmath92 , the set @xmath157 is countably infinite ; for @xmath160 , the set @xmath157 is uncountable of zero hausdorff dimension ; for all @xmath161 the set @xmath157 has positive hausdorff dimension @xcite .
we need some basic facts about @xmath162-expansions ( see e.g. @xcite ) , as well as some more recent results @xcite .
let @xmath163 .
we use the notation @xmath164 to indicate @xmath165 .
we will have @xmath166 , so the digits @xmath167 will always be in @xmath168 . given @xmath169 $ ] , the _ greedy expansion _ of @xmath10 in base @xmath170 is defined as the greatest sequence in @xmath159 in the lexicographic order @xmath171 on @xmath30 .
alternatively , the greedy expansion is given by the symbolic dynamics of the `` greedy '' @xmath162-transformation @xmath172\end{array } \right .. \ ] ] namely , the digit @xmath167 of the greedy expansion is @xmath57 if @xmath173 , and 1 otherwise .
the _ lazy expansion _ of @xmath10 in base @xmath170 is the smallest sequence in @xmath159 in the lexicographic order .
alternatively , it is given by the symbolic dynamics of the `` lazy '' @xmath162-transformation @xmath174 \\
\beta x -1 , & x\in ( \frac{{\lambda}^2}{1-{\lambda}},\frac{{\lambda}}{1-{\lambda } } ] \end{array } \right.,\ ] ] that is , the @xmath175-th digit is @xmath57 if @xmath176 $ ] , and 1 otherwise .
note that the functions @xmath177 and @xmath178 agree on @xmath179 , where @xmath180 $ ] ; see figure [ fig : greedylazy ] . a number @xmath169 $ ] has a unique expansion in base @xmath162 if and only if its greedy expansion coincides with its lazy expansion .
let @xmath181 be the function equal to @xmath177 and @xmath178 on the set of agreement .
in other words ,
@xmath182 \end{array}\right .. \ ] ] note that there is a `` gap '' @xmath183 in the domain of @xmath184 , so not every orbit is well - defined .
in fact , the infinite orbit @xmath185 exists if and only if @xmath10 has a unique expansion in base @xmath4 .
( 200,125 ) ( 0,25)(1,0)150 ( 25,0)(0,1)125 ( 25,25)(2,3)25 ( 50,62.5)(2,3)25 ( 50,25)(2,3)25 ( 100,25)(0,1)75 ( 25,100)(1,0)75 ( 75,62.5)(2,3)25 ( 50,25)(25,75)[b]@xmath186 ( 14,100)(0,0)@xmath156 ( 50,13)(0,0)[b]@xmath4 ( 75,7)(0,0)[b]@xmath187 ( 100,7)(0,0)[b]@xmath156 now we can define the sets which appear in theorems [ thm : class1 ] and [ thm : bias1 ] .
let @xmath188 where @xmath189 .
the statements in theorem [ thm : class1 ] about the size of @xmath91 follow from @xcite , as we now explain .
let @xmath190 observe that for @xmath191 and @xmath192 we have , for @xmath193 , @xmath194 also note that the digits @xmath2 of the expansion of @xmath195 are given by the symbolic dynamics of @xmath184 .
w. parry @xcite characterized the set of all sequences arising from greedy @xmath162-expansions as those which are lexicographically less than the greedy expansion of 1 , and the same holds for all their shifts ( a minor modification is necessary if the greedy expansion of 1 is finite ) .
the next lemma follows from the characterization of unique expansions as being both greedy and lazy .
denote by the bar the `` flip '' of `` reflection '' , i.e. , @xmath196 .
[ lem2 ] _ ( * ? ? ?
* lemma 4 ) _ let @xmath197 and let @xmath198 be the greedy expansion of 1 in base @xmath170 . then @xmath199 [ lem : nested - unique - exp ] for @xmath200 , we have @xmath201 we only need to show the middle inclusion .
let @xmath202 .
then @xmath203 first assume that @xmath204 . if @xmath205 , then @xmath206 since @xmath207 for @xmath208 .
if @xmath209 , then @xmath210 using that @xmath211 and @xmath212 .
the case @xmath213 reduces to the previous one by symmetry , by considering @xmath214 .
thus , @xmath215 , where @xmath216 let @xmath217 and define the frequency of @xmath218 s in @xmath219 by @xmath220 if the limit exists . in the next lemma
, we equip the sequence space @xmath30 with the metric @xmath221 . the hausdorff dimension on @xmath30
is calculated with respect to this metric .
( of course , one may change the power of the exponent from @xmath4 to any @xmath222 ; the base @xmath223 is a common choice . )
[ lem : freq ] * ( i ) * for all @xmath224 , @xmath225 * ( ii ) * for all @xmath226 there exists @xmath115 such that @xmath227 moreover , @xmath117 as @xmath118 .
also , @xmath228 furthermore , @xmath229 our proof uses techniques from @xcite .
the fact is stated in ( * ? ? ?
* example 17 ) with a sketch of proof ; we include a more detailed proof for completeness .
\(i ) in view of the monotonicity of @xmath230 , it is enough to prove the statement for @xmath231 .
the greedy expansion of 1 in base @xmath108 is @xmath232 .
if @xmath191 , then by lemma [ lem2 ] , @xmath233 it follows that @xmath219 is an arbitrary concatenation of the words in @xmath234 and their flips .
this implies that for all @xmath175 , @xmath235 and the claim follows .
\(ii ) if @xmath236 , then @xmath237 and hence the greedy @xmath162-expansion @xmath238 satisfies @xmath239 .
thus , there exists @xmath240 such that @xmath241 then @xmath230 contains arbitrary concatenations of words @xmath242 and @xmath243 , which can be written as @xmath244 .
this clearly allows one to get any frequency of 0 s in @xmath219 in the interval @xmath245 $ ] , as well as having no frequency at all , taking @xmath246 with @xmath247 .
moreover , since the set of arbitrary 0 - 1 sequences @xmath248 with a given frequency of 0 s in @xmath249 , has positive dimension , ( [ eq : dim2 ] ) follows . also , the set of arbitrary 0 - 1 sequences @xmath248 with no frequency of 0 s , has positive ( full ) dimension .
it remains to show that @xmath117 and @xmath250 as @xmath118 . to this end , consider the sequence @xmath251 of `` multinacci numbers '' , i.e. @xmath252 , so that @xmath253 .
clearly , @xmath254 as @xmath255 .
let @xmath256 .
then any sequence in @xmath30 without @xmath257 and @xmath258 belongs to @xmath230 . in particular
, @xmath230 contains arbitrary concatenations of @xmath259 and @xmath260 . taking sequences @xmath261
we can achieve any frequency of 0 s in @xmath219 in the interval @xmath262 on a set of positive dimension .
also , @xmath230 contains all sequences of the form @xmath263 , where @xmath264 are arbitrary .
a standard calculation yields @xmath265 since @xmath266 as @xmath267 , the proof is finished .
we start with a lemma which says that , for @xmath268 , the measure of of a ball centered at @xmath10 is comparable to the measure of the corresponding symbolic cylinder .
[ lem : meas - when - unique - exp ] for any @xmath269 there exists @xmath270 such that the following holds : if @xmath268 and @xmath271 is the unique sequence satisfying @xmath272 , then @xmath273 suppose that @xmath274 , and moreover , @xmath275 .
denote @xmath276 , and let @xmath277 .
it follows from ( [ eq : ss1 ] ) that @xmath278 we can apply ( [ eq : iter ] ) @xmath175 times for the ball @xmath279 to obtain @xmath280 where @xmath281 . now the upper bound in ( [ eq : ineq1 ] ) is immediate , and the lower bound follows from the fact that @xmath282 this is easy to see by self - similarity , again applying ( [ eq : ss1 ] ) : if @xmath283 , then applying ( [ eq : ss1 ] ) @xmath284 times we obtain one of the sets in the right - hand side containing the entire support of @xmath0 , so @xmath285 now the following is immediate from the last lemma and the definitions .
[ cor - cor3 ] for @xmath286 and @xmath287 with @xmath288 we have * ( i ) * if @xmath289 exists , then @xmath290 * ( ii ) * if @xmath291 does not exist , then @xmath292 . finally , the next lemma will allow us to transfer the dimension results of lemma [ lem : freq ] to the euclidean setting .
[ lem : bi - lipschitz ] the map @xmath293 is bi - lipschitz from the metric @xmath294 ( defined before lemma [ lem : freq ] ) to the euclidean metric on @xmath157 .
it is standard ( and very easy ) that @xmath295 is lipschitz on @xmath30 .
for the other direction , let @xmath296 and @xmath297 .
suppose @xmath298 .
then @xmath299 and @xmath300 are different , hence @xmath301 and @xmath302 are in different subintervals of @xmath179 .
thus , @xmath303 and the lemma follows .
the main claim is immediate from corollary [ cor - cor3 ] taking @xmath304 .
the statements ( i ) and ( ii ) about @xmath91 follow from the known results about @xmath157 @xcite mentioned above , together with lemma [ lem : nested - unique - exp ] .
finally , ( iii ) is a consequence of lemmas [ lem : nested - unique - exp ] , [ lem : freq](ii ) and [ lem : bi - lipschitz ] ( since bi - lipschitz maps preserve hausdorff dimension ) .
the statement ( i ) follows from lemmas [ lem : freq](i ) and [ lem : meas - when - unique - exp ] .
the statements ( ii ) and ( iii ) of the theorem follow from lemma [ lem : freq](ii ) , corollary [ cor - cor3 ] and lemma [ lem : bi - lipschitz ] , using again that bi - lipschitz maps preserve hausdorff dimension .
the proof of theorem [ thm : small - dim - bcs ] will be based on a combination of the potential - theoretic method with transversality arguments . to the best of our knowledge ,
this is the first instance in which transversality ideas are used to estimate the multifractal spectrum for a.e .
parameter .
theorem [ thm : small - dim - bcs ] will be an easy consequence of the following stronger technical result : [ thm : small - dim - bcs - intermediate ] fix @xmath305 . for almost all @xmath306 ,
@xmath307 if @xmath308 , then for almost all @xmath309 we have @xmath310 we first indicate how to complete the proof of theorem [ thm : small - dim - bcs ] ; the rest of this section is devoted to the proof of theorem [ thm : small - dim - bcs - intermediate ] . fix @xmath311 . by fubini , for almost every @xmath312 , the following holds for almost every @xmath313 such that @xmath314 : @xmath315 it follows from theorem [ thm : small - dim - bcs - intermediate ] applied to @xmath316 that @xmath317 has hausdorff dimension @xmath318 for almost every @xmath319 .
let @xmath7 be given by the relation , and observe that @xmath7 ranges between @xmath320 and @xmath58 ( the upper bound is due to the restriction @xmath314 ) .
let @xmath321 applying fubini again , we obtain from the above that , for almost every @xmath319 such that @xmath322 , the set @xmath131 has hausdorff dimension at least @xmath323 .
this concludes the proof .
we now start the proof of theorem [ thm : small - dim - bcs - intermediate ] .
the upper bound is standard and holds for all @xmath324 : [ lem : upper - bound - relative - bc ] with @xmath325 as in , @xmath326 clearly , @xmath327
\subset b\left(\pi_{\lambda}({\mathbf{i } } ) , { \mbox{\rm diam}}(\pi_{\lambda}[{\mathbf{i}}|n])\right ) = b\left(\pi_{\lambda}({\mathbf{i } } ) , \frac{{\lambda}}{1-{\lambda}}{\lambda}^n\right).\ ] ] hence @xmath328)}{-\log{\lambda}}.\ ] ] by the law of large numbers , the right - hand side equals @xmath329 for @xmath330-a.e .
@xmath219 , and this implies the lemma . in the remainder of this section we will find a lower bound for the @xmath331-typical value of @xmath332 , by employing transversality techniques .
unfortunately , as indicated earlier , we obtain results only for typical @xmath4 ( for a fixed pair @xmath333 ) . the following is a simple but key lemma , which enables the use of the potential - theoretic method in the calculation of the multifractal spectrum .
[ lem : potential - theoretic ] let @xmath8 and @xmath334 be borel probability measures on @xmath9 .
suppose that @xmath335 for some @xmath336 .
then @xmath337 for @xmath334-almost all @xmath10 .
this is immediate from the fact that @xmath338 which is well - known and easy to verify . in order to apply the previous lemma to bernoulli convolutions ,
we recall the following crucial transversality result ; see ( * ? ? ?
* corollary 2.9 ) , or @xcite for an alternative but explicit approach that yields a somewhat worse constant @xmath339 .
[ prop : transversality ] let @xmath126 .
there exists a constant @xmath340 such that the following holds .
let @xmath341 such that @xmath342 .
then @xmath343 for all @xmath269 .
the following lemma is standard .
[ lem : application - trans ] there exists @xmath340 such that for @xmath344 , @xmath345 and @xmath346 , @xmath347 we may assume that @xmath342 .
but then the lemma follows easily from proposition [ prop : transversality ] and fubini .
we are now able to conclude the proof of the theorem .
fix @xmath305 . in light of lemma
[ lem : upper - bound - relative - bc ] , we only need to prove that , for almost every @xmath319 , @xmath348 fix @xmath349 and @xmath350 .
let @xmath351 and let @xmath352 be a set of @xmath330-measure @xmath353 , on which @xmath354}{n } \rightarrow h_p^q \quad\text{uniformly}.\ ] ] by uniform convergence , there exists a constant @xmath355 such that @xmath356\neq \varnothing } \eta^p[{\mathbf{i } } ] \le c ' e^{-n(h_p^q-{{\varepsilon}})}.\ ] ] since @xmath357 and @xmath358 are arbitrary , by virtue of lemma [ lem : potential - theoretic ] , it is enough to show that @xmath359 after changing variables , we may rewrite @xmath360 lemma [ lem : application - trans ] and fubini yield the estimate @xmath361)\eta^q(\sigma\cap[{\mathbf{i } } ] ) \\ & \le c \sum_{n=0}^\infty { \lambda}_0^{-sn } \max _ { { \mathbf{i}}\in\{0,1\}^n : \sigma\cap [ { \mathbf{i}}]\neq \varnothing } \eta^p([{\mathbf{i}}])\\ & \le c c ' \sum_{n=0}^\infty { \lambda}_0^{-sn } e^{-n(h_p^q-{{\varepsilon}})},\end{aligned}\ ] ] where we used in the last line . since , according to the definition of @xmath362 , @xmath363 , the last series converges .
this completes the proof .
for @xmath364 , it is known that @xmath365 for @xmath45-almost all @xmath10 , and similar results hold for reciprocals of other pisot numbers @xcite . for the golden ratio , a sharp uniform lower bound for @xmath366
was found in @xcite . in the general case we can give uniform lower bounds for @xmath367 which hold for all values of @xmath65 .
however in most cases they are substantially less than @xmath58 , and we do not know whether it may happen that @xmath45 is absolutely continuous but @xmath368 for some @xmath153 .
( recall that in the biased case , this is possible by theorem [ thm : small - dim - bcs ] . )
the first part of the theorem follows from ( * ? ? ?
* proposition 2.2 ) .
for some specific self - similar measures closely related to the multifractal analysis , it also follows from ( * ? ? ?
* proposition 3.4 ) . to prove the second part ,
we first show that it holds for @xmath377 .
given @xmath379 , we will denote @xmath380 . for @xmath374 ,
let @xmath381 if @xmath382 , then @xmath383 and similarly ( or by symmetry ) @xmath384 hence @xmath385 is the only string @xmath219 of length @xmath386 whose first element is @xmath58 , and such that @xmath387 similarly , @xmath388 is the only string @xmath219 of length @xmath386 starting with @xmath57 , and such that @xmath389 we now fix @xmath153 , and note that there can be at most two distinct @xmath390 such that @xmath391 .
since the intervals are closed , we can choose @xmath392 such that for any @xmath153 there are still at most two distinct elements @xmath393 such that @xmath394 .
given @xmath395 , we can choose @xmath175 such that @xmath396 .
if we denote @xmath397 then the bernoulli convolution satisfies @xmath398 we now claim that @xmath399 for all @xmath10 .
we have already shown that @xmath400 for all @xmath10 , and we now proceed by induction .
we assume that for @xmath401 we have that @xmath402 for all @xmath10 .
pick @xmath403 and write @xmath404 , where @xmath405 and @xmath406 .
since @xmath407 , we have @xmath408 , and therefore there are at most @xmath409 choices for @xmath410 .
also , @xmath411 note that @xmath412 and , by our inductive hypothesis , for @xmath401 we have that @xmath413 for all @xmath414 .
hence we have that for each @xmath410 there are at most @xmath415 choices for @xmath416 .
therefore @xmath417 , as claimed .
let @xmath334 and @xmath8 be measures defined on an interval @xmath421 .
assume there are constants @xmath422 and @xmath423 such that , for any subinterval @xmath424 , we have @xmath425 and @xmath426
. then there exists @xmath427 such that , for any interval @xmath428 , we have @xmath429 .
the proof relies on standard facts on the multifractal spectrum of self - similar measures satisfying the open set condition
. see ( * ? ? ?
* chapter 2 ) for general background on self - similar measures , and @xcite and @xcite for their multifractal analysis .
in particular , it is well - known that the dimension spectrum @xmath434 is the legendre transform of the function @xmath435 given by @xmath436 moreover , @xmath437 is also the `` coarse '' multifractal spectrum of @xmath438
. see ( * ? ? ?
* chapter 11.1 ) for the relevant definitions and proofs .
the proof of the lower bound will involve ideas similar to those in lemma [ lem : freq ] .
we will fix @xmath439 and @xmath440 .
as in the proof of lemma [ lem : freq ] this means that any sequence in @xmath32 which does not contain @xmath257 or @xmath258 will be in @xmath441 .
let @xmath442 denote the set of all words in @xmath443 which do not consist only of @xmath57s or @xmath58s .
if we let @xmath444 $ ] , then the set of functions @xmath445 will yield an iterated function system satisfying the strong separation condition
. the attractor of the system will be denoted by @xmath446 , and will be a subset of @xmath447 ( recall and lemma [ lem : nested - unique - exp ] ) . given @xmath379 , let @xmath448 , where @xmath449 and @xmath450 .
let @xmath451 be the number satisfying @xmath452 clearly , @xmath453 as @xmath255 .
let @xmath454 be the self - similar measure satisfying @xmath455 it follows from lemma [ lem : meas - when - unique - exp ] and the strong separation condition for @xmath454 that @xmath456 for all @xmath457 such that the left - hand side is defined . hence @xmath458 however , the multifractal structure of @xmath454 is precisely known thanks to the strong separation condition ( see e.g. ( * ? ? ?
* theorem 11.5 ) ) . in particular , one can easily check that @xmath459 since @xmath453 as @xmath460 , we obtain the desired lower bound . it is known that @xmath468 for any compactly supported radon measure , for a proof see ( * ? ? ?
* theorem 3.3.1 ) .
( although the definition of coarse multifractal spectrum in @xcite is slightly different it uses general packings whereas we use mesh grids
it is a simple exercise to show that @xmath469 will be the same . )
we will denote @xmath470 .
we will use ideas and notation from proposition [ prop : unif - bound - near - ahalf ] .
in particular , we let @xmath471 , @xmath472 and @xmath473 be as in that proposition .
we fix @xmath374 and @xmath474 . without loss of generality , @xmath475 ( the upper bound @xmath476 is arbitrary
, any number smaller than @xmath58 will do ) . we claim that if @xmath481 are the centers of the intervals in the @xmath482-mesh of @xmath99 , then each @xmath483 belongs to at most @xmath484 of the sets @xmath485 , where @xmath486 is independent of @xmath175 . indeed ,
if @xmath487 , then @xmath488 with @xmath489 , whence , from the definition of @xmath473 , and using @xmath490 , @xmath491 and @xmath492 , as claimed .
it follows that @xmath493 since @xmath494 are precisely the @xmath495-measures of the @xmath496-mesh intervals of @xmath497 $ ] , it follows that @xmath498 where , recalling the definition of @xmath499 , @xmath500 on the other hand , it follows from that if @xmath501 , then @xmath502 contains the projected cylinder @xmath503 $ ] , where @xmath504 . since @xmath505
, we can therefore assume that each ball @xmath502 contains a cylinder @xmath506 $ ] , where @xmath507 depends only on @xmath472 ( and hence only on @xmath386 ) .
letting @xmath508 , and using that @xmath509 , we deduce that @xmath510 in other words , @xmath511 where @xmath512 combining and , we easily obtain that @xmath513 } \wtil{f}_{1/2,p}(\overline{\alpha}).\ ] ] recalling that @xmath514 , for which an explicit formula is known ( in particular , it is jointly continuous in @xmath23 and @xmath7 ) , the desired upper bound is achieved . j. c. alexander and william parry .
discerning fat baker s transformations . in _
dynamical systems ( college park , md , 198687 ) _ , volume 1342 of _ lecture notes in math .
_ , pages 16 .
springer , berlin , 1988 .
francois ledrappier . on the dimension of some graphs . in _ symbolic dynamics and its applications ( new haven , ct , 1991 ) _ , volume 135 of _ contemp .
math _ , pages 285293 .
soc . , providence , ri , 1992 .
yuval peres , wilhelm schlag , and boris solomyak .
sixty years of bernoulli convolutions . in _ fractal geometry and stochastics , ii ( greifswald / koserow , 1998 ) _ , volume 46 of _ progr .
_ , pages 3965 .
birkhuser , basel , 2000 .
boris solomyak .
notes on bernoulli convolutions . in _ fractal geometry and applications : a jubilee of benot mandelbrot .
part 1 _ , volume 72 of _ proc .
sympos . pure math .
_ , pages 207230 .
soc . , providence , ri , 2004 . | let @xmath0 be the distribution of the random series @xmath1 , where @xmath2 is a sequence of i.i.d .
random variables taking the values 0,1 with probabilities @xmath3 .
these measures are the well - known ( biased ) bernoulli convolutions .
in this paper we study the multifractal spectrum of @xmath0 for typical @xmath4 .
namely , we investigate the size of the sets @xmath5 our main results highlight the fact that for almost all , and in some cases all , @xmath4 in an appropriate range , @xmath6 is nonempty and , moreover , has positive hausdorff dimension , for many values of @xmath7 .
this happens even in parameter regions for which @xmath0 is typically absolutely continuous . |
to conclude , we present the main results of the paper . a bijective map : @xmath4 operators in @xmath5 matrix @xmath6 is constructed and associated with a star - product formalism . for @xmath1 these operators to form a basis in @xmath7 , conditions on matrix @xmath8 are derived . classification of possible matrices @xmath8 and related star - product schemes @xmath9 is accomplished .
this gives rise to a new approach of introducing bases in @xmath7 with desired properties .
one chooses a class of matrices and impose additional limitations .
once matrix @xmath8 is built , a corresponding basis ( set of operators ) in @xmath7 with expected properties appears .
a development of the paper is complemented by illustrating examples .
another substantial result is a series of propositions and corollaries which demonstrate peculiarities of dequantizers and quantizers , especially in a self - dual star - product scheme .
namely , it is proved that there exists no minimal tomographic star - product scheme with dequantizers in the form of povm effects and hermitian semi - positive quantizers . on applying this argument to self - dual schemes
, we have proved that ( i ) there exists no minimal self - dual star - product scheme with dequantizers in the form of povm effects and ( ii ) hermitian dequantizers and quantizers of a self - dual scheme must contain negative eigenvalues .
the achieved results can be useful for an analysis of the following problems which are of great interest for further consideration : symmetric but non - informationally complete structures of arbitrary rank , a relation between symmetric bases in spaces of different dimension , and specific bases in multipartite systems .
the authors thank the russian foundation for basic research for partial support under projects nos . 09 - 02 - 00142 and 10 - 02 - 00312 .
s.n.f . is grateful to the russian science support foundation for support under project
best postgraduates of the russian academy of sciences 2010 " .
thanks the ministry of education and science of the russian federation for support under project nos . 2.1.1/5909 , @xmath10558 , and 14.740.11.0497 .
v.i.m . was supported by nix computer company in the form of a gift ( computer ) provided by organizers of seminars on quantum physics and informatics in landau institute for theoretical physics , where v.i.m . was happy to deliver two lectures . | columns of @xmath0 matrices are shown to create different sets of @xmath1 operators acting on @xmath2-dimensional hilbert space .
this construction corresponds to a formalism of the star - product of operator symbols .
the known bases are shown to be partial cases of generic formulas derived by using @xmath0 matrices as a source for constructing arbitrary bases .
the known examples of the sic - povm , mubs , and the phase - space description of qubit states are considered from the viewpoint of the developed unified approach .
star - product schemes are classified with respect to associated @xmath3 matrices .
in particular , unitary matrices correspond to self - dual schemes .
such self - dual star - product schemes are shown to be determined by dequantizers which do not form povm .
[ sh ] * * [ cols="^ " , ] |
the existence of homo and heteroclinic cycles in systems with symmetry is no longer a surprising feature .
there are several examples of cycles arising in differential equations symmetric under the action of a specific compact lie group .
several definitions of heteroclinic cycles and networks have been given in the literature .
these objects are associated with intermittent dynamics and used to model stop - and - go behaviour in various applications . throughout the present article , we use the following definition valid for a finite dimensional system of ordinary equations ( ode ) : a _ heteroclinic cycle _
is a finite collection of equilibria @xmath0 of the ode together with a set of heteroclinic connections @xmath1 where @xmath2 is a solution of the ode such that : @xmath3 and @xmath4 .
when @xmath5 , we say that the set @xmath6 is a _
homoclinic cycle_. a _ homoclinic network _ is a connected finite union of homoclinic cycles associated to the same equilibrium point .
complex behaviour near a homo / heteroclinic network is often connected to the occurence of _
switching_. there are different types of switching , leading to increasingly complex behaviour near the network : * _ switching at a node _
@xcite characterised by existence of initial conditions , near an incoming connection to that node , whose trajectory follows any of the possible outgoing connections at the same node .
incoming connection does not predetermine the outgoing choice at the node . * _ switching along a heteroclinic connection _
@xcite , which extends the notion of switching at a node to initial conditions whose trajectories follow a prescribed homo / heteroclinic connection . *
_ infinite switching _
@xcite , which ensures that any sequence of connections in the network is a possible path near the network .
this is different from _ random switching _ in which trajectories shadow the network in a non - controllable way @xcite .
the absence of _ switching along a connection _ prevents _ infinite switching _ and , therefore , chaotic behaviour near the network .
the term _ switching _ has also been used to describe simpler dynamics where there is one change in the choices observed in trajectories .
this is the case described in @xcite . in this case , the network consists of two cycles and trajectories are allowed to change from a neighbourhood of one cycle to a neighbourhood of the other cycle .
this change is referred to as switching , although it is a very weak example of this phenomenon . in @xcite , the expression _ railroad switching _ is used in relation to switching at a node .
complex behaviour near a network can also arise from the presence of noise - induced switching , see @xcite .
we do not address the presence of noise in this note .
the authors of @xcite find a form of complicated switching ( possibly not infinite ) leading to regular and irregular cycling near a network .
there are several examples in the literature where the existence of infinite switching leads to chaotic behaviour near the network , see @xcite .
all the networks considered by these authors have at least one invariant saddle at which the linearized vector field has complex eigenvalues . in general ,
infinite switching is related with the existence of suspended horseshoes in its neighbourhood .
see for example the works @xcite . in these articles , the authors proved the existence of infinitely many initial conditions that realize a given forward infinite path .
these solutions lie on the sequence of suspended horseshoes that accumulate on the network .
so , infinite switching seems to be connected with the existence of suspended horseshoes .
the natural question is : * ( q1 ) * are there examples of homo / heteroclinic networks exhibiting infinite switching and without suspended horseshoes around it ?
the main goal of this note is to answer this question
. we will exhibit a class of vector fields whose flow has a homoclinic network exhibiting infinite switching and without suspended horseshoes around it .
the example is based on the most famous and rich examples in the dynamical systems theory the shilnikov model of a homoclinic cycle to a saddle - focus with negative saddle value @xcite .
although we deal with the classic shilnikov homoclinic loop , as far as we know , the combination _ @xmath7-symmetry _ and _ negative saddle - value _ of the saddle - focus is new .
this note provides an example of a simple attracting network exhibiting sensitivity with respect to initial conditions .
let @xmath8 be a compact three - dimensional manifold possibly without boundary and let @xmath9 the banach space of @xmath10 vector fields on @xmath8 endowed with the @xmath10 whitney topology with @xmath11 .
consider a vector field @xmath12 defining a system @xmath13 and denote by @xmath14 , with @xmath15 , the associated flow ( with initial condition @xmath16 ) . in this paper , we will be focused on an equilibrium @xmath17 of ( [ general ] ) such that its spectrum ( _ i.e. _ the eigenvalues of @xmath18 ) consists of one pair of non - real complex numbers with negative real part and one positive real eigenvalue .
this is what we call a _ saddle - focus_. a _ homoclinic connection _ associated to @xmath17 is a trajectory biasymptotic to @xmath17 in forward and backward times .
homoclinic network _ associated to @xmath17 consists of the equilibrium and a finite union of homoclinic connections associated to o. let @xmath19 be a homoclinic network associated to @xmath17 with @xmath20 .
[ def2 ] if @xmath21 , a finite path of order @xmath22 on @xmath23 is a sequence @xmath24 of homoclinic connections , where @xmath25 is an arbitrary map .
we use the notation @xmath26 for this type of finite path . for an infinite path ,
take @xmath27 .
let @xmath28 be a neighbourhood of the network @xmath23 and let @xmath29 be a neighbourhood of @xmath17 . for each homoclinic connection @xmath30 in @xmath23
, consider a point @xmath31 and a neighbourhood @xmath32 of @xmath33 .
the collection of neighbourhoods @xmath34 should be pairwise disjoint . given neighbourhoods as above
, we say that the trajectory of a point @xmath35 follows a finite path @xmath36 , if there exist two monotonically increasing sequences of times @xmath37 and @xmath38 such that for all @xmath39 , we have @xmath40 and : 1 .
@xmath41 for all @xmath42t_{1},t_{k+1}[$ ] ; 2 .
@xmath43 for all @xmath44 and @xmath45 for all @xmath39 ; 3
. for all @xmath46 there exists a proper subinterval @xmath47z_{j},z_{j+1}[$ ] such that , given @xmath42z_{j},z_{j+1}[$ ] , @xmath48 if and only if @xmath49 .
the notion of a trajectory following an infinite path can be stated similarly . along the paper , when we refer to points that follow a path , we mean that their trajectories do it . based in @xcite ,
we define : there is : [ def1 ] 1 . _
finite switching _ near @xmath23 if for each finite path and for each neighbourhood @xmath28 there is a trajectory in @xmath28 that follows it and 2 .
_ infinite switching _ ( or simply _ switching _ ) near @xmath23 by requiring that for each infinite path and for each neighbourhood @xmath28 there is a trajectory in @xmath28 that follows it .
an infinite path on @xmath23 can be considered as a pseudo - orbit of ( [ general ] ) with infinitely many discontinuities .
switching near @xmath23 means that any pseudo - orbit in @xmath23 can be realized . in @xcite , using connectivity matrices , the authors gave an equivalent definition of switching , emphasising the possibility of coding all trajectories that remain in a given neighbourhood of the network in both finite and infinite times .
our object of study is the dynamics around a special type of homoclinic network , for which we give a rigorous description here .
specifically we consider a family of vector fields in @xmath9 , @xmath11 , with a flow given by the unique solution @xmath50 of ( [ general ] ) satisfying the following hypotheses : 1 .
the point @xmath17 is a saddle - focus equilibrium where the eigenvalues of @xmath51 are @xmath52 and @xmath53 , where @xmath54 and @xmath55 .
2 . there is a trajectory @xmath56 biasymptotic to @xmath17 .
the vector field is @xmath7-symmetric under the action of @xmath57 .
hypotheses * ( h2 ) * and * ( h3 ) * imply the presence of an additional trajectory , say @xmath58 .
thus @xmath59 is a homoclinic network ; in particular @xmath60 in definition [ def2 ] .
we address the reader to @xcite for more details about equations with symmetry .
there are several papers in the literature dealing with the case where the inequality @xmath55 fails , all of them dealing with the complexity of solutions in a neighbourhood of @xmath23 , namely a sequence of suspended horseshoes accumulating on the network .
finitely many of these horseshoes survive under the addition of generic perturbing terms .
our main result says that although @xmath23 is attracting ( the statistical limit set associated to @xmath23 is the point @xmath17 ) , the approach to the network is chaotic .
[ main ] for a vector field @xmath12 whose flow satisfies * ( h1)**(h3 ) * , the following conditions hold : * there are no suspended horseshoes in the neighbourhood of @xmath23 ; * the network @xmath23 is asymptotically stable , in the sense that all trajectories starting in a small open neighbourhood of @xmath23 are attracted to the network ; * there is infinite switching near @xmath23 , realized by infinitely many initial conditions .
the proof of * ( a ) * and * ( b ) * may be found in the works @xcite .
see also @xcite .
the proof of * ( c ) * is addressed in [ sec6 ] of the present note . in particular , in a small neighbourhood of @xmath23 , @xmath61 , if @xmath62 is @xmath63-close to @xmath64 , the set of non - wandering trajectories of @xmath64 in @xmath61 consists of @xmath17 and one or two attracting limit cycles . using theorem [ main ]
, we conclude that the transient dynamics should visit the ghost of the homoclinic cycles ( in any prescribed order ) before falling on the basins of attractions of the periodic solutions .
in [ localdyn ] we linearize the vector field around the saddle - focus , obtaining an isolating block around it ; this section is concerned with introducing the notation for the proof of switching . in [ global ] , we obtain a geometrical description of the way the flow transforms a segment of initial conditions across the stable manifold of @xmath17 .
this curve is wrapped around the isolating block and accumulates on the unstable manifold of @xmath17 ( @xmath65-lemma for flows ) and , in particular , on the next connection .
the local stable manifold of @xmath17 crosses infinitely many times the previous curve .
the geometric setting is explored in [ sec6 ] to obtain intervals of the segment that are mapped by the flow into curves next to @xmath17 in a position similar to the first one .
this allows to establish the recurrence neeeded for infinite switching . for any infinite sequence of homoclinic connections ,
say : @xmath66 without using perturbation theory , we find infinitely many trajectories that visits the neighbourhoods of these connections in the same sequence .
throughout this note , we have endeavoured to make a self contained exposition bringing together all topics related to the proofs .
we have stated short lemmas and we have drawn illustrative figures to make the paper easily readable .
the behaviour of the vector field @xmath64 in the neighbourhood of the network @xmath23 is given , up to topological equivalence , by the linear part of @xmath64 in the neighbourhood of @xmath17 and by the transition map between two discs transversal to the flow in those neighbourhoods . in this section ,
we choose coordinates in the neighbourhood of @xmath17 in order to put @xmath64 in the canonical form and we assume that the transition map is linear . the main point is the application of samovol s theorem @xcite to @xmath67linearize the flow around @xmath17 , and to introduce cylindrical coordinates around the equilibrium .
there are no @xmath67-resonances here .
we use neighbourhoods with boundary transverse to the linearized flow . since @xmath17 is hyperbolic , by samovol s theorem @xcite , the vector field @xmath64 is @xmath68conjugate to its linear part in a @xmath69-small open neighbouhood around @xmath17 , @xmath70 .
we choose cylindrical coordinates @xmath71 near @xmath17 so that the linearized vector field can be written as : @xmath72 . for @xmath73 , on a segment @xmath74 there are infinitely many subsegments that are mapped by @xmath75 into @xmath76 , each one containing a point mapped into @xmath77 .
the small sub - segments contain smaller ones that are mapped into by @xmath78 into @xmath79 and the process may be continued forming a nested sequence .
, height=264 ] after a linear rescaling of the local variables , we consider a cylindrical neighbourhood of @xmath17 of radius @xmath80 and height @xmath81 that we denote by @xmath82 see figure [ neigh_vw ] .
their boundaries consist of three components : the cylinder wall parametrized by @xmath83 and @xmath84 with the usual cover @xmath85 and two disks ( top and bottom ) .
we take polar coverings of these disks @xmath86 where @xmath87 , @xmath88 and @xmath89 . by convention ,
the intersection points of @xmath23 with the wall of the cylinder has @xmath90 and @xmath91 angular coordinate .
the set of points in the cylindrical wall with positive ( resp .
negative ) second coordinate is denoted by @xmath92 .
( resp @xmath93 ) .
the similar holds for @xmath94 .
as depicted in figure [ neigh_vw ] , the cylinder wall of @xmath82 is denoted by @xmath95 .
note that @xmath77 corresponds to the circle @xmath96 .
the top and the bottom of the cylinder are simply denoted by @xmath94 .
the boundary of @xmath82 can be written as the disjoint union : @xmath97 where @xmath98 is the part of @xmath99 where the flow is not transverse .
it follows by the above construction that : let @xmath100 .
solutions starting : 1 . at @xmath95 go inside the cylinder @xmath82 in positive time ; 2 . at @xmath94
go outside the cylinder @xmath82 in positive time ; 3 . at
@xmath101 leave the cylindrical neighbourhood @xmath82 at @xmath102 .
if @xmath103 , let @xmath104 be the time of flight through @xmath82 of the trajectory whose initial condition is @xmath105 .
it only depends on @xmath106 and is given explicitly by @xmath107 in particular @xmath108 .
now , we obtain the expression of the local map that sends points in the boundary where the flow goes in , into points in the boundary where the flows goes out .
the local map @xmath109 near @xmath17 is given by @xmath110 where @xmath111 is the _ saddle index _ of @xmath17 .
observe that if @xmath112 is fixed , then @xmath113 the same expression holds for the local map from the other connected component of @xmath114 to @xmath115 ( where @xmath116 ) with the exception that the first coordinate of @xmath117 changes its sign .
in @xcite , the author obtains precise asymptotic expansions for the local map @xmath118 . in the present article ,
we omit high order terms because they are not needed to our purposes .
let @xmath119 . by the _
tubular flow theorem _
@xcite , solutions starting near @xmath120 follow one of the connections in @xmath23
. we may then define the transition map @xmath121 by flow box fashion and the return map to @xmath95 : @xmath122 hereafter , we concentrate our attention on initial conditions that do not escape from @xmath61 ; otherwise take a smaller subset in @xmath123 where the returm map is well defined .
the explicit expression for the return map is highly nonlinear since the distortion near the hyperbolic saddle - foci is tremendous . for our purposes
, it is not needed .
the coordinates and notations of [ localdyn ] will be used to study the geometry of the local dynamics near the saddle - focus .
this is the main goal of the present section but first we introduce the concept of a _ segment _ on @xmath95 .
let @xmath100 .
a _ segment _
@xmath124 : 1 .
[ segment w ] _ on _ @xmath79 is a smooth regular parametrized curve @xmath125\rightarrow \sigma^{in}_j$ ] that meets @xmath126 at the point @xmath127 and such that , writing @xmath128 , both @xmath129 and @xmath130 are monotonic and bounded functions of @xmath131 .
the definition of _ segment _ may be relaxed : the components do not need to be monotonic for all @xmath132 $ ] .
we use the assumption of monotonicity to simplify the arguments .
let @xmath133 , @xmath134 be a disc centered at @xmath135 and @xmath136 a line passing through @xmath137 . 1 .
a _ spiral _ on @xmath134 around the point @xmath137 is a smooth curve @xmath138 satisfying @xmath139 and such that if @xmath140 is its expression in polar coordinates around @xmath137 then : 1 . the map @xmath141 is bounded by two monotonically decreasing maps converging to zero as @xmath142 ; 2 .
the map @xmath143 is monotonic for some unbounded subinterval of @xmath144 and 3 .
a _ double spiral _ on @xmath134 around the point @xmath137 is the union of two spirals accumulating on @xmath137 and a curve connecting the other end points .
3 . given a _ spiral _
@xmath146 on @xmath134 around the point @xmath137 , an _
half circle _ bounded by @xmath136 is a connected component of @xmath147 .
the next result characterizes the local dynamics near the saddle - focus .
[ lemma3 ] let @xmath148 .
a _ segment _
@xmath124 on @xmath79 is mapped by @xmath149 into a spiral on @xmath150 accumulating on the point defined by @xmath151 .
the proof will be done for @xmath152 .
the other case is analogous .
let @xmath74 be a segment on @xmath153 .
write @xmath154 , where : * @xmath132 $ ] , * @xmath155 is an increasing map as function of @xmath131 and * @xmath156 .
the function @xmath149 maps the segment @xmath157 into the curve defined by : @xmath158= \left[(y^\star(s))^\delta , -\frac{\alpha}{e}\ln map @xmath159 is a spiral on @xmath160 accumulating on the point defined by @xmath161 because @xmath162 and @xmath163 are monotonic ( see remark [ rem1 ] ) and @xmath164 [ rem1 ] let @xmath148 .
the coordinates @xmath165 may be chosen so as to make the map @xmath166 increasing or decreasing , according to our convenience . from now on , we omit the dependence of @xmath167 on @xmath131 to simplify the notation .
in this section we put together the information about the return map . first note that if @xmath168 , we denote by @xmath169 and @xmath170 its topological closure and topological interior , respectively . in what follows we
remove the point @xmath171 from the @xmath172 because the return map is not defined at this point . from now on , let us fix @xmath61 , a small neighbourhood of @xmath23 .
figure [ neigh_vw ] illustrates the main idea of the following proof .
the next result shows that there are infinitely many points in @xmath173 which are mapped under @xmath75 into @xmath77 and that separate segments of initial conditions that follow the different connections @xmath56 and @xmath174 .
[ first lemma ] let @xmath175 be a rectangle in @xmath95 centered at one point of @xmath176 .
for any segment @xmath177 \rightarrow \mathcal{r}\cap \sigma^{in}_+$ ] , there is a family of intervals of the type @xmath178 $ ] such that for all @xmath179 , we have : as already said , we concentrate our attention on initial conditions that do not escape from @xmath61 . by lemma [ lemma3 ] ,
the image of @xmath183 under @xmath184 is a spiral accumulating on @xmath185 . in its turn , this spiral is mapped by @xmath186 into another spiral in @xmath95 accumulating on one point of @xmath176 .
the curve @xmath77 cuts transversely this spiral into infinitely many points .
let @xmath187 the points for which @xmath188 . by construction
, it is easy to see that if @xmath189 either @xmath190 or @xmath191 .
suppose , without loss of generality , that the first case holds .
then , by continuity of @xmath75 , for @xmath192 we get @xmath191 .
[ second lemma ] let @xmath175 be a rectangle in @xmath95 centered at one point of @xmath176 and let @xmath193 be as in lemma [ first lemma ] . then for sufficiently large @xmath194 , there are @xmath195 $ ] such that for all @xmath196 , we have : our starting point is the half - circle @xmath201 , @xmath179 , bounded by @xmath77 . since the image under @xmath75 of each half circle can be seen as the image of two connected segments , using lemma [ lemma3 ] the set @xmath202 is a double spiral accumulating on @xmath203 , which is mapped under @xmath186 into a double spiral accumulating on @xmath176 .
the line @xmath77 intersects this double spiral infinitely many times .
let @xmath204 the sequence of points such that @xmath197 .
for a given @xmath205 , the arguments used before may be used to conclude that there exists a sequence @xmath206 such that @xmath199)$ ] is an half - circle in @xmath92 bounded by @xmath77 and @xmath207)$ ] is an half - circle in @xmath93 bounded by @xmath77 . in the previous sections we have proved that a segment cutting transversely the stable manifold of @xmath17 contains subsegments @xmath193 that are mapped into new segments cutting transversely the stable manifold of @xmath17 . starting with a segment @xmath74 on @xmath153 ,
we may obtain , recursively , nested compact subsets containing initial conditions that follow any prescribed sequence of connections .
let @xmath208 .
we say that the path @xmath209 of order @xmath22 on the homoclinic network @xmath23 is inside the path @xmath210 of order @xmath211 if @xmath212 for all @xmath213 .
we denote this relation by @xmath214 . given a path of order @xmath215 , @xmath216 , we want to find trajectories that follow it .
let us fix a segment @xmath217 given by lemma [ first lemma ] and set that all initial conditions in @xmath218 follow the connection @xmath219 . take a closed subset @xmath220 of @xmath217 . by construction ,
all initial conditions starting in @xmath220 follow the connection @xmath219 .
the set @xmath221 is an half circle cutting transversely @xmath222 infinitely many times . by lemma [ second lemma ]
, one can obtain again sequences of points in @xmath223 , where a similar result to that in lemma [ first lemma ] can be stated for @xmath78 instead of @xmath75 .
take a closed subset @xmath224 of @xmath225 . by construction ,
all initial conditions starting in @xmath224 follow the path @xmath226 .
a recursive argument allows the construction of a compact set @xmath227 of initial conditions whose trajectories follow @xmath26 .
[ infinite2 ] we first need to introduce some extra terminology . given a path of order @xmath215 , @xmath228 we denote by @xmath229 the compact set @xmath230 obtained in the proof of proposition [ theoremfiniteswitching ] and we say that @xmath229 is an admissible set .
recall that all points in @xmath229 correspond to inifinitely many solutions following @xmath26 .
fix an infinite path @xmath232 , with @xmath233 . for
each @xmath234 define the finite path @xmath235 .
taking into account remark [ admissible ] it follows that there exists an infinite sequence of admissible sets @xmath236 such that @xmath237 for all @xmath21 .
since the sequence of compact sets @xmath236 is nested , @xmath238 .
any initial condition in @xmath239 gives a trajectory which follows @xmath240 .
the different solutions are distinguished by the number of revolutions around the isolating block of @xmath17 @xcite .
again by construction we find trajectories realising the required switching arbitrarily close to @xmath23 . based on theorem [ main ] , the answer to the question * ( q1 ) * is _ yes_. at this point , other questions arise , namely : _ are there other examples of heteroclinic networks where infinite switching holds without suspended horseshoes emerging in its neighbourhood ? _ | in general , infinite switching behaviour near networks is associated with the existence of suspended horseshoes .
trajectories that realize switching lie within these transitive sets . in this note , revisiting the equivariant shilnikov scenario , we describe an attracting homoclinic network exhibiting forward switching and without suspended horseshoes in its neighbourhood .
thus we provide an example of an asymptotically stable network exhibiting sensitive dependence on initial conditions . en gnral , le comportement de commutation autour des rseaux est associ lexistence de fers cheval en suspension .
trajectoires qui ralisent la commutation se situent dans ces ensembles transitifs .
dans cette note , visitant le scnario classique de shilnikov avec une symtrie , nous dcrivons un rseau homocline qui prsente commutation nayant pas des fers cheval en suspension autour de lui .
plus prcisement nous donnons un exemple dun rseau asymptotiquement stable avec sensible dpendance des conditions initiales . |
Subsets and Splits