article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
in electrodynamics , the integral equation method ( iem ) is a powerful method for the electromagnetic wave propagation @xcite . from a microscopic perspective
, a bulk material can be regarded as a collection of molecules ( or atoms ) , each of which is a scatter and emits radiation under the action of an external field .
the radiated retarded field and the exciting field interact to form the resultant transmitted field in the material .
this formulation is often expressed as an integral equation and naturally leads to the well known ewald - oseen extinction theorem which is a basis of the scattering theory @xcite .
the iem relates the macroscopic electromagnetic responses of the material with the microscopic features of its constituent units , thus gives much deeper physical insight into the interaction of wave with material in contrast to the conventional approach of maxwell theory @xcite . at the same time , the iem does not require boundary conditions and is advantageous in some situations , especially when it is not sufficient to describe the interaction by the macroscopic maxwell equations @xcite .
this approach has been widely used to study light scattering @xcite , the tip - sample interaction in near optics @xcite and so on . by now , the interactions of waves with semi - infinite @xcite , a slab @xcite or stratified @xcite isotropic dielectric media have been treated by the iem .
recently , one kind of artificial anisotropic dielectric - magnetic materials , metamaterials , has attracted considerable attention simulated by its fascinating properties @xcite . to understand the unique characteristics of metamaterials it is necessary and appealing to take a microscopic viewpoint . on this issue
, the iem has been successfully implemented to predict the electric and magnetic resonances in split - ring resonators @xcite , study the reflection of split - ring resonators @xcite and the imaging process of super lens @xcite , and explain the brewster phenomenon for te wave associated with metamaterials @xcite .
however , most previous work only consider the situation of semi - infinite materials and waves incident from vacuum .
then , how to treat the propagation of waves through stratified materials , especially for incidence from a material rather than vacuum ?
the latter problem is important because it will result in a completely new form of the extinction theorem , as shown in this paper .
the purpose of this paper is to use the iem to investigate the properties of wave propagation in stratified anisotropic dielectric - magnetic media .
our emphasis is to disclose the underlying physics of electromagnetic interaction with matter rather that to obtain the well - known phenomenal results .
on the basis of the superposition principle , we use hertz vector formulations of radiated fields to study the microscopic mechanism of propagation .
we arrive at new derivations of the dispersion relation , snell s law and reflection / transmission coefficients by self - consistent analyses .
moreover , we find two generalized forms of the extinction theorem . as an application
, we use the iem to study the wave propagation through a slab and discuss the physical process of photon tunneling , which are further verified through numerical simulations .
the results lead to a unified framework of the iem for the propagation of wave incident either from a medium or vacuum in stratified dielectric - magnetic materials .
in this section we take the microscopic viewpoint in which the matter consists of molecules that react to the exciting field like dipoles and formulate the radiated field in the form of a green s function integral of hertz vectors .
let a monochromatic electromagnetic field of @xmath0 and @xmath1 incident from an anisotropic material filling the semi - infinite space @xmath2 .
the @xmath3 plane is the plane of incidence and the schematic diagram is in fig .
[ schem ] .
since the material responds linearly , all the fields have the same dependence of @xmath4 which will be omitted subsequently for simplicity .
assume the reflected fields are @xmath5 and @xmath6 , and the transmitted fields are @xmath7 and @xmath8 .
the permittivity and permeability tensors of the @xmath9-th layer are simultaneously diagonal in the principal coordinate system , @xmath10 $ ] , @xmath11 $ ] .
schematic diagram for the reflection and refraction of te waves in stratified media.,width=302 ] inside the layers , the incident field drive the constituent unities to oscillate and radiate just as dipoles .
the radiated electric fields by electric and magnetic dipoles are decided by @xcite @xmath12 and the magnetic fields generated by dipoles are @xmath13 here @xmath14 and @xmath15 are the hertz vectors , which can be expressed as integrations of the green s function , @xmath16 @xmath17 is the dipole moment density of electric dipoles and @xmath18 is that of magnetic dipoles , which are related to the internal fields as @xmath19 , @xmath20 , where the electric susceptibility @xmath21 and the magnetic susceptibility @xmath22 . inside the layer , the dipoles produce forward waves as well as backward waves .
so , we assume the dipole moment densities @xmath17 and @xmath18 have the following forms @xmath23 where @xmath24 stands for the forward wave and @xmath25 stands for the backward wave @xcite .
the wave numbers are to be decided by the dispersion relation yet to be derived .
the green function is @xmath26 . to evaluate the hertz vectors ,
we firstly represent the green function in the fourier form .
then , substituting it into eqs .
( [ pi_e ] ) and ( [ pi_m ] ) and using the delta function definition and contour integration method @xcite , the hertz vectors of the @xmath9-th layer can be evaluated as right to the material left to the material where we have used the following notations @xmath30 note that the term with @xmath31 equals zero if @xmath32 contains @xmath33 , which can be found in the calculation .
then the hertz vectors of arbitrary layers can be found in eqs .
( [ pi_ne_int])-([pi_ne_right_left ] ) .
analogous to @xmath34 , @xmath35 can be obtained by interchanging @xmath36 and @xmath37 with @xmath38 and @xmath39 , respectively . inserting @xmath34 and @xmath35 into eqs .
( [ e_dipole_field ] ) and ( [ h_dipole_field ] ) , the radiated fields can be calculated out .
in this section , we deduce the expressions of radiated fields generated by constituent unities in the propagation of waves in stratified anisotropic dielectric - magnetic media .
then we use the superposition principle to show how all the fields are combined to yield the reflected and transmitted fields inside the individual layers . by a self - consistent analysis
we obtain the dispersion relation , snell s law and fresnel s coefficients .
moreover , we find two generalized forms of the extinction theorem . inside the @xmath9-th layer , the layer by itself and the other layers all radiate under the action of external fields .
their radiated fields add up to the real fields inside this layer , that is , @xmath40 here the radiated field from the left semi - infinite space @xmath41 the radiated field generated by the layer itself @xmath42}}{2q_{nz}(q_{nz}-k_{nfz})}{\bf q}({\bf q}_{nf},{\bm p}_{nf})\nonumber\\ & & + \frac{\exp{[i(k_{nfz}+q_{nz})z_n]}}{2q_{nz}(q_{nz}+k_{nfz})}{\bf q}({\bf q}_{nb},{\bm p}_{nf})+\frac{\exp{[-i(k_{nbz}+b_n)z_{n-1}]}}{2b_{nz}(b_{nz}+k_{nbz})}{\bf q}(-{\bf b}_{nb},{\bm p}_{nb } ) \nonumber\\ & & + \frac{{\bf q}({\bf k}_{nb},{\bm p}_{nb})}{k_{nbz}^2-b_{nz}^2 } + \frac{\exp{[-i(k_{nbz}-b_{nz})z_n]}}{2b_{nz}(b_{nz}-k_{nbz})}{\bf q}(-{\bm b}_{nf},{\bm p}_{nb}),\end{aligned}\ ] ] the radiated field generated by left layers @xmath43 @xmath44}-\exp{[i(k_{jfz}-q_{jz})z_{j } ] } } { 2q_{jz}(q_{jz}-k_{jfz})}{\bf q}({\bf q}_{jf},{\bm p}_{jf})\nonumber\\ & & + \sum_{j=1}^{n-1}\frac{\exp{[-i(k_{jbz}+b_{jz})z_{j-1}]}-\exp{[-i(k_{jbz}+b_{jz})z_{j}]}}{2 b_{jz}(k_{iz}+k_{jbz})}{\bf q}(-{\bf b}_{jb},{\bm p}_{jb}),\end{aligned}\ ] ] the field radiated by right layers @xmath45 @xmath46}-\exp{[i(k_{jfz}+q_{jz})z_{j-1 } ] } } { 2q_{jz}(k_{jfz}+q_{jz})}{\bf q}({\bf q}_{jb},{\bm p}_{jf})\nonumber\\ & & + \sum_{j = n+1}^{n}\frac{\exp{[-i(k_{jbz}-b_{jz})z_{j}]}-\exp{[-i(k_{jbz}-b_{jz})z_{j-1}]}}{2 b_{jz}(b_{jz}-k_{jbz})}{\bf q}(-{\bf b}_{jf},{\bm p}_{jb}).\end{aligned}\ ] ] the radiated field from the right semi - infinite space @xmath47 } } { 2q_{(n+1)z}(k_{(n+1)fz}+q_{(n+1)z})}{\bf q}({\bf q}_{(n+1)b},{\bm p}_{(n+1)f}),\end{aligned}\ ] ] and @xmath48 is an auxiliary function @xmath49\exp{(i{\bf k}\cdot { \bf r})}\ ] ] where @xmath50/\omega$ ] .
there are two cases for the layer .
* case 1 . *
the layer is an anisotropic dielectric - magnetic slab . using @xmath51/(\varepsilon_0\boldsymbol{\chi}_{ne})$ ] and inserting eqs .
( [ e^left.0_rad ] ) - ( [ e^right.n+1_rad ] ) into eq .
( [ e_n ] ) , we come to the following conclusions .
( iii ) . from terms with the phase factors @xmath60 or @xmath61 in eq .
( [ e_n ] ) follows the dispersion relation @xmath62 for te waves .
so @xmath63 .
( [ dispersioneh ] ) relates all the propagation modes in different layers .
* case 2 . *
the layer is vacuum . assuming the field in the layer @xmath64
, we arrive at @xmath65}}{2q_{(n-1)z}(q_{(n-1)z}-k_{(n-1)fz})}{\bf q}({\bf q}_{(n-1)f},{\bm p}_{(n-1)f})\nonumber\\&&-\frac{\exp{[-i(k_{(n-1)bz}+q_{(n-1)z})z_{n-1}]}}{2q_{(n-1)z}(q_{(n-1)z}+k_{(n-1)bz})}{\bf q}({\bf q}_{(n-1)f},{\bm p}_{(n-1)b}),\end{aligned}\ ] ] and @xmath66}}{2q_{(n+1)z}(q_{(n+1)z}+k_{(n+1)fz})}{\bf q}({\bf q}_{(n+1)b},{\bm p}_{(n+1)f})\nonumber\\&&-\frac{\exp{[-i(k_{(n+1)bz}-q_{(n+1)z})z_n]}}{2q_{(n+1)z}(q_{(n+1)z}-k_{(n+1)bz})}{\bf q}({\bf q}_{(n+1)b},{\bm p}_{(n+1)b})\end{aligned}\ ] ] from eqs .
( [ extinction_e_left ] ) and ( [ extinction_e_right ] ) , it seems that the field in the layer was only related with the adjacent regions .
however , it is only a mathematical impression because physically all the fields from the whole space contribute to the field in the layer , as shown in eq .
( [ e_n ] ) obviously . in region @xmath2
, the contributions of radiations from all regions form the real fields , yielding @xmath67 the field radiated by the material in the region @xmath2 can be obtained like eq .
( [ e^n_rad ] ) @xmath68 then we have the extinction of _ backward _ vacuum waves @xmath69 the incident and reflected fields have the following forms @xmath70 by solving the above equations one can obtain the dispersion relation like eq .
( [ dispersioneh ] ) and the reflected field .
if the region @xmath2 is vacuum , then @xmath71 eq . ( [ e_r_vacuum ] ) is well known in previous work in the case of incidence from vacuum @xcite .
differently , eq . ( [ extinction_e_0 ] ) is a completely new form of the backward extinction theorem because the region @xmath2 is not vacuum but a material in which the scatters are induced to generate radiated field . using it one can treat the problem of electromagnetic interaction for the incidence from either vacuum or a material .
actually , it is not difficult to obtain the reflected field and the transmitted field . using eqs .
( [ extinction_e_left ] ) , ( [ extinction_e_right ] ) , ( [ extinction_e_0 ] ) , and ( [ extinction_e_n+1 ] ) ( eqs . ( [ extinction_e_left ] ) , ( [ extinction_e_right ] ) , ( [ e_r_vacuum ] ) , and ( [ e_t_vacuum ] )
if the associated region is vacuum ) , one can derive the mutual relationship of @xmath79 in all regions @xmath80 the transformation matrix @xmath81 for te waves is @xmath82 } & r_{n{(n+1)}}\exp{[-i(k_{nfz}+k_{(n+1)fz})z_n]}\\ r_{n{(n+1)}}\exp{[i(k_{nfz}+k_{(n+1)fz})z_n ] } & \exp{[i(k_{nfz}-k_{(n+1)fz})z_n ] } \end{array}\right),\nonumber\\\end{aligned}\ ] ] where @xmath83 note that @xmath84 .
then , with eqs .
( [ e_i_r ] ) , ( [ e_t ] ) and ( [ e_n_n+1 ] ) we finally obtain the real reflected and transmitted fields in individual layers .
the magnetic fields can be obtained just as the corresponding electric fields after replacing the function @xmath85 with @xmath86 , @xmath87 where @xmath88/\omega$ ] . by now
, we have derived in a new way the reflected / transmitted fields and the reflection / transmission coefficients in stratified media .
although involving some seemingly complicated integrals that nevertheless have simple solutions , the iem leads almost miraculously to the correct results . obviously , eqs .
( [ e_n_n+1 ] ) and ( [ r_te ] ) are in agreement with the results obtained by the formal approach of maxwell theory @xcite .
it seems from eqs .
( [ extinction_e_left ] ) , ( [ extinction_e_right ] ) , ( [ extinction_e_0 ] ) , or ( [ extinction_e_n+1 ] ) as if the extinction only occurs on the interfaces @xcite
. in fact , however , the vacuum fields are extinct at every points in all regions , as shown evidently in eqs .
( [ e_n ] ) , ( [ e_0 ] ) or ( [ e_t ] ) . in the above analyses of this section
we do not presume any boundary condition .
instead , one can find that the result of eq .
( [ e_n_n+1 ] ) plays the role of boundary conditions in maxwell theory @xcite .
as an application of the iem , we discuss the case of a three - layer configuration . for the case of @xmath89 , by eqs .
( [ e_i_r ] ) and ( [ e_t ] ) we obtain the reflection coefficient @xmath90 and the transmission coefficient @xmath91 for te waves @xmath92 to further demonstrate the validity of the iem , we calculate the propagating fields in photon tunneling and compare the results by iem with those by solving maxwell equations plus boundary conditions @xcite .
consider a plane wave incident through a slab sandwiched between two prisms at the angle larger than critical angle .
we calculate the transmitted field in the second prism by the iem and the maxwell approach , denoted by @xmath93 and @xmath94 respectively , as shown in fig .
[ bket ] .
the field magnitude as a function of the width of the slab @xmath95 .
a te - polarized plane wave is incident on the air gap sandwiched by two prisms with the dielectric constant @xmath96 at the angle @xmath97 .
, width=302 ] from that figure , we can see the two results overlap each other , indicating that the iem is consistent with maxwell approach and thus is valid . in order to disclose the physical mechanism of photon tunneling ,
we calculate the radiated fields @xmath98 from the left to the second prism , which is the source of the transmitted field .
one can see that the amplitude of @xmath99 decreases exponentially as the width of the slab increases .
such a trend also occurs to @xmath100 .
this phenomenon can be understood easily as the transmitted field results from the radiated field induced by @xmath101 .
especially , we find that when @xmath98 approaches zero , the transmitted field is also close to zero .
in other words , photon tunneling can not occur without the radiated field @xmath99 from the left region . actually , detailed analyses show that the physical process of photon tunneling is that the vacuum field @xmath101 induce the dipoles in the second prism to radiate two kinds of waves ; one of the radiated fields extinct @xmath101 and the other forms the final transmitted field , leading to photon tunneling . based on the above results , we make numerical simulations of photon tunneling for slabs with different widths in fig .
[ beam ] . for completeness , we give results for both positive and negative indices .
field distributions for a gaussian beam through slabs with different widths @xmath95 .
( a ) @xmath102 , ( b ) @xmath103 , ( c ) @xmath104 , ( d ) @xmath105 for the slab with @xmath106 and @xmath107 ; ( e ) @xmath102 , ( f ) @xmath103 , ( g ) @xmath104 ; ( h ) @xmath105 for the slab with @xmath108 and @xmath109 .
the slab is sandwiched by two prisms with the dielectric constant @xmath96 and the incident angle is @xmath97 .
, title="fig:",width=151 ] field distributions for a gaussian beam through slabs with different widths @xmath95 .
( a ) @xmath102 , ( b ) @xmath103 , ( c ) @xmath104 , ( d ) @xmath105 for the slab with @xmath106 and @xmath107 ; ( e ) @xmath102 , ( f ) @xmath103 , ( g ) @xmath104 ; ( h ) @xmath105 for the slab with @xmath108 and @xmath109 .
the slab is sandwiched by two prisms with the dielectric constant @xmath96 and the incident angle is @xmath97 .
, title="fig:",width=151 ] field distributions for a gaussian beam through slabs with different widths @xmath95 .
( a ) @xmath102 , ( b ) @xmath103 , ( c ) @xmath104 , ( d ) @xmath105 for the slab with @xmath106 and @xmath107 ; ( e ) @xmath102 , ( f ) @xmath103 , ( g ) @xmath104 ; ( h ) @xmath105 for the slab with @xmath108 and @xmath109 .
the slab is sandwiched by two prisms with the dielectric constant @xmath96 and the incident angle is @xmath97 .
, title="fig:",width=151 ] field distributions for a gaussian beam through slabs with different widths @xmath95 .
( a ) @xmath102 , ( b ) @xmath103 , ( c ) @xmath104 , ( d ) @xmath105 for the slab with @xmath106 and @xmath107 ; ( e ) @xmath102 , ( f ) @xmath103 , ( g ) @xmath104 ; ( h ) @xmath105 for the slab with @xmath108 and @xmath109 .
the slab is sandwiched by two prisms with the dielectric constant @xmath96 and the incident angle is @xmath97 .
, title="fig:",width=151 ] field distributions for a gaussian beam through slabs with different widths @xmath95 .
( a ) @xmath102 , ( b ) @xmath103 , ( c ) @xmath104 , ( d ) @xmath105 for the slab with @xmath106 and @xmath107 ; ( e ) @xmath102 , ( f ) @xmath103 , ( g ) @xmath104 ; ( h ) @xmath105 for the slab with @xmath108 and @xmath109 .
the slab is sandwiched by two prisms with the dielectric constant @xmath96 and the incident angle is @xmath97 .
, title="fig:",width=151 ] field distributions for a gaussian beam through slabs with different widths @xmath95 .
( a ) @xmath102 , ( b ) @xmath103 , ( c ) @xmath104 , ( d ) @xmath105 for the slab with @xmath106 and @xmath107 ; ( e ) @xmath102 , ( f ) @xmath103 , ( g ) @xmath104 ; ( h ) @xmath105 for the slab with @xmath108 and @xmath109 .
the slab is sandwiched by two prisms with the dielectric constant @xmath96 and the incident angle is @xmath97 .
, title="fig:",width=151 ] field distributions for a gaussian beam through slabs with different widths @xmath95 .
( a ) @xmath102 , ( b ) @xmath103 , ( c ) @xmath104 , ( d ) @xmath105 for the slab with @xmath106 and @xmath107 ; ( e ) @xmath102 , ( f ) @xmath103 , ( g ) @xmath104 ; ( h ) @xmath105 for the slab with @xmath108 and @xmath109 .
the slab is sandwiched by two prisms with the dielectric constant @xmath96 and the incident angle is @xmath97 .
, title="fig:",width=151 ] field distributions for a gaussian beam through slabs with different widths @xmath95 .
( a ) @xmath102 , ( b ) @xmath103 , ( c ) @xmath104 , ( d ) @xmath105 for the slab with @xmath106 and @xmath107 ; ( e ) @xmath102 , ( f ) @xmath103 , ( g ) @xmath104 ; ( h ) @xmath105 for the slab with @xmath108 and @xmath109 .
the slab is sandwiched by two prisms with the dielectric constant @xmath96 and the incident angle is @xmath97 .
, title="fig:",width=151 ] from fig .
[ beam ] , one can see that the photon tunneling mostly occurs as the slab is thinner than a wavelength and the transmitted field increases as the slab width decreases . at the same time ,
positive or negative goos
hnchen shifts are confirmed in photon tunneling as observed in @xcite .
in summary , we have extended the iem to investigate the propagation of electromagnetic waves through stratified anisotropic dielectric - magnetic media . from a microscopic viewpoint in which the matter consists of molecular scatters , we used the superposition principle and the integral formulation of hertz vectors to analyze the interaction of field with matter .
we presented new derivations of the dispersion relation , snell s law and the reflection / transmission coefficients by self - consistent analyses . applying the iem
, we investigate the wave propagation through a slab and discuss the physical process of photon tunneling
. verified by numerical simulations , the results are in agreement with those obtained by macroscopic maxwell theory and disclose the underlying physics of wave propagation through stratified materials , which is the emphasis of the present paper .
moreover , we found two new forms of the extinction theorem generalized to incidence not from vacuum but from a material , which may enrich the theory of molecular optics @xcite .
we also demonstrated that the extinction occurs at an arbitrary location inside the materials , not just on the interfaces @xcite .
obviously , we recovered previous results for materials of positive refraction index @xcite .
more importantly , our results lead to a unified framework of the iem for wave propagation in stratified dielectric - magnetic materials .
the unified iem can be applied to other problems of propagation through dielectric - magnetic materials , including metamaterials .
this is because the units of metamaterials , whose dimensions are much less than the wavelength , react to exciting fields just as molecular scatters in conventional media @xcite . especially , the method can treat propagations of wave incident either from a medium or vacuum , in contrast with previous work @xcite in which only the incidence from vacuum is considered .
besides , the same framework not only enables the handling of arbitrary shaped objects made of continuous matter , but allows one to treat discrete particles . the associated coupling can be included in the iem without any formal difficulty @xcite .
it is hoped that the framework and the microscopic perspective of the iem may enable more unique properties in metamaterials to be understood and more applications to be developed . | we investigate the propagation of electromagnetic waves in stratified anisotropic dielectric - magnetic materials using the integral equation method ( iem ) .
based on the superposition principle , we use hertz vector formulations of radiated fields to study the interaction of wave with matter .
we derive in a new way the dispersion relation , snell s law and reflection / transmission coefficients by self - consistent analyses .
moreover , we find two new forms of the generalized extinction theorem . applying the iem
, we investigate the wave propagation through a slab and disclose the underlying physics which are further verified by numerical simulations .
the results lead to a unified framework of the iem for the propagation of wave incident either from a medium or vacuum in stratified dielectric - magnetic materials . |
over the past two decades , high - resolution observations with the _ hubble space telescope _ have shown that massive stellar clusters reside at the photometric and dynamical centers of most intermediate and low luminosity galaxies of all hubble types @xcite . with sizes in the range @xmath1 to @xmath2pc and masses in the range @xmath3 to @xmath4 , these nuclear clusters ( ncs )
have central densities up to @xmath5 , making them the densest stellar systems observed ( e.g. , * ? ? ?
* ; * ? ? ?
nscs are observed to be larger and brighter , and to follow different structural scaling relations than globular clusters .
the nsc half light radii scale with their total mass roughly as @xmath6 , while globular clusters have @xmath7 irrespective of their luminosity , albeit with a large scatter @xcite .
however , the nsc mass distribution overlaps with that of globular clusters at its low mass end , suggesting a possible connection between the two types of stellar clusters ( e.g. , * ? ? ?
* ; * ? ? ?
nscs have been studied in detail for large samples of galaxies in different environments .
observational work has revealed that the nucleation fraction of galaxies is similar for galaxies belonging to very different environments , including the virgo , fornax , and coma clusters as well as galaxies in the field @xcite . on average @xmath8 of all galaxies of all hubble types with magnitude in the range @xmath9
contain a well defined central cluster .
hence nscs appear to be ubiquitous structures in galaxies although they tend to be missing in galaxies brighter than magnitude @xmath10 and fainter than @xmath11 ( e.g. , * ? ? ?
* ; * ? ? ?
the real nucleated fraction could be however larger than the value constrained from observations given that the morphological complexity and high surface brightness often characterizing the center of galaxies can make nscs difficult to identify , especially in massive spheroids .
high resolution spectroscopic surveys have been used to place constraints on nsc ages and star formation histories @xcite .
the common finding emerging from these studies is that nscs are characterized by a complex star formation history with a mixture of morphological components and different stellar populations spanning a wide range of characteristic ages from @xmath2myr to @xmath2gyr .
cluster ages and masses are also found to depend on the host galaxy hubble type , with nscs in early - type spirals being older and more massive than those of late - type spirals .
the luminosity - weighted ages of nscs are typically smaller than the ages of host galactic disks , indicating that the growth of the nuclei is a continuous and ongoing process occurring during and after most of the host galaxy was formed @xcite .
age estimates for most nscs remain however susceptible to large uncertainties due to the fact that the light is often dominated by young stellar populations while at the same time the mass is dominated by the old stars .
the inability to infer spectroscopically the age of old cluster components can introduce a bias toward younger ages .
the bulk of the stellar population is typically found in an older and spatially more extended stellar component @xcite .
this appears to be the case for the milky way in which more than @xmath12 of the stars inside the inner parsec formed more than @xmath13 ago , while luminous massive stars dominate the central cluster light within @xmath14pc @xcite .
nsc old and young stellar populations also differ morphologically .
@xcite showed that the three edge - on late - type galaxies ic 5052 , ngc 4206 , and ngc 4244 have nuclei that are strongly flattened along the plane of their host galaxy disks . such nscs show evidence for young stellar components distributed onto a disk - like or ring structure superimposed on an older more spherical component . in the edge - on spiral galaxy ngc 4244 , old and young stellar components both show clear signs of rotation @xcite .
@xcite suggested that the presence of such multiple rotating morphological components and the presence of a flattened young cluster aligned with the major axis of the host galaxy point toward an in situ build up of nuclear clusters , whereby stars form episodically in compact nuclear disks and then lose angular momentum or heat vertically to form an older spheroidal structure .
it has been noted however that both rotation and the presence of young structural components does not exclude other formation mechanisms such as for example episodic accretion of young star clusters in the central part of the galaxy due to dynamical friction @xcite . in the case of ngc 4244 for example
, it is likely that accreation of star clusters has contributed at least in part to the growth of its nsc @xcite .
since the early studies , it has been realized that nsc masses correlate fairly well with galaxy properties such as bulge velocity dispersion , bulge and galaxy total luminosity @xcite .
the existence of such correlations suggests that the formation of nscs is intimately connected to the formation and evolution of the host galaxy ( e.g. , * ? ? ?
given that nscs and mbhs are found to coexist in some galaxies , that in these systems they have comparable masses , and that both follow tight correlations with galaxy properties , it is also natural to ask whether nscs and mbhs are somehow connected to each other .
nscs and mbhs are known to co - exist in galaxies with masses @xmath15 @xcite ; galaxies with masses lower than this value show clear evidence for nucleation but little evidence for a mbh .
conversely galaxies with masses above @xmath16 are dominated by mbhs but show no evidence for nucleation ( e.g. , * ? ? ?
* ; * ? ? ?
@xcite found that the nuclei and mbhs obey a similar scaling relation linking their mass to the virial mass of the host galaxy .
more recently , @xcite obtained nsc host galaxy correlations using a sample in which he excluded what were likely to be nuclear stellar discs from the galaxy sample of ferrarese et al . , while including an additional 13 nscs in galaxies with velocity dispersions out to about @xmath17 .
graham found that nsc and mbh masses follow different scaling correlations with host galaxy properties .
other authors confirmed this result , showing that the mass of nscs scales with the host - galaxy spheroid s velocity dispersion as @xmath18 , while the mass of mbhs follows the much steeper relation @xmath19 @xcite .
although this might suggest that mbhs and nscs did not form from the same mechanism @xcite , whether the formation of mbhs is connected to the evolution of nscs and whether the two types of central objects grow together or in competition from the same physical process remains unclear .
the observational findings mentioned above provided motivation for theoretical work aimed to understand how nscs form and how their evolution is linked to the evolution of their host galaxy .
two models have been suggested for the formation of nscs : ( i ) the cluster infall scenario , in which stellar clusters are driven to the galactic nucleus by dynamical friction , merge , and build up a nsc @xcite ; ( ii ) the nuclear star formation scenario , in which gas falls into the nucleus and forms stars ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . due to the inherent complexity of gas dynamics in star formation , which makes the second of these two processes difficult to model ,
theoretical work has been mostly directed toward the cluster infall scenario ( but see * ? ? ?
* ) . however , both dissipative and dissipationless processes are likely to play an important role in nsc formation @xcite .
theoretical studies have employed two different methodologies : @xmath20-body simulations and semi - analytical modeling .
@xmath20-body techniques are used to simulate the last stage of inspiral and merger of stellar clusters in the inner region of galaxies .
these models have shown that a star cluster merger scenario can explain without obvious difficulties the observed properties of nscs , including their density and velocity dispersion profiles ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
while these studies make reliable predictions about the aspect of nscs , they suffer from the fact that the adopted initial conditions are often not well motivated , hence the need to rely on semi - analytical models .
these are used to make predictions for the appearance of star clusters , of known mass and radius , in the center of galaxies , and to compute the accumulated mass at the center through dynamical friction migration over a hubble time . in semi - analytical models ,
the rate at which the nsc grows by accreting young clusters is estimated from empirical cluster formation rates , dynamical friction timescales and dissolution times ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
studies based on semi - analytical approaches have demonstrated that the nsc host galaxy property scaling relations and their half - mass radius mass relation are both consistent with formation by star cluster accretion @xcite .
all previous calculations assumed nsc formation to take place in isolated galactic spheroids , thus neglecting the role of galaxy evolution , mergers , and the role of in - situ star formation .
also , these former idealized attempts could not explore the details of the interplay between mbh and nsc evolution . in this paper
, we present a semi - analytical galaxy - formation model that allows us to shed light on exactly these points , i.e. it allows us to assess the role of galaxy mergers , mbh mergers and nuclear star formation on the growth of nscs .
we follow the formation and evolution of galaxies , mbhs and nscs along cosmic history , including the growth of nscs due to both central migration of stellar clusters and in - situ star formation , while also accounting for dynamical heating from ( binary ) mbhs .
the paper is organized as follows . in section
2 we introduce and discuss the numerical methods employed in our study .
section 3 describes the sample data to which our numerical results are compared . in sections 4 and 5
we describe the main results of our calculations and discuss some of their implications in section 6 .
we summarize in section 7 .
we study the formation of nscs and their co - evolution with mbhs along the merger history of their host galaxies by semianalytical techniques . first , we consider a purely dissipationless formation model in which stellar clusters form and migrate to the center of a galaxy through dynamical friction . we also consider a second semi - analytical model , which unlike the first follows the formation and merger history of galaxies ( both in their dark - matter and baryonic components )
this allows us to study the formation , evolution and disruption of nscs in a self - consistent way , accounting not only for the dynamical friction - driven inspiral of stellar clusters to the nucleus , but also for star formation in the nuclear gas , and for the effect of galaxy and mbh mergers .
we begin in this section by describing our methodology , highlighting in particular the simplifying approximations that are made in our models . in our semi - analytical model
@xmath21 a galactic nucleus forms through the orbital decay ( via dynamical friction ) and merger of star clusters in the central region of a galaxy .
this code adopts a rather idealized model of an isolated galactic spheroid and follows the consecutive inspiral of stellar clusters along their orbits until they decay into the galactic center or they are disrupted in the process .
this relatively simple approach has the advantage that the details of the tidal interaction of the clusters with the background galaxy and central mbhare easily included , and in a relatively precise manner . on the other hand ,
as previously stated , this model has the important limitation that galaxy evolution as well as dissipative processes that can lead to star formation episodes in the galactic center are neglected .
another basic assumption made in the @xmath21 model is that mbhs are already in place before nscs grow around them ; moreover we assume that all galaxies contain initially a mbh .
the code is essentially the same as that used by @xcite , and we direct the reader to that paper for a more detailed description . briefly , we represent the galaxy spheroid by a simple power - law density model : @xmath22 , where @xmath23 .
this latter expression assumes that the density of the galactic bulge follows a @xcite profile , @xmath24 , at @xmath25 .
given a mass for the central mbh , @xmath26 , we compute the galaxy density profile slope , scale radius , spheroidal mass and galaxy velocity dispersion by using the corresponding scaling relations that link these properties to @xmath26 @xcite . the galaxy effective radius , @xmath27 , was derived from the size - mass relation given by equation ( 32 ) of @xcite .
the ratio @xmath28 follows from equations ( 4 ) and ( 17 ) of @xcite , which give @xmath29 for @xmath30 .
we assign the stellar cluster masses using the cluster initial mass function ( cimf ) , @xmath31 @xcite , and limiting mass values of @xmath32 , @xmath33 .
the stellar clusters are assumed to form continuously over the age of the galaxy . as commonly done , we assume that the clusters have initially the same distribution as the stars in the galaxy , and that initially a fixed fraction @xmath34 of stars form in strongly bound stellar clusters .
this latter is approximately consistent with the typical cluster formation efficiency for the milky way found by @xcite and it is similar to the value adopted in previous studies @xcite .
gnedin et al .
showed that by adopting a power - law mass function @xmath31 and a fraction @xmath35 of clustered star formation , one can reproduce the observed density profile of clusters in m87 .
we assume that the central properties of a stellar cluster remain unchanged during inspiral and that @xmath36 , where @xmath37 is the cluster core radius and @xmath38 is the cluster tidal ( limiting ) radius given by @xcite : @xmath39 with @xmath40 the sum of galactic , mbh and nsc potentials , and @xmath41 a `` form factor '' that depends on the density distribution within the cluster .
the mass of a king model is then related to its tidal radius and velocity dispersion via the expression @xmath42 we note that although at first order the central properties of the clusters are not expected to change during inspiral , the stars in a nsc formed from cluster infalls should arguably end up with the same root - mean - square velocity as the galaxy host , which appears to be consistent with observations ( see * ? ? ?
. given a cluster of central velocity dispersion @xmath43 the time evolution of its orbital radius is @xmath44^{\frac{1}{3-\gamma}}\end{aligned}\ ] ] where @xmath45 is the initial separation from the center , for the coulomb logarithm we take @xmath46 , and @xmath47 is given in equation ( 1 ) of @xcite .
we approximately account for the dynamical dissolution of the clusters due to their collisional evolution and interaction with the external tidal field of the galaxy by only evolving clusters with dynamical friction time shorter than their dissolution time [ equation ( [ diss - t ] ) below ] .
the cluster orbits are evolved using equation ( [ sfc2 ] ) up to a maximum integration time of @xmath48 , and the nsc mass is obtained by summing up all the mass transported by the infalling clusters within a radius @xmath49 .
the nsc radius , @xmath49 , is computed from the recursive relation given by equations ( 14)-(16 ) of @xcite . as we evolve the cluster orbits , we compute their tidal disruption radius due to the galaxy , mbh and a pre - existing nsc gravitational fields , adopting a cluster core radius @xmath50 , roughly equal to the median value of the core radii listed in the harris compilation of galactic star clusters @xcite .
when the clusters reach their tidal disruption radius , the remaining core mass is dispersed around the radius of disruption .
when evaluating the final nsc density distribution we make the assumption that the stars from the disrupted clusters are isotropically distributed around the center .
we note that in reality the clusters will distribute initially in a disk - like structure , but this will morph into a more spherical distribution over a fraction of the nuclear relaxation timescale @xcite . a mbh binary formed during a galaxy merger
leaves an imprint on the galactic nucleus in the form of a mass deficit , @xmath51 , a decrease in the mass of the nucleus due to ejection of stars on intersecting orbits @xcite .
we account for this by subtracting a mass @xmath52 @xcite , where @xmath53 is the predicted number of major mergers after the last major accretion event , from the final accumulated mass inside @xmath54 .
the number of mergers @xmath53 is obtained from the galaxy mass using the distributions shown in figure ( 2 ) of @xcite , where in order to convert @xmath55luminosity to mass we used a constant mass - to - light ratio @xmath56 , as sometimes adopted in the literature for an old stellar population ( e.g. , * ? ? ? * ) .
finally we compare the resulting nsc density profile , obtained after @xmath2gyr of evolution , to the density profile , @xmath57 , of the galaxy , and define a galaxy as _ nucleated _ if at some radius the nsc stellar density was found to be higher than that of the background galaxy .
this latter condition is based on the fact that if the nsc densities are below the galaxy density at all radii , then it would be difficult to observationally distinguish the two components . our galaxy formation model ( hereafter @xmath58 )
is based on that of @xcite , which was further improved in @xcite .
more precisely , @xcite presented a semi - analytical galaxy formation model tracking the evolution of baryonic structures along dark matter merger trees .
these trees are produced with an extended press - schechter formalism , modified in order to reproduce the results of n - body simulations of dark - matter halos @xcite , while the baryonic structures include the hot , largely unprocessed intergalactic medium , galactic disks and spheroids ( in both their stellar and cold , chemically - enriched interstellar medium components ) , a low angular momentum reservoir of cold nuclear gas available for accretion onto the central mbh @xcite , and the mbh itself .
these baryonic components are interconnected by a plethora of interactions , schematically summarized in figure [ fig_model ] .
also included are environmental and tidal interactions between galaxies during mergers , following simple semi - analytical recipes @xcite .
moreover , @xcite later improved this model by refining the star formation law adding in particular an explicit dependence on the metallicity @xcite , as well as a distinction between pseudo - bulges forming from bar instabilities and classical bulges created by major mergers and by devising more realistic prescriptions for the evolution of the spins of mbhs with cosmic time . for this work ,
we have further ameliorated the model by including the formation and evolution of nscs ( cf .
figure [ fig_model ] ) .
more precisely , we consider two possible formation channels for these objects , namely one where nscs form from the dynamical friction - driven migration to the nuclear regions of star clusters created further out in the galaxy , and one in which nscs form _ in situ _ as a result of star formation in the nuclear regions .
note that we assume no high redshift seeds for the nscs , i.e. the nsc mass grows from zero at high redshifts to its present - day value through the two aforementioned channels ( this growth being modulated / hindered by the effect of galaxy and black - hole mergers , cf .
section [ sec : mergers ] ) . in the @xmath58 models described below
we define as nucleated any galaxy containing a central cluster more massive than @xmath59 , as this value corresponds approximately to the lower limit of the observed nsc mass distribution ( e.g. , * ? ? ?
we also define an early- ( late-)type galaxy as one with bulge to total mass ratio @xmath60 ( @xmath61 ) . for each of the @xmath58 models that we present , we simulate about 1300 galaxies with dark - matter masses ranging from @xmath62 to @xmath63 . to implement the dynamical friction - driven migration of star clusters to the central nucleus
, we first assume that star clusters are created during star - formation events with efficiency @xmath64 , both in galactic bulges and disks . in bulges , @xcite and
@xcite assumed a volumetric star - formation rate @xmath65 , while in disks they considered a vertically - averaged rate of star formation @xmath66 .
therefore , we assume that star clusters form in bulges and disks , and that the total mass of the star - cluster population ( in both the bulge and disk ) , @xmath67 , changes with rate @xmath68 where we choose @xmath69 , which is the typical value in the milky way @xcite .
we stress that plausible values
@xmath34 @xmath70 would only mildly impact the normalization of our results for the @xmath71 relation , and would not affect its slope significantly .
in addition we have tried a model with a variable @xmath64 set to 0.07 , 0.04 and 0.5 in disk , quiescent and starburst galaxies respectively ; these correspond approximately to the values of figure 4 in @xcite .
the results of this model were found to not significantly differ from those of models with a fixed @xmath64 .
thus , for the sake of simplicity , we present in what follows only the results of a model with fixed @xmath69 .
we then assume that the star clusters formed in the bulge are spatially distributed in the same way as the bulge stellar population , i.e. if that population has density profile @xmath72 , the volumetric number density @xmath73 of star clusters is assumed to satisfy@xmath74 where @xmath75 is the total number of star clusters in the bulge , and @xmath76 is the total bulge mass in stars .
note that following @xcite , we assume a hernquist distribution for @xmath72 ; we refer to @xcite for more details on the choice of the scaling radius for this density profile .
similarly , we assume that the disk s star clusters are distributed according to the same surface density profile as the overall stellar content of the disk , i.e. the superficial number density @xmath77 will satisfy @xmath78 where again @xmath79 is the total number of star clusters in the disk , and @xmath80 is the mass of the stellar disk .
again , for @xmath81 we follow @xcite and adopt an exponential profile ( see @xcite for more details on the calculation of the scaling radius for this distribution ) . as for the mass distribution of the newly formed star clusters ,
we assume a power - law mass function @xmath82 ( with @xmath83 bulge , disk ) .
combined with equations and , this gives the distribution functions for the bulge and disk star clusters @xmath84 where @xmath85 and @xmath86 are units of volume and surface .
we assume that these distribution functions are valid for masses @xmath87 for individual star clusters between @xmath88 and @xmath89 @xmath90 , and the normalization factor @xmath91 ensures that the integral over all masses and over the whole volume ( or surface ) is 1 .
note that in our runs we assume @xmath92 ; using the dissipationless model described in section [ mod1 ] , we find that reducing @xmath93 to @xmath94 has the effect of changing the normalization of the scaling relations by a factor @xmath95 , but did not change their overall slopes .
these distribution functions can then be used to calculate approximately how much mass is lost by the population of star clusters due to their collisional evolution and interaction with the external tidal field of the galaxy ( `` dynamical dissolution '' ) : @xmath96 with @xmath97 and @xmath98 @xmath99 being the halo s virial velocity ( e.g. , * ? ? ? * ) .
recent @xmath20-body simulations show that @xmath100 @xcite , which we adopt here . also , note that for simplicity we do not track separately the mass in star clusters in the bulge and that in the disk , but simply follow the evolution of the total mass in star clusters @xmath67 [ cf .
equation ] . in equation
[ and in equations and below ] we therefore simply assume that the mass of star clusters in the disk is @xmath101 m_{\rm gc}$ ] , and that the mass of star clusters in the bulge is @xmath102 m_{\rm gc}$ ] . in a similar fashion
, we account for the evaporation of star clusters in isolation through @xmath103 where @xmath104 note that because the evaporation time - scale does not depend on the star cluster s location ( since it describes evaporation in isolation ) , equation can be simplified to @xmath105 note that our models do not include the possibility of close encounters with giant molecular clouds , which in some cases could significantly reduce the clusters lifetime .
also they rely on necessarily simplified models for the galaxy . a more realistic model for the galactic potential might somewhat change the timescales introduced above
. however , we note that analytical models of nsc formation indicate that the final nsc mass is not much affected by the assumed slope of the galactic background density profile ( see section 4 of * ? ? ? * ) .
the total mass in star clusters also decreases because the individual star clusters fall toward the central nsc , under the effect of dynamical friction , i.e. @xmath106 the dynamical - friction timescale in the bulge is dominated by the interaction with the stellar background , and we therefore follow binney and tremaine ( 2008 ) [ equation ( 8.12 ) ] : @xmath107 where @xmath108 is the bulge velocity dispersion ( related to the halo s virial velocity by @xmath109 @xcite ) and we choose @xmath110 . as for the dynamical friction in the disk , we account for both the interaction with the stellar and gas component , i.e. @xmath111 where @xmath112 is the orbital angular moment of a star cluster of mass @xmath113 belonging to the disk ( and thus moving with velocity @xmath114 at a separation @xmath115 from the galactic center ) , while the angular momentum loss due to dynamical friction is expressed @xcite in terms of the surface densities @xmath116 and @xmath117 of the disk s stellar and gaseous components , the velocity dispersion of the stars in the disk @xmath118 , and the velocity dispersion / sound speed of the gas in the disk @xmath119 @xcite .
note that for @xmath118 we follow @xcite and assume @xmath120 , where @xmath121 is the maximum rotational velocity inside the disk . also observe that in equation we are summing the inverses of the timescales , so that the infall rates due to the dynamical friction from the gas and that due to the stars get summed in equation .
not all of the mass of the star clusters falling toward the nsc eventually accretes onto it , because star clusters get truncated due to the galactic tidal field , and tidally disrupted by the central mbh . to account for these effects
we assume that the rate of change of the nsc mass due to infall of star clusters is @xmath122 where clearly the major difficulty lies in computing the factor @xmath123 .
since it would be computationally prohibitive to evolve the infall of the individual star clusters self - consistently within our semi - analytical galaxy formation model , we calculate @xmath124 with the `` monolithic '' @xmath21 model of section [ mod1 ] , in order to derive an easy - to - use analytical expression . more specifically , approximating a star cluster as a king model ,
its tidal radius at a distance @xmath49 from the center of a galaxy containing a mbh and a nsc at its center is : @xmath125^{-1/2}~ , \nonumber\end{aligned}\ ] ] where @xmath126 is the total mass of the central massive objects ( i.e. , mbh plus nsc mass ) , and for the galaxy density profile we adopted the power - law model @xmath127 . note that to be consistent with the hernquist profile used in @xmath58 , we set @xmath128 in equation . also , the distance @xmath49 is to be set to the outer radius of the nsc , i.e. the distance from the galaxy center below which the star cluster is assumed to have become part of the nsc .
we assume @xmath129 , where @xmath130 is the nsc half - mass radius .
our detailed prescription for @xmath130 will be presented in equation below . for a king model , the truncated mass @xmath131 of the star cluster
is then related to its tidal radius via equation ( [ mt0 ] ) . using the @xmath21 model described in section [ mod1 ]
, we find that more than @xmath132 of the nsc mass comes from star clusters with initial mass @xmath133 , and that the ( mass - weighted ) mean central velocity dispersion of the star clusters contributing to the nucleus growth is @xmath134 km / s .
when setting @xmath135 , the average initial mass of these star clusters is @xmath136 .
assuming that all star clusters that decay to the center have similar central properties ( @xmath137 ) , and that these properties do not change during the infall , we have @xmath138 where @xmath139 is given by equations and . finally , to account for the possibility that a star cluster may undergo complete tidal disruption _ before _ it decays to a radius @xmath49 , we compare the tidal radius given by equation with the core radius @xmath140 pc of the star cluster .
if @xmath141 , we set @xmath142 . as mentioned above , nscs may also form by star formation in the cold gas accumulating in the galactic center . in the galaxy formation model of @xcite ( and in the improved version of this model used by @xcite ) , transfer of cold gas to a low - angular momentum nuclear reservoir
available for accretion onto the central mbh is assumed to be correlated with star - formation events in the galactic bulge through a relation @xcite @xmath143 where @xmath144 is the star formation rate in the gaseous bulge , and @xmath145
@xmath146 is a free parameter , which we set to @xmath147 in this paper ( as in @xcite ) . in the model of @xcite and @xcite , star formation in the bulge
is associated with major galactic mergers , and ( less importantly ) with bar instabilities in the galactic disk .
note that the physical mechanism responsible for the loss of angular momentum of the cold gas and its transfer to this nuclear reservoir may be the radiation drag caused by the newly formed stars @xcite , or the reshuffling / shocks of the gas as a consequence of disk instabilities or mergers . more generally , a correlation between bulge star formation and mbh accretion / growth is expected based on the @xmath148 relation for mbhs and on the parallelism between the quasar luminosity and star formation history @xcite .
the reservoir s gas is then made available to accrete onto the mbh on the viscous timescale . whenever the resulting viscous accretion rate exceeds @xmath149 ( @xmath150 being the eddington mass accretion rate and @xmath151 a free parameter ) , we truncate the accretion rate to that value . to allow for moderately super - eddington accretion in the case in which mbhs form from light popiii star seeds ( of mass @xmath152 at @xmath153 ) @xcite , we set @xmath154 for that seed model .
this is because some amount of super - eddington accretion is known to be needed to reconcile light mbh seeds with the quasar luminosity function at high redshift @xcite .
we also consider two `` heavy '' seed models ( namely that of @xcite and that of @xcite ) whereby mbhs grow from seeds of @xmath155 at @xmath153 , in which case we set @xmath156 . on top of these prescriptions , in this paper we assume that the gas in the nuclear reservoir also forms stars , which are assumed to contribute to the nsc . to compute the star - formation rate , we need to choose a size for the reservoir .
it is natural to assume that this size is comparable to the observed size of nscs .
more specifically , we assume that the reservoir is disk - like , and has a ( vertically - averaged ) exponential density profile , whose scale radius we choose such that the reservoir has the same half - mass radius @xmath157 as the nsc .
the latter is assumed to be @xmath158 with @xmath159 the dynamical mass of the nucleus ( including the gas in the nuclear region ) .
note that this scaling is inspired by observations of the size of nscs , which is found to scale with the square root of the luminosity @xcite .
once a density profile for the reservoir is specified , we apply a star formation law similar to that used by @xcite for galactic disks .
note that @xcite improved on @xcite by considering different modes of star formation in classical bulges forming from major mergers where star formation is assumed to take place in `` bursts '' of duration comparable to the bulge dynamical time as opposed to disks and pseudobulges arising from bar - instabilities of disks where star formation is assumed to happen in a `` quiescent '' mode , described by the prescription of @xcite . since star formation in the central region of our galaxy
is known to be weaker by at least a factor ten relative to what would be expected based on the observed gas densities @xcite , it seems more appropriate to use the `` quiescent '' star - formation mode for our nuclear reservoir . that the reservoir s feeding is triggered by star - formation events in the bulge , i.e. mainly by turbulence - driven bursts of star formation following major mergers .
] following therefore @xcite and @xcite , we assume that the star formation in the reservoir takes place on a timescale @xmath160 and involves only a fraction @xmath161 of the reservoir s gas , thus allowing us to write an approximate star formation rate in the reservoir as @xmath162 note that this simple prescription does not account for the spatial distribution of the star formation activity , e.g. the observational fact that younger stellar components seem to be more centrally concentrated than the rest of the nsc [ see for example @xcite ] .
the fraction @xmath161 of cold gas available for star formation is set by the fraction of molecular gas for metallicities @xmath163 ( in solar units ) , when star - formation happens in molecular clouds . at lower metallicities @xmath164 ,
star formation takes place in the atomic phase @xcite . in general
, @xmath161 is expected to decrease as the metallicity decreases , but recent observations of nearby spirals and dwarfs @xcite , as well as the small magellanic cloud @xcite show that it levels off at 2% @xcite .
we can thus summarize the dependence of @xmath161 on the metallicity by the explicit expression @xcite @xmath165^{-1/5 } , & ~{\rm ~if}~f_{\rm c } > 2\%,\\ 2\% , & ~{\rm otherwise } \end{array } \right .
\label{eq : fh2}\ ] ] with @xmath166 @xmath167 @xmath168 @xmath169 the timescale @xmath160 is instead given by @xcite @xmath170 with @xmath171 .
this expression is obtained by assuming that star formation happens in clouds , and the two branches appear according to whether the cloud density is determined by internal processes , or by the external pressure ( in galaxies with sufficiently high surface densities ) . also , note that the typical cloud mass does appear in this expression because it is related to the gas surface density by identifying it with the local jeans mass , and by assuming marginal gravitational stability of the reservoir . for the surface density of the reservoir , in both equation and
, we choose @xmath172 , @xmath173 being the scale radius of the reservoir s exponential surface density profile .
this is the central value of the density for an exponential surface density profile .
while this choice is quite arbitrary , we note that at high surface densities the combination @xmath174 has a weak dependence on @xmath175 , namely @xmath176 .
therefore , a different choice of @xmath175 by a factor 2 ( 5 ) only changes @xmath177 as given by equation by 0.1 ( 0.24 ) dex .
we stress that equation is admittedly a rough prescription for the star formation in the nuclear regions .
however , it seems to work reasonably well when compared to observations of the star formation rates in the central 500 pc of the milky way ( the `` central molecular zone '' ) .
indeed , in figure [ sf ] we compare the predictions of our star formation prescription [ where in equations and we identify @xmath175 with the surface gas density represented in the horizontal axis ] with measurements of the star formation in the central molecular zone at different angular scales ( and thus different average gas surface densities )
. fig2a.eps ( 16,43.5 ) a comparison between our star - formation prescription in the nuclear region denoted by a blue line , the width of the line representing the scatter of the predicted star - formation rate with metallicity , which we assume in this figure to vary between 1 and 4 solar units and the measured star formation in the milky way s central molecular zone ( cmz ) as reported by @xcite ; the average star formation rate in the central 150 pc of ngc 253 , as reported in table 6 of @xcite ; the milky way s star formation rate at galactocentric distance of @xmath178 kpc and that of ngc 6946 at galactocentric distances of @xmath179 and @xmath180 pc , as reported by @xcite in their figure 7 ; and the measured nuclear star formation in the star - forming galaxy sdp 81 at @xmath181 @xcite . for comparison , we also show by a dashed orange line the prediction of the kennicutt - schmidt law @xcite.,title="fig : " ] in the light of this reasonable agreement , we chose not to include any feedback effects ( agns or supernovae ) when considering star formation in the nuclear regions .
we stress however that both kinds of feedback are included when considering the bulge and disk star formation [ cf .
figure [ fig_model ] , and @xcite ] .
they therefore indirectly affect the star formation in the nuclear region , since the feeding of the reservoir depends on the larger - scale galactic evolution via equation .
[ [ delays - between - galaxy - and - black - hole - mergers . ] ] * delays between galaxy and black - hole mergers .
* + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + when two dark matter halos merge , the two galaxies , being more compact than the halos , initially retain their identity , and are slowly brought together by dynamical friction . during this evolution ,
environmental effects such as tidal stripping and tidal evaporation become important and contribute to remove mass from the smaller galaxy .
when the galaxies finally merge , within a few dynamical - friction time - scales after the halo merger , the mbhs of the two galaxies are expected to form a binary system .
this binary is slowly brought together by a variety of effects , i.e. dynamical friction at large separations , and at smaller @xmath182 pc separations by gas interactions ( if sufficient gas is present in the nuclear regions ) or by interaction with stars if the nuclear star cluster has a sufficiently dense core , or by triaxiality of the central potential . gas interactions ( and in particular planet - like migration ) might bring a mbh binary to merger on time - scales @xmath183 yr @xcite , while three - body interactions with stars might require timescales up to @xmath184 gyr @xcite or more .
in fact , it is not completely obvious that a mbh binary would merge at all ( within a hubble time ) in the presence of three - body interactions with stars alone .
this is known as the `` final - parsec '' problem @xcite , but recent numerical simulations suggest that in triaxial merger remnants ( such as those that would be expected from a recent galaxy merger ) mbh binaries might coalesce on timescales of a few gyr @xcite
. rotation in flattened galaxy models has also been suggested to help drive mbh binaries to coalescence , cf .
@xcite . also ,
if a mbh binary stalls at @xmath184 pc separation , a later galaxy merger may add another mbh to the system , forming a mbh triple .
at least for large @xmath185 mbh masses , triple mbh interactions may trigger binary mergers ( and possibly ejection of the smallest mbh ) on timescales @xmath186 yr @xcite .
because of the uncertainty about the mechanisms described above , we adopt here a set of minimal simplified prescriptions to estimate the `` delay '' between galaxy and mbh mergers .
these prescriptions are presented in the following . in gas - rich environments , defined by the criterion @xmath187
, we assume that the delay is simply given by the viscous time of the nuclear gas . at a distance @xmath115 from the central mbh ,
the viscous time is simply @xmath188 , @xmath189 being the kinematic viscosity .
the latter can be approximately expressed in terms of the gas velocity @xmath190 and the reynolds number @xmath191 as @xmath192 @xcite .
since the kinematic viscosity is huge for accretion disks , it is natural to assume that @xmath191 is the critical reynolds number @xmath193 marking the onset of turbulence .
indeed , laboratory experiments find that once this critical reynolds number is reached , the viscosity increases significantly @xcite .
assuming now that @xmath194 is the binary s influence radius and that @xmath195 , we can write the `` delay '' in gas - rich nuclear environments as @xcite @xmath196 where @xmath197 is the dynamical time at the influence radius . in this paper
we set @xmath198 .
this prescription does indeed yield delays @xmath199 yr in gas - rich environments , as expected . in gas - poor enviroments ( i.e. @xmath200 ) , three - body interactions with stars dominate and bring the mbhs together on a timescale @xcite @xmath201^{-1}\ , , \end{gathered}\ ] ] where @xmath202 , and @xmath203 is the radius at which gravitational - wave emission becomes dominant over three - body interactions with stars at driving the binary s evolution , i.e. @xmath204^{1/5}\,.\end{gathered}\ ] ] here , the density @xmath205 of the stellar background in which the binary moves is given by the maximum of the stellar - bulge density and the density of the nsc . for the latter , we use the average nsc density within its half - mass radius , i.e. @xmath206/[(8/3 ) \pi r_h^3]$ ] , where we decrease the mass of the nsc by twice the mass of the binary to approximately account for the mass deficit it causes on the nsc during its inspiral [ cf .
equation below ] . as for the stellar - bulge density
, our model assumes a hernquist profile , as mentioned earlier .
comparisons to n - body simulations @xcite show that the appropriate radius where this density profile needs to be evaluated in order to estimate @xmath203 correctly is the binary s mass influence radius , i.e. the radius at which the enclosed bulge mass ( in stars ) equals twice the binary mass . in order to model triple mbh systems when they form ,
we utilize the results of @xcite . by using numerical simulations
, @xcite found that mbh triples typically trigger the merger of the two more massive components on timescales of @xmath207 yrs ( on average ) for binary masses of @xmath208 ( and roughly comparable masses for the two components ) , while the lightest mbh may be ejected from the galaxy or be left wandering far from the galaxy s center , or less likely come back and coalesce with the remnant of the inner binary s merger .
moreover , the time - scales for the merger of the inner binary present a log - normal scatter of about @xmath209 dex around the median value of @xmath207 yrs , as shown in figure 8 ( upper panel ) in @xcite . to extend these results to arbitrary masses ,
we rescale these time - scales with the system s dynamical time at the binary s hardening radius , and we thus obtain an average `` delay '' due to mbh triple interactions given by @xmath210 whenever a triple forms , we then extract the delay with which the two more massive objects merge from a log - normal distribution centered on this timescale , and with r.m.s . of 1.4 dex .
for simplicity , we also assume that the lightest mbh is ejected from the galaxy and therefore lost to the subsequent galaxy evolution .
finally , @xcite noted that triple mbh interactions are less effective at driving mbhs to merger in low - mass systems , where the third incoming mbh is ejected from the galaxy before it can shrink the inner binary to separations at which gravitational - wave emission is important .
indeed , the escape velocity from a galaxy scales as the velocity dispersion , i.e. @xmath211 , where we defined the mass of the inner binary , @xmath212 , and we used the faber - jackson relation @xmath213 .
the separation @xmath214 at which a mbh binary is driven to coalescence by gravitational - wave emission within a hubble time @xmath215 can be obtained from eq .
( with the replacements @xmath216 and @xmath217 ) , and scales as @xmath218 ( with @xmath219 the symmetric mass ratio ) . in order for the third mbh to be able to shrink the inner binary to the separation @xmath214 , the velocity @xmath220 of the third mbh should be lower than @xmath221 when the binary s separation is @xmath214 .
the energy of the third mbh can be estimated simply by energy equipartition as @xmath222 , from which one gets @xmath223 , with @xmath224 .
this in turn gives @xmath225 .
the normalization of this ratio can be estimated using the results of @xcite , who find @xmath226 km / s and @xmath227 km / s for @xmath228 , @xmath229 and @xmath230 ( cf .
their figs . 5 and 11 ) , thus giving @xmath231^{1/8}}\,.\ ] ] therefore
, whenever this equation gives @xmath232 , we assume that the third mbh is ejected without driving the merger of the inner binary .
this implies that triple mbh interactions become ineffective at driving the merger of binary systems with total mass @xmath233 .
[ [ mass - deficit - caused - by - black - hole - binaries ] ] * mass deficit caused by black - hole binaries * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the presence of a mbh binary has profound effects on the nsc , causing a `` mass - deficit '' in the central regions of the galaxy . indeed , during the binary s inspiral , three - body interactions with the stars in the nuclear region result in an exchange of energy between the binary ( which becomes more and more bound ) and the stars , to which large velocities are imparted that are capable of ejecting them from the galactic nucleus .
also , when the mbh binary finally merges , the resulting mbh remnant acquires a kick velocity up to a few thousands km / s due to the anisotropic emission of gravitational waves @xcite , and this also contributes to remove stars from the galactic core .
a similar mass deficit may be caused by the ejection of the lightest mbh in a triple system .
we estimate the mass - ejection rate from the nsc due to the mbh binary s inspiral as @xmath234 where @xmath235 is the binary s mass ratio and @xmath236 is the the semi - major axis at which the binary becomes `` hard '' ( i.e. tightly bound ) .
the first term in the numerator of equation represents the mass scoured by the mbh binary before it becomes hard , where we have identified the ejected mass with the mass deficit as defined in @xcite ; the second term represents instead the mass ejected from @xmath236 to @xmath203 @xcite ; the explicit expressions for @xmath236 is given for instance in @xcite , i.e. @xmath237 while for @xmath203 we utilize equation .
also , in the denominator of equation , we note the presence of the timescale @xmath238 , computed via equations and . that timescale accounts for the fact that both terms in the numerator where computed for mbh binaries in gas - poor environments , hence the mass deficit at the numerator should be `` spread '' over the timescale characterizing stellar interactions .
of course , if the mbh binary lives in a gas - poor enviroment , in our model the inspiral lasts exactly @xmath239 , and the final mass deficit caused by the binary is @xmath240 $ ] . if the binary evolution is instead driven by gas interactions or by the formation of a mbh triple , that mass deficit is suppressed by a factor respectively @xmath241 or @xmath242 .
yr irrespective of the mechanism driving the binary s evolution prior to the gravitational - wave dominated phase
. in this case , the mass deficit is always @xmath240 $ ] .
this test confirms that our results for the nsc evolution are reasonably robust , and that the overall conclusions of this paper do not depend on our particular model for the delays . ]
in addition to the mass deficit caused during the inspiral , when the mbh binary finally merges the resulting mbh remnant acquires a kick , which , as mentioned above , can further remove mass from the nsc .
we estimate this mass deficit as @xcite @xmath243 where @xmath244 is the recoil velocity of the mbh remnant and @xmath245 is the escape velocity from the galactic center .
the latter can be easily calculated within our model , from the mass and density profiles of the bulge and nsc . as for @xmath244 , we follow @xcite and use the analytical formula of @xcite , which fits the results of numerical - relativity simulations . similarly , in the case of triple mbh systems , if the lightest mbh gets ejected from the system before it can cause the inner binary to merge [ i.e. , in our model , if the ratio @xmath246 given by equation is larger than one ] , we assume that the ejected mbh causes a mass deficit @xmath247 , with @xmath248 the ejected mbh s mass .
we stress that for simplicity we neglect the mass deficit caused by the recoil of the lightest mbh in the cases in which the triple interactions trigger the merger of the inner binary ( i.e. , in our model , when @xmath249 ) .
summarizing , the total mass deficit due to mbh coalescences is @xmath250 .
in addition to the effect of mbh binaries , we also account for the possible tidal disruption of nscs by mbhs during galaxy mergers . indeed ,
if a merger takes place between a galaxy ( `` 1 '' ) that contains a nsc ( but not a mbh ) and another galaxy ( `` 2 '' ) hosting a mbh ( and possibly a nsc ) , the nsc of `` galaxy 1 '' will be dragged by dynamical friction toward the mbh ( and nsc if present ) of galaxy 2 , and will therefore be tidally truncated / disrupted , as in the case of star clusters falling toward the nucleus of an isolated galaxy .
to model this effect , we assume that the nsc resulting from such a galaxy merger has mass @xmath251 the fraction @xmath252 accounts for the tidal truncation / disruption effects , and is calculated via equations , , and with @xmath253 taken to be the velocity dispersion of `` nsc 1 '' ( which one can compute in terms of its mass and size ) . note that calculating @xmath124 in the same way as for star clusters is justified , at least to first approximation .
indeed , as mentioned earlier , the star clusters that contribute the most to equation have average initial mass @xmath136 , which is of the same order of magnitude as a typical nsc mass . as in the case of star clusters ,
if @xmath254 ( with @xmath255 a typical nsc concentration parameter ) , we assume that `` nsc 1 '' is fully disrupted , and set @xmath142 . on the other hand ,
if both `` galaxy 1 '' and `` galaxy 2 '' have mbhs , we assume that the gravitational field of each massive bh protects its nsc from tidal truncation and disruption , and we set @xmath256 .
similarly , if neither galaxy contains a mbh , we set @xmath256 .
+ + + and with @xmath257 . in the low mass regime the nsc to galaxy mass ratio shows a weak correlation with @xmath258 , while in the more massive systems ( @xmath257 ) there is a clear trend with @xmath258 , with the nsc mass fraction decreasing rapidly with the galaxy mass .
this is consistent with the flattening of the scaling relations found at high galaxy masses in figure [ main ] .
[ mass - fr],width=268 ]
we compare our models to observational data in order to gain insights on the reliability of such models .
we then use our models to put constraints on the origin and evolution of nscs and to understand how their properties are linked to the evolution of mbhs and their host galaxies .
our sample of nsc objects was obtained by combining data from @xcite , @xcite , @xcite and by estimating nsc object masses of the galaxies in the fornax cluster catalog of @xcite .
@xcite give mass estimates for a total of 18 galaxies that cover hubble types from s0 to sm with dynamically determined nsc masses , and 15 sa and later type galaxies from @xcite with nsc mass estimates based on high - resolution spectroscopy .
velocity dispersions for 9 of these galaxies were obtained from table 2 of @xcite , the rest from the hyperleda database when available .
@xcite constructed their sample of nscs using photometric data from @xcite , @xcite , @xcite and @xcite .
table 1 of @xcite reports @xmath259 galaxies containing a nsc with a well determined mass , and total mass and velocity dispersion estimates for the majority of these .
the uncertainties on the nuclear object masses are given by @xcite , @xcite and @xcite as @xmath260 , @xmath261 , and a factor of two , respectively .
we additionally obtained estimates for the stellar masses of the full galaxy , @xmath258 , and the nsc , using the photometric data for the 43 early - type galaxies in the fornax cluster catalog of @xcite .
to determine , @xmath258 , we multiplied the total galaxy luminosity in the @xmath262 band given in the hyperleda database by a mass to light ratio of @xmath263 @xcite , based on the typical colors of the bulge population .
galaxy total b - magnitudes , @xmath264 , and velocity dispersions , @xmath108 , were also taken from the hyperleda database .
the nsc masses were obtained by multiplying the published luminosities by the appropriate mass - to - light ratio , obtained by using the empirical correlations between color and mass - to - light - ratio given in @xcite .
typical errors on @xmath265 are @xmath266 . after eliminating duplicate galaxies contained in more than one of the samples for each galaxy we included in the collective sample the mass estimate with the smallest uncertainty we collected a total of 89 galaxies with reliable estimates for both @xmath265 and @xmath108 , 127 objects with determined @xmath265 and @xmath258 , and 208 galaxies with measured nsc mass and host galaxy total b - magnitude . although we were not able to obtain all galaxy properties for every object from the literature ,
our data collectively represent the largest sample to date of nsc and host galaxy physical properties . in figure
[ main ] we plot the nsc masses as a function of the velocity dispersion of the galaxy spheroid as well as the galaxy mass , for early ( filled circles ) and late ( open circles ) type systems . in order to investigate the correlation between nscs and their host galaxy properties we use a locally - weighted linear regression model ( _ loess _ ) to fit non - parametric curves to the data @xcite .
_ loess _ does not require one to specify a global function of any form to fit a model to the data , as it combines multiple low - order polynomial regression models in a @xmath267-nearest - neighbor - based metamodel .
unlike a more `` standard '' simple linear regression analysis , no assumption is made that the data can be fitted by a straight line .
_ loess _ scatter - plot smoothing can therefore be used to reveal complex relationships that could be overlooked with traditional parametric estimation strategies .
the obvious trade - off is that in general it is not possible to express a _
loess _ model with a simple mathematical formula .
the smoothness of the _ loess _ regression function is determined by the amount of data points used for the local fits , a quantity controlled by the span parameter , @xmath41 . here
we select an optimal value of @xmath41 by using the generalized cross validation criterion ( gcv ; * ? ? ?
the basic idea is that the optimal representation of the data is obtained by adopting the smoothing parameter that minimizes the mean - squared error of the fit @xcite .
the solid thick black curves in figure [ main ] are @xmath268 interpolations obtained with the optimal smoothing parameters @xmath269 and @xmath270 for @xmath271 and @xmath258 respectively .
the @xmath272 variance bands of the _ loess _ curves are shown as thin black curves .
the first noteworthy point is that the @xmath268 curves shown in figure [ main ] are characterized by a significant bending at @xmath273 and @xmath274 , indicating nsc - host galaxy scaling relations that are flat or even declining for the most luminous galaxies .
moreover , the reconstructed variance bands follow the general trend of the @xmath268 curves , indicating that it is unlikely that the non - linear features present in the curves are due to random fluctuations in the data alone .
we tested whether or not a linear parametrization of the relations provides an adequate description of the data by using a @xmath275-test to compare the @xmath268 fits to simpler weighted linear fits @xcite .
the @xmath275 tests showed that the null hypothesis
i.e. , that the bivariate @xmath268 model yields no improvement in the fit over the linear regression can be rejected at a high level of significance .
for all three relations investigated here , the @xmath268 curves provide a better description to the data than linear models at @xmath276 confidence level .
this confirms that the non - linear features present in the functional dependence of @xmath265 on the galaxy properties are likely not due to noise variability in the data .
a complementary view is provided in figure [ mass - fr ] , where we plot the nsc mass fraction as a function of the total stellar mass of the host galaxy , i.e. , @xmath277 .
evidently , in the low mass regime ( @xmath278 ) the nsc to galaxy mass ratio shows a weaker correlation with total galaxy mass ( with kendall s rank correlation coefficient @xmath279 and associated probability value @xmath280 ) , while in the more massive systems ( @xmath257 ) there is a clear and steeper trend with @xmath258 , in the sense that the nsc mass fraction decreases rapidly with the galaxy mass .
this is in contrast with the results of @xcite , who claimed quite a steep decrease of the nsc mass ratio based on fits to the entire sample distribution .
we finally note that although our analysis demonstrates that current data are consistent with a significant bending of the nsc - host galaxy scaling relations , further observations and analysis , combined with theory , will be needed in order to refine our results . on the other hand , the broadening of the nsc mass distribution for the highest luminosity galaxies , as well as the presence of low mass nuclei ( @xmath281 ) in high mass galaxies ( @xmath282 ) certainly poses a serious concern for previous claims of the existence of fundamental correlations between nsc and host galaxy properties . moreover , as we show in the following , the particular form of these relations is in agreement with what is expected on the basis of our semi - analytical models of nsc formation . and
@xmath132 confidence bands are obtained through our fiducial galaxy formation model where all the relevant effects described in section [ gxev ] were included .
the blue lines correspond to a model in which the scouring effect of mbh binaries forming during galaxy mergers was ignored , i.e. , we set @xmath283 in these integrations . note how the predictions of the two models start to diverge at @xmath284 where mbhs and nscs have comparable masses in the model with @xmath283 ; at @xmath285 mbh mergers become efficient at eroding the surrounding nsc .
[ m - nsc ] ]
the red and blue lines in figure [ main ] show the resulting nsc mass as a function of @xmath108 obtained respectively with @xmath21 after @xmath2gyr of evolution and with @xmath58 at @xmath286 .
these curves give the median output at a given @xmath108 or @xmath258 as well as its @xmath287 and @xmath132 confidence - level regions , i.e. the regions containing respectively @xmath287 and @xmath132 of the nscs produced by our models at a given stellar mass .
the @xmath21 model median can be fit by @xmath288 and @xmath289 at @xmath290 and @xmath291 respectively .
this appears to be in good agreement with analytical expectations
@xcite finds @xmath292 , adopting an idealized isothermal sphere galaxy model .
similar scalings and a similar scatter around the median values are found with the galaxy evolution model @xmath58 , which takes into account both cluster inspirals and in - situ star formation .
the similarity between the results of the two models with and without in - situ star formation suggests that dynamical friction migration of star clusters is a fundamental process for the growth of the nuclei .
however , as we show below in section [ isvsm ] , dissipative processes also play an important role , contributing a significant fraction of the total nsc masses in our models .
figure [ main ] shows that at @xmath293 and @xmath274 , the scaling relations produced with @xmath21 ( i.e. , a purely dissipationless formation model ) appear to flatten toward the most luminous galaxies in agreement with the results of the data analysis of section [ data - an ] .
however , a more careful analysis of the plot also shows that such a model fails to explain the existence of nscs of mass @xmath294 which are clearly present in the observational data .
the reason for this discrepancy is that in @xmath21 we have assumed that all galaxies contain a mbh , and that these mbhs are in place at the center of all initial galaxy models before the nscs grow around them .
these assumptions are not fully supported by our galaxy formation model ( see section [ z - ev - sec ] ) and artificially lead to an enhanced mass removal from the stellar clusters accreting onto the central nsc , because of the tidal field of the mbh . the galaxy evolution model @xmath58 produces scaling relations at @xmath286 that are in remarkably good agreement with the observed correlations .
the scatter around the median value clearly increases at the high mass end of the nsc distribution .
a large population of nscs are produced at @xmath293 that are significantly underweight with respect to what the same model would predict by simply extrapolating the scaling correlations and scatter from low ( @xmath295 ) to high galaxy masses . in the following section we argue that the formation of such underweight " nscs and , consequently , the apparent bending of the scaling correlations can be explained in terms of the interaction of the nscs with their host galaxy mbhs . in our semi - analytical models
mbhs affect the formation and evolution of nscs in two important ways : ( i ) the strong tidal field of the central mbh accelerates the mass loss from stellar clusters as they enter its sphere of influence and eventually disrupts them as they come closer than a distance @xmath296 , where @xmath297 after a cluster is disrupted , most of its stars are dispersed around @xmath296 , which will limit the nuclear density within this radius .
thus , a preexisting mbhwill limit the amount of mass that can be transported to the center by decaying stellar clusters @xcite . as mentioned earlier [ cf .
equation ] , a similar effect applies to a nsc falling toward a central mbh after a galaxy merger .
( ii ) during galaxy mergers , mbh binaries form and harden by ejecting surrounding stars @xcite . by dynamically
heating the nsc , a mbh binary will lower the nsc central stellar density , or fully destroy the nsc when the binary s mass is significantly larger than the nsc mass @xcite .
the kick imparted to the merger remnant by asymmetric emission of gravitational wave radiation will remove additional mass from the galaxy nucleus @xcite . to demonstrate the role of mbhs in the evolution of the nuclei
, we performed additional simulations where the two dynamical effects due to mbhs mentioned above were not included .
( hereafter we refer to these models as models `` without mbh heating '' . ) the blue and red curves in the middle panels of figure [ main ] show the scaling correlations resulting from such models . when compared to our fiducial models ( upper panels ) , these new integrations produce steeper and significantly tighter @xmath71 and @xmath298 relations .
this indicates that the break in the nsc scaling relations , for which we found evidence in the observational data , can be attributed to the interaction of nscs with their companion mbhs .
simple arguments can indeed be used to understand why the break of the nsc scaling relations occurs at @xmath299 . from equation ( [ eq : rtidal ] ) , by requiring @xmath300pc , and adopting @xmath301 and @xmath302 , we find that mbhs more massive than roughly @xmath303 will suppress the further growth of a nsc via accretion of stellar clusters .
thus , as the mbh grows , the contribution of star cluster inspirals to nsc growth decreases and eventually stops for @xmath304 . using @xcite
: @xmath305 a mbh mass larger than @xmath306 corresponds to stellar spheroids with velocity dispersion @xmath307 .
this is consistent with the value of @xmath108 at which the nsc - host galaxy scaling correlations appear to flatten in the data sample .
+ + , containing a mbh ( @xmath308 ; solid lines ) , and containing a mbh plus a nsc ( dashed lines ) .
the right panel shows the fraction of nucleated galaxies , @xmath309 , that contain a central agn , identified here as a mbh with bolometric luminosity @xmath310 . ,
width=172 ] another argument leads us to identify @xmath311 as the critical value of mbh mass above which nscs are significantly affected .
the impact of a mbh binary on a central cluster depends on the @xmath312 mass ratio .
given that during a merger a mbh binary will eject a mass comparable to its total mass , a requirement for a mbh binary to affect significantly the nuclear structure of a galaxy is that its mass is of the order of or larger than the mass of the surrounding cluster , i.e. , @xmath313 .
using a standard weighted linear regression on our data gives the relation @xmath314 ; using equation ( [ sigmabh ] ) we find the new relation : @xmath315 thus , the mass of the mbh grows faster then the mass of the nsc , with the transition from nsc dominated to mbh - dominated galaxies occurring at @xmath311 or @xmath316 .
this is roughly the value of @xmath108 at which nscs appear to be significantly affected by their host galaxy mbhs in our models .
the total mass ejected from the center depends also on the number of stages in the merger hierarchy that have occurred since the mbhs first formed
i.e. , after @xmath317 mergers , the mass deficit is @xmath318 , with @xmath319 the final ( current ) black hole mass @xcite .
some low - mass galaxies today might have never experienced a major merger . by contrast , more massive galaxies form via mergers of primordial lower mass halos , and underwent an above average number of mergers .
in addition , since the occupation fraction of mbhs increases with galaxy mass , the scouring effect of mbh binaries is enhanced in the higher mass spheroids . from figure
[ main ] , we also see that the @xmath58 model generates two distinct populations of nscs at @xmath320 , which can be easily identified by looking at the model confidence bands .
nuclei with @xmath321 have masses that do not strongly correlate with their host galaxy mass and velocity dispersion the mass of these nscs lies significantly below the mass that we would obtain by extrapolating the scaling correlations from low to high galaxy masses .
this population of underweight nuclei is a result of the disruptive mbh binary mergers that have partially eroded the surrounding clusters .
nscs above the model median seem instead to follow the general trend outlined in the low mass galaxy region of the plot , where @xmath322 .
the host galaxies of such nuclei did not experience a major merger event since the epoch of the last major gas accretion event , so that their nscs remained essentially unaffected by mbh binaries up to the present epoch .
this picture is in agreement with expectations based on hierarchical models of galaxy formation for example , @xcite find that for a dark halo mass of @xmath323 only 31 , 53 and 69 per cent of these halos have experienced a major merger since @xmath3241 , 2 and 3 , respectively . we can now ask which of the two processes , tidal disruption of star clusters or mass ejection from mbh binaries , is responsible for the broadening of the scaling correlations found in the data . we find that both @xmath21 models with @xmath283 and ones without the tidal truncation / disruption of clusters by the central mbh produce scaling correlations that appear to flatten at @xmath273 .
the nsc scaling correlations produced by @xmath58 models in which we set @xmath283 ( but in which we included the tidal disruption of migrating clusters ) are instead at odds with observations , as they show no broadening of the nsc host galaxy property correlations for high velocity dispersion galaxies ; only @xmath58 models that do take into account the scouring effect of mbh binaries were found to be in good agreement with the observed relations .
we conclude that in @xmath58 the crucial ingredient to reproduce the @xmath265-host galaxy property scaling relations is the scouring effect of mbh binaries , while the tidal stripping of migrating clusters by central mbhs is only a secondary effect in these models .
we remark that the @xmath58 model has a clear advantage over the @xmath21 model , because the former follows the _ hierarchical _ evolution of nscs and mbhs along merger - trees , while the latter assumes a _ monolithic _ evolution .
although the results of the two models are generally in good agreement with one another , the differences outlined above are mainly a result of the implicit assumption made in the @xmath21 model that the nscs _ always _ grow around _ pre - existing _ mbhs .
this leads to an artificially enhanced mass removal from the stellar clusters accreting onto the central nsc , because of the mbh tidal field . to illustrate more clearly the role of mbh binaries in shaping their companion nsc properties , we compare in figure [ m - nsc ] the observationally constrained nsc to mbh mass ratio in galaxies containing both types of central objects to the same ratio obtained ( i ) in our fiducial @xmath58 model containing all the relevant effects described in section [ gxev ] , and ( ii ) a model in which we set @xmath283 , i.e. , in which the scouring effect of mbh binaries was not included in the calculation
clearly , the scouring effect due to mbh binaries described by equations ( [ ejmass ] ) and ( [ ejkick ] ) is the crucial ingredient to reproduce the observed correlation . also , note that the correlations produced by the two models start to diverge from each other at @xmath325 , in agreement with our predictions ( see also * ? ? ?
observationally , the frequency of nucleation in early type galaxies is found to increase sharply from zero for spheroids brighter than @xmath326 to @xmath327 for galaxies fainter than this magnitude @xcite . in a sample of 332 late type galaxies
, @xcite found that @xmath328 of these galaxies harbor a well defined nsc .
hence nscs are found in most galaxies of all hubble type , but tend to disappear in the brightest spheroids .
figure [ fract ] compares the frequency of nucleation obtained in our models to that of late and early - type galaxies , as derived from observations .
we define here early - type galaxies as systems with bulge - to - total mass ratio @xmath329 , and late - type galaxies as systems with @xmath330 . in agreement with observations ,
our models predict that almost all early - type galaxies that are less massive than @xmath331 contain a nsc , and that the frequency of nucleation is nearly zero for galaxies with mass larger than this value .
we note that the exact occupation number obtained through @xmath58 should be considered somewhat approximate the identification of a galaxy as nucleated in this model is uncertain given that we can not directly reconstruct the density profile of the nsc and compare it to the density profile of the galaxy background .
nevertheless , the occupation fraction is found to decrease significantly for early type galaxies with mass larger than @xmath332 , in good agreement with observational constraints .
the bottom panel of figure [ fract ] shows the frequency of nucleation in late type systems .
approximately @xmath333 of these galaxies were found to contain a nsc , also in fairly good agreement with observations .
the observed drop in the nucleation frequency at high galaxy masses is often attributed to the disruptive effects of mbh binaries forming during galaxy mergers @xcite and to the tidal disruption of migrating clusters at large galactocentric distances by central mbhs @xcite . in the right panels of figure [ fract ]
we show the nucleated fraction in models where the dynamical heating due to mbhs on migrating clusters and during mergers was not included . in these models
the percentage of nucleation remains close to @xmath334 regardless of galaxy luminosity , demonstrating that mbhs are responsible for the absence of nscs in the most luminous galaxies , as also argued in previous work .
nscs and mbhs are known to coexist in some galaxies across a wide range of masses and hubble types @xcite .
however , not all nscs contain a mbh ; an example is the m33 nsc that has no central mbh down to highly constraining limits @xcite .
studies indicate that at least some nscs can host a central mbh , but the overall occupation fraction of mbhs in nscs remains unknown .
the left panel of figure [ fract2 ] displays the fraction of all galaxies in the @xmath58 model that contain a mbh and the fraction of galaxies containing both a mbh and a nsc .
our models predict that a high fraction of galaxies of intermediate and low luminosity contain a central nsc , but that the nscs tend to disappear in massive galaxies .
thus , the total fraction of mixed systems with both a nsc and a mbhis roughly equal to the mbh occupation fraction for galaxies with mass @xmath335 , and is roughly equal to the nsc occupation fraction for galaxies more massive than @xmath332 . in the right panel of figure [ fract2 ]
we show the fraction of nucleated galaxies that contain a mbh with bolometric luminosity @xmath336 .
we find that the fraction of galaxies with a nsc that also have an active mbh increases from @xmath337 to @xmath338 going from @xmath339 to @xmath340 .
these results are in fairly good agreement with the observational results from @xcite , who found that about @xmath341 of nscs in their spectroscopic sample also host an agn . + + + , the nscs are typically more massive than mbhs . at @xmath286 nsc masses
are typically of the order @xmath342 , with only a few rare clusters having masses above @xmath343 .
note the transition from nsc dominated to mbh dominated galaxies for @xmath344 occurring at @xmath345 .
[ z - ev - m ] ] the lower panels of figure [ main ] show the scaling relations generated by a version of the @xmath58 model with no contribution from cluster inspirals . in these models , the only mechanism responsible for the formation and growth of the nuclei is local fragmentation of gas that leads to star formation and in - situ build up of a central compact cluster . the lower panels of figure [ main ] demonstrate that even a purely dissipative model provides a quite good description of the observed nsc - host galaxy scaling relations .
the median of the mass distribution as a function of @xmath108 can be fit by @xmath346 .
hence , in this model we obtain scaling relations that are consistent with the observation that nscs follow scaling relations that are shallower than the corresponding ones for mbhs .
we can derive analytical scaling relations for the in - situ formation model by considering the mbh and reservoir characteristic timescales of evolution .
let us consider a situation in which the mbh is hosted by a gas - rich nucleus [ resulting from a recent starburst , cf .
equation ] , i.e. @xmath347 .
the mbh will then grow approximately at the eddington rate , i.e. @xmath348 , with @xmath349 the salpeter timescale @xmath350 where @xmath351 is the radiative efficiency and @xmath352 the electron scattering opacity .
gas accumulates at the center of the galaxy mainly during starbursts [ cf .
equation ] .
bursts of star formation will take place on the dynamical timescale of the galactic bulge , i.e. @xmath353 with @xmath354 the galactic bulge scale radius . in the last expression
we have used the fact that both @xmath354 and @xmath108 are related to the total galaxy luminosity through the empirical correlations @xcite : @xmath355 and @xmath356 we can thus write , from equation , @xmath357 during the starburst event , and therefore @xmath358 where @xmath359 is the total mass of cold gas that falls to the nucleus during the star formation event , and @xmath360 and @xmath361 denote the changes in the mbh and stellar bulge masses during the starburst .
( note that eventually all the gas of the bulge is transformed into stars , if feedback is neglected , i.e. @xmath362 . ) now , if we assume that the mbh approximately satisfies the magorrian relation @xcite @xmath363 [ with @xmath364 @xcite ] , we can write @xmath365 where have assumed @xmath366 , because starbursts are typically triggered by major galaxy mergers , i.e. ones with mass ratio between the baryonic components of the two galaxies larger than @xmath367 . can be obtained from equation by assuming that the mbh satisfies the @xmath148 relation @xmath368 , if one observes that @xmath369 ( where we have used equation and assumed constant mass - to - light ratio ) . indeed , this latter relation yields @xmath370 , which replaced in equation gives @xmath371 , if one assumes @xmath372 in major galactic mergers .
] eventually all the gas of the reservoir will either form stars or accrete onto the mbh , i.e. @xmath373 . however , since we assumed @xmath347 , most of the nuclear gas will end up in the nsc rather than in the mbh , hence @xmath374 , and we can write , for @xmath375 , @xmath376 where we used @xmath377 and @xmath378 . and @xmath379 are essentially the rates of change of the mbh and nsc masses during a starburst ( time being parametrized by the host galaxy s velocity dispersion @xmath108 ) . ]
we can then integrate this equation to give @xmath380 where we have used equation and assumed a radiative accretion efficiency @xmath381 .
although very simplified , this model approximately reproduces the scaling relations generated by our dissipative @xmath58 model ( for @xmath382 ) in slope , and also ( within a factor of a few ) in normalization .
the mass growth of nscs is regulated by two processes : inspiral of star clusters and in - situ star formation . in order to quantify the amount of nsc
mass that is contributed by local star formation in our models , we computed the ratio of the median nsc mass obtained by including both cluster inspirals and in - situ star formation to that obtained with only the contribution of in - situ star formation .
we plot this quantity as a function of total galaxy mass in figure [ insitu - fr ] . in our models ,
about half of the mass of nscs is contributed by local star formation for galaxies less massive than @xmath383 , while for galaxies more massive than this value , the contribution from dissipative processes becomes progressively more important .
some provisional evidence for a preferential dissipative mode in galaxies of progressively larger masses was presented by @xcite and @xcite .
these authors showed that the brightest nuclei which reside in the most luminous hosts have colors that are significantly redder than expected on the basis of a star cluster merger scenario , suggesting that an additional process
e.g. , dissipational infall of metal - rich gas during mergers likely begins to dominate the formation of nuclei in galaxies of higher masses .
we finally note that the relative contribution of the two formation channels depends on the adopted value of star cluster formation efficiency @xmath64 which remains a quite uncertain parameter of our models .
however , given that plausible values of @xmath64 were found to only impact the normalization of our results for the empirical relations and to not affect their slope , assuming @xmath384 const we can simply express the relative contribution of the two formation channels using the general formulation : @xmath385 with @xmath386 the typical nsc mass that originates in - situ from episodes of star formation and @xmath387 the mass brought in by migrating clusters .
[ z - ev - sec ] the existence of a fundamental connection between mbhs and nscs was first suggested by @xcite . these authors showed that nscs and mbhs follow similar scaling relations with their host galaxy properties , and thus argued that they are two different manifestations of the same astrophysical type of system , which they called `` central massive object '' .
later , @xcite used a large sample of nucleated galaxies to show that the scaling relations of nscs and mbhs are quite different from each other , with the former following much shallower correlations with their host galaxy properties .
graham s findings might suggest that nscs and mbhs formed through different physical processes .
this could be the case if for example nscs originated elsewhere in the galaxy and then migrated to the center through dynamical friction processes ( e.g. , * ? ? ?
@xcite presented a first @xmath388 versus @xmath265 diagram and found , in agreement with our study , the existence of three different regimes : ( a ) nsc dominated nuclei at @xmath389 , ( b ) a transition region , and ( c ) mbh - dominated nuclei at @xmath307 .
@xcite argued that this is consistent with a picture in which black holes form inside nscs with a low - mass fraction .
they subsequently grow much faster than the nsc , destroying it completely when the ratio @xmath390 grows above @xmath391 .
@xcite also argued against mbh mergers as responsible for the disruption of the nuclei in the highest mass galaxies .
these authors pointed out that if nscs are disrupted during mergers , elliptical galaxies thought to be the product of galaxy mergers should rarely host a nsc .
the fact that most early - type galaxies have a nsc would therefore suggest that mergers do not play a pivotal role in leading to nsc disruption .
we do not agree with this interpretation . as discussed in section [ bh - eff ] , a mbh binary will eject from the galaxy center a mass comparable to its own mass , so that only mergers with mbh to nsc mass ratio larger than unity can significantly heat the nsc and make it susceptible to destruction during the merger event .
nscs can therefore survive and indeed grow during the morphological transformation from disk - dominated to bulge - dominated galaxies if the progenitor galaxy nscs are significantly more massive than their central mbhs . indeed , from the nsc and mbh scaling correlations , we find that nscs dominate the nuclei of galaxies with velocity dispersion @xmath392 , so that mergers of galaxies at the low end of the @xmath108 distribution will not lead to the complete disruption of the host galaxy nscs , while mergers of massive galaxies , characterized by a larger mbh to nsc mass ratio , will lead to their full disruption .
this picture is consistent with both the abundance of nscs in early type galaxies of intermediate luminosity , and with the fact that nscs tend to disappear in the brightest spheroids hosting the most massive mbhs .
+ + + to illustrate the simultaneous evolution of mbhs and nscs in our semi - analytical galaxy formation models , we plot in figure [ z - ev ] the mbh and nsc scaling correlations at different redshifts and the corresponding mass distributions in figure [ z - ev - m ] .
specifically , in order to calculate the median and confidence regions in our model we have have only considered the mbhs residing in bulge - dominated galaxies ( which we identify with ones having bulge to total mass ratio larger than @xmath393 ) .
are possible in the first place . ]
we find that at high redshifts , @xmath394 , the nuclei of galaxies are dominated by nscs . between redshift @xmath395 and @xmath1
the mbhs grow faster than the nscs , becoming by @xmath181 the dominant nuclear component in galaxies with @xmath396 .
after this point , the nsc scaling correlations start to flatten at high values of @xmath108 , as mbh binaries forming during mergers are now efficient at eroding the surrounding clusters .
in addition , the nscs can no longer grow efficiently in the most massive galaxies , since inspiraling stellar clusters are tidally disrupted at larger galactocentric distances in galaxies with progressively more massive mbhs . figure [ z - ev - m ] displays the redshift evolution of the mass distribution of mbhs and nscs . in these plots
we include all galaxies in our models regardless of their specific value of bulge to total mass ratio . at high redshift
the nscs are typically more massive than mbhs ; after @xmath397 the mass distribution of nscs intersects at about @xmath4 the distribution of mbhs .
this sets the transition between nsc dominated and mbh dominated galaxies seen also in the observational data .
we also compared our synthetic nsc mass distributions for early type galaxies with the observed mass distribution of nscs for the catalogs of early type galaxies of @xcite and @xcite .
anderson - darling and kolmogorov - smirnov statistical tests used to compare the observed to the mbh model cumulative distributions give @xmath398-values of @xmath399 and @xmath400 respectively , indicating that the hypothesis that the simulated and observed distributions are significantly different from each other can be rejected at a high level of confidence . for a comparison with the observed mbh mass function at @xmath286 ,
see instead @xcite and @xcite .
but for the heavy mbh seed scenario of @xcite.[z - ev - m - heavy ] , title="fig : " ] + in the previous sections , we have discussed the results of a galaxy formation model in which mbhs are evolved starting at @xmath153 from initial `` light '' masses @xmath401 .
this could be the case if mbhs formed as remnants of pop iii stars @xcite . in order to test the dependence of our results on the assumed scenario for the formation of mbhs , we evolved additional models in which halos are initially populated by mbh seeds of mass @xmath402 , which could for example form as a result of the collapse of massive protogalactic disks at redshifts @xmath403 @xcite . in particular , we consider a model in which these `` heavy '' seeds form with high halo occupation number @xcite , and one in which the halo occupation number is moderate ( namely the model of @xcite , where we set the critical toomre parameter below which the formation of a mbh seed becomes possible to @xmath404 ) .
figure [ z - ev - heavy ] shows the redshift evolution of the @xmath405 and @xmath406 relations in the case of the mbh heavy seed scenario of @xcite .
figure [ z - ev - m - heavy ] displays the corresponding mbh and nsc mass distributions from @xmath407 to @xmath286 .
a comparison of these plots with those in figures [ z - ev ] and [ z - ev - m ] demonstrates that the local mass distribution of nscs and their evolution from high redshift is not affected by the mbh seed model in any important way .
( of course , the mbh mass function in the heavy - seed models differs from that in the light - seed model at the low - mass end , which is dominated by mbhs that have not evolved significantly from their seeds .
however , at intermediate and high masses , the mbh mass functions are very weakly dependent on the seed model , since memory of the initial conditions has been lost due to accretion and mergers . )
note that also in the heavy - seed scenarios , we find that the nsc cumulative mass distributions at @xmath286 for early type galaxies are consistent with the observed mass distribution in @xcite and @xcite at a high level of significance ( similar conclusions hold for the seed model of @xcite ) .
the fact that our results about nsc evolution are not sensitive to the particular model chosen for the formation of mbh seeds is primarily a consequence of the high nsc to mbh mass ratio at high redshift . at @xmath408 , in both the heavy and light seed scenarios ,
the nscs are the dominant central component of galaxies ( with the exception of small nsc masses @xmath409 , at which mbhs may dominate if they are seeded with high halo occupation number at high redshift ) .
it is only after the peak of the quasar activity at @xmath410 that mbhs become massive enough to significantly affect the nscs at intermediate and high masses .
however , by this time the mbh mass distributions in different seed scenarios are very similar to each other ( again , with the exception of the low - mass end ) , and therefore the subsequent evolution of the nscs is also very similar .
the mass growth of the mbh population is in fact dominated by the mass accreted during the quasar epoch at @xmath397 , thus washing out the imprint of the initial conditions , cf .
also @xcite .
since to date the formation process of mbhs remains largely unconstrained , the fact that our results are robust against different initial seed formation scenarios is important if we want to make robust predictions about the overall evolution of the nsc population .
in this paper , we have presented a study aimed at understanding how the central regions of galaxies formed , and how the evolution of mbhs and nscs is connected to that of their host galaxy . in figure
[ nsc - ev - tr ] , we show examples of nsc and mbh evolution predicted by the @xmath58 model . these cases should be seen as qualitative because of the great variety of possible nsc and mbh histories that are possible within our model , but we have chosen examples that are roughly representative of the various possibilities mentioned below . in more detail , we consider the evolution of nscs back in time following their main progenitor , i.e. we start from the nsc hosted in the central galaxy at @xmath286 , and at each galaxy merger we follow its more massive nsc progenitor . when no nsc is present at @xmath286 in the central galaxy ( namely in the bottom - left panel ) , we follow the main - progenitor history of the mbh . as can be seen ,
nscs grow by the combined action of ( i ) stellar - cluster infall , which happens continuously throughout their history , but is particularly enhanced in starburst galaxies ( cf . the blue line in figure [ nsc - ev - tr ] , which denotes the mass in stellar clusters , and which shows a smooth evolution with superimposed spikes due to starbursts ) ; and ( ii ) in - situ star formation , which takes places for the most part in starburst galaxies formed by major mergers ( cf .
the spikes in the mass of the low - angular momentum reservoir available for nuclear star formation , i.e. the green line in figure [ nsc - ev - tr ] ) . note also that mbh mergers have a prominent effect on the mbh and nsc history , not only because mbh binaries erode or even completely destroy the nsc if the mbhs are sufficiently massive , but also because the merger can eject the remnant mbh from the galactic nucleus , when the kick velocity imparted by the anisotropic gravitational - wave emission is larger than the escape velocity from the central parts of the galaxy .
note instead that the apparent `` re - growth '' of the mbh after an ejection , which can be seen in figure [ nsc - ev - tr ] , is simply a consequence of our choice of following the nsc main - progenitor history : at certain galaxy mergers , the galaxy containing the main nsc ( but no mbh ) may merge with one carrying a mbh .
below we discuss in more detail how the evolution of the galaxy , and its merger history can give rise to the variety of galactic central structures observed today .
_ nucleated galaxies containing a mbh ( cf . , upper panels of figure [ nsc - ev - tr ] ) .
+ _ if the mbh mass does not grow above @xmath411 throughout the evolution of the galaxy , the central nsc can not be significantly eroded during galaxy mergers by inspiraling mbh binaries , because of the large nsc to mbh mass ratio . moreover
, stellar clusters will migrate to the center essentially undisturbed by the tidal field of the central mbh .
if during the last merger event the mbh remnant is retained , the end product will be a galaxy containing both a central nsc and a mbh ; an example of such a galaxy is the milky way which hosts a mbh of mass @xmath412 @xcite and a nsc of mass @xmath413 @xcite .
a handful of external galaxies are also know to contain both a nsc and a mbh , which are often found to have comparable masses @xcite .
these galaxies lie near the transition region between mbh dominated and nsc dominated systems . as the mbhs of these galaxies grow , mbh mergers will partially destroy their host nscs in the process ( e.g. , upper right panel of figure [ nsc - ev - tr ] ) .
_ nucleated galaxies without mbh ( cf . ,
middle panels of figure [ nsc - ev - tr ] ) .
+ _ after a merger a mbh can receive a `` kick '' due to gravitational recoil with velocities as large as a few thousands @xmath414 , which may eject the mbh from its host galaxy .
the ejected mbh will carry a cluster of bound stars , a hypercompact stellar system that can appear similar in size and luminosity to star clusters or ultracompact dwarf galaxies @xcite .
if the ejected mbh has a mass @xmath415 , the nsc will remain virtually unaffected .
if from that point on the galaxy evolves passively without experiencing mergers which might bring a new mbh to the center , the evolution s end product will be a nucleus containing a nsc but no mbh .
examples of such type of evolution are illustrated in the middle panels of figure [ nsc - ev - tr ] .
a similar end product can be attained if a mbh seed never formed and the galaxy evolved without experiencing major mergers during its entire evolution .
we note in passing that whether a mbh seed will form might depends on the properties of the central cluster at high redshift @xcite .
if the accumulated cluster has a high velocity dispersion @xmath416 , this might seed the growth of a central mbh because kinematic heating from binary stars is insufficient to prevent complete core collapse , which might then lead to runaway mergers and the formation of a central mbh seed of @xmath417 . in galaxies hosting clusters with initial velocity dispersion less than @xmath418
, binaries will prevent core collapse and consequently the formation of a mbh seed .
galaxies with clusters with initial velocity dispersions below this limit might therefore never form a mbh .
such evolutionary path might lead to the formation of galactic nuclei such as those of m33 or ngc205 , which lack a mbh and host a low velocity dispersion nsc @xcite .
_ mbh dominated galaxies without nsc ( cf . ,
lower panels of figure [ nsc - ev - tr ] ) .
+ _ after the mbh mass grows above @xmath419 , any major merger will be highly disruptive for a central nsc ; mbh binaries forming during major mergers in this high mbh mass regime will carve out the galactic center of stars destroying a pre - existing nsc , eventually producing a central mass - deficit @xcite . the accretion of a nucleated dwarf galaxy by a low - density giant galaxy , could bring a nsc to the center of the latter galaxy .
however , @xcite showed that the secondary galaxy is disrupted during the merger by the giant galaxy mbh tidal field , producing a remnant with a central density that is only slightly higher than that of the giant galaxy initially .
moreover , as shown in @xcite , after a nsc is disrupted in a dry merger it will have a hard time to regrow , because the mbh tidal field is effective at disrupting migrating clusters .
thus , once the galactic mbh mass grows above @xmath419 , the central nsc is likely to be disrupted and the central core is likely to be preserved during the subsequent evolution of the galaxy . _
low mass spheroids without nsc .
+ _ nscs tend to disappear in galaxies fainter than @xmath420 @xcite . while the purely dissipationless formation model @xmath21 predicts that the nsc occupation fraction decreases substantially in low mass systems in agreement with observations , the @xmath58 model results in a high occupation fraction of nscs in low mass galaxies , and it is therefore in tension with observational findings .
we believe that the reason for this discrepancy is that discreteness effects are not accounted for in @xmath58 as argued in the following . in this latter model , we compute the flux of mass accreting onto the nucleus by _ averaging _ over the initial mass function of the stellar cluster population . in @xmath21 , we simulate instead the inspiral of each of the formed clusters , drawing their masses from the same initial mass function .
because of the initial small number of stellar clusters in low mass galaxies , and because of the steep power law that we adopt for their initial mass function , some of these galaxies will host no cluster that is massive enough to decay to the center before being disrupted in the process .
this suggests that the lack of nscs in spheroids less luminous than @xmath420 is a consequence of the initial low number of massive clusters in these galaxies .
clearly , this effect can not be accounted for in the @xmath58 model , which does not follow the evolution of individual stellar clusters but rather the average mass infall to the nucleus .
our analysis focuses on the mass evolution of nscs and ignores effects due to dynamical relaxation that can change the density profile of the clusters over time .
we have shown that the merger history of galaxies after @xmath397 plays a key role in determining the structure of nscs observed today and their disappearance in the brightest galaxies . in our models ,
we did not account for relaxation processes that can rebuild an eroded cusp as stars diffuse into the mbh sphere of influence .
even if a nsc is fully disrupted during a merger , a stellar cusp of mass of order @xmath341 the mass of the black hole and extending out to one tenth its sphere of influence will reform over the relaxation time @xcite . although cusp regrowth is not expected to have an important effect on the mass evolution of the nscs , it is worth discussing it , as the ability of a mbh to shape a galactic nucleus and the observational consequences depend on whether cores are regenerated or preserved after mergers .
@xcite performed @xmath20-body simulations to investigate the evolution of merging star clusters with seed black holes .
using black holes masses of @xmath421 their host nsc ones , they investigated how the binary coalescence dynamically heats and destroys the surrounding clusters . while @xcite argued that this mechanism provides a pathway to explain the observed reduction in the nucleus - to - galaxy stellar mass ratio as one proceeds from dwarf to giant elliptical galaxies
, they did not discuss how their results would be affected by two - body relaxation after the mbh binary merges , and between mergers . during galaxy mergers ,
cores are carved out in the stellar distribution , with radii of order the influence radius of the massive binary .
this makes nsc susceptible to disruption . on the other hand ,
if the supply of stars is continuously replenished , the damage to the nsc can be much smaller .
furthermore , even if a large core is formed , a stellar density cusp can regrow via energy exchanges between stars moving in the gravitational field of the mbh remnant .
black holes with masses of @xmath421 that of the nsc , such as those considered in @xcite , correspond to low mass galaxies with a short ( @xmath422yr ) relaxation time . from equation ( [ nsc - bh ] )
one finds that such a low mass ratio corresponds to a mbh mass of @xmath423 . using equation @xmath424 ,
this corresponds to a host stellar spheroid velocity dispersion of @xmath425 . in such low mass galaxies ,
the cusp will regenerate itself in about @xmath426gyr from the infall of a second black hole , as collisional relaxation drives the stellar distribution toward its steady state form @xcite .
we conclude that in low mass spheroids the effect of mbh binaries on the host nscdensity distribution is expected to be minimal , and their imprint on the nuclear properties is expected to be erased after a short time .
the probability that a galaxy of the size of the milky way experienced a major merger ( i.e. , with mass ratio larger than 0.3 ) after @xmath427 is about @xmath428 @xcite .
galaxies of such luminosities have relaxation times that are longer than the age of the galaxy , but which are short enough that two - body relaxation might have affected their central clusters .
these galaxies might have nuclei that are still evolving toward their steady state after they were dynamically heated by a mbh binary .
thus , the results of our models imply that `` under - massive '' nscs in galaxies with bulge mass @xmath429 might still carry an imprint of the merger history of their host galaxy , in the form of a density core that extends out to a small fraction of the central mbh influence radius .
whether these nscs will turn out to harbor parsec - scale cores will be addressed in the future with high resolution imaging , which will enable to resolve the crowded stellar environment of nscs @xcite .
nscs and mbhs represent important central components of galaxies that bear witness to their formation history .
our aim has been to generate a numerical framework to better understand how the central structures of galaxies formed , and how the evolution of mbhs and nscs is connected to that of their host galaxies .
we have studied the evolution of nscs in a cosmological context , by taking into account the growth of mbhs by merging subhalos containing both mbhs and nscs .
a semi - analytical galaxy formation model is applied to follow the evolution of dark matter halos along merger trees , as well as that of the baryonic components .
specifically , the evolution of dark matter halos along the merger trees includes their baryonic components such as hot gas , stellar and gaseous bulges , and stellar and gaseous galactic disks .
we study the simultaneous evolution of mbhs and nscs in our galaxy formation model .
the main results of our work are summarized in the following : \(1 ) the mass growth of nscs is seen to be regulated by inspiral of star clusters , in - situ star formation as well as galaxy and mbh mergers .
we have found that both cluster inspirals and in - situ star formation contribute a significant fraction of the total mass of nscs .
\(2 ) we found that in - situ star formation ( as well as growth through migrating clusters ) , generates nsc - host galaxy scaling relations that are much shallower than the same correlations for mbhs , in agreement with observations .
\(3 ) in our model , the merger history of galaxies after @xmath430 plays a key role in determining the structure of the nscs observed today .
core - depletion due to gravitational slingshot of host galaxy stars by inspiralling mbhs forming during galaxy mergers has a negligible impact on the evolution of nscs in low mass galaxies , while it leads to their full disruption in galaxies more massive than @xmath431 . in galaxies of intermediate luminosity ( @xmath432 )
, mbh mergers cause the partial dissolution of the clusters .
such partially eroded clusters appear at @xmath286 significantly underweight relative to linear nsc - host spheroid scaling correlations .
the presence of an under - massive nsc population results in an increased scatter of the nsc - host galaxy empirical correlations at high galaxy luminosities , for which we find evidence in observational data .
\(4 ) we identify @xmath311 as the critical value of mbh mass above which nscs are significantly affected .
once the galactic mbh mass grows above this value , mbh binary mergers become highly disruptive for the central nsc leading to the formation of a central low density core . for galaxies more massive than the milky way , after such low density core forms , the central galactic regions are likely to remain depleted of stars throughout the subsequent evolution of the galaxy .
\(5 ) our models predict that the fraction of nucleated early type galaxies containing an accreting mbh ( with bolometric luminosity @xmath310 ) increases from @xmath433 at @xmath434 to @xmath435 at @xmath436 .
these fractions are found to be much smaller for late type galaxies , for which only @xmath437 of nucleated galaxies with @xmath438 also contain an active mbh . among galaxies of all types , the fraction of mixed systems containing both a mbh and a nsc is about @xmath439 at @xmath440 .
\(6 ) we found that the results shown in this paper are quite robust when we change the cluster formation efficiency , @xmath441 ; when we allow @xmath64 to vary with galactic properties ( e.g. we set it to 0.07 , 0.04 and 0.5 in disk , quiescent and starburst galaxies respectively ; * ? ? ? * ) ; when we change the mbh seed model ( i.e. light - seed models vs heavy - seed models , with several halo occupation numbers at high redshift ) as well as other details of our model , such as merger - tree resolution , initial redshift of the simulations , prescriptions for agn feedback , etc . in conclusion
, our study supports a scenario in which black holes form inside nscs with a low - mass fraction , so that nscs are initially the dominant central component of galaxies .
after their formation , nscs and mbhs grow in parallel on their own scaling correlations with host galaxy properties , with nscs dominating in low mass spheroids and mbhs dominating in high mass galaxies .
the simple fact that nscs and mbhs grow on different scaling correlations explains the well known transition from mbh- to nsc - dominated galaxies as one proceeds from dwarfs to giant ellipticals , without the need of invoking competitive feedback processes from young nscs and/or agn activity .
during the course of this work , we have benefited from conversations with several colleagues , including m. colpi , d. merritt , n. murray , and a. sesana .
we acknowledge support from a ciera postdoctoral fellowship at northwestern university ( to f.a . ) ; from the european union s seventh framework programme ( fp7/people-2011-cig ) through the marie curie career integration grant galformbhs pcig11-ga-2012 - 321608 ( to e.b . ) ; from erc project 267117 ( dark ) hosted by universit pierre et marie curie - paris 6 and at jhu by national science foundation grant oia-1124403 ( to j.s . ) .
f.a . acknowledges hospitality from the institut dastrophysique de paris , where the early plan for this work was conceived . e.b and j.s .
acknowledge hospitality from the lorentz center ( leiden , nl ) , where part of this work was carried out .
computations were performed on the gpc supercomputer at the scinet hpc consortium , as well as on the horizon cluster at the institut dastrophysique de paris . | studying how nuclear star clusters ( nscs ) form and how they are related to the growth of the central massive black holes ( mbhs ) and their host galaxies is fundamental for our understanding of the evolution of galaxies and the processes that have shaped their central structures .
we present the results of a semi - analytical galaxy formation model that follows the evolution of dark matter halos along merger trees , as well as that of the baryonic components .
this model allows us to study the evolution of nscs in a cosmological context , by taking into account the growth of nscs due to both dynamical friction - driven migration of stellar clusters and star formation triggered by infalling gas , while also accounting for dynamical heating from ( binary ) mbhs .
we find that in - situ star formation contributes a significant fraction ( up to @xmath0 ) of the total mass of nscs in our model . both nsc growth through in - situ star formation and through star cluster migration
are found to generate nsc host galaxy scaling correlations that are shallower than the same correlations for mbhs .
we explore the role of galaxy mergers on the evolution of nscs , and show that observational data on nsc host galaxy scaling relations provide evidence of partial erosion of nscs by mbh binaries in luminous galaxies .
we show that this observational feature is reproduced by our models , and we make predictions about the nsc and mbh occupation fraction in galaxies .
we conclude by discussing several implications for theories of nsc formation . |
it is well - known that a spin-@xmath12 particle associated with a no self - conjugate field has @xmath13 cp - odd permanent electromagnetic moments . in particular , a charged @xcite or neutral vector particle @xcite has two electromagnetic moments , namely , the electric dipole moment ( edm ) and the magnetic quadrupole moment ( mqm ) .
the scrutiny of these properties may provide relevant information for our knowledge of cp violation , which still remains a mysterious phenomenon whose experimental validity has been well established via some flavor - changing processes such as the mixing of the @xmath14 @xcite and @xmath15 @xcite hadrons .
there is no yet conclusive evidence that the origin of cp violation in @xmath16 mixing is the cabibbo - kobayashi - maskawa ( ckm ) phase @xcite , which is the only source of cp violation in the electroweak sector of the standard model ( sm ) , but recent results from @xmath15 factories at slac and kek strongly suggest @xcite that the dominant contribution to @xmath17 mixing arises from such a phase indeed .
this means that , as far as @xmath15 hadron physics is concerned , there is no much room left to detect new sources of cp violation . on the other hand ,
diverse studies @xcite show that the ckm phase has a rather marginal impact on flavor - diagonal processes such as the electric dipole moments of elementary particles , which means that they could be highly sensitive to new sources of cp violation .
in fact , neither the fermions nor the @xmath0 gauge boson can have edms at the one - loop level because a cp - violating phase can not arise at this order since the corresponding amplitudes depend only on the absolute value of the ckm matrix elements .
it has been shown @xcite that the edm of both quarks and @xmath0 boson vanishes also at the two - loop level and appears first at the three - loop level @xcite . in the case of charged leptons , without the presence of right - handed neutrinos the edm is still more suppressed and it can only be generated at the four - loop level or higher orders @xcite . however , the existence of new sources of cp violation in this sector is expected due to the recent discovery of neutrinos masses and lepton mixing @xcite . in contrast , as far as the @xmath0 mqm is concerned , it has been shown @xcite that it receives a tiny contribution at the two - loop level in the sm .
apart from the mere theoretical interest , the study of the edm and mqm of the @xmath0 boson is important from the experimental point of view because they can induce large contributions to the edm of the fermions @xcite , such as the electron or the neutron , which may be at the reach of low - energy experiments .
it is thus worth investigating new sources of cp violation beyond the sm .
although these properties of the @xmath0 boson can only be generated through loop effects within renormalizable theories , they may receive large contributions in many sm extensions @xcite .
the only class of models which can generate these quantities at the one - loop level are those involving both left- and right - handed currents with complex phases @xcite . in principle , this one - loop generated effect would contribute dominantly to these @xmath0 properties , but it could be strongly suppressed due to the presence of a tiny complex phase , as occurs in left - right symmetric models(lrsm ) @xcite due to experimental constraints on the @xmath18 mixing . although the presence of at least one fermionic loop involving a dirac trace is required to generate a term proportional to the levi - civita tensor in the @xmath19 vertex , the combination of fermion and scalar fields may supply a potential source of cp violation in flavor - diagonal processes , provided that their interactions involve both scalar and pseudoscalar couplings .
the simultaneous presence of these types of higgs - fermion couplings violate cp invariance , which in turn can induce a trilinear higgs-@xmath0 vertex with similar cp property at the one - loop level .
a @xmath20 vertex including a linear combination of cp - even and cp - odd couplings is enough to generate a cp - odd component in the on - shell vertex @xmath19 , which would correspond to a two - loop effect in a renormalizable theory .
although this source of cp violation is generated at the two - loop level , it could give a contribution to the cp - odd electromagnetic moments of the @xmath0 boson larger than those induced by other alternative sources .
this mechanism does not depend crucially on the existence of a complex phase since it is a direct consequence of the presence of higgs - fermion couplings that violate cp invariance in the fundamental lagrangian , which contrasts with the case of the ckm phase or that arising from left- and right - handed charged currents .
indeed , it is not necessary to go beyond the fermi scale to introduce the most general renormalizable cp - violating @xmath21 vertex .
it arises for instance in the yukawa sector of the type - iii two - higgs doublet model ( thdm ) @xcite , where the @xmath20 vertex is induced at the one - loop level as a linear combination of cp - even and cp - odd couplings , which in turn generate the most general on - shell @xmath19 vertex including both cp - even and cp - odd dynamical structures at the two - loop level .
although there are many types of feynman diagrams contributing to the @xmath0 cp - odd properties in this model @xcite , the contribution from the one - loop @xmath20 vertex generated by the yukawa coupling @xmath22 differs from any other source as it leads to a finite and gauge invariant result by itself @xcite .
the same type of effect can arise from heavy fermions that may be present in several sm extensions , and we will focus on this possibility .
although sm extensions composed by very complex higgs sectors are common , it is worth emphasizing that most of them contain at least one sm - like higgs boson @xmath23 .
such a higgs boson is sm - like in the sense that it is expected to be relatively light , with a mass of order of the fermi scale , and with tree level couplings presenting small deviations from the sm that vanish in some appropriate limit . in this work ,
we are interested in studying this class of deviations , which may lead to the appearance of cp - odd electromagnetic properties of the @xmath0 boson .
more specifically , we will concentrate on the most general @xmath3 vertex including both cp - even and cp - odd components . instead of focusing on a specific model , it is convenient to parametrize this class of effects in a model independent manner via the effective lagrangian technique @xcite
. it is assumed that these effects are induced by particles that are much heavier than the fermi scale and can thus be integrated out in the generating functional .
this framework is quite appropriate to describe any physics phenomenon that is absent or very suppressed in the sm .
apart from the advantages of working in a model - independent fashion , this approach has some additional advantages from the technical point of view .
in particular , a two - loop calculation , as the one we are interested in , can be treated as a one - loop effect .
below , we will discuss the structure of the effective @xmath3 vertex and its implications on the edm and mqm of the @xmath0 boson .
the rest of the paper is organized as follows . in sec .
ii , the @xmath4-invariant effective lagrangian description of the @xmath3 vertex is presented and used to determine its impact on the on - shell @xmath19 vertex at the one - loop level within the effective theory , whereas secs .
iii and iv are devoted to discuss our results and present our conclusions , respectively .
a scheme that is well suited to analyze new physics effects lying beyond the fermi scale consists in introducing @xmath4-invariant operators of dimension higher than four that modify the sm dynamics .
in particular , the structure of the tree level @xmath3 vertex can be modified by introducing a dimension - six effective lagrangian given by @xmath24 where @xmath25 is the @xmath26 field strength , @xmath27 , and @xmath28 is the standard model higgs doublet . the parameters @xmath29 and @xmath30 parametrize the details of the underlying physics , and
they could be determined once the fundamental theory is known . on the other hand , @xmath31 and @xmath32
are new different physics scales , the latter being associated with cp - violating effects .
such anomalous higgs-@xmath0 interactions have already been considered in the literature @xcite .
for instance , they were introduced to analyze the cp structure of the @xmath3 coupling at the cern large hadron collider ( lhc ) @xcite , and more recently at next linear colliders ( nlc ) @xcite . also , it is worth mentioning that this vertex already arises at the one - loop level within the context of the sm provided that at least one @xmath0 boson is off - shell @xcite . for the purpose of this work ,
it is enough to concentrate on the first and last terms of the lagrangian ( [ el ] ) .
our main goal is to study the impact of this anomalous @xmath3 vertex on the cp - odd electromagnetic properties of the @xmath0 boson , which are generated at the one - loop level in the context of effective lagrangians .
the most general on - shell @xmath19 vertex can be written as a linear combination of cp - even and cp - odd electromagnetic gauge structures @xcite : @xmath33 where @xmath34 ( @xmath35 ) is the cp - even ( cp - odd ) component : @xmath36 + 2\delta \kappa ( q_\beta g_{\alpha \mu}-q_\alpha g_{\beta \mu})+\frac{4\delta q}{m^2_w}p_\mu q_\alpha q_\beta,\\ \gamma^o_{\alpha \beta \mu}&=&2\delta \widetilde{\kappa}\epsilon_{\alpha \beta \mu \lambda}q^\lambda+\frac{4\delta \widetilde{q}}{m^2_w}q_\beta \epsilon_{\alpha \mu \lambda \rho}p^\lambda q^\rho.\end{aligned}\ ] ] the notation and conventions used in these expressions are shown in fig .
the @xmath37 and @xmath38 form factors define the cp - conserving electromagnetic moments of the @xmath0 boson , the magnetic dipole moment @xmath39 and the electric quadrupole moment @xmath40 , through the following relations @xmath41 on the other hand , @xmath5 and @xmath6 determine the cp - violating edm and mqm : @xmath42 we now turn to show that a @xmath3 vertex involving a linear combination of both cp - even and cp - odd couplings induces the @xmath43 and @xmath7 form factors at the one - loop level . as already mentioned , this is a two - loop or higher order effect since the effective operator @xmath44 can only be generated at one - loop or higher orders by the underlying theory @xcite .
ignoring the term associated with the coefficient @xmath45 , the higgs-@xmath0 interaction can be written , in the unitary gauge , as @xmath46,\ ] ] where @xmath47 .
we have introduced the definition @xmath48 , being @xmath49 gev the fermi scale .
notice that , due to @xmath4-invariance , the @xmath19 vertex receives contributions from both the trilinear @xmath3 and quartic @xmath50 couplings .
the feynman rules needed to calculate the edm and mqm are shown in fig .
[ fig2 ] , with the @xmath51 tensor given by @xmath52 in the unitary gauge , the cp - violating @xmath3 vertex contributes to the @xmath5 and @xmath7 form factors through the feynman diagrams shown in fig .
it is worth noting that the effective lagrangian given in eq .
( [ el ] ) generates nonrenormalizable dimension - five @xmath3 and @xmath53 vertices , which yield a divergent contribution to @xmath43 . after evaluating the feynman diagrams of fig .
[ fig3 ] , we arrive at an ultraviolet divergent amplitude , which is due to the presence of a nonrenormalizable interaction , thereby requiring a renormalization scheme . we have used the @xmath54 scheme with the renormalization scale @xmath55 , which leads to a logarithmic dependence of the form @xmath56 .
after some algebra , the cp - odd form factors can be written in terms of two - point passarino - veltman scalar functions : @xmath57,\end{aligned}\ ] ] @xmath58,\ ] ] where @xmath59 , @xmath60 , @xmath61 , and @xmath62 .
notice that the @xmath7 form factor is ultraviolet finite , which is consistent with the fact that @xmath1 and @xmath2 satisfy the relation @xmath63 in any renormalizable theory as @xmath7 can not arise at the one - loop level @xcite , but at higher orders .
since our calculation represents a two - loop or higher order effect in the fundamental theory , the contribution to @xmath7 must be necessarily finite , in accordance with renormalization theory .
the trilinear @xmath19 vertex .
the black circle denotes anomalous contributions.,width=144 ] feynman rules for the vertices @xmath3 , @xmath53 , and @xmath19 in the unitary gauge.,width=240 ] feynman diagrams contributing to the @xmath19 vertex in the unitary gauge.,width=384 ]
we turn now to our numerical results .
the edm and mqm of the @xmath0 boson depend on three free parameters : the coupling constant @xmath30 , the new physics scale @xmath32 , and the higgs boson mass @xmath64 .
as already mentioned , the effective operator @xmath65 can only be generated at one - loop or higher orders by the fundamental theory @xcite .
assuming that it is induced at the one - loop level , the @xmath30 parameter must contain a factor of @xmath66 along with a @xmath67 coupling for each gauge field . from these considerations , it is reasonable to assume that @xmath68 , where @xmath69 is a dimensionless loop function , whose specific structure depends on the details of the underlying physics . since the cp - violating effects are expected to be of decoupling nature , @xmath70 is expected to be of the order @xmath71 at most . in order to make predictions , we will adopt a somewhat optimistic scenario , which consists in assuming that @xmath72 . we will thus make predictions under the assumption that @xmath73 has the following form @xmath74 we now would like to analyze the behavior of @xmath1 and @xmath2 as a function of @xmath64 and @xmath32 . the dependence of @xmath1 ( @xmath2 ) on the higgs boson mass is shown in fig . [ fig4 ] ( [ fig5 ] ) for @xmath64 ranging between 120 gev and 200 gev and for @xmath751 , 3 and 5 tev . from these figures
we can observe that @xmath1 and @xmath2 are not very sensitive to the higgs boson mass .
in fact , for @xmath76 tev , @xmath1 ranges between @xmath77 and @xmath78 e@xmath9 cm , whereas @xmath2 goes from @xmath79 to @xmath80 e@xmath9@xmath11 . as compared to the values obtained for @xmath76 tev , @xmath1 and @xmath2 are decreased by a factor of @xmath81 when @xmath82 tev and @xmath83 when @xmath84 tev . finally , the behavior of the @xmath0 moments is shown in fig . [ fig6 ] as a function of @xmath32 and for @xmath85 gev .
the electric dipole moment @xmath1 as a function of @xmath64 for @xmath86 , 3000 , and 5000 gev.,width=240 ] the same as in fig . [ fig4 ] but now for the magnetic quadrupole moment @xmath2.,width=240 ] the @xmath1 and @xmath2 dependence on @xmath32 for @xmath85 gev.,width=240 ] it is worth comparing our results with those obtained in other scenarios . to begin with
, we would like to discuss the sm predictions for @xmath1 and @xmath2 . as already noted ,
the lowest order nonzero contribution to @xmath1 arises at the three - loop level , whereas @xmath2 appears up to the two - loop order . at the lowest order ,
@xmath1 has been estimated to be smaller than about @xmath87 e@xmath9 cm @xcite . as far as @xmath2 is concerned
, it has been estimated to be about @xmath88 e@xmath9@xmath11 @xcite .
in contrast , some sm extensions predict values for @xmath1 that are several orders of magnitude larger than the sm one . it must be emphasized here that all these studies have focused only on @xmath1 .
for instance , a value of @xmath89 e@xmath9 cm was estimated for @xmath1 in lrsm @xcite , and similar results were found in supersymmetric models , which induce this moment via one - loop diagrams mediated by charginos and neutralinos @xcite .
also , a nonzero @xmath1 can arise through two - loop graphs in multi - higgs models @xcite .
explicit calculations carried out within the context of thdms show that @xmath90 e@xmath9 cm @xcite . a similar value was found in the context of the so - called @xmath91 models , which induce a nonzero @xmath1 via two - loop graphs similar to the ones of thdm @xcite . from these results
, we can conclude that our model - independent estimation for @xmath1 lies within the range of the predictions obtained from most popular renormalizable theories .
it is well known that the @xmath0 edm and mqm can induce important contributions to the edm of light fermions .
this fact was exploited by the authors of ref .
@xcite , who used the experimental upper bound on the neutron edm , @xmath92 e@xmath9 cm , to obtain the upper bound @xmath93 e@xmath9 cm .
our prediction for @xmath1 , which is eight orders of magnitude larger than that of the sm , is consistent with this upper bound .
as far as @xmath2 is concerned , currently there is no indirect experimental upper bound , but we would like to emphasize that our result is quite remarkable as predicts that new physics effects are capable of enhancing @xmath2 up to fifteen orders of magnitude above the sm contribution .
as far as the direct observation of the cp - odd structure of the @xmath19 vertex is concerned , the prospect of the nlc and clic @xcite have triggered the interest in the @xmath94 reaction as an efficient tool to produce large quantities of @xmath0 boson pairs , which would allow one to study new physics effects on the @xmath95 ( @xmath96 ) vertex .
the possibility of extracting cp - odd asymmetries from these colliders has been examined by several authors , mainly in a model - independent approach via effective lagrangians @xcite .
these studies suggest that an effect in @xmath1 at the level of @xmath97 e@xmath9 cm may be at the experimental reach .
also , it is expected that careful studies on the polar and azimuthal distributions of lepton - antilepton pairs produced in @xmath0 decays may further enhance the constraints on the size of @xmath1 @xcite .
the origin of cp violation is a fascinating open problem worthwhile of both theoretical and experimental attention .
although flavor - diagonal cp - violating processes , such as the electric dipole and magnetic quadrupole moments of the @xmath0 boson , are not induced by the ckm mechanism , their presence can not be dismissed since they may represent a genuine effect of physics lying beyond the fermi scale .
this is suggested by many sm extensions predicting the existence of new sources of cp violation that may give sizeable contributions to the static quantities of the @xmath0 boson .
as far as renormalizable theories are concerned , the yukawa sector of several sm extensions seems to be a potential source of cp violation since it may involve the presence of higgs bosons with both scalar and pseudoscalar couplings to the fermions . in this work
, we examined such a possibility via the model - independent approach of effective lagrangians .
we assumed the existence of new heavy fermions that generate the most general @xmath3 vertex at the one - loop level .
such a cp - violating nonrenormalizable interaction was parametrized through an @xmath4-invariant operator and combined with the usual cp - even sm coupling to calculate the one - loop contribution to the cp - odd structure of the on - shell @xmath19 vertex .
the fact that the dimension - six operator can only be generated at the one - loop level means that the cp - odd property of the @xmath19 vertex is a two - loop effect in the fundamental theory .
explicit expressions for the form factors @xmath5 and @xmath7 , which define the edm and mqm of the @xmath0 boson , were derived .
an appropriate renormalization scheme was adopted in order to renormalize the divergent form factor @xmath43 . as for the form factor @xmath7 ,
its nondivergent nature reflects the fact that it can not arise at the one - loop level in any renormalizable theory . assuming reasonable values for the unknown parameters of the effective theory , the estimated values of the cp - odd electromagnetic moments are @xmath98 e@xmath9 cm and @xmath99 e@xmath9@xmath11 .
these values are eight and fifteen orders of magnitude above the respective sm contributions . to our knowledge ,
this model - independent estimation for the size of the @xmath2 is the first one obtained in theories beyond the sm . on the other hand ,
the value predicted for @xmath1 is of the same order of magnitude or larger than those predicted by some sm extensions , and it is consistent with the existing indirect upper bound derived from the neutron electric dipole moment . for a review on the phenomenological aspects of thdm - iii
see : m. sher , arxiv : hep - ph/9809590 ; d. atwood , l. reina and a. soni , econf * c960625 * , lth093 ( 1996 ) ; d. atwood , l. reina and a. soni , phys .
d * 55 * , 3156 ( 1997 ) .
w. a. bardeen , r. gastmans and b. lautrup , nucl .
b * 46 * , 319 ( 1972 ) ; k. j. f. gaemers and g. j. gounaris , z. phys .
c * 1 * , 259 ( 1979 ) ; k. hagiwara , r. d. peccei , d. zeppenfeld and k. hikasa , nucl .
b * 282 * , 253 ( 1987 ) ; u. baur and d. zeppenfeld , nucl .
b * 308 * , 127 ( 1988 ) ; _ ibid _ b * 325 * , 253 ( 1989 ) .
i. vendramin , nuovo cim .
a * 105 * , 1649 ( 1992 ) ; t. h. west , phys . rev .
d * 50 * , 7025 ( 1994 ) ; t. kadoyoshi and n. oshimo , phys . rev .
d * 55 * , 1481 ( 1997 ) ; n. oshimo , nucl.phys .
suppl . * 59 * , 231 ( 1997 ) .
a. queijeiro , phys .
b * 193 * , 354 ( 1987 ) ; f. boudjema , k. hagiwara , c. hamzaoui and k. numata , phys .
d * 43 * , 2223 ( 1991 ) .
see also , j. ellison and j. wudka , ann .
nucl . part .
* 48 * , 33 ( 1998 ) . | the possibility of nonnegligible @xmath0 electric dipole ( @xmath1 ) and magnetic quadrupole ( @xmath2 ) moments induced by the most general @xmath3 vertex is examined via the effective lagrangian technique .
it is assumed that new heavy fermions induce an anomalous cp - odd component of the @xmath3 vertex , which can be parametrized by an @xmath4-invariant dimension - six operator .
this anomalous contribution , when combined with the standard model cp - even contribution , lead to cp - odd electromagnetic properties of the @xmath0 boson , which are characterized by the form factors @xmath5 and @xmath6 .
it is found that @xmath5 is divergent , whereas @xmath7 is finite , which reflects the fact that the latter can not be generated at the one - loop level in any renormalizable theory .
assuming reasonable values for the unknown parameters , we found that @xmath8 e@xmath9 cm , which is eight orders of magnitude larger than the sm prediction and close to the upper bound derived from the neutron electric dipole moment .
the estimated size of the somewhat less - studied @xmath2 moment is of the order of @xmath10 e@xmath9@xmath11 , which is fifteen orders of magnitude above the sm contribution . |
the study of cataclysmic variables ( cvs ) in our galaxy is plagued by the same problem that afflicts so many other areas of astrophysics : uncertainty in the distances to , and hence luminosities of the objects being studied . a second and
no less severe problem is the space density and spatial distribution uncertainties caused by our parochial view of the milky way .
while we can detect erupting galactic classical novae several kpc from the sun , most other cvs are located closer than about 500 pc .
the reason is simple : novae often achieve @xmath1 @xmath2 , rivaling the most luminous objects in our galaxy for weeks at a time , while dwarf novae ( dn ) , nova - like variables and their magnetic cousins rarely exceed 10 @xmath2 .
astronomers have succeeded in cataloguing barely a thousand galactic cvs in over a century of searching . only for a handful
are ironclad distances published from hubble space telescope parallaxes @xcite .
expansion parallaxes for classical novae contribute another 20 or so reasonably secure distances and luminosities ( see @xcite for a summary ) .
it would clearly be of enormous benefit to cv science if hundreds or thousands of cataclysmic variables , all at the same distance , could be located .
accurate luminosity functions , and bias - free period distributions and eruption frequencies would become available to confront inadequately - constrained theoretical models .
new sub - classes of cvs might be discovered , and systematic variations in outburst properties and binary orbital distributions would likely be uncovered .
the situation is improving for cvs in a few special places : the crowded cores of globular clusters .
recent detections of dozens of cv candidates in 47 tuc @xcite , allow , for the first time , the comparative study of many cataclysmics , _ all at the same distance_. unfortunately , hst and chandra are essential for followup studies , and these telescopes are two of the rarest commodities in astrophysics .
it may be decades before the newly discovered cvs are fully characterized .
worse , many of these cvs probably formed by tidal captures and/or evolved violently under the influences of passing stars .
only one globular dn is easily resolved and studied from the ground : v101 in m5 @xcite .
its rather long ( 5.79 hour ) period @xcite warns us that it may be anomalous .
cvs in the cores of globular clusters have much to teach us , but they re likely to be very different from field cvs .
the outlook is potentially more promising in the magellanic clouds . as in the case of globular clusters
, all cvs discovered in the clouds are at nearly the same distance , so direct luminosity comparisons are meaningful .
spatial densities of field lmc and smc stars are roughly comparable to those in the neighborhood of the sun , rather than those typical of globular cluster cores , so tidal encounters almost never happen .
cvs in the clouds must form and evolve via ordinary binary evolution , just as in the field in our own galaxy .
it is certainly true that the metallicities of the red dwarf companions to the white dwarfs in magellanic cvs will usually be lower than those in the galaxy .
lower metallicity is a much less drastic effect @xcite than dynamical interactions @xcite in globular clusters , so direct comparisons between galactic and magellanic cvs should be extremely fruitful .
finally , the clouds must be home to @xmath3 or more cvs ( see below ) , an enormous sample likely to contain every subtype of cv we know , and perhaps some that we do nt yet recognize . before carrying out galaxy - lmc population comparisons , we must , of course find cvs in the clouds . since 1897
about 35 erupting classical novae have been spotted in the lmc and smc .
accurate ( 1 - 2 arcsecond ) positions for most of these , and recovery of some in deep @xmath4 , @xmath5 and @xmath6 images has recently occurred @xcite .
the current sample of quiescent cloud novae is important to follow - up , but will grow in number only very slowly .
this is because all classical novae must recur @xcite with inter - eruption times of at least @xmath7 yr .
thus the many thousands of classical novae that exist in the clouds will only slowly reveal themselves , via eruptions , over many millennia . as dwarf novae
are a substantial sub - population of cvs which reveal themselves through outbursts every few weeks to months , _ almost all _ dn in any given field should be identifiable , at least in principle , in a deep survey of order 6 - 12 months in length .
luminous erupting dn should achieve @xmath8 , @xmath9 near maximum ( see section 6 ) and thus be detectable except in the most crowded lmc fields .
surveying the entire lmc and smc often enough to find nearly all erupting dn is a daunting observational program that can and will eventually be undertaken . in this paper
we report a much more modest but realistic feasibility study to demonstrate , for the first time , the existence of erupting dn in the lmc .
we also derive a crude estimate of the total number of luminous dn in the lmc . in section 2 we present the observational strategy and database , while the photometry calibrations are described in section 3 .
completeness tests are detailed in section 4 , and the six erupting dn candidates are shown in section 5 .
simulations of erupting dn in the lmc are compared to these candidates in section 6 .
we discuss whether the candidates could be other kinds of variables , and the implications for cv numbers and space densities if the candidates are true dn in section 7 .
we briefly summarize our results in section 8 .
to maximize our chances of detecting erupting lmc dn we requested the longest dark run with the largest telescope ( the 4-m)and widest field imager ( mosaic2 8k @xmath0 8k ) available in 1999 at ctio .
we were awarded the dark nights of 1999 december 2 , 4 , 6 , 8 , 10 and 12 .
all nights were clear , with seeing mostly in the range @xmath10 arcsec . because dn are usually remarkably blue ( @xmath11 is a good rule - of - thumb ; though much redder values do occur @xcite ) we opted to image our chosen lmc field in @xmath4 and @xmath6 filters .
three to eight images were obtained each night in each filter , with total exposure times ( typically ) of 2.5 hours in @xmath4 and 1 hour in @xmath6 .
a detailed log is given in table 1 .
the plate scale was 0.270 arcsec / pixel , and our total field of view was 1362 arcmin@xmath12 = 0.38 deg@xmath12 .
the lmc field we chose is centered at ra ( j2000 ) 05:33:36.7 and dec ( j2000 ) -70:33:44 .
the field of view was chosen because five erupting classical novae ( the novae of 1948 , 1970a , 1970b , 1981 and 1988a ) have been observed in the area covered by our ccd array . . .
more than any other location in or near the lmc .
we were thus able to monitor these five old novae `` for free '' while searching for erupting dwarf novae .
the results of the old nova monitoring will be reported elsewhere @xcite .
an overall view of the location of this field within the lmc is shown in figure 1 , with a magnified view of the eight ccd fields superposed .
the field is at the southeastern end of the lmc bar , and crowding is significant . the large surface density of stars hampered efforts to detect faint dn , but was critical to maximize the chances of detecting at least a few bright erupting dn .
the individual mosaic2 @xmath13 images were combined using the `` montage2 '' routine contained within the stand - alone daophot package . on a given night , individual frames were taken with a total vertical dither of @xmath14 pixels .
these frames were run through daophot s matching program `` daomaster '' to derive the subpixel frame - to - frame shifts .
these shifts were passed to montage2 , which produced a sky subtracted median image of the individual frames .
the calibrations were derived from a set of standard magnitudes taken from the central region of m67 as given in @xcite . to ensure that the central region of m67 was observed on all eight chips ,
the cluster was observed over two nights on the `` right hand '' side of the array and then later on the `` left hand '' side .
however , the `` left hand '' side was observed on the fifth night , and was therefore unusable due to poor seeing .
nevertheless , the zero - points between chips were consistent to within .05 magnitudes , and the zero - points derived for the `` right hand '' side were used for the entire data set . due to the variable seeing
, there also exist very subtle differences in the photometry on a night - to - night basis .
thus , in addition to the global zero - point calibration uncertainty described above , there also exist night - to - night zero - point corrections on the order of .05-.1 mag .
these corrections were found by first searching for the least variable ( most constant ) stars that returned valid photometry for all of the six nights .
this was done in an iterative manner to locate stars with variability at least three times less than the average of the ensemble .
once this goal was achieved , a night - to - night average was calculated , and this was subtracted from the photometry of the returned candidates .
none of these rather small uncertainties has any effect on the candidate detections and conclusions of this paper .
accurate photometry and astrometry of any source is limited not only by the source s brightness , but also by the degree of crowding in the field
. simulations were performed to determine the efficiency of recovery of a collection of synthetic stars within a representative sub - region of the dataset as a function of brightness .
these simulations use the actual point spread function for a given night and filter , yielding a measure of the effective plate limit and the overall detection sensitivity of these observations and data reductions .
first , the same small sub - region of the entire field was chosen in both the @xmath4 and @xmath6 bands and from each of the six nights .
the stars in this 500 @xmath0 500 pixel sub - image showed crowding typical of the rest of the dataset ( @xmath15 4000 stars in the sub - image ) . moreover ,
the star brightness range in the overall dataset was well matched by the range in this sub - image .
next , in order to generate a set of synthetic stars , the `` addstar '' routine contained within daophot was used to place 100 copies of the point - spread function randomly across an image and at a set brightness .
this was repeated 10 times , for a total of 1000 synthetic stars at a given brightness , night , and filter .
then `` allstar '' returned photometry on all the stars from these 10 runs .
the fraction of the input synthetic stars that were recovered gives the level of completeness for that given brightness .
a synthetic star was considered `` recovered '' if it was measured within @xmath16 magnitudes of its input value .
the entire procedure was then repeated with a different brightness for the input synthetic stars . a total magnitude range of @xmath17 for both the @xmath4 and @xmath6 filters was covered in increments of 0.25 magnitudes each .
the results are shown in figures 2a and 2b .
each of the completeness curves reflect well the seeing of each night .
for example , the completeness for the third night ( best seeing ) is 70% at @xmath18 , and 40% at @xmath19 , typical of an lmc dn near eruption maximum . on the fifth night ( the worst seeing night ) the completeness at these same magnitudes is barely 2 - 3% .
fortunately every night s imagery except that of night 5 was deep enough and complete enough for us to expect to see at least @xmath20 of all erupting dn at or near peak brightness .
for a field with such a high stellar density ( @xmath21 stars / arcmin@xmath12 ) , the construction of a good point spread funtion ( psf ) is hampered most by neighboring stars crowding the psf stars .
much effort was expended to automate the search for variables entirely , comparing lists of psf photometry stars on successive nights . night - to - night seeing
changes produced many candidates in the automated search that were rejected upon visual inspection .
rejection of a candidate invariably occurred because the `` candidate '' revealed itself to be an artifact created by overlapping psfs of several stars , changing with the seeing from night to night .
the dwarf - nova candidates were ultimately found by blinking rapidly through the best four of the six nightly frames to look for any subtle changes in the appearance of the stellar field .
dwarf novae should rise from invisibility to easily detectable in the 48 hours between observations . to look for changes on a fairly detailed level ,
each chip was visually divided into 32 sub - regions , and blinked rapidly . brightening from previously empty regions of the sky
is easily seen . a star that is ramping up or down in brightness by @xmath22 mags
can also be easily detected since the transition between the first and last night is very prominent .
artificial dwarf novae placed in the frames were visually detected with efficiencies similar to the completeness curves of figure 2 . to be recognized as a candidate , a variable had to be visible in both the @xmath4 and @xmath6 medianed frames of at least one of the six nights , and to have varied by over one magnitude in both filters between any two nights .
this selection method initially revealed 14 dwarf nova candidates from the entire dataset .
a second criterion was then applied to weed out short - period , large amplitude variables : no candidate could vary significantly ( more than about 1.0 magnitudes ) from frame to frame during any one night .
it is certainly true that the light output of many cvs `` flickers '' on timescales of seconds to hours , and that some cvs undergo short , deep eclipses .
we nevertheless adopted this conservative approach to eliminate non - cv , short - period eclipsing binaries .
inspection of each night s individual images confirmed that 7 candidates were likely eclipsing binaries , and one candidate was so crowded that we can not say for sure that it is a real variable .
the eclipsing variables may be useful for lmc distance determinations or other projects , and so we supply their finder charts , and coordinates in figure 8 and table 3 , respectively .
we are thus left with six good candidate erupting dn .
we define a good candidate as one that is seen in both @xmath4 and @xmath6 on at least one night ; on all images in each filter of that night ; and whose variability characteristics are consistent with those of at least one well - studied galactic dwarf nova ( see next section ) .
the nightly images of the six dn candidates are shown in figure 3 , and their photometry is presented in figure 4 .
the poisson error is the dominant photometric error in these plots , although readnoise and sky noise also have been incorporated into the error bars shown .
we defer a discussion of these candidates and their implications until after the next section , where we simulate expected dn images and light curves .
to complement the completeness tests described in section 4 , simulations of eruptions of several well known and characterized galactic dwarf novae , artificially placed in the actual lmc data , were needed .
eruptions of ss cyg , u gem , ss aur , ar and , and em cyg were all simulated in a relatively uncrowded but otherwise representative patch of the lmc field ( an uncrowded field was chosen to highlight the photometric appearances of lmc dn , unhindered by the crowding or incompleteness addressed in figure 2 ) .
the images from these simulations are shown in figure 5 .
the three dwarf novae ss cyg , u gem and ss aur have accurately determined parallaxes @xcite and , using their well tabulated apparent magnitudes @xcite , the absolute magnitudes are easily calculated .
the absolute magnitudes for ar and and em cygni were obtained from @xcite .
these five objects were chosen to cover the range of absolute magnitudes seen for erupting and quiescent dn .
our simulated dwarf novae all reflect the rise , plateau , and decline times from @xcite , start at quiescence on the first night of observations and attain maximum brightness on the second observing night ( 48 hours later ) , have all been placed at the distance of the lmc and , with one exception noted below , use mean @xmath23 colors from @xcite .
we have assumed that the lmc dn are dimmed by @xmath24 = 0.3 and @xmath25 = 0.5 , in accord with @xcite .
psf photometry was performed on these frames ; the resulting light curves and retrieved photometry are shown in figure 6 .
an additional simulated ss cygni eruption sequence ( labeled `` ss cyg sequence 2 '' in figure 6 ) incorporated more detailed information about this object s color evolution with time . during its eruption , ss cygni becomes distinctly more red , reaching @xmath26 before attaining maximum brightness @xcite .
this second ss cyg sequence utilizes these colors , and has been shifted by one day in phase .
that is , the first simulated observation of this second sequence catches the dwarf nova on the rise .
( `` ss cygni sequence 1 '' shows the artificial dn in quiescence on the first night ) .
this second ss cyg sequence was included because it mimics rather well the overall observed light curves of dn candidates shz1 , shz4 and shz6 , though the latter two objects are still redder than the simulated dn .
we regard this difference as minor because reddening does vary strongly , and on small spatial scales , across the lmc @xcite .
it is clear from figure 5 that we are unlikely to have been able to detect erupting dn like u gem , ss aur , ar and or em cyg all are too intrinsically faint to be detected , in our crowded fields
. if our candidates are really dn they are much more likely to be ss cygni , or possibly su uma - type cvs in superoutburst .
comparison of the six candidates shown in figure 3 with the lmc - dn simulation images shown in figure 5 ( particularly `` ss cyg sequence 2 '' and `` su uma superoutburst '' ) demonstrates that the brightness and variability behaviors of our candidates are broadly consistent with those expected of luminous erupting dwarf novae in the lmc .
this is further supported by comparison of the measured and simulated light curves of figures 4 and 6 , respectively .
however , it is by no means certain that all or even some of the six variables shown in figure 3 are really lmc erupting dwarf novae .
what are the other possible variables that might mimic lmc dn behavior ?
amongst these are : chance superpositions of background supernovae or classical novae , gamma ray bursts ( grb ) , microlensing events , milky way variables along the line of sight to the lmc , and non - cv lmc variables .
the microlensing hypothesis can immediately be discarded because of the non - grey light curve behaviors , non - symmetric shapes of the light curves and ( in most cases ) overly long time at maximum light .
grb afterglows typically achieve r or i @xmath27 one day after outburst .
while the peak brightness of these variables is in accord with the grb hypothesis , we can eliminate this possibility because none of the candidates declines quickly enough to be a grb .
large area , multi - epoch surveys for faint variables ( e.g. @xcite ) show that the surface density of our variables is far too high for them to be field rr lyraes , classical novae or supernovae . the brightness and moderately blue colors of the candidates
rule out other types of galactic variables .
rr lyraes in the galaxy or lmc would be considerably brighter , and flare stars do nt match the observed nightly brightness profiles and/or blue colors .
a final possibility to consider is that some of our candidates might be galactic or lmc eclipsing binaries .
galactic binaries would have to be rather low luminosity systems ( say @xmath28 ) and thus much redder than observed to appear at @xmath29 .
eclipsing lmc systems would range in luminosity from @xmath30 ( shz6 ) to @xmath31 ( shz3 and shz5 ) , corresponding to a - type main sequence stars .
the amplitudes of light variation in shz1 , shz3 , shz4 , shz5 and shz6 are greater than 0.7 magnitude and thus preclude equal mass ( and brightness ) binaries , but a hot pre - white dwarf + a cooler companion star could mimic the observed light curves for shz1 , shz5 and shz6 .
as noted earlier , we see no sign of significant variability for any of the six candidates on the individual frames taken during the two to five hours of observations during each of the six nights of the run .
the candidates light curves support their tentative identifications as dn , though some may turn out to be other kinds of variables
. to be absolutely certain of these objects identities will require challenging follow - up observations .
four possible confirmation techniques are the following : \1 ) spectra near quiescence ( to demonstrate the presence of balmer emission lines ) would be definitive proof , but the dn are then expected to be near 25th magnitude ... beyond the likely capability of even an 8 meter telescope in such crowded fields .
\2 ) imagery every night or two for several months with a 4 meter class telescope should reveal repeated eruptions separated by weeks to months .
such a program would also distinguish erupting cvs from lmc eclipsing binaries , where the emergence of a hot , blue star from eclipse seen only at one epoch can be confused with a genuine dn eruption .
this can be done , but will be very demanding of large telescope time .
\3 ) ultraviolet imagery of the fields of the six candidate dwarf novae with the hubble space telescope ( hst ) might reveal uv - bright objects , as the spectral energy distributions of most cvs rise sharply into the uv .
4 ) several hours of time - resolved uv or optical photometry , again with hst , might reveal the flickering and/or orbital modulation characteristic of cvs .
space density estimates of cvs near the sun suggest @xmath32stars@xmath33 @xcite .
space density estimates of all types of stars in the solar neighborhood yield @xmath34stars@xmath33 , suggesting that about 1 cv exists in the galaxy - and probably in the lmc - for every @xmath7 stars .
the lmc displays a v - band luminosity of -18.1 , corresponding to a mass of @xmath35 @xmath36 @xcite and a population of a few times @xmath35 stars .
if the galaxy and lmc manufacture cvs with similar efficiencies and rates then we estimate a total lmc cv population today of a few times @xmath3 objects .
about half of known cvs are dn @xcite which suggests a total lmc dn population of order @xmath3 .
these dn are spread across the 10 @xmath0 10 degree surface of the lmc .
erupting classical novae are observed to be distributed quite uniformly across the face of the lmc @xcite , consistent with them belonging to an old population .
a surface density of 10,000 dn per square degree across the lmc is thus expected .
the size of our field of view was 1362 arcmin@xmath37 = 0.38 deg@xmath37 , which should include 0.38 @xmath0 10,000 = 3,800 dn .
the average time between eruptions for 21 well studied galactic dn is 29 days @xcite .
the length of our observing run ( 11 nights ) suggests that any dn erupting on the nine nights between nights 2 and 10 , inclusive , could have been detected ( if it became sufficiently luminous ) .
hence ( 9/29 ) @xmath0 3800 @xmath15 1200 dn eruptions should have occurred in our field of view that were ( at least in principle ) detectable . the 4 meter telescope seeing varied significantly over the six nights on which we observed , as clearly seen in the completeness curves of figure 2 . only on the second ,
third and sixth nights were conditions good enough to allow straightforward detection ( with 20% to 40% completeness due to crowding ) of dn reaching @xmath38 . assuming ( conservatively ) that we missed half of all dn because only 3 of 6 observing nights were `` good '' , and that we missed 80% of all eruptions even on the three `` good '' nights because of crowding
, we might still have expected to have detected @xmath39 erupting dn .
the apparent brightness of a dwarf nova depends both on the underlying binary system inclination and orbital period .
nearly face - on disks and longer orbital periods ( which have larger and brighter disks ) will dominate a magnitude limited sample @xcite .
high inclination systems will be up to 2 magnitudes fainter than those seen nearly face - on , and will therefore be lost in a sample that is detecting only the brightest tip of the distribution .
longer period systems will have larger and brighter disks , and thus might be expected to dominate our sample .
however , the number of short - period dn ( @xmath40 hr ) is twice the number of long - period systems ( @xmath41 hr ) , and it is these short - period systems that undergo su uma - like superoutbursts .
su uma superoutbursts typically last @xmath42 weeks , and system brightnesses might match those observed for our six candidates .
this will be clearer when parallaxes are available for at least a few su uma cvs .
the fact that we detected at most six dn when ( an admittedly simple ) model predicts 120 suggests that many erupting systems were too faint to be detected .
this is in accord with the model light curves of figure 6 , and reinforced by the simulated images of figure 5 showing that dn like u gem , ss aur and em cyg would all have been missed in our survey , and even ( relatively luminous ) ar and would have been possible but challenging to detect .
what fraction of dn are at least as luminous as ss cyg or su uma at maximum ?
unfortunately , the answer is unknown , because the distances to all but four galactic dn are too uncertain to yield a believable luminosity distribution .
( we note that the canonical literature distance to the prototypical dn , ss cyg , was in error by a factor of two until hst parallaxes became available in 2000 ) .
if our six candidates are all bona fide erupting dn in the lmc ( of either ss cygni or su uma type ) this suggests that 95% of all dn are less luminous than these prototypical dwarf novae .
this percentage rises if fewer than six of our candidates are really dn .
we have carried out an 11 day - long survey for erupting dwarf novae in the large magellanic cloud , during which we imaged 0.38 deg@xmath37 in @xmath4 and @xmath6 filters on alternate nights .
artificial dn and completeness tests confirm that we imaged faint enough ( to 23rd mag on the better nights ) to detect ss cygni - like eruptions .
six candidates were isolated from approximately a million stars observed with the ctio 8k @xmath0 8k mosaic2 ccd imager .
if these are true lmc dn , then they are amongst the most luminous of outbursting dwarf novae in the clouds ; ss cygni - like outbursts or possibly su uma - like superoutbursts .
we estimate the total lmc dn population as @xmath43 , and the lmc surface density as @xmath44 dn / deg@xmath12 .
this suggests that at least 95% of erupting lmc dn are fainter than the six candidates we present , and were thus missed .
an extra magnitude in sensitivity and improvements in spatial resolution will likely lead to the discovery of thousands of dn in the magellanic clouds .
we are indebted to the cerro tololo observatory directors robert williams and malcolm smith for the generous allocation of observing time for this project .
we also thank tony moffat , darragh o donohue , brian warner and an anonymous referee for valuable comments . while the time coverage we have of the seven variables noted in section 5 is far too sparse for any attempt at period determination
, we include their coordinates in table 3 and finder charts ( in @xmath4 and @xmath6 , minimum and maximum ) in figures 7 and 8 .
lllllcc 12.03.99 & @xmath45 & @xmath46 & 1.6 `` & 1.4 '' & 19% & 9% + 12.05.99 & @xmath47 & @xmath48 & 1.3 `` & 1.2 '' & 52% & 27% + 12.07.99 & @xmath47 & @xmath49 & 1.3 `` & 1.0 '' & 68% & 37% + 12.09.99 & @xmath50 & @xmath51 & 1.5 `` & 1.2 '' & 14% & 38% + 12.11.99 & @xmath52 & @xmath53 & 1.8 `` & 1.7 '' & 2% & 3% + 12.13.99 & @xmath54 & @xmath46 & 1.3 `` & 1.1 '' & 54% & 26% + lll shz1 & 5:30:42.60 & -70:47:32.5 + shz2 & 5:31:38.25 & -70:38:33.7 + shz3 & 5:34:42.07 & -70:49:23.6 + shz4 & 5:35:10.66 & -70:28:02.0 + shz5 & 5:35:47.75 & -70:24:10.2 + shz6 & 5:35:24.47 & -70:23:55.4 + lll v1 & 5:33:25.77 & -70:52:05.7 + v2 & 5:30:45.36 & -70:50:42.9 + v3 & 5:33:28.53 & -70:46:08.2 + v4 & 5:33:02.97 & -70:35:57.6 + v5 & 5:32:33.25 & -70:28:48.5 + v6 & 5:31:01.26 & -70:18:11.6 + v7 & 5:36:39.89 & -70:37:53.5 + | we report the first likely detections of erupting dwarf novae ( dn ) in an external galaxy : the large magellanic cloud .
six candidates were isolated from approximately a million stars observed every second night over 11 nights with the ctio 8k @xmath0 8k mosaic2 ccd imager .
artificial dwarf nova and completeness tests suggest that we are seeing only the brightest of the lmc dn , probably ss cygni - like cvs , but possibly su uma - type cataclysmics undergoing superoutbursts .
we derive crude but useful limits on the lmc dn surface density , and on the number of dn in the lmc . many thousands of cataclysmic variables in the magellanic clouds can be discovered and characterized with 8 meter class telescopes . |
tidal dwarf galaxies ( tdgs ) are objects resembling actively star forming dwarf galaxies and are assembled from the debris ( tidal tails and bridges ) launched into the igm by violent galaxy interactions in which at least one member is a gas rich galaxy .
they are composed of stars and gas from the outskirts of one or both of the parent galaxies involved in the interaction .
the recent surge of interest in tdgs started with papers by mirabel , lutz , & maza ( 1991 ) on the superantennae and mirabel , dottori , & lutz ( 1992 ) on the antennae ( see also the review by duc & mirabel 1999 ) .
several groups of authors have since embarked on the exciting topic of tdgs as witnessed by these proceedings ( see e.g. , the contributions by duc et al . and by braine et al .
this volume ) .
currently outstanding questions are : ( i ) are tdgs really recycled objects made of collisional debris or pre - existing galaxies involved in a three
body interaction ? ( ii ) are tdg genuine density enhancements in the tidal tails or are they merely due to projection effects along the line of sight ? ( iii ) do tdgs form self gravitating entities or are they simple transient condensations ? ( iv ) are tdgs dark matter ( dm ) dominated , like galaxies in general , and dwarf galaxies in particular , or are they nearly devoid of dm , as theory predicts ? ( v ) finally do tdgs leave the potential well of their progenitors and hence constitute a sizeable fraction of the known dwarf galaxy population or do they eventually fall back and merge , leaving no trace ?
all these questions have actually been raised for the particular tdg candidate identified in the northern tidal tail of arp245 , an interacting system composed of two spiral galaxies , ngc 2992 and ngc 2993 . although arp245n was observed at many wavelengths and
is one of the best studied tdg candidates , its nature as a tidal object or as a real entity have been challenged .
smith & struck ( 2001 ) argued that tdg arp245n could actually be a preexisting edge - on disk galaxy that is interacting with the other two galaxies .
hibbard et al .
( this volume ) point out that this system is viewed from an unfavorable perspective , making the projection effects particularly severe .
the combination of high resolution kinematical and morphological data is critical to tackle all these issues .
we have therefore carried out hi and co interferometric observations of the system .
ngc2992/3 is a relatively nearby system , at an adopted distance of 31 mpc ( v@xmath2kms@xmath3 ) .
its prominent northern tidal tail hosts a tidal dwarf galaxy candidate which because of its proximity can be studied in detail ( fig . 1 ) .
the system was observed by us in hi with the nrao very large array ( vla ) at an intermediate angular resolution of @xmath4 ( duc et al .
2000 ) . toward ngc2992
, the hi shows a peak at the location of the tidal dwarf candidate located at the tip of the tidal tail emanating from ngc2992 .
hi is seen in absorption against the radio continuum from the nucleus .
ngc2992 is classified as a seyfert 1.9 galaxy and in the light of h@xmath5 a biconical structure can be seen which extends out into the halo . toward ngc2993
, the hi map has a ringlike structure which in its western extension has no optical counterpart . within ngc2993 and arp245n ,
hii regions are concentrated within hi clumps and trace star forming regions .
numerical simulations of the ngc 2992/3 collision indicate that we see the system @xmath6myr after closest approach ( duc et al .
2000 ) . as reported by braine et al .
( 2000 , 2001 ) , arp245 was detected with the iram 30m dish in both the co(1@xmath00 ) and the co(2@xmath01 ) transitions .
the observations also revealed that the co emission is extended along the tdg . the h@xmath7 mass , assuming the standard ( galactic ) co to h@xmath7 conversion factor of n@xmath8@xmath9/kkms@xmath3 is @xmath10m@xmath11 .
an oxygen abundance of 12+log(o / h)=8.6 was measured in the hii regions of the tdg candidate .
this high metallicity excludes the hypothesis that the tdg is in fact a preexisting dwarf galaxy .
it is also unlikely that a more massive galaxy was directly involved in the interaction .
indeed , the morphology of the system matches very well that of the numerical model in which only two galaxies are interacting .
the field containing the interacting pair ngc2992/3 was observed with the vla in its b configuration on 24 april and 7/8 may 2001 for a total of almost 10 hours .
data calibration and reduction followed standard procedures using the classic aips data reduction package .
the final data products have a resolution of @xmath12 ( at 5kms@xmath3 velocity resolution ) and reach an rms noise of 0.5 mjybeam@xmath3 per channel .
we observed the tidal dwarf near ngc2992 in the co(1@xmath00 ) transition using the owen s valley radio observatory s mm array ( ovro ) in the e , c and l configurations for a total of 9 tracks from october 2001 through may 2002 .
the equatorial e configuration was needed to improve the beam shape for this @xmath13 source .
about 40 hours were spent on source .
data were recorded using a correlator setup resulting in velocity resolutions of 5kms@xmath3 ( after online hanning smoothing ) with a total bandwidth of 320 kms@xmath3 .
a datacube was produced using the miriad software package which was cleaned to a level of about twice the rms noise ( noise : 16mjybeam@xmath3 in a 5kms@xmath3 wide channel ) .
the final beam size is @xmath14 .
figure 1 shows an integrated hi map of the b
array vla data only .
it should be noted that an interferometer acts as a spatial filter and that the b array is sensitive to structures with typical sizes of between @xmath15 and @xmath16 . as a result , applying
a clean algorithm can leave some low level striping which will disappear once the new data are incorporated with the intermediate resolution maps .
the right panel in figure 1 shows as contours overlaid on the hi map , the co and h@xmath5 emission .
co is clearly detected in the ovro observations and is found to be resolved .
the total integrated flux is @xmath173jykms@xmath3 ( @xmath18kkms@xmath3 ) , corresponding to a total molecular gas mass , assuming a standard conversion factor of n@xmath19@xmath9/kkms@xmath3 , of m@xmath20m@xmath21 .
this is about 25% of the single dish flux which implies that most of the molecular gas is distributed smoothly across the region .
arp245n is a typical tdg candidate in the sense that it is a major hi concentration associated with recent star formation which resides near the tip of the tidal tail .
because the latter is seen close to edge on , the question thus arises whether the apparent concentration is a genuine feature ( hibbard & mihos , this volume ) .
bournaud et al .
( 2003 , 2004 ; see also the contributions by amram et al .
, this volume ) have run extensive numerical models and shown the characteristic shape in position velocity space of a _ bona fide _ tdg and that of a spurious feature .
they show that the kinematical signature of projection effects is a change in the sign of the velocity gradient along the tail before reaching its apparent tip . for curved tails that are extended enough in 3d space , a loop - like feature
may even be seen in a position
velocity ( pv ) diagram along the tails .
figure 2 shows such a pv diagram along the tidal tail connecting the tdg candidate and ngc2992 using the intermediate and high resolution vla hi data .
the signal was actually integrated over a band with a width similar to that of the tail .
the loop expected for projection effects is clearly seen on the figure .
the part of the tail that is seen curving back towards ngc 2992 ( as seen projected on the sky ) is actually consistent with our earlier numerical simulations of the system ( duc et al .
, 2000 ; see the face on view in their fig .
the hi morphology of the tidal tail , as seen at high resolution in fig . 1
, may also give some clues as to the geometry of the system .
its u like shape could be interpreted as being due to bending of the tidal tail near its apparent tip .
the tail is actually not seen perfectly edge on ( as indicated , in the optical , by the large width of the stellar tail ) .
thanks to the higher spatial resolution provided by the vla in its b configuration , we can hence resolve the projection effects which was impossible with the early c array data . on this map
, the hi column density seems to peak in the part of the tail which points back to ngc 2992 .
this is where ovro detected the bulk of the molecular gas and where the brightest hii regions in the tail are found .
the velocities of all these phases match .
the spatial and velocity coincidence between the co , h@xmath5 and hi emission peaks at this location in the tail is a strong indication that a genuine condensation is present there and that this is likely the progenitor of a tidal dwarf galaxy . at the same position
, a pv diagram perpendicular to the tidal tail shows a small scale velocity gradient similar to that expected for a rotating body ( see fig . 3 ) .
the peak to peak velocity range is 100kms@xmath3 .
a word of caution is warranted here , though . as duc et al .
( 2000 ) mentioned , the simulated pv diagram along the same direction in the numerical model shows a similar gradient .
further simulations are required to disentangle the embedded tdg from the rest of the tail .
we should then be able to determine its dynamical mass and , comparing it with the luminous mass ( corresponding to the hi condensation in the b array ) , probe its dark matter content .
not taking into account the line of sight crowding , and considering all the matter present at the apparent tip of the tail , one derives a dynamical mass similar to the luminous one and equal to @xmath22m@xmath11 . for the above - mentioned reasons , these are most likely overestimates . in summary , a first analysis of new high resolution hi and co datacubes tends to support the existence of a bound entity within the northern tail of arp245 .
however , they also show the kinematical signature expected when part of the tail is bending away along the line of sight , and eventually back to ngc 2992 .
because of these projection effects , the size and mass of the embedded tdg candidate derived from low resolution data are probably overestimates . | we present high resolution interferometric observations of the cool atomic and cold molecular ism of the tdg candidate arp 245n , an object resembling a dwarf galaxy in the northern tidal tail of the interacting system ngc2992/3 .
we observed the hi line with the nrao vla and the co(1@xmath00 ) transition with the ovro millimeter interferometer at @xmath1 angular resolution ( 750pc linear resolution ) .
these datacubes offer the required spatial and velocity resolution to determine whether the mass concentration near the tip of the tail is a genuine feature , and hence a good tdg candidate , or an artefact caused by a fortuitous alignment of our line of sight with the direction of the tail .
a preliminary analysis seems to confirm that arp245n is a self gravitating entity .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
magnetic nanoparticles , or nanoscale magnetic systems , have generated continuous interest since late 1940s as the investigation of their properties turned out to be challenging from both scientific and technological point of view . in 1949 , in a pioneering work @xcite , nel set the pace towards understanding of the magnetic behavior of nanoparticles , leading to an important development of fundamental theories of magnetism and modeling of magnetic materials , as well as remarkable technological advances , e.g. , in the area of information storage and data processing , fostering the development of magnetorecording media with increasingly higher densities .
nanoparticles , as compared to bulk materials , possess very important novel properties such as enhanced remanence and giant coercivity , as well as exponentially slow relaxation at low temperature due to anisotropy barriers , which ensures great stability of the information stored .
however , nanoparticles become _ superparamagnetic _ @xcite at finite temperatures for very small sizes , and this is an impairment to the information storage . on the other hand , discovery of superparamagnetism that results from thermally
activated crossing of the anisotropy energy barrier by the magnetic moment of the particle opened a rich area to the application of nonequilibrium statistical mechanics .
there are mainly two types of nanoparticle samples : i ) assemblies of , e.g. , cobalt , nickel or maghemite , nanoparticles with volume distribution and randomly oriented axes of magnetocrystalline anisotropy ; ii ) isolated single particles of cobalt or nickel that can be probed by the micro - squid technique @xcite .
while most of the experiments have been done on assemblies , isolated particles are more important both as units of information storage and as a physical system
. in the investigation of the static properties of magnetic nanoparticles a great deal of work up to date has been based on the monte carlo ( mc ) technique .
in addition to numerous simulations of the ising model , this technique has been used with the more adequate classical heisenberg model to simulate idealized isotropic models with simple cubic ( sc ) lattice and spherical shape in ref .
magnetic nanoparticles with realistic lattice structure were recently simulated in ref .
@xcite taking into account the surface anisotropy ( sa ) and ddi .
on the other hand , in most of theoretical approaches to the dynamics of a small magnetic particle the latter is considered as a single magnetic moment .
this is the _ one - spin approximation _ that is only valid for particles that are not too large and thus are single - domain , and not too small to be free from surface effects . letting apart spin tunneling ( see , e.g. , ref .
@xcite ) that becomes important for extremely small sizes such as those of molecular magnets , the magnetic moment can overcome the anisotropy - energy barrier and thus reverse its direction , at least in two ways : either under applied magnetic field that suppresses the barrier , or via thermal fluctuations .
the former , at zero temperature , is well described for particles with the uniaxial anisotropy by the stoner - wohlfarth model @xcite .
thermally activated crossing of the energy barrier is described by the nel - brown model @xcite and its extensions ( see , e.g. , refs .
@xcite ) . at elevated temperatures , rotation of the magnetization in materials with strong anisotropy
is always accompanied by changing its magnitude .
this results in a shrinking of the stoner - wohlfarth astroid as described by the modified landau theory @xcite , and ( qualitatively ) confirmed by experiments @xcite .
both stoner - wohlfarth and nel - brown models have been confirmed by experiments on individual cobalt particles @xcite .
however for magnetic particles with strong surface anisotropy magnetization switching occurs as the result of successive switching of individual ( or clusters of ) spins inside the particle @xcite .
such deviations from the one - spin approximation have been observed in metallic particles @xcite , @xcite , and ferrite particles @xcite , @xcite . deviations from the one - spin approximation and temperature effects
lead to the absence of magnetization saturation at high fields @xcite , @xcite , @xcite , shifted hysteresis loops after cooling in field , and field dependence of the magnetization at very low temperatures .
the latter effect has been clearly identified in dilute assemblies of maghemite particles @xcite of @xmath0 nm in diameter .
in addition , aging effects have been observed in single particles of cobalt and have been attributed to the oxidation of the sample surface into antiferromagnetic coo ( see @xcite and references therein . ) .
it was argued that the magnetization reversal of a ferromagnetic particle with antiferromagnetic shell is governed by two mechanisms that are supposed to result from the spin frustration at the core - shell interface of the particle .
some of the above - mentioned novel features are most likely due to magnetic disorder at the surface which induces a canting of spins inside the particle , or in other words , an inhomogeneous magnetic state .
this effect was first observed with the help of mssbauer spectroscopy by coey @xcite and later by morrish and haneda ( see @xcite for a review ) , and later by pren et al .
@xcite ( see also the recent article @xcite ) . to sum up , the picture of a single - domain magnetic particle with all spins pointing into the same direction is no longer valid when one considers the effect of misaligned spins on the surface , which makes up to @xmath1 of the total volume in a particle of @xmath0 nm in diameter .
one of our goals is thus is to understand the effect of surfaces on the thermodynamics and magnetization profiles in small systems , and subsequently upon their dynamics .
this requires a microscopic approach to account for the local environment inside the particle , microscopic interactions such as spin - spin exchange , ddi , and the magneto - crystalline bulk and surface anisotropy .
as this task is difficult owing to the large number of degrees of freedom involved , one has to gain a sufficient understanding of static properties before proceeding to the dynamics .
there are three main effects that distinguish magnetic particles from bulk magnets and that were investigated in a series of our recent publications : 1 .
finite - size effect in isotropic magnetic particles with idealized periodic boundary conditions .
the spin - wave spectrum of such particles is discrete and there is the mode with @xmath2 that corresponds to the global rotation of the particle . as a result
, the standard spin - wave theory fails and one has to distinguish between @xmath3 and _ intrinsic _ magnetizations .
boundary effect , i.e. , pure effect of the free boundary conditions at the surface in the absence of the surface anisotropy .
this effect leads to the decrease of the particle s intrinsic magnetization and it makes the latter inhomogeneous at @xmath4 .
3 . effect of surface anisotropy that changes the ground state of the particle and makes the intrinsic magnetization inhomogeneous even at @xmath5 .
this paper is organized as follows : in sec .
[ sec : basic ] we describe the hamiltonian and introduce the basic notions of the induced and the intrinsic magnetizations . in sec .
[ sec : mspwtse ] , we first study inhomogeneities in small magnetic particles of a box shape induced by pure boundary effects in the absence of surface anisotropy , i.e. , the effect of free boundary conditions ( fbc ) , and compare their influence with that of the finite - size effects in an idealized model with periodic boundary conditions ( pbc ) .
we consider spins as @xmath6-component classical vectors . at
first we present analytical and numerical results in the whole range of temperatures in the limit @xmath7 where the problem simplifies .
then for the classical heisenberg model , @xmath8 at low temperatures we formulate the modified spin - wave theory accounting for the global rotation of the particle s magnetization .
the results are compared with those of our mc simulations . in sec . [
sec : mspwse ] we consider round - shaped systems and include surface anisotropy . in the case of the sa much weaker than the exchange interaction we study the problem perturbatively in small deviations from the perfectly ordered , collinear state .
then we investigate the hysteretic properties and the behavior of the magnetization as a function of temperature and applied field by mc simulations .
the last section summarizes the results and points out open problems .
within the classical approximation it is convenient to represent the atomic spin as the three - component spin vector @xmath9 of unit length on the lattice site @xmath10 .
we will consider the hamiltonian that in general includes the exchange interaction , magneto - crystalline anisotropic energy , zeeman energy , and the energy of dipolar interactions ( ddi ) @xmath11 where @xmath12 and @xmath13 is the value of the atomic spin .
for materials with uniaxial anisotropy @xmath14 in eq .
( [ hamgeneral ] ) reads @xmath15 with easy axis @xmath16 and constant @xmath17 .
this anisotropy model can be used to describe the surface effect if one attributes the same easy axis and the same anisotropy constant @xmath18 for all core spins and different easy axes and anisotropy constants to surface spins . within the simplest transverse surface anisotropy ( tsa )
model all surface spins have the same anisotropy constant @xmath19 whereas their easy axes are perpendicular to the surface , see .
e.g. , refs .
more realistic is nel s surface anisotropy ( nsa ) model @xcite , @xmath20 where @xmath21 is the coordination number of site @xmath10 that for the surface atoms is smaller than the bulk value @xmath22 and @xmath23 is the unit vector connecting the site @xmath10 to its nearest neighbors @xmath24 one can check that for the simple cubic ( sc ) lattice the contributions from the bulk spins in ( [ nsa ] ) @xmath25 are irrelevant constants
. for materials with magneto - crystalline cubic anisotropy @xmath26 reads @xmath27 for @xmath28 , the energy @xmath29 has minima for six orientations of the type [ 100 ] and maxima for eight orientations of type [ 111 ] .
if one discards @xmath29 and @xmath30 the hamiltonian ( [ hamgeneral ] ) can be generalized for @xmath6-component spin vectors .
this is useful as in the limit @xmath31 the problem simplifies while retaining important physics , see below .
magnetic particles of _ finite size _ do not show magnetic ordering at nonzero temperatures at @xmath32 as the global magnetization of the particle can assume all possible directions ( superparamagnetism ) .
it is thus convenient to define two magnetizations , @xmath33 and @xmath34 the first being the magnetization induced by the magnetic field and the second being a measure of the short - range order in the particle . omitting the factor @xmath35 that can be restored later , we first define the magnetization of a microscopic spin configuration @xmath36 where @xmath37 is the number of magnetic atoms in the system .
the thermodynamic average of @xmath38 yields what we call the _ induced magnetization _ @xmath39 the _ intrinsic magnetization _ is related to the spin correlation function : @xmath40 if the temperature is low and there is no surface anisotropy , all spins in the particle are bound together by the exchange interaction and @xmath38 behaves as a rigid `` giant spin '' , @xmath41 . if a magnetic field @xmath42 is applied , @xmath38 exhibits an average in the direction of @xmath43 which leads to a nonzero value of the induced magnetization @xmath44 for isotropic @xmath6-component vector models ( including the ising model , @xmath45 @xcite the latter is given by the well known formula @xmath46 where @xmath47 is the langevin function [ @xmath48 for the isotropic heisenberg model and @xmath49 for the ising model ] .
an important question is whether eq .
( [ spmrelation ] ) remains valid at elevated temperatures where @xmath50 .
we have shown that the _ superparamagnetic relation , _ eq . ( [ spmrelation ] ) , becomes exact for @xmath51 but otherwise it contradicts the exact relation @xmath52 one also can introduce the local intrinsic magnetization @xmath53 according to @xmath54 this quantity is smaller near the boundaries of the particle than in the core for the model with fbc because of boundary effects at @xmath55
finite - size magnetic systems with free boundary conditions ( fbc ) present a spatially inhomogeneous many - body problem . in this section
we shall only deal with finite - size versus boundary effects , leaving the more profound effects of the surface anisotropy for the next section .
one of the interesting problems here is the interplay between boundary effects due to fbc and the `` pure '' finite - size effects . in systems of hypercubic shape
, the latter can be singled out by using artificial periodic boundary conditions ( pbc ) .
the standard mean - field approximation ( mfa ) and spin - wave theory ( swt ) are inappropriate for finite systems because of the goldstone mode associated with the global rotation of the magnetic moment in zero field .
appropriate improvements include the so - called @xmath7 model ( see sec .
[ sec : asm ] ) operating at all temperatures and the modified spin - wave theory for finite magnets at low temperatures ( see sec . [
sec : swtmc ] ) . also the mc routine should incorporate global rotations of spins , in addition to the metropolis algorithm of individual spin rotations .
one can improve upon the mfa by taking into account correlations in a wide temperature range for bulk and finite magnets by replacing 3-component spin vectors in ( [ hamgeneral ] ) by @xmath6-component ones and taking the limit @xmath51 .
this model was introduced by stanley @xcite who showed that in the bulk its partition function coincides with that of the exactly solvable spherical model ( sm ) @xcite . on the other hand , for spatially inhomogeneous and anisotropic systems
the @xmath56 model is the only physically acceptable model of both ( see , e.g. , refs .
so far , the @xmath7 model was only applied to spatially inhomogeneous systems in the plane geometry @xcite . in ref .
@xcite we extended it to _ finite _ box - shaped magnetic systems with free and periodic boundary conditions . in the mfa
the curie temperature of the @xmath6-component model is @xmath57 @xmath58 , where @xmath59 is the zero fourier component of @xmath60 .
it is convenient to use @xmath61 as the energy scale and introduce the dimensionless variables @xmath62 for the nearest - neighbor ( nn ) interaction @xmath60 with @xmath22 neighbors , @xmath63 is equal to @xmath64 if sites @xmath10 and @xmath65 are nearest neighbors and zero otherwise . in the bulk the @xmath7 model is described by two coupled nonlinear equations for the magnetization @xmath33 and the so called gap parameter @xmath66 : @xmath67 where @xmath68 is the lattice green function and @xmath69 is the fourier transform of @xmath70 the curie temperature is defined by @xmath71 and is @xmath72 i.e. , @xmath73 where @xmath74 is the watson integral ( @xmath75 for the sc lattice ) .
the system of equations describing the inhomogeneous @xmath7 model can be obtained using the diagram technique for classical spin systems @xcite in the limit @xmath7 and generalising the results of ref .
@xcite for spatially inhomogeneous systems to include the magnetic field @xmath76 .
this is a system of equations for the average magnetization @xmath77 directed along the field@xmath78 gap parameter @xmath79 and correlation functions for the remaining `` transverse '' spin components labeled by @xmath80 , i.e. , @xmath81 ( all transverse correlation functions are the same ) .
this system of equations has the form @xmath82 where @xmath83 is the dyson matrix and @xmath84 is the kronecker symbol . solving this system of equations consists in determining @xmath85 and @xmath86 as functions of @xmath87 from the first two _ linear _ equations and inserting the solutions into the third nonlinear equation ( the constraint equation ) that leads to a system of nonlinear equations for all @xmath88 that is in general subject to numerical solution . combining eqs .
( [ defmatr ] ) results in @xmath89 that yields @xmath90 where @xmath91 and is @xmath92 the langevin function for @xmath93 . alternatively eq .
( [ mvsm ] ) can be derived from eq .
( [ mviam ] ) replacing @xmath94 and neglecting @xmath95 in the limit @xmath7 and then solving the resulting algebraic equation for @xmath96 one can find in the literature formulae of the type @xmath97 , where the saturation magnetization @xmath98 is usually associated with the bulk magnetization at a given temperature ( see , e.g. , refs .
@xcite ) . in our case , eq .
( [ mvsm ] ) is exact and @xmath50 is defined by eq .
( [ binfdef ] ) . for large sizes @xmath37 ,
( [ mvsm ] ) describes two distinct field ranges separated at @xmath99 where @xmath100 in the range @xmath101 the total magnetic moment of the system is disoriented by thermal fluctuations , @xmath102 . in the range @xmath103
, the total magnetic moment is oriented by the field , @xmath33 approaches @xmath104 , and both further increase with the field towards saturation ( @xmath105 ) due to the suppression of spin waves in the system .
this scenario is inherent to all @xmath106 models @xcite .
having established the superparamagnetic relation , eq .
( [ mvsm ] ) we are left with the problem of calculating @xmath107 . for the pbc the solution becomes homogeneous and one obtains ( [ bulkeqs ] ) where in @xmath68 the integral is replaced by a sum over discrete wave vectors @xcite , whereas @xmath108 } .$ ] for the model with fbc analytical solution is only possible at low and high temperatures . at @xmath109 small deviations from the collinear state with @xmath110 can be described by the modified swt for arbitrary @xmath111 see sec . [
sec : swtmc ] .
the method for solving the @xmath7 model consists in obtaining the correlation functions @xmath86 and magnetization @xmath85 from the first two linear equations in eq .
( [ defmatr ] ) , substituting them into the third equation of eq .
( [ defmatr ] ) , and solving the resulting system of nonlinear equations for the gap parameter @xmath88 numerically .
[ sef_t3d ] ( left ) shows the temperature dependence of the intrinsic magnetization @xmath104 , eq .
( [ binfdef ] ) , and local magnetizations @xmath53 , eq .
( [ mlocaldef ] ) , of the @xmath112 cubic system with free and periodic boundary conditions in zero field . for periodic boundary conditions , @xmath104 exceeds the bulk magnetization at all temperatures .
in particular , at low temperatures this is in accord with the positive sign of the finite - size correction to the magnetization , eqs .
( [ mlowth0 ] ) and ( [ deltan ] ) .
local magnetizations at the center of the faces and edges and those at the corners decrease with temperature much faster than the magnetization at the center .
one can see that below the bulk critical temperature @xmath104 is smaller than the bulk magnetization .
this means that the boundary effects suppressing @xmath104 are stronger than the finite - size effects that increase @xmath104 .
this is also seen from the low - temperature expression of @xmath104 given in eq .
( [ mlowth0 ] ) , see also fig . [ sef_del ] .
[ sef_t3d ] ( right ) shows the magnetization profile in the direction from the center of the cube to the center of a face .
it is seen that perturbations due to the free boundaries extend deep into the particle , whereas the mfa predicts , on the contrary , a fast approach to a constant magnetization when moving away from the boundary @xcite .
this is a consequence of the goldstone mode which renders the correlation length of an isotropic bulk magnet infinite below @xmath113 .
mc simulations of the classical heisenberg model @xcite , @xcite yield a similar result ( see fig .
[ barce3 ] right ) .
we have shown that the critical indices for the magnetization at the faces , edges , and corners are higher than the bulk critical index @xmath114 for the present @xmath115 model .
the critical index at the face @xmath116 is the most studied surface critical index ( see , for a review , refs .
@xcite ) . the exact solution of bray and moore @xcite for the correlation functions at criticality in the @xmath115 model and application of the scaling arguments yield the value @xmath117 ( see table ii in @xcite ) .
exact values of the edge and corner magnetization indices , @xmath118 and @xmath119 , seem to be unknown for @xmath115 .
cardy @xcite used the first - order @xmath120-expansion to obtain @xmath121 for the edge with an arbitrary angle @xmath122 . for @xmath123 and @xmath115 in three dimensions the result for the edge critical exponent reads @xmath124 . to estimate the magnetization critical indices in our model we have performed a finite - size - scaling analysis ( see , for a review , ref .
@xcite ) assuming the scaling form @xmath125 and plotting the magnetization times @xmath126 vs @xmath127 . here
@xmath128 is the critical index for the correlation length in the bulk and @xmath129 , where @xmath130 is the bulk curie temperature .
our results for the system with @xmath131 and @xmath132 merge into single `` master curves '' for @xmath133 , @xmath134 , and @xmath135 , which have been obtained by fitting @xmath136 at @xmath137 , i.e. , @xmath138 .
note that our value 0.86 for the surface magnetization critical index @xmath116 is substantially lower than the value @xmath117 following from scaling arguments .
this disagreement is probably due to corrections to scaling which could be pronounced for our small linear sizes @xmath131 and @xmath139 .
a more efficient way for obtaining an accurate value of @xmath116 is to perform a similar analysis for the semi - infinite model .
the latter was considered analytically and numerically for @xmath140 and @xmath32 in @xcite .
we also mention the monte carlo simulations of the ising model @xcite which yield @xmath141 , @xmath142 , and @xmath143 . the field dependence of @xmath104 and @xmath33 at fixed temperature , as obtained from the numerical solution of eqs .
( [ defmatr ] ) for cubic and square systems is shown in fig .
[ sef_h ] . naturally the numerical results for @xmath33 confirm eq .
( [ mvsm ] ) which describes both the effect of orientation of the system s magnetization by the field and the increase of @xmath104 in field . using the zero - field value of @xmath104 in eq .
( [ mvsm ] ) leads to a poor result for @xmath33 as shown by the dashed curve for the 10@xmath144 system in fig .
[ sef_h ] .
the field dependence of the particle s magnetization similar to that shown in fig .
[ sef_h ] for the cubic system was experimentally obtained for ultrafine cobalt particles in @xcite , as well as in a number of previous experiments .
the curves for the square system in fig .
[ sef_h ] illustrate the fact that in two dimensions thermal fluctuations are much stronger than in 3@xmath145 , which leads to lower values of both @xmath104 and @xmath33 at a given temperature .
the bulk magnetization @xmath146 in two dimensions vanishes at zero field and it thus goes below the intrinsic magnetization @xmath104 in the low - field region . as briefly discussed in sec .
[ sec : basic ] , and in more details in the previous section , it is important to investigate the precise relation between the induced magnetization @xmath147 of eq .
( [ mdef ] ) and intrinsic magnetization @xmath104 of eq .
( [ mdef ] ) for the more realistic heisenberg model , @xmath148 . to do so
, we have developed a finite - size spin - wave theory that yields analytical results at low temperatures .
we also performed simulations with the improved monte carlo technique @xcite . in the absence of sa , at low temperatures
all spins in the particle are strongly correlated and form a `` giant spin '' @xmath38 defined in eq .
( [ mvecdef ] ) which behaves superparamagnetically .
in addition , there are internal spin - wave excitations in the particle that are responsible for @xmath149 at @xmath150 .
these excitations can be described perturbatively in small deviations of individual spins @xmath9 from the global direction of @xmath38 .
thus we use @xmath151 with @xmath152 and insert an additional integration over @xmath153 in the partition function , @xmath154 and first integrate over the magnetization magnitude @xmath155 .
thus we reexpress the vector argument of the @xmath156-function in the coordinate system specified by the direction of the central spin @xmath157 @xmath158 \right ) .\ ] ] then after integration over @xmath155 one obtains @xmath159 \right ) \mathrm{e}^{-\mathcal{h}_{\mathrm{eff}}/t } , \label{zndef}\ ] ] where @xmath160 is the partition function for the fixed direction @xmath161 and @xmath162 . \label{heff}\ ] ] in eq .
( [ zndef ] ) , the @xmath156-function says that the sum of all spins does not have a component perpendicular to @xmath38 .
this will lead to the absence of the @xmath2 component of the transverse spin fluctuations .
that is , the global - rotation goldstone mode that is troublesome in the standard spin - wave theory for finite systems , has been transformed into the integration over the global direction @xmath161 in eq .
( [ zndef ] ) .
@xmath160 is computed at low temperature by expanding @xmath163 up to bilinear terms in the transverse spin components @xmath164 : @xmath165 \delta _
{ ij}-j_{ij},\end{aligned}\ ] ] where @xmath166 is the zero - field ground - state energy .
next , upon computing the resulting gaussian integrals over @xmath167 , one obtains @xmath168 ^{(d-1)/2 } , \label{zndeterm}\ ] ] where for particles of cubic shape @xmath169 the prime on the product in ( [ zndeterm ] ) means omitting the @xmath170 mode .
results for particles of arbitrary shape can be found in ref .
@xcite . in eq .
( [ zndeterm ] ) the exponential factor corresponds to rigid spins whereas the second factor describes spin - wave corrections .
the latter makes the angular dependence of @xmath160 more complicated .
differentiating @xmath171 with respect to @xmath172 yields the induced magnetization @xmath173 then the intrinsic magnetization @xmath104 can be obtained from the exact relation ( [ mviam ] ) , and the validity of the superparamagnetic relation ( [ spmrelation ] ) can be checked .
now we consider particles of cubic shape ( @xmath174 ) at low temperatures .
for both pbc and fbc , we have obtained the following correction to the magnetization in zero field @xcite @xmath175 where @xmath176 is the sum without the @xmath2 term .
the results for pbc and fbc differ only by the values of the discrete wave vectors in eq .
( [ mlowth0 ] ) @xcite : @xmath177 where @xmath178 this subtle difference is responsible for much stronger thermal fluctuations in the fbc model due to boundary effects .
the limit @xmath179 of @xmath176 is the so - called watson s integral @xmath180 of eq .
( [ bulkeqs ] ) .
the difference between @xmath176 and @xmath181 has different signs for pbc and fbc models @xcite @xmath182 therefore , fig .
[ sef_del ] shows that the coefficient in the linear-@xmath183 term in eq .
( [ mlowth0 ] ) is smaller than in the bulk for the pbc system and greater for the fbc system .
that is , boundary effects suppress the intrinsic magnetization at low temperatures while finite - size effects lead to its increase .
[ sef_del ] shows that boundary effects render a larger contribution than the finite - size effects , making the net magnetization well below that of the bulk .
the field dependence of @xmath33 and @xmath104 is defined by the expansion of the lattice green function for small gaps , @xmath184 @xmath185 where for the sc lattice @xmath186 and the numerical results for @xmath187 can , for @xmath188 , be fitted as @xmath189 the spin - wave gap @xmath190 depends on the temperature and field ; for @xmath191 [ see eq .
( [ hvdef ] ) ] the gap approaches its zero - field value while for @xmath192 one has @xmath193 thus eq .
( [ pgtilexp ] ) defines one more crossover in field , the crossover between its first and second lines at @xmath194 in the field range @xmath195 the discreteness of the lattice can be neglected and the bulk result @xmath196 is reproduced . in the most interesting region
@xmath197 one obtains @xmath198 on the other hand , eq .
( [ mresult ] ) describes a crossover from the quadratic field dependence of @xmath104 at low field , @xmath199 to the linear dependence at @xmath200 note that for @xmath201 where @xmath202 and a rigid magnetic moment would saturate , @xmath33 continues to increase linearly as @xmath203 this is due to the field dependence of the intrinsic magnetization @xmath104 . at higher fields
there is another crossover to the standard spin - wave theory expression for @xmath104 .
approximate expressions for @xmath104 in the different field ranges are @xmath204 a simple analysis shows @xcite that the superparamagnetic relation ( [ spmrelation ] ) is a very good approximation for not too small systems , @xmath205 in the whole range below @xmath206 the deviation from eq .
( [ spmrelation ] ) is controlled by the small parameter @xmath122 of eq .
( [ mresult ] ) . above @xmath207
however , deviations from eq .
( [ spmrelation ] ) are large , except for the model with @xmath208 in the close vicinity of @xmath113 , there is a crossover to the high - temperature form of eq .
( [ spmrelation ] ) given by the function @xmath209 of eq .
( [ binfdef ] ) .
the modified swt developed above can be applied to study inhomogeneities in the fbc model .
the local _ intrinsic _ magnetization defined by eq .
( [ mlocaldef ] ) shows stronger temperature dependence near the boundary than the averaged @xmath210 of eq .
( [ mlowth0 ] ) .
the biggest effect of the surface is naturally attained at the corners of the cube where @xmath211 , at @xmath32 @xcite . here
we apply our monte carlo technique that accounts for the global - rotation goldstone mode to compute the induced and intrinsic magnetizations , and to investigate the superparamagnetic relation between them , for the heisenberg model with fbc .
our results for ising model can be found in @xcite .
first of all , in fig . [ fig - hth ]
we compare theoretical predictions of our analytical calculations within spin - wave theory for the heisenberg model with our mc results at @xmath212 where @xmath213 is the actual bulk curie temperature . for the small size @xmath214 the square - root field dependence of the magnetization ( third line of eq .
( [ mfinalres ] ) ) does not arise and finite - size corrections are very important . for @xmath104
one should use eq .
( [ mresult ] ) with @xmath215 given by eq .
( [ mlowth0 ] ) with numerically computed @xmath216 and @xmath217 for the fbc model [ cf .
( [ deltan ] ) and ( [ cn ] ) ] .
this yields @xmath218 and @xmath219 .
the theoretical dependence @xmath220 is practically a straight line which goes slightly above the mc points .
this small discrepancy can be explained by the fact that the applicability criterion for our analytical method , @xmath221 is not fully satisfied at @xmath222 , and a better agreement is achieved at lower temperatures . for comparison
we also plot the theoretical @xmath220 for the model with periodic boundary conditions .
here one has @xmath223 and @xmath224 , thus @xmath225 and @xmath226 , so @xmath220 goes noticeably higher and with a much smaller slope .
the quadratic field dependence of @xmath104 in the region @xmath227 is not seen at this low temperature since the value of @xmath228 is very small and thus much more accurate mc simulations would be needed .
we also plot in fig .
[ fig - hth ] the analytical result for the field dependence of @xmath33 @xcite which favorably compares with our mc data .
[ fig - h ] ( left ) shows the intrinsic magnetization @xmath104 and induced magnetization @xmath33 versus the scaled field @xmath229 for different temperatures .
we see that the particle s magnetic moment is aligned and thus @xmath230 for @xmath231 , if @xmath232 . at @xmath233
the field aligns individual spins , and this requires @xmath234 , i.e. , @xmath235 .
the quadratic dependence of @xmath220 at small fields manifests itself strongly at elevated temperatures .
the results of fig .
[ fig - h ] ( right ) show that the superparamagnetic relation of eq .
( [ spmrelation ] ) with @xmath50 is a very good approximation everywhere below @xmath113 , for the heisenberg model .
this also holds for the ising model @xcite . on the other hand , above @xmath236 eq .
( [ spmrelation ] ) with the function @xmath209 of eq .
( [ binfdef ] ) is obeyed .
the difference between these limiting expressions decreases with increasing number @xmath6 of spin components and disappears in the spherical limit ( @xmath7 ) .
surface anisotropy causes large deviations from the bulk behavior that are much stronger than just the effect of free boundaries .
also sa exerts influence upon the coercive field .
here we first consider magnetic structures and hysteresis loops at zero temperature induced by a strong transverse surface anisotropy ( tsa ) .
then for the more physically plausible nel s surface anisotropy ( nsa ) we investigate analytically and numerically its contribution to the effective anisotropic energy of the particle in the case when the nsa is weak in comparison to the exchange .
after that we study the effect of the nsa at finite temperatures using the mc technique .
in the end of this section we investigate the thermal and spatial dependence of the magnetization of a maghemite particle . here
we study the effect of tsa on the hysteresis loop and the angular dependence of the switching field .
we construct an effective stoner - wohlfarth ( sw ) astroid for a single - domain spherical particle with free surfaces , a simple cubic ( sc ) crystal structure , ferromagnetic exchange @xmath237 uniaxial anisotropy in the core @xmath18 , and the tsa of strength @xmath238 . using the hamiltonian ( [ hamgeneral ] ) without the ddi at @xmath239
we solve the coupled landau - lifshitz equations ( lle ) for each spin in the particle @xcite until a stationary state is attained . in @xcite
the same method was used for studying hysteresis loops in nanoparticles with a random bulk or surface - only anisotropy . here
we address the question of whether one can still use the simple sw model for a nanoparticle endowed with strong surface effects .
we show that it is so as long as @xmath240 otherwise switching of the particle s magnetization occurs via the reversal of clusters of spins , invalidating the simple sw model .
we consider @xmath238 and exchange coupling on the surface as free parameters since there are so far no definite experimental estimations of them , whereas the core parameters are taken as for the bulk system . in the sequel
, we use the reduced parameters , @xmath241 .
[ hyst_ks_a45 ] shows that when @xmath242 becomes comparable with @xmath65 , the hysteresis loop exhibits multiple jumps , which can be attributed to the switching of different spin clusters containing surface spins whose easy axes make the same angle with the field direction .
the hysteresis loop is characterized by two field values : one that marks the limit of metastability , called the _ critical field _ or the saturation field , and the other that marks the magnetization switching , and is called the _ switching field _ or the coercive field .
this progressive switching of spins is illustrated in fig .
[ struct_t_d10_j0_k1 ] for simplicity for the non - interacting case . for small values of @xmath243
the tsa model renders hysteresis loops and limit - of - metastability curves that scale with the sw results for all values of the angle @xmath244 between the core easy axis and the applied field , the scaling constant being @xmath245 , see in fig .
[ fig2_jap ] . for larger values of @xmath246 , but @xmath247 , we still have the same kind of scaling but the corresponding constant now depends on the angle @xmath248 between the core easy axis and field direction .
this is reflected by a deformation of the limit - of - metastability curve ( see fig . [
fig3_jap]a ) .
more precisely , the latter is depressed in the core easy direction and enhanced in the perpendicular direction .
however , there is still only one jump in the hysteresis loop implying that the magnetization reversal can be considered as uniform .
for @xmath249 , there appear multiple steps in the hysteresis loop associated with the switching of spin clusters .
it makes the hysteresis loop both qualitatively and quantitatively different from those of the sw model , as the magnetization reversal can no longer be considered as uniform . in addition , in the present case , there are two more new features : the values of the switching field are much higher than in sw model , and more importantly , its behavior as a function of the particle s size is opposite to that of the previous cases ( compare figs .
[ fig2_jap]a and [ fig3_jap]b ) .
more precisely , in this case one finds that this field increases when the particle s size is lowered .
this is in agreement with the experimental observations in nanoparticles ( see , e.g. , @xcite for cobalt particles ) .
the whole situation is summarized in fig .
[ hcvsks2 ] where we plot the critical field @xmath250 as a function of @xmath251 for different values of the surface - to - core ratio of the exchange coupling . for large values of @xmath252 surface spins
are aligned along their radial easy axes , and because of strong exchange coupling they also drive core spins in their switching process , which requires a very strong field to be completed .
the value of @xmath242 where @xmath250 jumps up ( e.g. , @xmath253 for @xmath254 ) marks the passage from a regime where scaling with the sw results is possible ( either with a @xmath244-dependent or independent coefficient ) to the second regime where this scaling is no longer possible because of completely different switching processes . to estimate @xmath238 and the critical field , consider a 4 nm cobalt particle of fcc crystal structure , for which the lattice spacing is @xmath255 , and there are 4 cobalt atoms per unit cell .
the ( bulk ) magneto - crystalline anisotropy is @xmath256 erg / spin or @xmath257 erg / cm@xmath144 , and the saturation magnetization is @xmath259 emu/@xmath260 .
the switching field is given by @xmath261 .
for @xmath262 , @xmath263 and @xmath264 , so @xmath265 t . on the other hand
, @xmath263 means that the effective exchange field experienced by a spin on the surface is of the order of the anisotropy field , i.e. @xmath266 . then using @xmath267mev we get @xmath268erg / spin , or using the area per surface spin ( approximately @xmath269 ) , @xmath270erg/@xmath271 .
for the case of @xmath272 , @xmath273 and @xmath274 , which leads to @xmath275 t and @xmath276erg / spin or @xmath277erg/@xmath271 .
we calculate the contribution of the nsa @xcite to the effective anisotropy of magnetic nanoparticles of spherical shape cut out of a simple cubic lattice .
the effective anisotropy arises because of deviations of atomic magnetic moments from collinearity and dependence of the energy on the orientation of the global magnetization with respect to crystallographic directions .
we show @xcite that the result is second order in the nsa constant , scales with the particle s volume , and has cubic symmetry with preferred directions @xmath278 $ ] .
as shown many times before , as the size of the magnetic particle decreases , surface effects become more and more pronounced . in many cases
surface atoms yield a contribution to the anisotropy energy that scales with particle s surface , i.e. , to the _ effective _ volume anisotropy decreasing with the particle s linear size @xmath279 as @xmath280 as was observed in a number of experiments ( see , e.g. , refs .
@xcite , @xcite ) .
the @xmath281 surface contribution to @xmath282 is in accord with the picture of all magnetic atoms tightly bound by the exchange interaction while only surface atoms feel the surface anisotropy .
this is definitely true for magnetic films where a huge surface contribution to the effective anisotropy has been observed .
the same holds for cobalt nanoclusters of the form of truncated octahedrons @xcite where contributions from different faces , edges , and apexes compete , resulting in a nonzero , although significantly reduced , surface contribution to @xmath282 .
however , for symmetric particle s shapes such as cubes or spheres , the symmetry leads to vanishing of this first - order contribution . in this case one
has to take into account deviations from the collinearity of atomic spins that result from the competition of the surface anisotropy and exchange interaction @xmath283 .
the resulting magnetic structures ( for the simplified radial sa model ) can be found in @xcite , @xcite , @xcite ( see also fig .
[ nsa - structure ] for the nsa ) . in the case
@xmath284 deviations from collinearity are very strong , and it is difficult , if not impossible , to characterize the particle by a global magnetization suitable for the definition of the effective anisotropy . for @xmath285
the magnetic structure is nearly collinear with small deviations that can be computed perturbatively in @xmath286 .
the global magnetization vector @xmath287 [ this is the same as the vector defined in ( [ mdef ] ) ] can be used to define the anisotropic energy of the whole particle .
the key point is that deviations from collinearity , and thereby the particle s energy , depend on the orientation of @xmath287 , even for a particle of a spherical shape , due to the crystal lattice .
to illustrate this idea , we neglect the bulk anisotropy and the ddi in the hamiltonian ( [ hamgeneral ] ) . for a sc lattice eq .
( [ nsa ] ) reduces to @xmath288 where @xmath289 are the numbers of available nearest neighbors of the atom @xmath10 along the axis @xmath290 one can see that the nsa is in general biaxial . for @xmath291 and @xmath292
the @xmath122-axis is the easy axis and the @xmath293-axis is the hard axis .
if the local magnetic moments @xmath9 are all directed along one of the crystallographic axes @xmath122 , then the anisotropy fields @xmath294 are also directed along @xmath122 and are thus collinear with @xmath9 . hence , at least for @xmath285 , there are no deviations from collinearity if the global magnetization @xmath287 is directed along one of the crystallographic axes . for other orientations of @xmath287 , the vectors @xmath295 and @xmath296 are not collinear , since then at least two components of @xmath9 are non zero , and the transverse component of @xmath296 with respect to @xmath9 causes a slight canting of @xmath9 and thereby a deviation from the collinearity of magnetizations on different sites .
this adjustment of the magnetization to the surface anisotropy lowers the energy .
as we shall see , this effect is strongest for the [ @xmath2971,@xmath2971,@xmath2971 ] orientations of @xmath287 . for both signs of @xmath298
these are easy orientations , whereas [ @xmath299,$ ] [ @xmath300,$ ] and [ @xmath301 $ ] are hard orientations .
to solve the problem numerically , we have to fix the global magnetization of the particle in a desired direction @xmath302 ( @xmath303 ) by using the energy function with a vector lagrange multiplier @xmath304 : @xmath305 to minimize @xmath306 we numerically integrate the evolution equations @xmath307 \right ] , \qquad \mathbf{f}_{i}\equiv -\partial \mathcal{f}/\partial \mathbf{s}_{i } \nonumber \label{lleqs } \\ \mathbf{\dot{\lambda } } & = & \mathbf{\partial } \mathcal{f}/\partial \mathbf{% \lambda } { = -}\mathcal{n}\left ( \mathbf{\nu } -\mathbf{\nu } _ { 0}\right ) .\end{aligned}\ ] ] starting from @xmath308 and @xmath309 until the stationary state is attained and an energy minimum is found .
our numerical results for the magnetic energy of spherical particles as a function of the orientation of the global magnetization are shown in fig .
[ nsa - etheta ] ( left ) .
[ nsa - etheta ] ( right ) shows differences between the basic directions [ 001 ] , [ 011 ] , and [ 111 ] .
it is seen that @xmath310 for @xmath311 limit , i.e. , @xmath312 scale with the particle s volume @xmath313 . to analytically solve the problem in the continuous limit , we replace in eq .
( [ nsasc ] ) the number of nearest neighbors of a surface atom by its average value @xmath314 is the @xmath122-component of the normal to the surface @xmath315 . the surface - energy density can then be obtained by dropping the constant term and multiplying eq .
( [ nsasc ] ) by the surface atomic density @xmath316 : @xmath317 .
\label{esdef}\ ] ] at equilibrium the landau - lifshitz equation reads @xmath318 for small deviations from collinearity one can seek for its solution in the form @xmath319 is the solution of the internal neumann boundary problem @xmath320 .
\label{psineumann}\end{aligned}\ ] ] this equation can be solved with the help of the green function @xmath321 ( see ref .
@xcite ) , and the final result for the second - order energy contribution is @xmath322 taking into account the cubic symmetry and computing numerically a double surface integral one can write the result of eq .
( [ efinal ] ) as @xmath323 where @xmath324 this defines the large-@xmath325 asymptotes in fig . [ nsa - etheta ] ( right ) shown by the horizontal lines .
the analytical results above is valid for particle sizes @xmath325 in the range @xmath326 the lower boundary is the applicability condition of the continuous approximation . since the surface of a nanoparticle
is made of atomic terraces separated by atomic steps , each terrace and each step with its own form of nsa [ see eq .
( [ nsasc ] ) ] , the variation of the local nsa along the surface is very strong . approximating this variation by a continuous function according to eq .
( [ zavr ] ) requires pretty large particle sizes @xmath327 .
this is manifested by a slow convergence to the large-@xmath325 results in fig .
[ nsa - etheta ] ( right ) . the upper boundary in eq .
( [ napplreg ] ) is the applicability condition of the linear approximation in @xmath328 . for @xmath329 deviations from the collinear state
are strong , and the effective anisotropy of a magnetic nanoparticle can not be introduced . as we have seen in eq .
( [ ekappa ] ) , the contribution of the surface anisotropy to the overall anisotropy of a magnetic particle scales with its volume @xmath330 .
this surprising result is due to the penetration of perturbations from the surface deeply into the bulk .
if a uniaxial bulk anisotropy @xmath18 was present in the system , perturbations from the surface would be screened at the bulk correlation length ( or the domain - wall width ) @xmath331 .
then for @xmath332 the contribution of the surface anisotropy to the overall anisotropy would scale as the surface : @xmath333 . for not too large particles , @xmath334 ,
contributions of both anisotropies to the anisotropic energy are additive and scale as the volume .
if the bulk anisotropy is cubic , both contributions have the same cubic symmetry [ c.f .
( [ ca_xyz ] ) and ( [ ekappa ] ) ] , and the experiment should provide a value of the effective cubic anisotropy different form the bulk value @xcite .
for the uniaxial bulk anisotropy , the two contributions have different functional forms .
even if the bulk anisotropy is dominant so that the energy minima are realized for @xmath335 , the surface anisotropy makes the energy dependent on the azimuthal angle @xmath336 .
this modifies particle s energy barrier by creating saddle points and strongly influences the process of thermal activation @xcite . for small deviations from the cubic or spherical shape ,
i.e. , for weakly elliptic or weakly rectangular particles , there should emerge a corresponding weak first - order contribution @xmath337 that would add up with our second - order contribution .
for an ellipsoid with axes @xmath338 and @xmath339 , the anisotropy energy scales with the surface , @xmath340 [ cf .
( [ ekappa ] ) ] , so that @xmath341 can be large even for @xmath342 the nel constant @xmath298 is in most cases poorly known .
however , for metallic co @xcite quotes the value of surface anisotropy @xmath343 erg / cm@xmath144 , i.e. , @xmath344 k. this is much smaller than @xmath345 k , which makes our theory valid for particle sizes up to @xmath346 , according to eq .
( [ napplreg ] ) . for this limiting size
one has @xmath347 that is large for nearly spherical particles , @xmath348 . in this section ,
we consider more realistic model of round - shaped ( spherical or ellipsoidal ) nanoparticle of simple cubic or spinel crystalline structure , uniaxial or cubic anisotropy in the core , and transverse or nel anisotropy on the surface , described by the hamiltonian defined by eqs .
( [ hamgeneral ] ) , ( [ uaa ] ) , ( [ ca_xyz ] ) , and ( [ nsa ] ) . using various techniques explained above , we compute the magnetization , induced and intrinsic , as a function of temperature and applied magnetic field , for different values of the surface anisotropy constant and exchange coupling @xcite@xcite .
we shall mainly focus on novel features stemming from the combination of anisotropy , field and temperature effects on the magnetization . in our simulations we consider the core ferromagnetic coupling @xmath349 and uniaxial anisotropy @xmath350 on the surface we adopt @xmath351 , while the anisotropy is given by the nel expression ( [ nsa ] ) with constant @xmath352 .
we ignore the ddi for simplicity .
in particular , we are interested in how anisotropy affects the superparamagnetic relation ( [ spmrelation ] ) that has been shown to hold at all temperatures below @xmath236 for isotropic systems . as no analytical calculations are possible here , we resort to the monte carlo technique with global spin rotations . in fig .
[ magt_n9_neel ] ( left ) we plot the core , surface and net magnetizations of a nanoparticle of @xmath353 spins , as functions of temperature in zero magnetic field .
these results do confirm what was obtained from the spherical model [ see fig .
[ sef_t3d ] ( left ) ] for isotropic box - shaped systems , namely that boundary effects suppress the magnetization , and here we see that this effect is enhanced by the sa.a drastic effect of the sa is clearly seen in the field dependence of the scaled induced magnetization @xmath33 as shown in fig .
[ magt_n9_neel ] ( right ) at @xmath354 . in the presence of a strong sa the superparamagnetic relation ( [ spmrelation ] )
is no longer valid , and @xmath355 strongly deviates from the langevin function . in this section
we deal with the ferrimagnetic maghemite nanoparticles ( @xmath359-fe@xmath357o@xmath358 ) having spinel crystalline structure , summarizing the results of @xcite , @xcite .
this time we include the ddi in the hamiltonian of eq .
( [ hamgeneral ] ) .
the bulk anisotropy in such materials is cubic , still we are using a uniaxial anisotropy to simplify the study of effects that are of more interest to us here .
we consider maghemite particles of various sizes ( @xmath360 that correspond to a radius of 2 - 3.5 nm ) and with the physical properties in the core ( spinel crystal structure with lacuna , exchange and dipolar interactions , anisotropy constant , etc . ) as those of the bulk , except that the anisotropy is taken as uniaxial .
we use the tsa model with @xmath361 erg/@xmath271 .
all spins in the core and on the surface are identical but interact via different couplings depending on their locus in the lattice .
we assume that the exchange interactions between the core and surface spins are the same as those inside the core .
although we treat only the crystallographically `` ideal '' surface , we do allow for a scatter in the exchange constants on the surface .
in contrast , in refs .
@xcite it was assumed that all exchange interactions are the same but there was postulated the existence of a fraction of missing bonds on the surface . in fig .
[ barce3 ] ( left ) , we see that the surface magnetization decreases more rapidly than the core magnetization as the temperature increases .
moreover , it is seen that even the ( normalized ) core magnetization per site does not reach its saturation value of @xmath362 at very low temperatures , which shows that the magnetic order in the core is disturbed by the surface .
this may also be due to lacuna in the spinel structure . in fig .
[ barce3 ] ( right ) we plot the spatial evolution of the local magnetization from the center to the border of the particle , at different temperatures . at all temperatures it decreases with distance from the center . at high temperatures ,
the local magnetization exhibits a temperature - dependent jump , and then continues to decrease .
this indicates that there is a radius within which the magnetization assumes relatively large values .
this result agrees with that of @xcite where this radius was called the _ magnetic radius_.
we have demonstrated by different analytical and numerical methods the importance of accounting for the magnetization inhomogeneities in magnetic nanoparticles , especially in the presence of sa .
the latter makes the magnetization inhomogeneous even at @xmath5 and in general modifies the relation between the intrinsic and induced magnetizations .
it also changes the magnetization switching mechanism , since for large sa the particle s spins switch cluster - wise . for small sa
we were able to calculate the spin canting in the particle analytically and to obtain a novel second order contribution to the particle s overall anisotropy .
it remains to generalize this result for nonzero bulk anisotropy .
another important task is to study dynamical implications of the many - body effects in magnetic nanoparticles .
99 l. nel , c. r. acad .
paris * 228 * ( 1949 ) 664 ; ann .
geophys . * 5 * ( 1949 ) 99 .
w. wernsdorfer , adv .
chem . phys . * 118 * ( 2001 ) 99 .
v. wildpaner , z. phys .
b * 270 * ( 1974 ) 215 .
h. kachkachi et al .
j. b * 14 * ( 2000 ) 681 .
e. m. chudnovsky and j. tejada , _ macroscopic quantum tunneling of the magnetic moment _ , cambridge university press , 1998 .
h. w. schumacher et al .
* 90 * ( 2003 ) 17201 ; 17204 .
e. c. stoner and e. p. wohlfarth , philos .
london , ser . a * 240 * ( 1948 ) 599 ; ieee trans
mag-27 * ( 1991 ) 3475 .
w. f. brown , phys . rev .
* 130 * ( 1963 ) 1677 ; ieeem * 15 * ( 1979 ) 1196 .
w. t. coffey et al . , adv .
* 117 * ( 2001 ) 483 .
h. kachkachi and d. a. garanin , physica a * 300 * ( 2001 ) 487 . h. kachkachi and m. dimian , phys .
b * 66 * ( 2002 ) 174419 ; m. dimian and h. kachkachi , j. appl . phys .
* 91 * ( 2002 ) 7625 .
r. h. kodama et al .
* 77 * ( 1996 ) 394 ; r.h .
kodama and a.e .
berkovitz , phys .
b * 59 * ( 1999 ) 6321 .
j. t. richardson et al .
* 70 * ( 1991 ) 6977 .
j. p. chen et al . , phys .
b * 51 * ( 1995 ) 11527 .
a. ezzir , _ proprits magntiques dune assemble de nanoparticules : modlisation de laimantation _ , universit paris - sud , orsay 1998
. m. respaud et al . , phys .
b * 57 * ( 1998 ) 2925 .
e. tronc et al . ,
j. mag . mag . mat .
* 221 * ( 2000 ) 110 .
w. t. coffey et al . , adv .
chem . phys .
* 117 * ( 2001 ) 483 .
k. haneda , can .
* 65 * ( 1987 ) 1233 .
p. pren et al . , hyperfine int .
* 93 * ( 1994 ) 1049 .
s. mrup , j. mag . mag . mat . * 266 * ( 2003 ) 110 .
a. aharoni , _ introduction to the theory of ferromagnetism _ ,
oxford science pubs . 1996 .
v. shilov , _ effects of surface anisotropies of the ferromagnetic resonance in ferrite nanoparticles _ , ph.d . thesis of university paris vii , 1999 .
l. nel , j. phys .
radium * 15 * ( 1954 ) 225 .
h. e. stanley , phys .
* 20 * ( 1968 ) 589 ; phys . rev .
* 176 * ( 1968 ) 718 .
t. n. berlin and m. kac , phys . rev .
* 86 * ( 1952 ) 821 .
d. a. garanin , z. phys .
b * 102 * ( 1997 ) 283 ; j. phys . a * 29 * ( 1996 ) l257 ; * 29 * ( 1996 ) 2349 ; * 32 * ( 1999 ) 4323 ; phys . rev .
e * 58 * ( 1998 ) 254 .
d. a. garanin and v. s. lutovinov , solid state commun .
* 50 * ( 1984 ) 219 .
d. a. garanin , j. stat . phys .
* 74 * ( 1994 ) 275 ; phys .
b * 53 * ( 1996 ) 11593 .
m. e. fisher and v. privman , phys .
b * 32 * ( 1985 ) 447 ; commun . math . phys .
* 103 * ( 1986 ) 1986 . k. binder , in _ phase transitions and critical phenomena _ , eds . c. domb and j. l. lebowitz , academic press , new york 1983 , vol.8 , pp.75267 ; h. w. diehl , ibid 1986 , vol .
10 , pp . 75267 .
a. j. bray and m. a. moore , phys .
* 38 * ( 1977 ) 735 .
j. l. cardy , j. phys .
a * 16 * ( 1983 ) 3617 . k. binder , in _ computational methods in field theory _ , eds . h. gausterer and c. b. lang , springer , berlin 1992 .
, d. a. garanin , phys . rev .
e * 58 * ( 1998 ) 254 .
m. pleimling and w. selke , eur .
j. b * 5 * ( 1998 ) 805 .
h. kachkachi and d. a. garanin , eur .
j. b * 22 * ( 2001 ) 291 .
d. a. dimitrov and wysin , phys . rev .
b * 50 * ( 1994 ) 3077 .
d. a. garanin and h. kachkachi , phys .
* 90 * ( 2003 ) 65504 . c. chen , o. kitakami , s. okamoto , and y. shimada , j. appl .
* 86 * ( 1999 ) 2161 .
m. jamet et al . , phys .
* 86 * ( 2001 ) 4676 .
y. labaye et al . , j. appl
* 91 * ( 2002 ) 8715 .
d. a. garanin et al , phys .
e * 60 * ( 1999 ) 6499 .
d. s. chuang et al . , phys .
b * 49 * ( 1994 ) 15084 .
h. kachkachi et al . , j. mag .
mag . mat .
* 221 * ( 2000 ) 158 . | we present a review of recent advances in the study of many - body effects in magnetic nanoparticles .
considering classical spins on a lattice coupled by the exchange interaction in the presence of the bulk and surface anisotropy , we investigate the effects of finite size , free boundaries , and surface anisotropy on the average and local magnetization for zero and finite temperatures and magnetic fields .
superparamagnetism of magnetic particles necessitates introducing two different , induced and intrinsic , magnetizations .
we check the validity of the much used relation between them within different theoretical models .
we show that the competition between the exchange and surface anisotropy leads to spin canting dependent on the orientation of the average magnetization with respect to the crystallographic axes and thus to a second - order effective anisotropy of the particle . we have also investigated the switching mechanism of the magnetization upon varying the surface anisotropy constant .
some cases of more realistic particles are also dealt with . _ fine particle systems , surface anisotropy , classical spin models _ 75.50.tt - 75.30.pd - 75.10.hk |
magnetic field passively advected by an incompressible velocity field obeys the induction equation @xmath4 where @xmath5 is the diffusivity .
we will assume that the diffusivity is negligibly small ( large prandtl number ) , and the magnetic field is therefore frozen into the fluid .
the incompressibility of the velocity field is not relevant for further calculations ; for the compressible case , eq .
( [ induction ] ) holds for @xmath6 , where @xmath7 is the density of the fluid .
the random velocity field is assumed to be gaussian in the eulerian frame : @xmath8 where @xmath9 is an arbitrary function regular at @xmath10 , this is the so - called batchelor regime @xcite .
the solution of the eq .
( [ induction ] ) is @xmath11 where @xmath12 and @xmath13 are initial values of the fluid - particle coordinate and of the magnetic field .
the object @xmath14 in the integrand is the _
lagrangian _ velocity field .
the _ eulerian _
velocity has the form @xmath15 . to make use of the formula ( [ correlator ] )
, we need to stop " the moving point @xmath16 in the function @xmath17 .
this is done differently in the eulerian and lagrangian frames ( see fig.[trajectories ] ) .
in the lagrangian frame , @xmath18 is fixed and we need to express @xmath19 in terms of @xmath20 .
the general expression is @xmath21 . for small @xmath22
, we expand this expression up to the second order in @xmath23 , and iterate once with respect to @xmath24 .
we obtain @xmath25 where all the velocities are taken at the point @xmath18 , and we use the short - hand notation @xmath26 and @xmath27 . to obtain the equation for the second moment of @xmath28 , we substitute the expansion ( [ velocity_expansion_lagrange ] ) into the formula ( [ differential ] ) , raise the latter to the second power and average using formula ( [ correlator ] ) .
we assume that @xmath22 is much smaller than the inverse growth rate of the second moment , and that the correlation time @xmath29 is much smaller than @xmath22 .
assuming isotropic and spatially homogeneous initial distribution , @xmath30 , @xmath31 , we find the equation for the second - order moment @xmath32 in the lagrangian frame .
for the incompressible velocity field , it takes the form @xmath33 to derive this equation we assumed that @xmath34 , where @xmath35 is a @xmath0 function smeared over the correlation time @xmath29 . in particular , we have chosen the box " regularization : @xmath36 for @xmath37 $ ] , and @xmath38 otherwise .
the coefficients in the correction terms in the formula ( [ growth_lagrangian ] ) are given by the following integrals : @xmath39 all the derivatives of @xmath40 in formula ( [ growth_lagrangian ] ) are taken at @xmath41 ; the subscripts denote derivatives with respect to the corresponding components of @xmath42 , and @xmath43 is the space dimension .
one can easily check that the first - order @xmath29 correction to the growth rate is negative .
it is important to note that this correction is not universal since the integrals ( [ i_1]-[i_3 ] ) depend on the chosen regularization @xmath35 of the @xmath0 function .
in the eulerian frame , the point @xmath44 is fixed in the formula ( [ differential ] ) , and we need to express @xmath45 in terms of @xmath46 . by analogy with the expression ( [ velocity_expansion_lagrange ] ) , we write @xmath47 .
the expansion now takes the form @xmath48 where we use the short - hand notation @xmath49 . to calculate the second moment of @xmath28
, we have to use the expression ( [ differential ] ) : @xmath50 that should be expanded up to the fourth order in @xmath51 .
then , we substitute the expansion ( [ velocity_expansion_euler ] ) for @xmath52 .
finally , assuming isotropic and spatially homogeneous initial distribution , @xmath30 , @xmath31 , and making use of ( [ correlator ] ) , we can find the equation for @xmath53 .
we do not present the detailed calculation here since for the incompressible velocity field , the answer coincides with the lagrangian case ( [ growth_lagrangian ] ) , as it should .
the appearance of the term @xmath54 in eq .
( [ growth_lagrangian ] ) shows that for the finite correlation time @xmath29 , the velocity field @xmath55 can not be treated as linear in the calculation of the growth rate .
this is obvious since a lagrangian particle is swept by the finite distance @xmath56 before the velocity field becomes decorrelated .
the appearance of @xmath57 ( or , equivalently , @xmath58 ) is a manifestation of the absence of the galilean invariance . for the velocity field with finite correlation time ,
such invariance is broken as it can be seen from eq .
( [ correlator ] ) .
the batchelor advection , i.e. the advection by a smooth velocity field is a good model for many physical problems .
it is usually a valid approximation when one is interested in the correlators of the advected fields on the scales @xmath59 that are much smaller than the viscous scale of the velocity field @xmath60 .
an additional simplification arises when the correlation time of the velocity field @xmath29 is much smaller than the inverse growth rates of the advected fields .
mathematically , such a limit is described by the @xmath0-correlated in time velocity field .
in such a case , the growth rates depend only on @xmath61 , as a consequence of the scale invariance . in this case one can substitute the linear velocity field @xmath62 , @xmath63 , for the real field , and thus reduce the field - theoretical problem to the quantum mechanics of the fluctuating matrix @xmath64(t ) ; for details and references , see @xcite .
we have demonstrated that the possibility of such a reduction is an artifact of the @xmath0-correlated velocity field , and have presented a simple method for calculating the finite - time corrections to the growth rates .
the criterion of the applicability of the quantum mechanical reduction is not only @xmath65 , but also @xmath66 . for @xmath67 ,
the universal growth rates do not exist , they are determined by the statistics of the velocity field on the integral scale and by the form of the short - time regularization .
this fact is sometimes overlooked in the literature ; in the present note we have tried to clarify the question .
the general formalism allowing to systematically find the finite - time corrections to the dynamo growth rates , has been developed in @xcite by a different method .
a simple discussion of the stochastic calculus applied to the dynamo problem can be found in @xcite . | the batchelor passive advection is an advection by a smooth velocity field .
if the velocity field is a @xmath0-correlated in time random gaussian process , then the problem is reduced to quantum mechanics of fluctuating velocity gradient @xmath1 . for the finite - time correlated velocity field ,
such a reduction does not exist . to illustrate this point ,
the second moment of a passively advected magnetic field is considered , and the stochastic calculus is used to find finite - time corrections to its growth rate .
the growth rate depends on large scale properties of the velocity field . moreover , the problem is not universal with respect to the short - time regularization : different regularizations give different answers for the growth rate .
+ pacs numbers : 47.27.gs # 1sec .
[ # 1 ] # 1section [ # 1 ] # 1appendix [ # 1 ] # 1appendix [ # 1 ] # 1([#1 ] ) # 1([#1 ] ) # 1eq . ( [ # 1 ] ) # 1equation ( [ # 1 ] ) # 1fig .
[ # 1 ] # 1figure [ # 1 ] # 1tab .
[ # 1 ] # 1table [ # 1 ] # 1ref
. # 1reference @xmath2 ) @xmath3 ] psfig.sty |
activated dynamics of granular materials , such as sand , exhibits a fascinating range of behaviors @xcite .
for example , sand can behave like a fluid when driven or poured from a container .
when at rest , it adopts a solid state , such as dry sand on a beach . in the limit of slow driving ,
a combination of these behaviors ensue , as recent experiments @xcite demonstrate . in these experiments
@xcite , a granular bed of glass beads of various sizes is poured into a wide and deep dish , so that the open surface on top is relatively smooth . objects mounted rigidly on a device that measures the support force are vertically plunged in or withdrawn from the bed .
as the object is extracted , particles displaced by the object behave as a solid , moving up en - masse .
particles slide down intermittently , relaxing into open holes , especially around the object itself . at short times , jamming and avalanching , or stick - slip , motion is prevalent , while over long periods a convective , turbulent motion occurs indicating fluid - like behavior .
apparently similar phenomena arise when an object is plunged slowly into a bed , although the forces on the object are significantly larger for the relatively shallow beds considered .
the obtained data in ref .
@xcite shows various interesting scaling behaviors , regardless of the extraction or plunging of the object .
for instance , the variations of the observed average forces as a function of depth for different object diameters @xmath0 can be collapsed by a simple rescaling of the force with its buoyancy and of the depth with @xmath0 , independent of the grain size in the bed .
intuitively , one would expect that plunging down through such a granular bed would always involve larger forces than withdrawing the same object . on the way down
, the object has to work against force chains @xcite communicating the effect of the rigid boundary conditions at the bottom ( and sides @xcite ) of the bed , while on the way up , force chains terminate at the open , force - free boundary at the top of the bed . hence , it comes as a surprise that in the experiment @xcite the force as a function of the depth of the object within the bed increases with a steeper power - law for withdrawing than for plunging .
this suggests that at large depths ( much larger than attainable in those experiments ) the force of withdrawing will eventually cross - over and exceed the force of plunging .
as the experiment ( and our model here ) shows , an object slowly withdrawn from a granular bed with a two - dimensional geometry has to lift up a bulk of interlocked material whose weight grows like @xmath1 , @xmath2 , with increasing depth @xmath3 ( measured down from the open boundary ) .
in contrast , it appears that a plunging object , failing to displace any bulk of material against the rigid boundary below , descends by _ locally rearranging _ material , for which relevant forces merely increase with @xmath3 in as much as the local pressure in the bed depends on @xmath3 .
accordingly , in the experiment plunging forces increase only slightly more than linear with @xmath3 .
unfortunately , our model does not allow for the study of plunging , for which the consideration of frictional forces appears essential .
in this report , we propose a simple two - dimensional model for the withdrawal of objects from granular beds .
inspired by the `` tetris '' model @xcite , it ignores many characteristics of granular materials , such as friction , to drastically simplify simulations .
results show that the simulation model exhibits both fluid and solid behaviors , as well as the characteristic stick - slip motion of granular materials .
additionally , we reproduce the same data collapse found in the withdrawal experiment , with similar scaling exponents . 2.9truein
in this model , dimers cover a link and the two adjacent lattice sites and are placed on a square - lattice that is tilted by @xmath4 with respect to the direction of gravity .
these dimers , once dropped , retain their orientation permanently in the lattice , leaning either right or left .
no two dimers can overlap , neither completely on the link nor partially on any of the adjacent lattice sites .
when pushed or chosen for an update , dimers move according to the rules depicted in fig .
[ update ] .
these minimally irregularly - shaped particles lead to complex , frustrated configurations with many uncovered sites , merely due to the excluded volume effect .
2.5truein on the computer , we initially place dimers into the lattice , starting from a rigid bottom , row - by - row , in random sequential order within each row to avoid super - rough surfaces generated by purely random deposition @xcite .
at each site , dimers can initially choose random orientations .
placements forbidden due to the excluded volume are rejected , which leads to the formation of voids .
the system s width , which has periodic boundaries , is always chosen sufficiently large to avoid boundary effects .
once initialized , we insert from the bottom an object spanning a number @xmath0 of lattices spacings . after each upward move ( by one lattice spacing ) of the object ,
the granular bed of dimers is updated , following the rules in fig .
[ update ] .
[ lift ] shows the complete cycle , beginning with the object movement .
starting from the surface of the object , particles are tagged if they need to move up due to a displacement of the object or other particles below them .
this process ends only at some void or , more likely , at the open surface .
the weight of all the tagged particles ( marked green in fig . [ lift ] ) constitute the force needed to move the object at that time .
after that force is recorded , all tagged particles are moved up in parallel .
then , random sequentially in each row from the bottom to the top , particles are allowed to relax by sliding into available voids below them , according to the rules in fig .
[ update ] .
this relaxation process may have to be repeated , until all particles again have a stable support . at this point
, the object once again moves up and all preceding steps are reiterated .
in fig . [ liftscal ] we plot the average force on the object , given by the lifted mass , as a function of the depth @xmath3 .
we find data collapse of the average force vs. depth by rescaling with the object diameter @xmath0 in the same way as was found for the experimental data @xcite .
the force varies minutely faster than quadratically with depth , indicating that the object lifts a wedge of bulk material with a slightly outward - curving interface .
we let the location @xmath5 of the interface as a function of the distance @xmath6 from the path of the object ( in two dimensions ) have the functional form of a `` poweriod '' @xcite , @xmath7 where @xmath8 and @xmath9 are constants .
in this case , the mass , the area within poweroid , scales with the depth of the object @xmath10 , with @xmath11 . in fig .
[ lift ] , the interface seems to be almost linear , @xmath12 , consistent with a quadratic relation , @xmath2 .
the experiments find an exponent of @xmath13 for lifting a long horizontal bar of width @xmath0 , which is comparable to our two - dimensional geometry .
this indicates that the lifted wedge has a more parabolic cross - section , @xmath14 .
it appears that frictional forces , absent in the model , allow steeper overhangs on both sides of the interface between moving and stationary particles .
2.2truein 2.2truein fluctuations between the forces on the object during its upward motion seem to be entirely uncorrelated , both in the experiments as well as in our simulations . in fig .
[ fluctuations ] we merely show one generic instance of the changes in mass lifted ( relative to the total mass lifted at that point ) between consecutive upward steps .
both sets of data show the same random characteristics , and the corresponding correlation functions are essentially trivial .
this suggests that the interface between moved and stationary particles fluctuates randomly and without long - term memory between consecutive moves .
the model also exhibits stick - slip behavior without any additional assumptions about friction or other forces .
the irregular shape of the particles permits the creation of voids behind the withdrawing object by jamming the passage of other particles into the wake , as fig .
[ stickslip ] shows .
thereby , the weight resting on the object remains high ( `` stick '' ) until the jam is resolved by moving the object further up and particles can pour in to fill the void ( `` slip '' ) .
this stick - slip behavior occurs throughout a run , creating large fluctuations in the mass lifted ( see fig .
[ fluctuations ] ) .
1.7truein 2.7truein in addition to measuring the force on the object , we have tracked tracer particles to study the transport of particles during the upward motion of the object in more detail . as a result
, we find fluid - like behavior over these longer periods . as fig .
[ tracer ] indicates , there exists a convective - like motion among the particles .
this convective zone does not seem to penetrate far from the track of the ( slow - moving ) object .
we have introduced a simple model for vertical granular drag and presented some preliminary simulation results which convey the major aspects of granular motion , including the solid , fluid and stick - slip behaviors .
the strength of this model lies in its simplicity . without major assumptions about forces such as friction or electrostatics
, this simulation replicates , to good agreement , experimental data . in particular
, the model reproduces the essential scaling collapse found for the corresponding experiments @xcite .
future extension of this model could involve the study of fast - moving objects , where the velocity could be tuned by prematurely stopping the updates in the relaxation process .
another interesting application of this model concerns motion of an object at any angle , but only above the horizontal .
essential for the applicability of our current model is that any stress induced by the motion can be relieved at the open surface .
we would like to thank stephan koehler for sharing his time explaining his experiment as well as sharing his data .
tomasz kott would like to thank emory university for its hospitality and the howard hughes medical institute for supporting this research through grant # 52003727 .
this work was also supported under grant # 0312510 of the division of materials research at the national science foundation . | motivated by recent experiments on objects moved vertically through a bed a glass beads , a simple model to study granular drag is proposed .
the model consists of dimers on a slanted two - dimensional lattice through which objects are dragged very slowly to obtain full relaxation between moves .
such an approach avoids complications due to static friction in more realistic off - lattice models , and provides for fast simulations at large system sizes .
the upward motion of objects of various diameters embedded in the lattice is simulated and close resemblance with the experiments is found . |
current efforts in observational cosmology are directed towards characterizing the energy content of the universe .
the discovery of the accelerating expansion of the universe inferred from the hubble diagram for sn type ia @xcite combined with constraints from cosmic microwave background radiation from the wilkinson microwave anisotropy probe @xcite , from cosmic shear observations @xcite , cluster baryon fractions @xcite , galaxy surveys @xcite suggests that @xmath0 per cent of the energy in the universe is in the form of dark energy .
dark energy is best described by the equation of state that relates its pressure , @xmath1 , to its energy density , @xmath2 , via @xmath3 , with @xmath4 corresponding to the case of the cosmological constant .
massive foreground clusters often produce many sets of highly distorted arcs and multiple images of background galaxies @xcite .
the lensing effect is determined by the mass distribution , the precise configuration of the lens and source with respect to the observer , and the angular diameter distances between the lens , source and observer .
when at least two sources at distinct redshift are strongly lensed in a cluster , the relative positions of the arcs depend on the ratios of angular diameter distances . in principle therefore , cosmological parameters can be constrained from the lensing configuration .
constraining the geometry and matter content of the universe using multiple sets of arcs in cluster lenses has been explored by @xcite .
recently , @xcite has suggested using multiple images produced by galaxy lenses instead of clusters and attempted an application to the case of the lensing system hst14176 + 5226 .
@xcite showed that the cosmological sensitivity of the angular size - redshift relation could be exploited using sources at distinct redshifts and developed a methodology to simultaneously invert the lens and derive cosmological constraints .
@xcite applied the technique to the cluster cl0024 + 1654 and found support for an accelerating universe .
@xcite showed that the recovery of cosmological parameters was feasible with 3 sets of multiple images for a single simulated cluster .
@xcite then applied the technique to the lensing cluster abell 2218 and using 4 multiple image systems at distinct redshifts , find @xmath5 and @xmath6 assuming a flat universe .
recently , a double einstein ring has been discovered and applied to cosmography @xcite . as the system has only two sources ,
only weak cosmography constraints were obtained .
there have been several studies that have explored the cross - sections of multiple image formation using simulated clusters from cosmological _
n_-body simulations .
@xcite and @xcite studied the efficiency of numerical clusters to produce multiple images in different dark energy models .
they found the expected abundance of arcs with a given length - to - width ratio and the lensing cross - sections depend on the equation of state of dark energy .
these statistical studies of multiple image formation concluded that strong lensing offers a plausible method to discriminate amongst various dark energy models .
in this paper we study two basic aspects of the non - statistical approach to cosmography with cluster strong lensing ( csl ) , which have yet to be addressed .
these are : 1 ) how many clusters must be individually marginalized , with the resulting @xmath7 from each cluster added together to form a final @xmath7 for cosmology recovery ( i.e. stacked ) , in the ideal case to obtain cosmological constraints comparable with other methods and 2 ) the required accuracy of the observational parameters , i.e. the source redshift and image positions . by studying these two aspects separately from the subtleties involved with cluster mass distribution reconstruction , we can make definitive statements on the type of cluster surveys required for a first application of csl .
we do not model the lensing effects induced by the distribution of substructure along the line of sight .
this additional complication has been examined in earlier work by @xcite and @xcite , who both used ray tracing methods through @xmath8-body simulations . simulating a representative shell of the universe centered at @xmath9
, @xcite found that approximately 10%-15% of the lensing mass was to be found in substructure along the line of sight , for about one third of lensing configurations with a source redshift above @xmath10 .
in particular , it was shown that substructure along the line of sight is limited to a finite number of lens planes , typically one or two additional planes , when line of sight substructure played a significant role in lensing .
they have shown explicitly that @xmath11 of the contribution to the convergence arises from secondary lens planes . for cluster
strong lensing the critical factor is not the error induced in the convergence but rather the perturbations caused in the positions of the multiple images due to the presence of matter in additional lens planes .
a detailed and exhaustive set of cosmological simulations that will enable the matter distribution of a typical line of sight to be averaged over many representative lines of sight is needed to accurately quantify this effect . in @xcite
it was shown that systematic errors in the cosmology recovery could result , if the line of sight substructure was not taken into account .
the outline of this paper is as follows : we begin with an introduction to the basics of strong lensing and dark energy in section 2 .
then in section 3 we discuss image families , and define the cosmological family ratio , which csl constrains . the methodology used to generate smooth simulated clusters
is described in section 4 and section 5 , along with the @xmath7 minimization scheme used in the cosmology recovery .
we present the results of recovering the input cosmology in section 6 , and study the dependence of the recovery on lens parameters in section 7 .
we then study the allowed errors in the source redshift and image positions in section 8 .
this is followed by a discussion of the simplifying assumptions made in this work in section 9 .
we conclude with a summary of our technique , and the implications for future survey strategies .
the gravitational lens equation is a mapping from the source plane at redshift @xmath12 to the image plane ( or lens plane ) at redshift @xmath13 and is given by , @xmath14 where @xmath15 is the angular position of the image , @xmath16 is the unobserved source position , @xmath17 is the deflection angle which depends on @xmath18 the reduced angular diameter distance , @xmath19 the mass distribution of the lens and @xmath15 . the cosmological dependence is contained in @xmath18 , which we call the reduced angular diameter defined by : @xmath20 , where @xmath21 is the angular diameter distance between redshift @xmath22 and @xmath23 , and @xmath24 , @xmath25 and @xmath26 are the observer - lens , lens - source and observer - source angular diameter distances respectively . in a flat homogeneous friedmann - robertson - walker cosmology
the angular diameter distance is given by , @xmath27 where @xmath28 is the speed of light ; @xmath29 is the present day value of the hubble constant , and @xmath30 and @xmath31 are the fractional contributions of the total matter and dark energy respectively , in units of the critical density and @xmath32 is the dark energy equation of state .
note that we do not consider evolving dark energy models in our simulations , since there is currently no observational evidence for such a time evolution in the dark energy equation of state @xcite . to further explore the dependence of the lens equation on cosmological parameters
, we note that the deflection angle can be written in terms of the gradient of the reduced gravitational potential @xmath33 : @xmath34 .
the reduced gravitational potential can be related to the projected surface mass density along the line of sight , @xmath35 , through the poisson equation , where the projected gravitational potential @xmath36 is defined by @xmath37 in the thin lens approximation the projected gravitational potential is a solution of : @xmath38 .
the total deflection @xmath39 is therefore determined both by the cosmological parameters ( via @xmath40 ) and the projected surface mass density of the intervening lens . for a given configuration of the source , lens and observer ,
the critical value of the surface mass density is : @xmath41 the region where the projected surface mass density of a lens exceeds this critical value defines the strong lensing regime .
background sources that are aligned behind this region are likely to be multiply imaged , magnified and distorted .
there are hence two fundamental equations which define the gravitational lensing dependence on the angular diameter distances @xmath42 .
the first is the lens mapping , equation ( [ lensequation ] ) , which defines the configuration of the images produced by a lensing cluster , and the second is equation ( [ eqn : sigmacrit ] ) which defines the strong lensing regime .
the cosmological dependence of strong lensing is discussed further in section [ sec : cosmofamilyratio ] .
consider a lensing configuration where at least two sources have been strongly lensed by a cluster .
once the multiple images of each source have been identified , we can write the source and the associated image positions with a family index @xmath43 .
the lens equation is then , @xmath44 where @xmath45 denotes the position of the @xmath46 image of family @xmath43 . when there are @xmath47 multiple images detected in family @xmath43 , we can sum over all the images in equation ( [ lenseqintrofinot ] ) .
substituting into equation ( [ projgravpot ] ) for the reduced gravitational potential allows the summed equation to be written in terms of the line of sight projected gravitational potential as follows : @xmath48 note that equation ( [ lenscos1 ] ) is a non - linear function of the image positions , since the sum of the image positions depends on the projected cluster potential gradient . given a set of input cosmological parameters , the lens redshift and a source redshift for a single family of images , i.e. @xmath49 , the angular diameter distance ratio is specified and constant .
the cluster potential is conventionally normalized by the square of the velocity dispersion , which we denote by @xmath50 . in equation ( [ lenscos1 ] )
the reduced angular diameter distance @xmath18 and the cluster mass normalization @xmath51 are degenerate when recovering cosmological parameters .
_ therefore , cosmological parameters can not be disentangled from the cluster mass normalization with only one family of multiple images . _
possible solutions to remove this degeneracy are considered by @xcite .
they show an additional prior on the mass profile is required and this must come from observations independent of strong lensing . for parametric models ,
this is typically a total mass determination from the detected x - ray emission of the cluster .
now consider the case when two strongly lensed sources are observed as multiple image families in the same cluster lens at two redshifts .
let family 1 have a total of @xmath52 observed images and family 2 have @xmath53 observed images .
we require that at least two images from each are observed , i.e. @xmath54 . taking the ratio of family 1 to family 2 ( equation [ [ lenscos1 ] ] ) we obtain , @xmath55 note that we now have a ratio of two factors of @xmath56 .
this clearly means that the degeneracy between the cosmological parameters and the cluster potential normalization is broken when 2 families of multiple images are observed in a cluster lens . in equation ( [ lenscos2 ] ) , the term in braces is a function of the angular diameter distances only .
this quantity is fundamental to the determination of cosmological parameters through csl and we define this term as the family ratio and denote it by @xmath57 : @xmath58 where @xmath13 is the lens redshift , @xmath59 and @xmath60 are the two source redshifts , and @xmath61 is the angular diameter distance defined in equation ( [ angulardiameter ] ) .
the cosmological family ratio can be written in terms of the reduced angular diameter distance , @xmath62 , where the subscripts 1 and 2 here denote the family index .
we plot @xmath57 in fig .
[ fig : familyratio ] for a fixed lens redshift of @xmath63 and differing source redshifts . since the non - evolving dark energy component with @xmath64
can also be interpreted as the cosmological constant @xmath65 , we plot the contours of the cosmological family ratio in both the @xmath66-@xmath65 plane in addition to the @xmath66-@xmath32 plane .
note that the contours in the @xmath66-@xmath65 plane are not as sensitive to the lens - source configuration as they are in the @xmath66-@xmath32 plane .
as shown in the previous section , the cosmological family ratio , @xmath57 , depends on the source redshifts and the lens redshift ( see fig . [
fig : familyratio ] ) .
it is therefore clear that the distribution of sources in magnitude and redshift as well as the cluster redshift distribution are key inputs in our simulations .
therefore we base our simulations of strong lensing clusters on high resolution observational data from _ hst _ images .
we note that although the cluster normalization is irrelevant to our method for constraining the cosmological parameters , we still construct this quantity rigorously .
this ensures the simulations produce lensing clusters with cross - sections similar to those of observed clusters . in order to simulate the lensed images
produced by a massive cluster , it is necessary to characterize the number density , @xmath8 , of the sources behind the lens .
we consider background point sources for which the number density is a function of magnitude and redshift , i.e. @xmath67 .
furthermore , space - based positional accuracy is assumed throughout this paper . we must therefore draw redshifts and magnitudes from a space - based source distribution .
the hubble deep field ( hdf ) @xcite and hdf - south ( hdf - s ) wfpc2 @xcite are the ideal data sets to derive the source distribution .
it is also possible to use the recent cosmos data to derive the redshift and magnitude source distributions , see @xcite .
in particular we have checked that our source redshift distribution , which is derived below , is equivalent to that of @xcite .
we employ the wfpc2 hdf and hdf - s photometric redshift catalogues derived by the stony brook collaboration @xcite and characterize the hdf sources by a distribution function .
this method is chosen since we wish to analyse an ensemble of clusters , where each cluster lenses a unique source configuration . characterizing the source distribution by a function
is preferable in our case , since we do not want to bias the results by lensing large scale structures such as voids or filaments .
note that there is evidence of a weak cluster in the hdf @xcite . to model the source distribution from the hdf and hdf - s data ,
we first make two cuts to the catalogues .
initially , we drop all objects with magnitudes greater than @xmath68 .
next we select objects in the redshift interval @xmath69 .
the lower bound is chosen because the lensing efficiency is negligible for low source redshifts .
the upper redshift bound is chosen because the photometric redshift determination is unreliable for @xmath70 and most observed multiply imaged sources behind massive clusters have redshift @xmath71 .
this yields @xmath72 objects when the catalogues are merged .
the corresponding projected source number density is @xmath73arcmin@xmath74 for the ensemble catalogue .
we now bin the merged catalogue in magnitude from @xmath75 in intervals of @xmath76 and in redshift from @xmath77 in intervals of @xmath78 .
these data are then fit with functions of the form , @xmath79\,\exp[-\,c\,(z - d)^{t } ] \\ & & \times\exp[-\,g\,(m - b)^{u}(z - d)^{v}],\end{aligned}\ ] ] where the exponents in the exponential terms , @xmath80 , can take an even integer value .
the constants @xmath81 and @xmath43 are determined in the fitting procedure .
we simply use the lowest @xmath7 fit to construct a probability distribution function in the simulations .
this is @xmath82 , with the other parameters given by : @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87 and @xmath88 .
notably , the @xmath7 of the above fit is approximately one half of the @xmath7 that is obtained when the binned magnitude and redshift data are decoupled and fitted on an individual basis , i.e. @xmath89 and @xmath90 .
it is therefore clear that to avoid a systematic bias in the source generation , a joint fit in magnitude and redshift space is required .
deep acs imaging of abell 1689 has allowed identification of more than 30 families of multiply images systems @xcite .
it is clear that abell 1689 is the ideal candidate to compare with our source distribution .
we base our analysis on @xcite , however our conclusions are unchanged if we use the analysis of @xcite . with the lens model as a guide
@xcite have identified multiple images to @xmath91 .
this means images with magnitudes comparable to and less than the magnitude limit of our hdf catalogues , @xmath92 , are detected in the acs observations .
this means we can compare our source distribution to the properties of abell 1689 directly . in order to compare the source characteristics of abell 1689 with our source distribution ,
we need to lens this distribution through a mass model . @xcite
have published an navarro , frenk , & white ( nfw ) fit ( see section [ sec : nfw ] ) to approximate the true mass profile . using a radially averaged ( and therefore circular ) mass profile
, they provide the tangential and radial critical curves , computed at redshift @xmath93 .
they also determine the nfw scale radius @xmath94kpc and we adopt @xmath95kpc .
we then adjust an nfw fit until the two critical curve locations are reproduced .
since we are using a smooth profile , the properties of abell 1689 we compare our source distribution with can not depend on the small scale details of the mass distribution .
there are two quantities which satisfy this requirement : 1 ) the total number of lensed sources and 2 ) their redshift distribution
. a test of the hdf constructed source distribution would be the ability to reproduce these two quantities for simulated lenses similar to abell 1689 .
note that we do expect some systematic effects due to the lack of assumed substructure in the modeling .
in particular , the fact that abell 1689 is non - circular , and is known to have images which are influenced by substructure , means any comparison between the actual data and our model ( which is circular , and has no structure ) will exhibit residuals . to compare our simulations with the abell 1689 sources , we take @xmath96 as the source redshift for each image from table 2 of @xcite . also , because the hdf catalogues used to derive the source distribution
have been cut at @xmath97 , we drop all sources with redshift @xmath98 .
the abell 1689 source redshifts are then binned in intervals of @xmath99 from @xmath100 to @xmath97 . for proper comparison with the simulations
, the bin counts are normalized by the total number of families in abell 1689 .
we simulate the smooth abell 1689 nfw profile for 500 source configurations generated with our source distribution .
we then lens the sources independently for each configuration , using the procedure explained in section [ sec : clusterandimage ] .
multiply imaged sources are identified and binned identically as the abell 1689 sources .
the average number of multiply imaged sources in the 500 simulations is @xmath101 per simulation .
the minimum and maximum number of families are 7 and 28 respectively and of the 500 configurations 102 exhibit 20 or more families of images .
to @xmath97 there are 27 families in the @xcite analysis .
although the mean number for our generated images is lower than that of the 27 strongly lensed sources in abell 1689 by @xmath102 , we do not consider this a serious problem .
the total number of strongly lensed sources is within factor of 2 in 69 per cent of the 500 realizations .
we now examine the redshift distribution of the lensed sources in the simulations .
the configurations are combined to produce a distribution of redshift probabilities in each bin .
this allows the mean and standard deviation to be computed , see fig .
[ figabell ] .
our simulated redshift distribution lies within 1@xmath103 of the abell 1689 sources , except for the @xmath104 bin for which there are no observed multiply imaged sources in abell 1689
. finally we discuss the image properties themselves .
since the mass model does not include substructure , we do not expect to reproduce the source multiplicity distribution or the image magnitude distribution of abell 1689 accurately . for a circular nfw mass profile , lensed sources with multiplicity of at most 3
can only be produced , whereas higher multiplicities are observed in abell 1689 .
however , as shown in fig .
[ figabell ] , the image magnitude distribution as generated through our source distribution does match that of abell 1689 , and the overall shape is reproduced .
both 1 ) the total number , and the 2 ) redshift distribution of the simulated sources are consistent with the source characteristics of abell 1689 . given the results of this comparison , we conclude that the hdf derived source population distribution is a faithful representation of the actual source distribution behind massive lensing clusters . in order to generate a set of massive , strong lensing clusters in our simulations , we need to derive redshifts from an observed cluster redshift distribution .
to do so , we use the massive cluster survey ( macs ) @xcite sample as our template .
we choose this sample since it is x - ray luminosity selected , which is a robust proxy for mass , and the _ hst _ follow - up of these clusters has demonstrated that a large majority of these are strong lenses @xcite .
we use the binned macs cluster redshifts ( see fig . 9 in @xcite ) to assign @xmath105 to each redshift bin .
the @xmath106 bin is omitted in the analysis since the cross - section for multiple imaging by clusters in this redshift range is negligible .
an exponential fit is performed and the result is @xmath107 $ ] .
the fit is then normalized , integrated and inverted to form a probability distribution .
the final probability distribution returns a cluster lens redshift for use in the simulations in the range @xmath108 and can be expressed analytically as @xmath109 $ ] , where @xmath110 is a random number between 0 and 1 .
the nfw density profile is derived from fits to cluster scale dark matter haloes that form in _
n_-body simulations of a cold dark matter dominated universe @xcite .
this density profile is given by , @xmath111 where @xmath112 is orientated along the line of sight and the radius @xmath113 is orthogonal to the line of sight .
there are two defining profile parameters for the nfw model , the characteristic density @xmath114 and the scale radius of the profile @xmath115 . the corresponding characteristic velocity dispersion @xmath50 is defined as , @xmath116 in our simulations we will calculate the total projected mass along the line of sight within the radius @xmath117 , where @xmath28 is the concentration parameter . for @xmath118 this is given by , @xmath119\ ] ] where @xmath120 ( see @xcite for further details on the profile )
. strong and weak lensing analysis of massive cluster lenses have determined that the total mass of such clusters within @xmath121 1 mpc , lies in the range , @xmath122 , @xcite .
this mass range is the starting point for our construction of simulated smooth cluster lenses .
consider a cluster scale halo with an nfw profile . to generate the halo we first draw a total cluster mass from the range @xmath123 uniformly .
we then assign profile parameters to characterize the mass distribution .
a concentration is drawn randomly from the range @xmath124 @xcite .
strong lensing models using nfw fitting profiles find that the typical scale radius lies in the range @xmath125kpc . from the total mass , @xmath28 and @xmath115 , we compute the characteristic velocity dispersion of the nfw halo .
note that we ignore any scaling of @xmath114 with cosmology for simplicity .
finally , with the halo structure determined , the ellipticity is then drawn randomly from @xmath126 .
in this section we discuss the details of our simulations .
firstly , the procedure used to generate the lensing clusters and multiple image catalogues are discussed , and we explain how the source magnitude and redshift distribution and the cluster redshift distribution are used in the simulations .
we then discuss our mass modeling assumptions .
finally , our marginalization method used to construct the confidence regions for cosmological parameter recovery are presented . to perform our analysis
, we generate clusters that will produce the requisite number of families of multiple images .
the methods used to generate the cluster lenses are described in detail in section [ haloconstruction ] .
we choose all clusters that produce multiple images which can be detected via relative astrometry in the high resolution channel ( hrc ) of acs .
this means the effective einstein radius @xmath127 , which we compute at @xmath128 , must be less than @xmath129 .
in addition to this upper bound we also apply a lower bound of @xmath130 .
this bound restricts the simulations to massive clusters that plausibly exhibit image configurations that are measurable with the hrc .
we note there is no effect on the simulations if the maximum einstein radius condition is dropped , as long as the same pixel resolution is maintained
. this may be the case if observations were taken with the wide field camera on _ hst _ and the images were processed with sub pixel dithering .
once an appropriate cluster has been generated a source catalogue is generated .
this catalogue is drawn from the hdf distribution derived in section [ sec : sourcedistconstruct ] . a redshift and magnitude
is assigned for each source using the distribution derived ( numerically ) from equation ( [ eqn : hdffit ] ) .
each source is also assigned a random position in the field of view @xmath131 .
the number of sources in the catalogue for which magnitudes , redshifts and positions are generated , is dictated by the projected source number density of the ensemble hdf catalogue which is @xmath132 arcmin@xmath74 and the field of view .
the source catalogue is then lensed with the code lenstool @xcite .
the lensed catalogue is then sorted into single images and multiple image families . at this point
we have a complete simulated data set that is equivalent to a known observed sample of ideal lensing clusters .
observationally , this can be accomplished efficiently through high resolution space - based images of known massive clusters or deep ground based observations of clusters .
we note that the macs cluster survey provides a catalogue of clusters selected through their x - ray luminosity .
this cluster catalogue provides a useful starting point for observing programs to identify strongly lensing massive clusters .
such a systematic survey is currently underway @xcite on _ hst _ acs .
the image catalogues we use in our recovery of the cosmological parameters assume _ hst _ quality spatial resolution and spectroscopic redshifts .
we first examine the ideal case , where the image positions and redshifts are known exactly . in section 8
we expand our simulations to examine the effects of errors in the image positions and source redshifts .
however , before considering possible sources of error , a best case scenario must be established when cluster lenses are stacked . in the simulations ,
a magnitude limit is specified @xmath133 , when constructing the image catalogue from the lensed sources and is taken to be , unless otherwise stated , @xmath134 in the ab(8140 ) band @xcite .
this ensures that all images will have measurable spectroscopic redshifts .
therefore , all families in the lensed image catalogue which do not have two images with magnitudes @xmath135 are removed at this point . applying
the magnitude cut prior to the final selection of families in the image catalogue is a stringent requirement , since we only use images with @xmath135 in a given family . in the case of observational data , although it may not be possible to identify all members of a given image family spectroscopically , it is often possible to use properties like the image parity , morphology , photometric redshifts , or a combination of these to make the identification . finally , since core images have been detected in lensing studies @xcite
, we do not remove such images unless they do not pass the magnitude cut . in the spirit of our simulations , where we are investigating the ideal case of stacking strong lensing clusters
, we use the same form of the mass profile that was used to generate the image catalogue .
the centre of the lens is fixed at the origin and the ellipticity and lens redshift are taken as known from the original cluster lens . in section
[ sec : lensdep ] , we describe how the profile characteristics of the mass modelling influence the cosmology recovery . it is shown in section [ sec : lensdep ] that for the nfw profile , the velocity dispersion and scale radius should be considered unknown in the cosmology recovery .
we note that the prior of a known mass profile is not the case when using observational data , where the underlying mass model is unknown .
a variety of mass model priors will need to be considered in practise , since the mass distribution can be complicated , see for example the bullet cluster of @xcite .
when observational data is analysed the cosmological parameter bias from the modelling of realistic clusters must be understood .
these additional aspects present significant computational challenges when stacking a large number of clusters and we defer their study for the time being .
the @xmath7 minimization we use is computed in the source plane for computational speed , since the lens equation is many to @xmath136 when mapping from the image plane to the source plane .
our measure is given by , @xmath137 where @xmath138 is the source position corresponding to image @xmath139 from the @xmath46 family as mapped back through the lens equation ( equation [ [ lensequation ] ] ) , and @xmath140 is the standard deviation squared in the source plane .
the mean source position , @xmath141 , is defined as @xmath142 and is the average over all source positions of images belonging to the @xmath46 family .
therefore @xmath143 is the difference between the observed image positions as viewed from the source plane . in the calculation of equation ( [ eqn : chi2 ] ) ,
@xmath140 is computed using the assumed error in the image plane of @xmath144 . using the known image position , we sample at eight image error points : @xmath145 , @xmath146 , @xmath147 and @xmath148 , and map each point back to the source plane using the lens equation .
we then calculate the difference between @xmath138 and each of the mapped error points .
the eight @xmath149 are first squared before averaging , i.e. @xmath140 , because this over estimates the standard deviation squared relative to the reverse order . therefore using this averaging method leads to a _ conservative estimate _ of the final @xmath7 and thus the confidence intervals ( cis ) .
the minimization routine , which uses a parabolic optimization in combination with parameter space bounds , is designed to ensure the @xmath7 surface of each cluster is accurate to at least @xmath150 .
this accuracy is sought because we want the final likelihood accurate to @xmath151 per cent .
after this is achieved we form the 1@xmath103 to 4@xmath103 cis .
.recovery properties on the @xmath66@xmath32 plane .
the bounds are the limits of the @xmath152 confidence contours , with the input cosmology @xmath153 subtracted .
@xmath154 is the area enclosed by the @xmath152 contour ( see equation [ [ eqn : areaenclosed ] ] ) . [ cols="^,^,^,^,^,^",options="header " , ] we investigate the cosmology recovery properties by considering cluster samples with different family numbers . to obtain the recovery properties , we proceed as outlined in sections [ simulations ] and [ sec : analsim ] . at this stage
the input parameters to the simulations are the number of clusters to stack , the total number of families in each cluster and the magnitude limits for the multiple images .
the recovery properties are computed for each cluster , in the specified cosmological parameter space .
the constraints from each cluster in the sample are then stacked to obtain the final recovery properties .
we consider observations of 5 and 10 clusters , with 2 , 3 and 5 multiple image families .
the magnitude limit , @xmath133 , is @xmath155 and @xmath156 for the 2 , 3 and 5 multiple image families respectively and the recovery properties of the input cosmology are computed in the @xmath66-@xmath157 and @xmath66-@xmath32 planes .
the results of stacking the first 5 generated clusters and then all 10 are shown in fig . [ fig5and10clusters ] and characterized in table [ tab : simproperties ] for the @xmath30-@xmath32 plane . in the @xmath66-@xmath158 plane
we observe that the parameter constraints obtained by _ stacking _ lensing clusters is insensitive to the value of the cosmological constant when marginalizing over @xmath66 .
the @xmath66 constraints are also not stringent , with @xmath66 constrained to lie between approximately 0.1 and 0.50 for the 5 family 10 cluster case .
these properties of the cosmology recovery in the @xmath30-@xmath158 are expected . by examining fig .
[ fig : familyratio ] , one can see that the contours passing through the input cosmology are degenerate for different source redshift configurations .
this means that the constraints derived from the family ratio will be tightest around the contour defined by the input cosmology . however , in the @xmath66-@xmath32 plane ( lower two rows of fig .
[ fig5and10clusters ] ) the family ratio is sensitive to the dark energy equation of state . in fig .
[ fig : familyratio ] , we see that the @xmath57 contours passing through the input cosmology are quasi - orthogonal for different source redshift configurations .
therefore clusters with distinct source redshift planes will have different constraint directions in the @xmath66-@xmath32 plane .
this is also the case for different lens redshifts given the same two source redshifts however the effect less pronounced .
when _ stacking the cluster sample _ the orthogonality of different family ratios leads to _ closed _ contours in the @xmath30-@xmath32 plane , for the 5 family 10 cluster simulation .
we then examine the constraints obtained when 40 and 100 clusters are stacked for 2 , 3 and 5 multiple image families .
the magnitude limits are the same as for the 5 and 10 cluster runs but only the @xmath66-@xmath32 plane is considered .
the results are shown in fig .
[ fig40and100clusters ] and table [ tab : simproperties ] .
note that 40 clusters with 5 families and 100 clusters with only 3 families , produce almost identical cis .
the role of the lens parameters in the optimal retrieval of the input dark energy equation of state is examined in this section .
we address the issue of the @xmath7 dependence on the marginalization over combinations of the dominant lens parameters .
the marginalization for each parameter set is computed on the @xmath66-@xmath32 plane in a flat universe , for an input cosmology @xmath159 .
we also consider the constraints derived from a full multi - dimensional marginalization and compare to the corresponding simple @xmath7 minimization .
a massive nfw halo is considered in this section and substructure is omitted from the computations for the purpose of simplicity .
the nfw halo is fixed at the origin and the parametrization we use is typical of massive cluster lenses @xcite and the lens parameters we use are : @xmath160 km s@xmath161 , @xmath162kpc and @xmath163 , with the lens redshift @xmath164 .
two sources are lensed and they are located at @xmath165 and @xmath166 respectively , with redshifts @xmath167 and @xmath168 .
sources 1 and 2 have multiplicities of 5 and 3 respectively .
first we consider the case where one has complete knowledge of the gravitational lens , i.e. all lens parameters are known exactly .
the result of the marginalization with all lens parameters fixed is shown in the upper left panel of fig .
[ fig : nfwarray ] for the nfw lensing cluster .
note that the confidence region for the nfw profile is non - zero .
in fact , the existence of a non - zero confidence region is a general feature of the lensing systems in our simulations .
based on fig .
[ fig : nfwarray ] it is clear that the intrinsic confidence region of a lensing system represents a fundamental limit on the determination of the cosmological parameters from a single configuration . even in the optimal scenario , i.e. complete knowledge of the lens system , one must consider the confidence region and not the best fitting cosmology in a rigorous statistical treatment .
the best - fitting cosmological parameters , as for example employed in @xcite , can not be used to characterize the cosmological parameter recovery , since they do not take into account the limitation of the intrinsic confidence region . noting this property , we have used cis exclusively in this work . in practice
, one does not have prior information on all the parameters defining the mass profile of an observed lensing cluster .
for the nfw lensing profile , there are three combinations of the lens parameters to consider marginalizing over when recovering the cosmology .
these are : the velocity dispersion @xmath50 , the scale radius @xmath115 and both @xmath50 and @xmath115 .
the results of each marginalization are given in fig .
[ fig : nfwarray ] . to 4@xmath103 contours in the @xmath30-@xmath32 plane are shown , and the cross marks the input cosmology @xmath169 .
the parameter(s ) in the upper right corner of each panel specify the marginalization variable(s ) . ] for the nfw profile , the shape of the confidence contours changes for the different marginalization parameters .
for the case of marginalization over @xmath115 alone , in which @xmath50 is held fixed , closed confidence contours are obtained in the @xmath30-@xmath32 plane .
this is distinct from the contours derived for the marginalization over @xmath50 alone , which yields a confidence region which follows the contours of constant cosmological family ratio @xmath57 . in fig .
[ fig : nfwarray ] , the simultaneous marginalization of @xmath115 and @xmath50 expands the 4@xmath103 confidence region asymmetrically .
in addition , note that all other confidence contours qualitatively follow the contours for the family ratio .
therefore it is obvious that optimization over both the characteristic density and scale radius is required when recovering cosmology constraints from the nfw profile .
an additional comment on fig .
[ fig : nfwarray ] is warranted with regard to the cosmology recovery .
it is obvious from the panels in fig .
[ fig : nfwarray ] that when only @xmath170 or only @xmath50 are allowed to vary , that tighter cosmological constraints are obtained , compared to the case where both @xmath170 and @xmath50 are marginalized over simultaneously ( bottom right panel of fig .
[ fig : nfwarray ] ) .
it is therefore natural to ask , what are the cosmology recovery properties in the cases where only @xmath170 or @xmath50 are marginalized for stacked clusters ? however , when considering actual cluster lenses , if we were to marginalize over say only @xmath170 , we would have to fix the velocity dispersion .
this would require an additional prior , and in actual cluster lenses this prior is not typically available .
it may be possible to constrain the total cluster mass through x - ray observations , and in this case , one would use equation ( [ eqn : nfwtotalmass ] ) to apply the priors to _ both _ the velocity dispersion and the scale radius .
however , in general , when using an nfw profile to model an actual strong lensing cluster , both @xmath170 and @xmath50 are not known a priori @xcite .
the reader should therefore note that although stronger cosmological constraints could be obtained if @xmath170 or @xmath50 were known a priori , this is not the case in actual cluster models because separate priors on @xmath170 or @xmath50 can not be found . to this end , we have only presented the marginalization over both @xmath170 and @xmath50 in section [ sec : recovery ] .
up to this point , we have considered a simple minimization routine of the lens parameters with respect to the @xmath7 estimator .
we have done this over the cosmological plane and for different lens parameter combinations for the nfw profile .
however , we now examine the validity of this assumption for the nfw lensing configuration in section [ sec : marglens ] . we now perform a full multi - dimensional marginalization over the lens parameters , @xmath115 and @xmath50 , for the nfw lensing configuration .
our adopted method is as follows : first , we find the optimized lens parameters which gives the minimum @xmath171 at each cosmology point . from the optimized lens parameters ,
we then compute a @xmath7 surface extending to @xmath172 .
the parameter spacing on the @xmath7 surface is @xmath78arcsec for @xmath115 and @xmath173 km s@xmath161 for @xmath50 .
the @xmath7 surface is then converted into a likelihood surface using , @xmath174 which is integrated to obtain the marginalized likelihood at a given cosmology input .
this likelihood is then converted to a @xmath7 surface , using @xmath175 . to quantify the difference between the multi - dimensional marginalization and the simple optimization method
, we consider an integral which characterizes the enclosed area , @xmath176 where @xmath53 defines the confidence contour , and @xmath177 the region enclosed by the given contour . for the multi - dimensional marginalization over @xmath115 and @xmath50 , we have approximately , @xmath178 , @xmath179 , @xmath180 and @xmath181 .
whereas for the simple @xmath7 minimization @xmath182 , @xmath183 , @xmath184 and @xmath185 .
the difference between the enclosed area is largest for @xmath154 , which is @xmath186 per cent .
importantly , the 1@xmath103 contour from the simple optimization method encloses an area larger than the full multi - dimensional marginalization .
both methods recover a @xmath187 at the input cosmology . given the maximum @xmath188 per cent difference between the enclosed areas from the multi - dimensional marginalization and the simple @xmath7-optimization , we are justified in using the simple optimization method .
with sufficient multiple images to constrain the cluster potential , one expects only one global minimum in the @xmath7 optimization surface .
this is the case for the nfw lensing configuration considered here , and for similar lensing configurations one expects the optimization method to accurately reproduce the confidence contours derived from a full multi - dimensional marginalization .
we have also simulated smooth cluster lenses that contain observational noise .
the noise sources we have considered are positional errors and redshift error .
these two sources are the two major observational parameters which will determine the type of survey required for the application of csl .
our simulations are restricted to the 5 families 10 cluster case ( see bottom right panel in fig . [ fig5and10clusters ] ) in the @xmath30-@xmath32 plane .
our objective is to examine the levels of noise which do not lead to relaxation of the cis , when compared to our simulations of the ideal case .
we use this approach since this will represent a tolerance threshold where parameter constraints will weaken .
first we consider redshift accuracy .
we have re - run the 5 families 10 cluster simulation with gaussian redshift noise with the following standard deviations @xmath189 and @xmath190 . when adding the noise to the source redshift , we draw from the appropriate gaussian and add the resultant value to the known redshift .
note that all multiple images of a given family receive the same noise contribution to their redshift .
our results are as follows .
we find that photometric redshift accuracy , i.e. @xmath191 , leads to significant relaxing of the cis and in some cases does not recover the input cosmology correctly . of the remaining standard deviation simulations , all recover the input cosmology with increasingly tighter constraints . for @xmath192 ,
there is negligible confidence contour relaxation when compared to the ideal simulation .
next we move onto image positions . following a similar procedure to the redshifts
, we find that a positional accuracy of @xmath193 pixels of the _ hst _ acs is required to recover the confidence region in the @xmath66-@xmath32 plane .
this means that space based imaging of the strong lensing clusters will be required .
we have also considered image position error in combination with redshift error , and find that the combination @xmath192 and @xmath193 pixels in acs does not lead to relaxation of the confidence contours compared to the ideal case .
this means that space based imaging of strong cluster lenses , followed up by spectroscopic redshift determination of the multiple images , is the appropriate survey method to pursue for the application of csl to cosmography .
as with any other method used to constrain cosmological parameters , csl has a number of sources of error one must consider .
our goal has been to examine observational strategies under the most optimal conditions in this first detailed study .
our results demonstrate that by stacking lensing clusters with several multiple image families , confidence regions for the input cosmological parameters can be recovered , as shown in figs [ fig5and10clusters ] and [ fig40and100clusters ] . in earlier work
, @xcite raised the issue of parameter bias introduced by the unknown cluster mass profile , and the presence of substructure .
they concluded from extracting the best - fitting cosmological parameters ( rather than confidence regions ) that this technique produces @xmath194 per cent errors for a single cluster .
we are not disputing the importance of correctly modelling the mass profile and substructure .
however , in our simulations we have focused on the recovery properties of the ideal scenario , and the errors cosmography csl can tolerate in an observational data set .
@xcite has also investigated the recovery bias including line of sight density fluctuations using a single numerical cluster .
it should also be pointed out that any parameter recovery bias resulting from substructure along the line of sight ( fig .
4 of @xcite ) , will be minimized if uncorrelated lines of sight are used when constructing the cosmological constraints .
also , bias due to line of sight substructure could possibly be accounted for , by performing redshift surveys behind cluster lenses and modeling any detected structure along the line of sight in multiple lens planes . removing the systematic parameter bias due to substructure along
the line of sight is an important topic that requires further investigation and is left for future work .
since we have only used positional accuracy in our cosmological recovery , we should ask : what additional @xmath7 measures can we add to our current recovery measure of equation ( [ eqn : chi2 ] ) that will further constrain the input cosmology ?
the first obvious candidate is a higher order shape estimators for multiple images , since we have ignored all shapes and simply used a point approximation in our simulations . in real clusters
multiply imaged sources will have their shapes distorted as they are strongly lensed and this will give additional information about the lensing mass distribution , see for example @xcite .
another improvement may come from the possibility of using other observational maps of clusters to infer additional properties of the cluster .
for example , a total mass constraint can be obtained from x - ray cluster observations .
in this paper , we have examined the recovery of input cosmological parameters using the cosmological family ratio for several sets of multiple images in stacked simulated smooth lensing clusters . for a given lensing configuration ,
the constraints on the @xmath30-@xmath32 plane are tighter than those on the @xmath66-@xmath65 plane .
we find that the recovery of input parameters depends both on the number of multiple image families with known redshifts and the number of clusters stacked .
as shown in figs [ fig5and10clusters ] and [ fig40and100clusters ] , the confidence contours tighten as we increase the number of families from 2 - 5 , and as the total number of stacked clusters is increased from 1 - 100 .
for example , the recovery is comparable in terms of area enclosed within the 1@xmath103 - 4@xmath103 contours for the case with 5 families using 40 clusters and 3 families using 100 clusters . as we demonstrate , both strategies stacking fewer clusters with a larger number of image families or a larger number of clusters with fewer families lead to comparable constraints .
currently there are about 40 clusters with high resolution _ hst _
acs images , of which a handful ( about 7 - 10 ) have more than 3 families of multiple images with measured redshifts . proposed future space missions , the james webb space telescope and jdem will enlarge the existing lensing cluster samples by a factor of 20 or more , while simultaneously detecting a larger number of families on average for each cluster through very deep imaging .
determining spectroscopic redshifts for multiple image families is likely the limiting step in terms of the observational effort for the csl method .
if 10 clusters with 3 multiply imaged sources are stacked , the dark energy equation of state in a flat universe can be constrained to @xmath195 at 2@xmath103 where @xmath66 has been marginalized over , see fig . [ fig5and10clusters ] .
this is comparable with current methods @xcite .
therefore to be competitive with current methods , given that actual massive cluster lenses will present sub - optimal recovery conditions relative to those considered here , an initial application of csl to actual lensing clusters will require at least 10 clusters with 3 or more multiply imaged sources with abell 1689 quality _ hst _ acs images .
such constraints may be obtained only if the source redshifts are determined spectroscopically .
these factors indicates that initial observations should focus on massive clusters that exhibit large lensing cross sections .
once these strong lensing clusters are imagined to abell 1689 quality with the _ hst _ , those with 3 or more multiply imaged sources should be given priority when determining which lensing clusters to follow up spectroscopically .
we acknowledge useful discussions with jean - paul kneib . allen s. w. , schmidt r. w. , ebeling h. , fabian a. c. , van speybroeck l. , 2004 , mnras , 353 , 457 bacon d. j. , refregier a. r. , ellis r. s. , 2000 , mnras , 318 , 625 biesiada m. , 2006 , phys .
d , 73 , 023006 blandford r. d. , narayan r. , 1992 , ara&a , 30 , 311 brada m. et al .
, 2006 , apj , 652 , 937 broadhurst t. et al . , 2005 ,
apj , 621 , 53 bullock j. s. , kolatt t. s. , sigad y. , somerville r. s. , kravtsov a. v. , klypin a. a. , primack j. r. , dekel a. , 2001 , mnras , 321 , 559 casertano s. et al . , 2000 , aj , 120 , 2747 comerford j. m. , meneghetti m. , bartelmann m. , schirmer m. , 2006 , apj , 642 , 39 cooray a. r. , 1999 , apj , 524 , 504 dalal n. , hennawi j. f. , bode p. , 2005
, apj , 622 , 99 ebeling h. , edge a. c. , henry j. p. , 2001 ,
apj , 553 , 668 ebeling h. , 2005 , hst prop .
10491 efstathiou g. et al . , 2002 ,
mnras , 330 , l29 fernndez - soto a. , lanzetta k. m. , yahil a. , 1999 , apj , 513 , 34 gabasch a. , goranova y. , hopp u. , noll s. , pannella m. , 2008 , mnras , 383 , 1319 gavazzi r. , treu t. , koopmans l. v. e. , bolton a. s. , moustakas l. a. , burles s. , marshall p. j. , 2008 , apj , 677 , 1046 golse g. , kneib j .-
, soucail g. , 2002 , a&a , 387 , 788 hinshaw g. et al . , 2009 ,
apjs , in press ( arxiv:0803.0732 ) jullo e. , kneib j .-
, limousin m. , elasdttir . , marshall p. j. , verdugo t. , 2007 , njph , 9 , 447 kaiser n. , 2000 , apj , 537 , 555 kneib j .-
, 1993 , phd thesis , universit paul - sabatier , toulouse komatsu e. et al . , 2009 , apjs , in press ( arxiv:0803.0547 ) leonard a. , goldberg d. m. , haaga j. l. , massey r. , 2007 , apj , 666 , 51 link r. , pierce m. j. , 1998 , apj , 502 , 63 limousin m. , kneib j .-
, natarajan p. , 2005 ,
mnras , 356 , 309 limousin m. et al . , 2007 , apj , 668 , 643 lombardi m. et al . , 2005 , apj , 623 , 42 macci a. v. , 2005 , mnras , 361 , 1250 meneghetti m. , bartelmann m. , dolag k. , moscardini l. , perrotta f. , baccigalupi c. , tormen g. , 2005 , a&a , 442 , 413 meneghetti m. , jain b. , bartelmann m. , dolag k. , 2005 , mnras , 362 , 1301 natarajan p. , de lucia g. , springel v. , 2007 , mnras , 376 , 180 navarro j. f. , frenk c. s. , white s. d. m. , 1997 , apj , 490 , 493 paczynski b. , gorski k. , 1981 , apj , 248 , l101 perlmutter s. et al . , 1999 , apj , 517 , 565 riess a. g. et al . , 1998 , aj , 116 , 1009 riess a. g. et al . , 2004 ,
apj , 607 , 665 riess a. g. et al . , 2007 , apj , 659 , 98 sahni v. , starobinsky a. , 2006 , ijmpd , 15 , 2105 seljak u. et al . , 2005 , phys .
d , 71 , 103515 semboloni e. et al .
, 2006 , a&a , 452 , 51 sereno m. , 2002 , a&a , 393 , 757 smith g. p. , kneib j .-
, smail i. , mazzotta p. , ebeling h. , czoske o. , 2005 , mnras , 359 , 417 soucail g. , kneib j .-
, golse g. , 2004 , a&a , 417 , l33 spergel d. n. et al . , 2003 , apjs , 148 , 175 spergel d. n. et al . , 2007 , apjs , 170 , 337 tonry j. l. et al . , 2003 ,
apj , 594 , 1 van waerbeke l. et al . , 2000 ,
a&a , 358 , 30 villumsen j. v. , freudling w. , da costa l. n. , 1997 , apj , 481 , 578 wambsganss j. , bode p. , ostriker j. p. , 2005
, apj , 635 , l1 winn j. n. , rusin d. , kochanek c. s. , 2004 , nature , 427 , 613 williams r. e. et al .
, 1996 , aj , 112 , 1335 wittman d. m. , tyson j. a. , kirkman d. , dellantonio i. , bernstein g. , 2000 , nature , 405 , 143 yahata n. , lanzetta k. m. , chen h .- w .
, fernndez - soto a. , pascarelle s. m. , yahil a. , puetter r. c. , 2000 , apj , 538 , 493 | by stacking an ensemble of strong lensing clusters , we demonstrate the feasibility of placing constraints on the dark energy equation of state .
this is achieved by using multiple images of sources at two or more distinct redshift planes .
the sample of smooth clusters in our simulations is based on observations of massive clusters and the distribution of background galaxies is constructed using the hubble deep field .
our source distribution reproduces the observed redshift distribution of multiply imaged sources in abell 1689 .
the cosmology recovery depends on the number of image families with known spectroscopic redshifts and the number of stacked clusters .
our simulations suggest that constraints comparable to those derived from other competing established techniques on a constant dark energy equation of state can be obtained using 10 to 40 clusters with 5 or more families of multiple images .
we have also studied the observational errors in the image redshifts and positions .
we find that spectroscopic redshifts and high resolution _
hubble space telescope _
images are required to eliminate confidence contour relaxation relative to the ideal case in our simulations .
this suggests that the dark energy equation of state , and other cosmological parameters , can be constrained with existing _
hubble space telescope _
images of lensing clusters coupled with dedicated ground - based arc spectroscopy .
[ firstpage ] cosmological parameters gravitational lensing clusters |
in this paper we continue our study @xcite of the temporal aspects of tunneling considered in our approach as a combined stochastic process to consist from two alternative elementary ones , transmission and reflection , evolved coherently . in @xcite we have introduced the transmission and reflection times provided that a particle is in a localized state .
these time scales were extracted from the time dependence of the average positions of transmitted and reflected particles .
it is evident that this way of timing a particle is not applicable in the limiting case , when the particle is in some stationary state described by a non - normalized wave function .
this case is of great importance , since it is connected to the scattering of uniform beams of identically prepared particles , noninteracting with each other . to find a time spent by transmitted and reflected particles in the barrier region , in the stationary scattering problem , is the main aim of the present paper .
how to obtain the wave functions for transmission and reflection , for a given potential and initial wave function , is presented in @xcite . here
we expose both the functions in the form to be more suitable for our purpose .
let setting the scattering problem be such as in @xcite .
in particular , we will suppose that the rectangular potential barrier of height @xmath0 is localized in the interval @xmath1 $ ] .
then the wave functions for transmission , @xmath2 , and reflection , @xmath3 , can be written for @xmath4 as follows ( @xmath5 is the energy of a particle ) . for @xmath6 @xmath7 for @xmath8 @xmath9 for @xmath10 @xmath11 for @xmath12 @xmath13 where @xmath14 , @xmath15 , @xmath16,@xmath17 @xmath18 @xmath19 is the mass of a particle .
besides , @xmath20 @xmath21 @xmath22 @xmath23 @xmath24 @xmath25 in its turn , @xmath26 @xmath27 one can easily check that @xmath28 and @xmath29 , @xmath30 where @xmath31 and @xmath32 are the coefficients of transmission and reflection , respectively ( @xmath33 ) ; @xmath34 . thus , @xmath35 .
note , both the stationary wave functions are non - normalized and , hence , the standard timing procedure becomes usefulness in timing a particle with the well defined energy . at the same time the properties of these wave functions
provide another natural way for timing a particle in the barrier region .
indeed , the density of the probability flux , @xmath36 , for @xmath2 is everywhere constant and equals to @xmath37 .
this means that the velocity of this flux , @xmath38 , can be written as @xmath39 taking into account exps .
( [ 1 ] ) - ( [ 4 ] ) for @xmath2 , we see that outside the barrier region @xmath40 . in the barrier region ,
the velocity of the flux decreases exponentially when it approaches the midpoint @xmath41 .
one can easily show that @xmath42 , but @xmath43 .
thus , any selected infinitesimal element of the flux passes the barrier region for the time @xmath44 , where @xmath45 one can easily show that @xmath46.\end{aligned}\ ] ] then , taking into account exps .
( [ 1 ] ) - ( [ 4 ] ) ) , we obtain @xmath47 @xmath48 @xmath49=-\frac{1}{\kappa^2}(\kappa^2+k^2)t\sinh(2\varphi).\ ] ] eventually , we have @xmath50;\end{aligned}\ ] ] where @xmath51 similarly , for @xmath52 @xmath53;\end{aligned}\ ] ] where @xmath54 @xmath55 , if @xmath56 ; otherwise , @xmath57
as is seen , the above time scale for transmission ( [ 5 ] ) ( though the way of introducing is different ) coincides by form with the well - known bttiker s dwell time ( see @xcite ) . however , unlike the latter , exp .
( [ 5 ] ) describes only transmitted particles .
doing in the spirit of the bttiker s dwell - time concept , we introduce the dwell time for reflection , @xmath58 , as @xmath59 where @xmath60 is the incident probability flux ( note , the probability flux for @xmath61 itself is zero ) . considering the expressions for @xmath61
one can easily show that for @xmath62 @xmath63 for @xmath52 @xmath64
as was shown in @xcite , the phase - time and old dwell - time concepts , @xmath65 , do not imply the passage to the case of a free particle , that is , when @xmath66 . at the same time
, our concept of the asymptotic times for transmission and reflection ( see @xcite ) ensures this passage .
one can easily check that the above dwell time for transmission implies this passage , too .
note , both the characteristic times are nonnegative quantities .
let @xmath67 be the time spent by a free particle in the interval @xmath1 $ ] : @xmath68 .
then one can show that for barriers , in the limit @xmath69 ( @xmath70 ) , we have @xmath71;\end{aligned}\ ] ] for wells , @xmath72.\end{aligned}\ ] ] thus , not only for @xmath73 , but also for @xmath74 @xmath75 when @xmath76 remind ( see @xcite ) , the @xmath77 in the latter case .
remind , the bttiker s dwell time for @xmath62 read as ( see @xcite ) @xmath78 for @xmath52 @xmath79 it is useful to compare formally a new dwell - time scale ( to describe transmitted and reflected particles separately ) with bttiker s dwell time to describe all particles jointly .
figs.1 and 2 show the @xmath5-dependence of @xmath80 and @xmath81 for @xmath82 , @xmath83 and @xmath84 , @xmath85 , respectively . figs .
3 and 4 show the dependence of @xmath65 , @xmath44 and @xmath86 on the barrier s width @xmath87 for @xmath84 , @xmath88 and @xmath84 , @xmath89 , respectively . in all cases , @xmath90 where @xmath91 is the mass of an electron .
as is seen , @xmath44 and @xmath65 coincide at the points of resonance .
however , this result is expected .
besides , @xmath92 is more smooth function than @xmath93 moreover , except for some neighborhoods of resonant points , @xmath94 our results show that the most essential difference between these time scales takes place for @xmath62 ( see figs.2 and 4 ) . in particular , fig.4 shows that @xmath44 increases exponentially when the width of an opaque potential barrier increases .
at the same time @xmath95 ( as well as the well - known phase tunneling time and our asymptotic times for transmission and reflection ) tends , in this case , to some constant value . as regards @xmath58 ( see ( [ 30 ] ) , ( [ 31 ] ) ) , in this limit @xmath96 note also , @xmath97 for wells ( @xmath98 ) , when @xmath99 however , in the case of resonance , when @xmath100 @xmath101 at @xmath102
as is known , the just mentioned property of the phase and dwell tomes is excepted to lead to superluminal velocities of a particle in the barrier region ( the well - known hartman effect ) .
however , we see that the behavior of @xmath44 , in contrast to the bttiker s dwell time , denies the existence of the hartman effect for a particle with the well - defined energy .
however , from our study it follows that this effect does not appear also for a particle whose initial state is described by the wave packet of any width .
fig.5 shows the expectation value of the particle s position as a function of @xmath103 .
it was calculated for the transmitted wave packet providing that @xmath104 , @xmath105 , @xmath106 . at @xmath107 the ( combined )
state of the particle is described by the gaussian wave packet peaked around @xmath108 ; its half - width @xmath109 ; the average energy of the particle @xmath110 .
this figure displays explicitly the difference between the exact and asymptotic transmission times . while the former gives just the ( average )
time spent by a particle in the barrier region , the latter describes , in fact , the ( average ) lag or outstripping of a transmitted particle with respect to a free particle whose average velocity equals to the asymptotic average velocity of a scattered particle . in the case considered ( see fig.5 ) the exact transmission time equals approximately to @xmath111 , the asymptotic one is of @xmath112 , and @xmath113 . as is seen , the dwell time for transmission and exact transmission time , both evidence that , though the asymptotic transmission time is small for this opaque barrier , transmitted particles spend much time in the barrier region .
note , the possibility of accelerating a particle in the region of an opaque potential barrier , to superluminal velocities , is doubtful _ a priori_.
the point is that the average energy of a particle , @xmath114 , is constant for this static potential barrier .
thus , for the average kinetic ( @xmath115 ) and potential ( @xmath116 ) energies of a particle we have , for any @xmath103 , @xmath117 where @xmath118 is the average kinetic energy of transmitted particles at @xmath107 .
it is evident that @xmath119 if @xmath56 .
hence , at the stage of scattering , @xmath115 may only decrease .
of course , this fact does not at all mean that the average velocity of a particle , at the stage of scattering , can not be larger than its asymptotic value .
the point is that @xmath120 @xmath121 is the particle s momentum .
thus , @xmath122 may increase due to diminishing the wave - packet s width in the momentum space .
our calculations show that this takes place near the barrier region ( see fig.5 ) .
thus , when approaching the left boundary of the barrier , the particles is accelerated , on the average . as a result , it may overcome the potential barrier .
when the particle enters the barrier region , it almost stops .
the region of an opaque potential barrier serves as a storage of particles . behind the barrier
the average velocity of particles may first exceed its own asymptotic value .
however , far from the barrier , when the width of the wave packet , in the momentum space , backs to its initial ( asymptotic ) value , the same does the average velocity of particles .
note , the more is the part of @xmath123 in the average kinetic energy of a particle at @xmath107 , the larger may be the grows of the particle s velocity near the barrier region .
however , one has to bear in mind that this velocity can not essentially exceed the value , @xmath124 .
so that , as it follows from our approach , tunneling a particle through an opaque potential barrier must not be accompanied by the hartman effect .
so , in our approach , the state of the whole quantum ensemble of tunneling particles is considered as a coherent superposition of the states of the subensembles of transmitted and reflected particles . for a given potential and initial state of a particle
, there is an unique pair of solutions to the one - dimensional schrdinger equation , which describe both the subensembles at all stages of tunneling . for each of these subensembles
we have introduced three time scales .
two of them ( the exact and asymptotic tunneling times ) describe a completed scattering process .
the third time scale ( the dwell time ) relates to the stationary tunneling problem .
it should be stressed that the asymptotic ( transmission and reflection ) times differ from the well - known phase asymptotic times .
the dwell times for transmission and reflection differ from the bttiker s dwell time .
analysis of tunneling a particle through an opaque rectangular barrier , carried out in the framework of our approach , does not confirm the existence of the hartman effect . at the end we have to point to some shortcomings of the above time concepts .
so , the concept of the exact tunneling times does not give explicit expressions for these quantities ; besides , it is not suitable for timing a particle with the well - defined energy ( this becomes more evident in the case of the reflection of wide ( in @xmath125-space ) wave packets , when the expectation value of the particle s position simply does not `` enter '' the barrier region ) .
the main shortcoming of the dwell - time concept is that it is valid only for the stationary scattering process .
the our next purpose is to present the concept of the larmor times for transmission and reflection . by our opinion ,
this concept is of great importance in solving the tunneling time problem . 861 n. l. chuprikov , wave functions and characteristic times for transmission and reflection , quant - ph/0405028 .
n. l. chuprikov , a comparative analysis of tunneling time concepts : where do transmitted particles start from , on the average ? , quant - ph/0501067 .
m. buttiker , 1983 phys .
b * 27 * 6178
@xmath81 and @xmath80 versus @xmath5 , for @xmath83 and @xmath126 . fig .
5 . the @xmath103-dependence of the average position of transmitted particles ; the initial ( full ) state vector represents the gaussian wave packet peaked around the point @xmath108
, its half - width equals to @xmath130 the average kinetic particle s energy is @xmath131 @xmath104 , @xmath105 . | as was shown in quant - ph/0405028 , the state of a tunneling particle can be uniquely presented as a coherent superposition of two states to describe alternative sub - processes , transmission and reflection . in this paper , on the basis of the stationary wave functions for these sub - processes , we give new definitions of the dwell times for transmission and reflection . in the case of rectangular potential barriers the dwell times are obtained explicitly .
in contrast with the well - known bttiker s dwell - time , our dwell time for transmission increases exponentially , for the under - barrier tunneling , with increasing the barrier s width . by our approach the well - known hartman effect is rather an artifact resulted from an improper interpretation of the wave - packet tunneling and experimental data , but not a real physical effect accompanying the tunneling phenomenon . |
along with the successful run of lhc at 8tev , many researchers now are anxiously waiting judgment on whether the 125gev possible excess displayed by atlas and cms experiments is the standard model ( sm ) higgs or some other particle(s ) or just a statistical fluctuation .
if before the end of 2012 the excess is shown either not to be the sm higgs or just to disappear as more data accumulates , the sm higgs then will be simply ignored by the lhc experiments .
this will be a very cheering result , since it offers the strongest direct experiment proof that the electroweak symmetry breaking mechanism predicted by sm is not correct and new physics beyond sm is needed , despite any discovery of new physics yet .
particle physics will enter a golden age full of excitements with prospects from various investigation ventures and potential discoveries of predicted or unexpected new physics .
however , if by the end of this year , data from the atlas and cms experiments support the discovery of the sm higgs , sm then may become a perfect model not only theoretically but also experimentally . while celebrating the great success ,
one disadvantage that particle physicists avoid talking about , emerges more and more realistically , i.e. , no new physics appears . some have shown that , with the 125gev higgs , sm survives all the way to the planck scale@xcite .
particle physics in this scenario then has a bleak future without new phenomena emerging , apart from those predicted by sm .
the conventional basis of unclear electroweak symmetry breaking mechanism that new physics searchers relied on may no longer exist .
there are two different schemes to deal with this bad situation .
the first is to insist on searching new physics in electroweak symmetry breaking sector of sm and it is already taken by many researchers . considering that the final verification of sm higgs will need many years of further experimental effort , along with ever - increasing amounts of data from the experiment
, one can gradually narrow various higgs couplings to ordinary quarks , leptons and gauge bosons to their sm values to examine the possible left new physics spaces @xcite .
the second is that we propose and emphasize in this paper . assuming the 125gev access is exactly sm higgs , we search new physics evidence in alternative fields instead of continuously precision check of the higgs profile .
two apparent such facts are the existence of dark matter predicted from cosmology and massive neutrinos from their oscillation experiments . however
, dark matter maybe only participates in gravitational interactions and a simple generalization of sm can introduce neutrino masses .
these two evidences are not strong enough to ensure the existence of non - trivial new physics . to raise hopes for a future for those new physics believers
, we have to find some extra evidence or clues requiring new physics .
we know that sm fits experiments very well up to now ; based on present experimental data , we do not have strong evidence for new physics .
we need to lower expectations to investigate effects that are not so strong .
conventionally , the veracity of sm is tested by precision measurements in the electroweak sector .
a well - known scheme is the global electroweak sm fit in which a series of observables at the z - pole and off z - pole are chosen to fit experiment results with theoretical predictions by tuning the input parameters @xmath8 , @xmath9 , @xmath10 , and other quantities which are irrelevant for this work .
the complete global fit for z - pole observables in @xcite shows that the deviation of electroweak observables is smaller than 1@xmath11 .
the exceptions are following seven observables : @xmath1 with 2.7@xmath11 , @xmath2 with 1.8@xmath11 , @xmath12 with 1.8@xmath11 , @xmath13 with 1.5@xmath11 , @xmath14 with 1.5@xmath11 , @xmath15 with 1.4@xmath11 , and @xmath16 with 1.3@xmath11 . in latest improved results with gfitter @xcite , there are still seven observables with the deviation of more than 1@xmath11 , especially @xmath1 and @xmath17 with the deviation of more than 2@xmath11 .
since this is already the best fit of sm which leads unknown higgs mass @xmath18gev , any further variation of @xmath19 will worsen the result .
this can be further seen from @xmath19 dependence of separate fitting for above five observables .
gfitter shows that @xmath1 and @xmath20 prefer smaller while @xmath21 , @xmath22 and @xmath23 prefer larger @xmath19 .
so within sm itself , no matter what value of @xmath19 , it is impossible to further improve the present global electroweak sm fitting result . if we insist on reducing the deviation values for these observables , the only way is to add in new physics . in this sense ,
the moderately large deviations in the global electroweak sm fit are the evidences that we seek re - demonstrate the existence of new physics
. improving global electroweak sm fits can then provide a measure to judge the new physics models . in the literature
, we do not find many works treating the global electroweak sm fits in this way .
one possible reason is that the deviation of the fit is not so large , and researchers have turned their attention to other observables , such as the oblique parameters @xmath7 , @xmath24 and @xmath25 @xcite , especially @xmath7 which often gives very strong constraints on new physics models .
in fact , many years ago , when @xmath26 received a very large deviation @xcite , much work appeared trying to explain the result in terms of various new physics models .
along with improvements in experimental results , the deviation in @xmath26 was reduced over time ; researchers were then no longer interested in those kinds of discussions .
now , assuming that the sm higgs is confirmed , what we want to do is to re - initiate the investigation in this direction . with this application of global electroweak fitting to new physics models in mind
, we examine below the simplest abelian extension of sm that can be used as an example of the new physics beyond sm . in this scenario , the conventional sm gauge group @xmath27
is extended to @xmath28 .
the neutral heavy gauge boson associated with the extra broken @xmath29 group is usually called the @xmath3 boson , assumed here to be the lightest new physics particle beyond sm .
many new physics models have one such @xmath3 boson ( for details see ref.@xcite ) as a remnant of new physics interactions . for this paper , we take @xmath3 as a simplest case of new physics to study its effects on the global electroweak fit .
we find three references in the literature that investigated such effects of a @xmath3 particle @xcite . in ref.@xcite ,
the author performed a complete electroweak fits to z extensions of sm and pointed out that presence of @xmath3 may solve @xmath30 anomaly problem .
in ref.@xcite , it was shown that a @xmath3 based on gauged @xmath31 lepton number can improve the fit for the anomalous magnetic moment of the muon . in ref.@xcite ,
the author claimed that a @xmath3 , with suppressed couplings to the electron compared to the @xmath0 boson and couplings to the @xmath32 quark , provides an excellent fit to @xmath1 and @xmath2 and significant improvements in @xmath22 and @xmath26 at @xmath332gev scale . to make a relatively large contribution ,
the @xmath3 mass was fitted close to @xmath0 boson mass , otherwise the contribution from @xmath3 exchange vanishes unless @xmath3 is near the @xmath0-pole .
in ref.@xcite , only the simplest @xmath34 mixing was considered , while in the analysis of refs.@xcite , effect from @xmath35 mixing was ignored .
this is ok for the fittings for those observables out of z - pole , where mixing effect is small .
but for the z - pole observables we focused in this paper , fitting result is sensitive to the @xmath35 mixings . to avoid the unnatural @xmath3 mass near @xmath10 and include in important @xmath35 mixing effect
, we consider the most general model - independent @xmath3 boson described by an electroweak chiral lagrangian given in our previous paper @xcite .
in fact , from an effective field theory point of view , all coefficients appearing in the electroweak chiral lagrangian @xcite are free parameters which need to be fixed either from experiments or from underlying theoretical models .
our global fit can be seen as a kind of procedure to extract out experiment constraints on these coefficients .
in contrast to ref.@xcite , in this fit the @xmath3 mass needs not to close to z - pole due to some extra free parameters not fixed by the global fit .
direct exchange at the @xmath0-pole discussed in ref.@xcite then becomes enough small to be neglected in our work .
other @xmath3 bosons effects include modifications of @xmath0 couplings through @xmath35 mixings and @xmath3 couplings to sm fermions .
they are the main sources to cause the change . in next section
, we will discuss how to manipulate them and obtain our improved global electroweak fitting result .
this is the main part of the paper
. we will first discuss @xmath35 mixings and @xmath3 couplings to sm fermions separately , then describe our global fitting procedure .
finally give our fitting result and make some discussions .
for @xmath3 mixings with electroweak neutral bosons @xmath0 and @xmath4 , usually a lighter @xmath3 is possible at larger mixing angles ; smaller mixing angles arise only for heavier @xmath3 ( see ref.@xcite ) .
more detailed results depend on the form of the mixings set by the models . considering that the experimental constraint for the @xmath3 mass has already reached the tev energy region ,
we assume here a small mixing ( order of @xmath36 ) @xmath3 . based on a model independent chiral lagrangian , we have already classified and parameterized the most general @xmath35 mixings in ref.@xcite . to include in @xmath3 mixings effects ,
we define the rotation matrix between the gauge eigenstates of the neutral gauge bosons of @xmath37 and the corresponding mass eigenstates as follows : @xmath38 with weinberg angle @xmath39 and @xmath40 .
after rotation by @xmath25 , the neutral gauge fields will be diagonalized into their mass eigenstates . for the @xmath3 coupling to sm fermions , we need to determine the @xmath29 charge assignments of the sm fermions .
these are related to following neutral current interactions @xmath41 where @xmath42,@xmath43 and @xmath44 are the respective @xmath45 , @xmath46 and @xmath29 gauge coupling constants and @xmath47f_i \nonumber\\ j^\mu_x&=&\sum_i\bar{f}_i\gamma^\mu[y'_{il}p_l+y'_{ir}p_r]f_i\label{jxdef}\end{aligned}\ ] ] are neutral currents corresponding respectively to the third component gauge boson @xmath48 of weak isospin , hypercharge gauge boson @xmath49 and an extra @xmath29 gauge boson @xmath50 . in addition , @xmath51 are the left / right - handed @xmath29 charges for the sm fermion field @xmath52 with flavor index @xmath53 ; the electroweak gauge symmetry @xmath45 constrains these to eighteen charges in the most general case : three left - handed quark charges for each of the three generations @xmath54 , @xmath55 , @xmath56 ; six right - handed quark charges for each quark @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath61 , @xmath62 ; three left - handed lepton charges for each generation @xmath63 , @xmath64 , @xmath65 ; and six right - handed lepton charges for each leptons @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 . in the mass eigenstates basis ,
the neutral currents become @xmath72 with @xmath73 being the renormalized electric charge incorporating @xmath3 contributions and @xmath74 . for the rotation matrix ( [ umatrix ] ) , the neutral currents in the mass eigenstates basis are related to those in the interaction eigenstates basis as : @xmath75 to keep the gauge symmetry , the assignments of the charges must satisfy the anomaly cancelation conditions .
a generation independent solution has been discussed in our previous paper @xcite , in which only two charges are free due to cancelations of @xmath76 ^ 2u(1)'$ ] , @xmath77 ^ 2u(1)'$ ] , @xmath78 ^ 2 $ ] , @xmath79 ^ 2u(1)'$ ] , @xmath80 ^ 3 $ ] and the mixing gravitational - gauge anomalies . to obtain the most general fitting result , we focused on making general generation dependent charge assignments .
considering that some charges are very small in the fitting result , we further discuss a case in which these small charges vanish . in all , two cases arise : 1 .
the anomaly cancels only when all three generations are included .
we find that there are twelve free charges .
2 . within case 1
, we further stipulate that the charges for the second - generation quarks , the first - generation left - handed lepton , and right - handed electron vanish , i.e. @xmath81 . six free charges exist .
to simplify the fitting procedure and give prominence to our physical result , we performed a global electroweak fit by fixing the sm contribution with its predictions given by pdg2010 @xcite , and only investigated the possible @xmath3 tree order contributions . a more subtle alternative fit without fixing sm contributions , which will need more complex sm computations , will be discussed elsewhere .
to realize this simplified global electroweak fit , we took three precisely measured observables , @xmath8 , @xmath9 and @xmath10 , as input parameters with which to express the three sm three parameters : @xmath42 , @xmath43 and electroweak vacuum expectation value @xmath82 in the sm lagrangian .
other sm parameters , such as fermion masses @xmath83 , higgs mass @xmath84 , and strong coupling constant @xmath85 , are irrelevant in the fit to fix sm contributions . at tree level , we fix @xmath9 .
the fine structure constant @xmath86 and z boson mass @xmath10 are corrected by modifications including the respective contributions to @xmath87 and @xmath88 @xmath89 with vacuum expectation value @xmath90 and @xmath0 mass @xmath91 in sm .
an electroweak observable @xmath92 , depending on @xmath9 , @xmath93 and @xmath94 , can be divided into two parts : one is @xmath95 coming from sm fitting values in @xcite and depending on @xmath9 and sm @xmath8 and @xmath10 , and the other is @xmath96 coming from a @xmath3 correction depending on mixing parameters @xmath97 introduced in ( [ umatrix ] ) , sm fermion @xmath29 charges @xmath98 and constrained by the anomaly cancelation condition , i.e. , @xmath99 note , some of @xmath97 such as @xmath100 , @xmath101 , @xmath102 and @xmath103 do not enter into the formulae to those z - pole electroweak observables .
therefore these @xmath97s are irrelevant to our global fit and we will not discuss their values in this paper . the difference between the present experimental data and sm fitting results will provide a narrow space for the @xmath3 correction @xmath104 .
we solve for the minimum of @xmath105 function @xmath106 by tuning @xmath97 and @xmath98 to obtain our global electroweak fitting results , the smaller the @xmath105 , the better the results .
here , @xmath107 and @xmath108 are experimental values and errors , respectively .
we consider all @xmath0-pole observables listed in @xcite and @xcite ; their definitions and expressions at @xmath0-pole appear in the literature @xcite .
the fitting result on the mixing parameters and sm fermion @xmath29 charges is given in table.[tab : parameters ] .
.fitting result on the mixing parameters @xmath97 and sm fermion @xmath29 charges in @xmath3 effective theory . [ cols="^,^,^",options="header " , ] the value of sm @xmath105 is @xmath109 .
@xmath110 is not decided by fit independently .
@xmath111 is the coefficient of @xmath112)^2 $ ] in chiral effective theory given in ref.@xcite .
it comes from the relation @xmath113 to keep photon massless .
the global fit result is given in fig.[fig.pull ] .
pull ] we see that the best fit is case 1 , for which the largest two deviations in sm fit , @xmath1 and @xmath2 , are obviously reduced .
the largest two deviations in case 1 is 1.2@xmath11 for @xmath2 and 1.1@xmath11 for @xmath114 , and deviations for other observables are below @xmath115 .
this implies that the global fit of the electroweak observables does support case 1 .
the @xmath3 mass can not be fixed in our fit due to following relation @xmath116\end{aligned}\ ] ] with @xmath117gev and the coefficient @xmath118 of the term @xmath119)^2 $ ] in chiral effective theory @xcite . due to the undetermined parameters @xmath44 , @xmath101 , @xmath103 , and @xmath111
, @xmath120 keeps free in our result .
note heavy @xmath3 with @xmath121 leads large negative @xmath111 for small couplings @xmath44 and small mixing parameters @xmath101 and @xmath103 .
considering that case 1 predicts that some charges take very small values , we annul these charges by further assuming in case 2 null charges for the second - generation quarks , the first - generation left - handed lepton , and right - handed electron . for case 2 ,
the largest deviation is still @xmath6 for @xmath114 with @xmath105 value of @xmath122 and we find relatively large couplings to the third generation which is matched with result of ref.@xcite .
of course , one can continuing annulling more charges , but we find then either the absolute value of the largest deviation is larger than @xmath123half of the largest deviation for sm , or we may have @xmath3 mixings of order @xmath124 .
we drop these cases anymore , because in further loosening the deviation value , the largest absolute value becomes @xmath5 signifying that all charges vanish and we have just reverted to sm .
large @xmath3 mixings may yield light @xmath3 @xcite or contradict with electroweak precision data .
it should be noted that the mixing parameters @xmath97 used in this paper are not and usually are larger than the conventional mixing angle used in literature .
the conventional @xmath34 mixing angle @xcite is defined @xmath34 mixing angle as @xmath125m_z^2}{m_{z'}^2-m_z^2}.\label{theta'}\end{aligned}\ ] ] here , @xmath126 is sm @xmath0 mass and @xmath127 stands for mixing correction to @xmath0 mass . in our fit , @xmath127 is the order of less than @xmath128 ( corresponding to @xmath0 mass pull less then @xmath129 ) .
then the order of @xmath130 is @xmath131 , @xmath132 and @xmath36 for @xmath133 10tev , 1tev and @xmath134gev respectively . and
in case 1 and 2 the numerator of ( [ theta ] ) @xmath135 $ ] always is @xmath136 , which just matches our fitting results given in table.[tab : parameters ] . except the magnitudes for the mixing parameters @xmath97 , in case 2 , there appear some null charges which implies that @xmath3 does not couple to the corresponding fermions .
the implications of these decouplings on the new physics model building need further investigations . with the above global fitting results
, we can further check effects on the oblique parameter @xmath7 .
the experiments prefer small negative values , whereas many unsuccessful new physics models contribute it positive corrections .
thus we have a test for our global fit . the @xmath3 correction contribution to @xmath7
is @xmath137\label{s}\end{aligned}\ ] ] for which we find a contribution of @xmath138 for case 1 and @xmath139 for case 2 .
i.e. , our fitting result does push @xmath7 parameter to its experiment preferred negative region .
to summarize , we pointed out and emphasized that even if lhc experiments would confirm sm higgs , the global electroweak fit always implies the existence of new physics . mixing of a hypothetical @xmath3 with @xmath0 and @xmath4 at order of @xmath36 and couples to sm fermions with suitable anomaly free charge assignments will really improve global fitting results .
it can change @xmath105 substantially from @xmath109 of sm to @xmath140 , reduce the largest deviation from @xmath5 in sm to @xmath6 in our scenario , suppress all observables below @xmath115 except @xmath2 and @xmath114 , and make @xmath3 extra contributions to the oblique parameter @xmath7 to @xmath138 .
global electroweak fits do not prefer @xmath3 couplings to the second - generation quarks , the first - generation left - handed lepton , and right - handed electron .
our global fit is done for z - pole observables which is sensitive to @xmath35 mixing and leads typical correction at order of @xmath36 .
this is only the part of the most general global fitting program which includes the fits for both z - pole and off z - pole observables , but any of new physics models must at least match z - pole observables .
usually the new particles beyond sm are heavier than @xmath10 , which implies that their exchange effects at z - pole are small .
to more detail analyze , the correction is at most at order of @xmath141 , with typical new physics particle mass @xmath142 at order of 1tev .
this is one order of magnitude smaller than the correction from our @xmath35 mixing .
therefore if there are not enough big mixings among new physics particles and sm electroweak gauge particles , we can not expect generally sufficient large improvements on global electroweak fit at z - pole from exchange of new physics particles . among various new physics particles ,
the most important particle which mixes with the sm electroweak gauge particle is @xmath3 .
the other possible charged vector particles @xmath143 are less important due to their relative simple mixings with @xmath144 , ref.@xcite gives the simliar result .
further the loop correction from these new physics particles usually suppressed by loop factor @xmath145 which may be at the same order of our @xmath35 mixing depending on the detail value of coupling of the loop . for the global fit coming from the loop corrections of new physics particles and from those observables off the z - pole scale , we will leave the our corresponding discussions in future investigations .
this work was supported by national science foundation of china ( nsfc ) under grant no.11147192 , 11075085 and 11005084 , specialized research fund for the doctoral program of high education of china no.20110002110010 , fundamental research funds for the central universities , and scientific research foundation of guangxi university grant no .
xbz100686 .
1 m. shaposhnikov , arxiv:0708.3550 . c. englert , t. plehn , m. rauch , d. zerwas , p.m. zerwas , phys .
b*707 * , 512 ( 2012 ) ; m. rauch , arxiv:1205.2506 ; m. klute , r. lafaye , t. plehn .
m. rauch , d. zerwas , arxiv:1205.2699 .
k. nakamura et al .
( particle data group ) , j. phys .
g*37 * , 075021 ( 2010 ) .
m. baak , m. goebel , j. haller , a. hcher , d. ludwig , k. mnig , m. schott , j. stelzer , arxiv:1107.0975 . j. steinberger , cern - open-97 - 013 .
p. langacker , rev .
phys . * 81 * , 1199 ( 2008 ) .
p. langacker , g. paz , l .- t .
wang , and i. yavin , phys .
lett . * 100 * , 041802 ( 2008 ) .
f.del aguila , j.de blas and m.perez-victoria , jhep * 09 * , 033 ( 2010 ) j.heeck and w.rodejohann , arxiv:1107.5238[hep - ph ] r. dermisek , s - g .
kim , and a. raval , arxiv:1201.0315[hep - ph ] .
y. zhang , s .- z .
wang , q. wang , jhep * 03 * , 047 ( 2008 ) .
hewett , t.g.rizzo , phys
* 183 * , 193 ( 1989 ) .
y. zhang , q. wang , jhep * 07 * , 012 ( 2009 ) .
y. zhang , qing wang , chin .
c*36 * , 298 ( 2012 ) .
the lep electroweak working group , phys .
rep . * 427 * , 257 ( 2006 ) . | among the @xmath0-pole observables , @xmath1 and @xmath2 suffer moderately - large standard deviations from the standard model predictions .
fine - tuning of the unknown higgs mass only reduces the deviation of one of them at the expense of increasing the deviation of the other observable .
if we take this fact seriously , the result can be interpreted as independent experimental evidence of existing new physics beyond sm , even if a 125gev higgs on the lhc is finally confirmed to be sm higgs .
we show in this paper that the existence of a @xmath3 boson mixing with @xmath0 and @xmath4 and with generation - dependent anomaly - free charge assignments , helps to suppress @xmath1 and @xmath2 at @xmath0-pole simultaneously and dose reduce the largest deviation from @xmath5 in sm predictions to @xmath6 in our scenario .
the global electroweak fit does not prefer @xmath3 coupling to the second - generation quarks and the first - generation left - handed lepton and right - handed electron .
the fitting result also supports a negative contribution to the @xmath7 parameter from the @xmath3 boson .
pacs numbers : 12.60.cn ; 14.70.pw ; 12.15 lk key words : global fit ; @xmath3 ; anomaly free ; chiral effective theory |
one of the challenges of strong interaction physics nowadays is to understand the phase diagram of quantum chomodynamics ( qcd ) .
particular interest is devoted to the high density and low temperature region because of possible applications for the physics of compact stars .
these investigations became very exciting since it has been reported that the energy gaps of color superconducting quark matter could be as big as @xmath4 mev @xcite and thus being of the same order as the constituent quark masses dynamically generated in the chiral symmetry breaking transition .
gaps this large could not only have a direct influence on the quark matter equation of state ( eos ) but , due to the bcs relation @xmath5 critical temperatures would imply that quark matter if formed early in the protoneutron star evolution should be color superconducting .
dense and cold matter has been broadly studied for the flavor symmetric case ( for reviews , see @xcite ) . from first principles ,
the limit of asymptotic densities can be studied and it was found that the color - flavor - locking ( cfl ) phase , in which quarks of all three colors and flavors participate in the condensation , is energetically favorable ( see @xcite ) . using the simple nambu - jona - lasinio model @xcite , it was found that for intermediate densities the 2sc phase was favored @xcite .
therefore , a typical phase diagram calculated for the flavor symmetric case shows the low temperature and high density domain being in one of the color superconducting quark matter phases .
however , for compact star applications the situation is much more complex : equilibrium w.r.t .
weak interactions ( @xmath6-equilibrium ) , color and charge neutrality impose additional constraints to dense matter .
alford and rajagopal pointed out the difficulty of the 2sc phase to achieve charge neutrality and therefore they claimed that this phase would not be realized in compact stars @xcite . without solving the gap equation , they treated the strange quark mass as a parameter in their calculations and concluded that the cfl phase would be favorable against 2sc .
more recently , steiner , reddy and prakash @xcite found within an njl model calculation that the 2sc phase may be realized only in a very small region of the phase diagram at lower densities which eventually would be covered by the hadronic phase if it was properly take into account .
again , the 2sc phase seemed not to be reliable , basically because the charge neutrality condition imposes a strong constraint to the quark chemical potentials .
the discussion of the existence of a mixed phase of 2sc quark matter and normal quark matter ( nq-2sc ) by increasing the asymmetry between the up and down quarks number densities was first performed by bedaque @xcite . within the njl model ,
a detailed study of possible neutral mixed phases in @xmath6-equilibrium has been performed by neumann , buballa and oertel @xcite considering global neutrality between the phases . in accordance with the earlier expectations
, these authors showed that at very low densities - where the model is probably to fail - the normal quark matter component is dominant , at intermediate ones the 2sc phase and at still higher densities , the cfl phase .
moreover , the latter phase and the strange quarks is supossed to appear at very high densities ( nearly above @xmath7 mev @xcite ) that barely reached in the very inner core of compact stars and thus leading to no observable effects .
unfortunately , in the intermediate density region we have no rigorous calculations and we have to rely on phenomenological models suitable , however , for the investigation of qualitative features of the phase diagram . the above mentioned and widely used njl has the great advantage that chiral symmetry breaking and diquark condensation are treated at the same mean - field level as phase transitions of an interacting field theory . besides well - known limitations of the njl - type models ( no confinement , non - renormalizable ) there is some arbitrariness in the choice of the model parameters even if the strategy is followed that vacuum properties of low - lying hadrons should be reproduced ( e.g. the pion mass , pion decay constant and the chiral condensate ) , see e.g. @xcite for reviews . in this work
we use a chiral quark model that can represent the nonlocality of the quark interactions in a more realistic way via formfactor functions . furthermore , we vary the softness of these functions and analyze its consequences in the phase structure .
we investigate the parameter regime and obtain that no pure 2sc itself does occur under compact star constraints for low temperatures ( below @xmath8 mev ) .
we also study the relevant region of the phase diagram for compact star applications and obtain that neither gapless 2sc ( g2sc @xcite ) exists ; its occurence is limited to a narrow window for higher temperatures ( above 40 mev ) before the second order phase transition to the deconfined phase takes place . for rather strong coupling constants in the diquark channel ( @xmath9 0.86 )
a mixed phase nq-2sc is likely to occur in the interior of compact stars .
for particular values of @xmath0 the corresponding equation of state has been already succesfully used for hybrid star configurations , explaining the mass - radius relation of very compact objects like rx j185635 - 3754 @xcite or a huge release of energy in the evolution of a protoneutron star with trapped antineutrinos @xcite . nevertheless , for weak and intermediate coupling constants ( even for @xmath10 , the value suggested by the fierz transformation of a massive gluon exchange or two - flavor instanton induced interaction @xcite ) , the 2sc phase is unlike to occur .
if this is the case , this result could have important consequences for the neutron star phenomenology : if 2sc does not occur inside compact stars , then other patterns of diquark pairing satisfying color and charge neutrality can become energetically more favorable than normal quark matter .
examples are spin-1 condensates , e.g. color - spin - locking ( csl ) phases @xcite the influence of the small gaps ( @xmath11 kev ) that are expected for these phases on compact star cooling has recently been investigated and they could realize the appropriate cooling phenomenology @xcite .
we use a nonlocal chiral model for quark matter in the 2sc phase to study dense neutral matter in compact stars following @xcite .
formfactor functions @xmath12 model the nonlocality of the quark interaction in the momentum space .
we assume that this four fermion interaction is instantaneous and therefore the formfactors depend only on the modulus of the three - momentum @xmath13 . @xmath14 and @xmath15 are the coupling constants in the scalar meson and the diquark channels , respectively .
while @xmath14 is fixed together with the current quark mass and range parameters of the formfactors ( see below ) from hadron observables , @xmath15 is a free parameter of the approach which we vary using the parameter @xmath16 . in the mean field approximation
the grand canonical thermodynamic potential is a function of the temperature @xmath17 and of the chemical potentials @xmath18 for the quark with flavor @xmath19 ( @xmath20 ) and color @xmath21 ( @xmath22 ) given by @xmath23 where the effective potential @xmath24 is a function of the order parameters of the theory : the mass gap @xmath25 and the diquark gap @xmath26 .
the constant @xmath27 is chosen such that the pressure of the physical vacuum vanishes .
the matsubara frequencies for fermions are given by @xmath28 and the nambu - gorkov inverse quark propagator can be obtained as @xmath29 where @xmath30 is the chemical potential matrix and the elements of @xmath31 are the dynamical masses of the quarks given by @xmath32 . here , @xmath33 is the usual matrix in the dirac space and @xmath34 and @xmath35 are antisymmetric tensors in the flavor and color spaces respectively .
the nonlocality of the interaction between the quarks in both channels quark - antiquark ( @xmath36 ) and quark - quark ( @xmath37 ) is implemented via the same formfactor functions @xmath12 . in our calculations
we use the gaussian ( g ) , lorentzian ( l ) and cutoff ( njl ) formfactors defined as @xmath38^{-1},\\ \label{nf } g_{\rm njl}(p ) & = & \theta(1 - p/\lambda_{\rm njl})~.\end{aligned}\ ] ] the parameter sets ( quark mass @xmath39 , coupling constant @xmath14 , interaction range @xmath40 ) for the above formfactor models ( see tab .
[ par ] ) are fixed by the pion mass @xmath41 mev , pion decay constant @xmath42 mev and the mass gap @xmath43 mev at @xmath44 @xcite . in order to describe asymmetric two flavor quark matter @xcite in a neutral phase @xcite we introduce the baryon chemical potential @xmath45 , the chemical potential for the isospin asymmetry @xmath46 and the chemical potential for the color asymmetry @xmath47 .
these chemical potentials are conjugate to the conserved quantities in dense matter in neutron stars : baryon number density , electric and color charge density , respectively ( see next subsection ) .
we express @xmath48 in terms of @xmath49 : @xmath50 for the 2sc pairing pattern we assume that @xmath51 in which the red ( @xmath52 ) and the green ( @xmath53 ) quarks are paired .
the dispersion relation of the unpaired color ( blue ) is @xmath54 . as has been shown in @xcite for the 2sc phase the relation @xmath55 holds and therefore @xmath56 so that the quark thermodynamic potential can be written more explicitly as @xmath57 + \nonumber\\ & & \omega\left[\epsilon_b\left(e(p)+\frac{1}{3}(\mu_b+\mu_i)-\frac{2}{3}\mu_8\right)-\mu_i , t\right]+ \nonumber\\ & & \omega\left[\epsilon_b\left(e(p)-\frac{1}{3}(\mu_b+\mu_i)+\frac{2}{3}\mu_8\right)+\mu_i , t\right]+ \nonumber\\ & & \omega\left[\epsilon_b\left(e(p)+\frac{1}{3}(\mu_b+\mu_i)-\frac{2}{3}\mu_8\right)+\mu_i , t\right ] \ }
\nonumber\\ & & -\frac{2}{\pi^2}\int^\infty_0dpp^2\ { \omega\left[\epsilon_r\left(e(p)-\frac{1}{3}(\mu_b+\mu_i)-\frac{1}{3}\mu_8\right)-\mu_i , t\right]+ \nonumber\\ & & \omega\left[\epsilon_r\left(e(p)+\frac{1}{3}(\mu_b+\mu_i)+\frac{1}{3}\mu_8\right)-\mu_i , t\right]+ \nonumber\\ & & \omega\left[\epsilon_r\left(e(p)-\frac{1}{3}(\mu_b+\mu_i)-\frac{1}{3}\mu_8\right)+\mu_i , t\right]+ \nonumber\\ & & \omega\left[\epsilon_r\left(e(p)+\frac{1}{3}(\mu_b+\mu_i)+\frac{1}{3}\mu_8\right)+\mu_i , t\right ] \}~ , \label{ome9}\end{aligned}\ ] ] where following the notation in @xcite @xmath58= t\ln\left [ 1+\exp\left(-\frac{{\rm x}}{t}\right)\right]+\frac{{\rm x}}{2}~.\end{aligned}\ ] ] the factor @xmath59 in the last integral in ( [ ome9 ] ) comes from the degeneracy of the red and green colors ( @xmath60 ) .
the total thermodynamic potential @xmath61 contains the quark contribution @xmath62 and the contribution @xmath63 of the leptons @xmath64 that are treated as a massless , ideal fermi gas : @xmath65 the conditions for the local extremum of @xmath62 correspond to coupled gap equations for the two order parameters @xmath66 and @xmath26 , @xmath67 .
the global minimum of @xmath62 represents the state of thermodynamic equilibrium from which the equation of state can be obtained by derivation . the stellar matter in the quark core of compact stars
consists of @xmath68 quarks and leptons @xmath69 under the conditions of * @xmath6-equilibrium : @xmath70 which in terms of chemical potentials reads ( @xmath71 , @xmath72 ) @xmath73 * charge neutrality : @xmath74 which could also be written as @xmath75 and * color neutrality : @xmath76 here the number densities @xmath77 are defined in relation to the corresponding chemical potentials @xmath78 as @xmath79 where the index @xmath80 denotes the particle species and @xmath81 and @xmath82 .
the solution of the color neutrality condition shows that @xmath47 is about 5 - 7 mev in the region of relevant densities ( @xmath83 mev ) .
since the isospin asymmetry is independent of @xmath47 we consider @xmath84 in our following calculations .
[ gequations ] we solve the gap equations for stellar matter under the constraints of the previous subsection .
the results are plotted in the fig .
[ gequations ] for three different strengths @xmath0 of the coupling constant in the diquark channel at zero temperature . for the commonly used value of @xmath85 the solutions with lorentzian and njl formfactors show a normal quark matter phase at intermediate densities and a superconducting phase shifted to higher densities .
the solution with gaussian formfactor exhibits no diquark condensation at all . for @xmath86 superconductivity appears immediately at the chiral restoration with large gaps for the three formfactors .
however , in this case a mixed phase construction is necessary to neutralize the matter ( see next section and fig . [ q_branches ] ) . for an even stronger diquark coupling constant , @xmath87 , a pure superconducting phase with large diquark gaps occurs and the threshold for the chiral restoration transition
is shifted to lower densities than for @xmath10 and @xmath88 for the three formfactors .
[ q_branches ] to achieve the charge neutrality condition we apply a mixed phase construction between the subphase with diquark condensation and the subphase of normal quark matter using the gibbs conditions for phase equilibrium .
the temperature @xmath17 , the chemical potentials @xmath89 and the pressure @xmath90 should be equal for the subphases involved , in particular , @xmath91 where the subscripts denote the normal ( @xmath92 ) and the superconducting ( @xmath93 ) subphases . in the 2sc mixed phase we construct , the charge neutrality is satisfied as a global constraint : the positively charge superconducting phase coexists with the negatively charged normal quark matter phase in an homogeneous mixture .
surface tension and coulomb effects are calculated to be small enough not to affect the results qualitatively ( up to a value of @xmath94 mev/@xmath95 on the pressure @xcite ) and we disregard them in this work .
the charges of the subphases ( states corresponding to the local minima in the thermodynamic potential , i.e. solutions of the gap equations ) are shown in fig . [ q_branches ] for selected values of @xmath45 as a function of @xmath46 for the gaussian formfactor at zero temperature .
the upper ( lower ) branch corresponds to the subphase with ( without ) diquark condensation , the jump represents the transition between the subphases where the end points are states with the same pressure . for a given @xmath45
, we can see that increasing the asymmetry @xmath96 in the system ( the mismatch between the up and down quark fermi seas ) the diquark condensation is disfavored .
when comparing the charges for different @xmath0 , we found that for @xmath97 ( three lines with jumps that intersect the @xmath98 line ) , the superconducting branches are positively charged and the normal ones are negatively charged . at @xmath98
we obtain the coexistence of the subphases at the same pressure and the occurrence of a mixed phase . for @xmath10 , ( at fixed @xmath99 mev ) the lower
branch crosses the @xmath98 line and neutral quark matter is just normal . at the same @xmath45 but in case of the very strong coupling constant @xmath100 we obtain charge neutrality for a pure superconducting phase .
[ phase_diag_mui ] to explore the relation between asymmetry and @xmath6-equilibrium in neutral matter we plot in fig .
[ phase_diag_mui ] the solution of the gap equations under the condition @xmath98 in the @xmath46 vs. @xmath101 plane for the gaussian formfactor at @xmath102 .
the results are shown for different values of @xmath0 and the lines for the @xmath6-equilibrium ( [ b_equil ] ) correspond to different values of the chemical potential for the neutrinos and antineutrinos .
for a fixed @xmath0 , e.g. @xmath97 , two branches are shown : states in a phase with diquark condensation on the right ( @xmath103 , for large @xmath101 ) and states in a normal quark matter phase ( @xmath92 , for large negative @xmath101 ) on the left .
the plateau in between them shows that the system undergoes a phase transition from the superconducting to the normal phase ( and the two subphases coexist in a mixed phase ) when the asymmetry increases or the number of electrons decreases .
the stellar matter in compact objects should satisfy both conditions which are simultaneously fulfilled at the intersection of the corresponding lines .
therefore , for the situation in which there are no trapped leptons in the star ( @xmath104 ) , a mixed phase is preferable for @xmath105 for the gaussian formfactor .
for lower values of the coupling constant diquark condensation is not possible
. a similar analysis shows , for the two other formfactors , that the values of @xmath0 at which superconductivity is possible are smaller but its appearance is shifted to higher densities , see fig .
[ chi ] . the volume fraction that is occupied by the subphase with diquark condensation
is defined as @xmath106 and is plotted in fig .
[ chi ] as a function of @xmath45 at @xmath102 . for a strong coupling like @xmath100
, we obtain that @xmath107 once quark matter appears and therefore a pure superconducting phase for the three formfactors is realized . for intermediate strength of @xmath97 ,
the fraction of the superconducting subphase is between 0.6 and 0.9 depending on the formfactor and on the density . for @xmath10
, we obtain that no diquark condensation is possible for the gaussian formfactor and for the lorentzian and njl formfactors the percentage of superconducting matter is less than @xmath108 , whereby the onset of color superconductivity is shifted to higher baryon chemical potentials and thus possibly becomes irrelevant for compact stars . in fig .
[ chi_lepton ] the contribution of the leptons to the volume fraction is shown .
while muons are negatively charged and act in favor of the charge neutrality - increasing the fraction of diquark condensation - trapped antineutrinos increment the asymmetry in the system and produce the opposite effect .
we explore the relevant region of the phase diagram of qcd for the gaussian formfactor of our model . for strong coupling constant @xmath97 ,
a rich structure of color superconducting quark matter phases is found above the first order chiral transition ( solid line on the left ) as it is shown in fig .
[ phasediag ] . for the quark matter sector
we obtain that at low temperatures ( up to @xmath109 mev depending on @xmath45 ) the mixed phase nq-2sc is energeticaly favored with increasing volume fraction @xmath110 of the 2sc phase as the temperature increases ( solid lines caracterize equal values of @xmath110 indicated over the lines ) . at intermediate temperatures a large region of pure 2sc phase is obtained . at higher temperatures a narrow window of g2sc
occurs before the second order phase transition to a deconfined phase takes place .
nevertheless , for weak and intermediate coupling constants ( e.g. for @xmath10 mev ) this structure of superconducting phases disappears and the normal quark matter phase is preferable . for higher values of the chemical potential , @xmath111 mev , we expect the strange quark to appear and cfl phase to dominate ( see @xcite for a recent three flavor phase diagram with a similar parametrization within an njl model ) .
the number density @xmath77 of the particle species @xmath80 in the mixed phase are given by @xmath112 and shown as functions of @xmath45 at zero temperature in fig .
[ numbers ] . from these graphs
we can see that increasing @xmath0 the number of electrons in the system increases by order of magnitudes and the onset of the phase transition from the vacuum is shifted to lower densities .
the energy density @xmath113 in the mixed phase is given in a similar form as @xmath114 we evaluate the quark matter eos at @xmath102 within this nonlocal chiral model and show the results in fig .
[ eos ] . [ numbers ] [ eos ] as in @xcite we can display the results for the pressure in the form of a bag model @xmath115 where @xmath116 is the ideal gas pressure of quarks and @xmath117 a _ density dependent _ bag pressure , see fig .
[ bag ] . for @xmath10 ,
@xmath118 mev/@xmath95 is nearly a constant function of the density ; for @xmath97 we found a large pressure effect of the diquark condensate and therefore a lower and strongly density dependent @xmath119 .
finally , for @xmath100 the pressure of the condensate is so huge that the bag pressure reaches negatives values .
[ bag ] .parameter sets ( @xmath40 , @xmath120 , @xmath121 ) of the nonlocal chiral quark model for the different formfactors . [ cols="^,^,^,^",options="header " , ] [ par ]
within our study of the equation of state and phase structure for two - flavor quark matter at low temperature under compact star constraints we found that the occurence of a 2sc phase is sensitive to variations of both the formfactor of the nonlocal quark interaction and the strength of the diquark coupling constant .
our results suggest that for standard values of the coupling @xmath1 either the 2sc phase does not occur ( gaussian formfactor ) or it exists only in a mixed phase with normal quark matter with a volume fraction less than @xmath2 , occuring at high baryon chemical potentials @xmath3 mev and most likely not relevant for compact stars .
we obtain that neither the g2sc phase occurs in compact stars ; it appears in a narrow window at higher temperatures and for strong coupling constants ( e.g. @xmath97 ) .
this result suggests that other pairing patterns like the csl phase with relatively small gaps of the order of @xmath11 kev may be realized .
this hypothesis is in accord with a recent description of modern cooling data using hybrid star models .
we thank our colleagues for their discussions and interest related to our work , in particular j. berdermann , m. buballa , m. huang , i. shovkovy , n.n .
scoccola and d.n .
the research of d.n.a . has been supported by landesgraduiertenfrderung ( mv - germany ) , h.g . was supported by the virtual institute of the helmholtz association under grant no .
vh - vi-041 and by the daad partnership programme between the universities of yerevan and rostock .
d.n.a . and d.b .
acknowledge the hospitality of tandar laboratory , cnea , buenos aires during the final stages of this work and the daad - antorchas programme for financial support of their visit under grant no .
d/04/27956 .
k. rajagopal and f. wilczek , `` the condensed matter physics of qcd , '' arxiv : hep - ph/0011333 . m. g. alford , ann .
nucl . part .
* 51 * ( 2001 ) 131 .
t. schfer and f. wilczek , phys .
d * 60 * ( 1999 ) 114033 .
s. reddy and g. rupak , phys .
c * 71 * ( 2005 ) 025201 .
h. grigorian , d. blaschke and d. n. aguilera , phys .
c * 69 * ( 2004 ) 065802 .
d. n. aguilera , d. blaschke and h. grigorian , astron .
astrophys . * 416 * ( 2004 ) 991 .
t. schfer , phys .
d * 62 * ( 2000 ) 094007 .
h. grigorian , d. blaschke and d. voskresensky , phy . rev .
* c71 * ( 2005 ) , in press , arxiv : astro - ph/0411619 . s. m. schmidt , d. blaschke and y. l. kalinovsky , phys .
c * 50 * ( 1994 ) 435 .
o. kiriyama , s. yasui and h. toki , int .
j. mod .
e * 10 * , 501 ( 2001 ) . | we study the phase structure and equation of state for two - flavor quark matter at low temperature under compact star constraints within a nonlocal chiral quark model .
we find that the occurence of a two - flavor color superconducting ( 2sc ) phase is sensitive to variations of both the formfactor of the interaction and the ratio @xmath0 between the coupling constants in the diquark and the scalar meson channels .
our study suggests that for standard values of the coupling ratio @xmath1 either the 2sc phase does not occur ( gaussian formfactor ) or it exists only in a mixed phase with normal quark matter ( nq-2sc ) with a volume fraction less than @xmath2 , occuring at high baryon chemical potentials @xmath3 mev and most likely not relevant for compact stars .
we also present the relevant region of the phase diagram for compact star applications and obtain that no gapless 2sc occurs at low temperatures .
pacs number(s ) : 04.40.dg , 12.38.mh , 26.60.+c , 97.60.jd |
it is well established that massive stars form in dense cores within giant molecular clouds ( gmcs ) , while low - mass stars form both in dark clouds and gmcs .
the formation process of an isolated low - mass star is fairly well understood ; the gravitational collapse of a molecular cloud into a protostellar core and the subsequent accretion via a circumstellar disk surrounded by an infalling envelope ( e.g. , shu , adams , & lizano 1987 and references therein ) , followed by the dissipation of momentum into its surroundings through molecular outflows ( lada 1985 ; fukui et al . 1986 , 1993 ; fukui 1989 ; and references therein ) .
the current paradigm is now being tested by observations , and it has become clear that some modifications to the details of protostellar formation are needed ( e.g. , andr et al . 1993 ; onishi et al . 1996 , 1998 ; evans 1999 and references therein ) . concerning massive star formation ,
an empirical scenario of the later stages of the massive star formation process has been presented ( e.g. , kurtz 2000 and references therein ) ; dense ( @xmath24 10@xmath25 @xmath17 ) , compact ( @xmath26 0.1 pc ) , and hot ( @xmath27 100 k ) regions called hot cores are formed in gmcs , then the uv radiation from newly formed stars within hot cores ionizes the surrounding material to produce ultra - compact h ii regions ( uchiis ) , and finally the central stars become visible , such as the trapezium cluster in orion . on the contrary ,
evolution before the formation of hot cores remains unclear , although some attempts to detect sources at the earliest stages of massive star formation have been made ; surveys toward iras point sources and maser sources ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have revealed the presence of the dense , compact , and massive cores ( density @xmath27 10@xmath28 @xmath17 , temperature @xmath27 100 k , luminosity @xmath27 10@xmath18 @xmath19 ) sometimes called high - mass protostellar objects ( hmpos ; * ? ? ?
* ) , some of which are not accompanied by uchiis . although sources without uchiis may be younger than the rest , the presence of iras point sources and maser sources themselves stongly indicates that stars are already formed within hmpos .
it is quite natural to suppose that dense , compact , massive , _ starless _ cores ( density @xmath24 10@xmath28 @xmath17 , size @xmath26 1.0 pc , mass @xmath27 10@xmath13 @xmath14 ) are the precursors of the hmpos and hot cores ( e.g. , * ? ? ?
systematic studies of these massive _ starless _ cores had not been carried out until recently , due to the difficulty in selecting targets from the existing database .
it is to be noted that large - scale surveys for possible candidates for massive _ starless _ cores ( density @xmath24 10@xmath18 @xmath17 ) have already been made ; e.g. , @xcite surveyed the l1630 cloud ( orion b ) in cs ( @xmath4 = 21 ) . the importance of the study of _ starless _ cores has been revealed in a series of unbiased surveys for dense cores in low - mass star - forming regions , especially from a viewpoint of setting observational constraints on the evolutional time scales of the theoretical model @xcite .
the same is expected for the massive star forming regions and thus unbiased surveys for massive dense cores in massive star - forming regions are also crucial .
one of the most important tasks in the study of massive star formation at the moment is to identify massive dense cores which have not yet experienced massive - star formation but are capable of future massive - star formation .
high - resolution observations of these cores will give observational constraints on the initial phases of the evolutional model of a massive star ; massive stars are formed by direct accretion onto a central protostar through an accretion disk ( e.g. , * ? ? ? * ) , which is simple extension of low - mass star formation , or by collisions of intermediate - mass protostars within a young cluster ( e.g. , * ? ? ?
. moreover , such observations are the only method to reveal the initial conditions of massive star forming cores , since the properties of the cores change rapidly soon after the formation of massive stars due to strong uv radiation and stellar winds .
surveys for hot cores and hmpos toward iras point sources and maser sources are , however , biased toward regions which have experienced star formation , and thus it is not possible to obtain information on the initial conditions of the massive star forming cores
. therefore studies of dense cores which have not yet experienced massive - star formation are awaited .
unbiased surveys for massive dense cores over entire gmcs in molecular lines as well as in the far - infrared and ( sub-)millimeter continuum may be the most effective methods to search for massive _ starless _ cores which are capable of future massive - star formation .
far - ir and submillimeter / millimeter continuum observations are best suited to detect radiation from the dust in _ starless _ cores , since the radiation from cold dust ( @xmath29 @xmath26 20 k ) peaks at far - ir ( @xmath30 150 @xmath31 m ) wavelengths with gradual decrease toward longer wavelengths .
recent development of sensitive instruments , such as bolometer array receivers , makes ( sub-)millimeter continuum observations more effective for carrying out a large - scale survey for massive dense cores .
an unbiased survey of entire gmcs with ( sub-)millimeter continuum has not yet been made , but several surveys in small fields surrounding iras point sources are now in progress for the purpose of detecting massive dense cores located near the iras point sources .
so far , regions of @xmath24 10@xmath32 @xmath12 10@xmath32 around @xmath24 200
iras point sources have been observed at a resolution of 824@xmath33 @xcite .
based on these surveys , a few massive dense cold cores without mid - ir ( msx ) and far - ir ( iras ) counterparts have been found @xcite .
molecular line observations have the invaluable merit of drawing kinematic information within the cores , which is impossible in continuum observations .
the empirical relation between the luminosity of the forming star and the line width of the parent dense core has been found by many authors from observations in nh@xmath34 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) as well as in c@xmath3o ( e.g. , * ? ? ?
* ) , which makes molecular line observations still more important .
we , the nanten group , are now conducting unbiased surveys for dense cores in the c@xmath3o ( @xmath4 = 10 ) line with the nanten 4-meter millimeter telescope .
@xcite compiled a sample of 174 c@xmath3o dense cores in low - mass star - forming regions of taurus @xcite , l1333 @xcite , ophiuchus north @xcite , @xmath35 oph @xcite , chamaeleon @xcite , lupus @xcite , corona australis @xcite , the pipe nebula @xcite , and the southern coalsack @xcite including intermediate - mass cluster - forming regions such as @xmath35 oph , cha i , r cra , and lupus 3 , and found that cores with active star formation tend to have larger column density , number density , and mass , and tend to be gravitationally more bound .
observations toward massive star forming regions are also in progress ; we have detected 112 cores so far in the unbiased surveys for dense cores in gmcs such as orion a , orion b , cepheus ob3 , vela , s35/s37 , and centaurus @xcite .
the beam size of the nanten telescope , 2@xmath367 ( @xmath240.3 pc at the distance of the nearest gmc , orion b , @xmath37 = 400 pc ) , is not necessarily sufficient to resolve massive _ starless _ cores .
however , it is still meaningful to select _ candidates _ for massive _ starless _ cores ( density @xmath38 @xmath17 , size @xmath24 a few pc , mass @xmath30 a few @xmath39 @xmath14 , which are estimated from the physical parameters of massive _ starless _ cores ) , which can be traced by c@xmath3o ( @xmath4 = 10 ) observations with nanten , since only a few massive _ starless _ cores have been identified so far . in this paper , we report the results of an unbiased survey for dense cores in the gmc associated with the @xmath0 carinae nebula ( the @xmath0 car gmc hereafter ) , with the aim of obtaining samples of massive-_starless_-core candidates which are possible sites for future massive star formation .
the @xmath0 car gmc has been one of the most active star - forming regions in the galaxy at least until the formation of the youngest stellar cluster associated with the gmc ( @xmath243 myrs ago , e.g. , * ? ? ?
eight stellar clusters ( trumpler 14 , 15 , 16 , collinder 228 , bochum 10 , 11 , ngc 3293 , and ngc 3324 ; hereafter we abbreviate trumpler 14 and trumpler 16 as tr 14 and tr 16 , respectively ) are associated with the gmc , which contains more than 64 o - type stars including one of the most massive stars , @xmath0 car @xcite .
when we estimate the star formation activity from the richness of the clusters by counting the number of the most massive stars ( e.g. , stars with the spectral type of o3 ) since the total number of the cluster members can not be counted , the @xmath0 car gmc ( with five o3 stars ) is estimated to be the second most active star - forming region after ngc 3603 with six o3 stars @xcite .
its proximity to the sun ( @xmath37 @xmath24 2.5 kpc for the @xmath0 car gmc compared to 6.9 kpc for ngc 3603 ) makes the @xmath0 car gmc the best site for the detailed study of massive star formation , because much higher spatial resolution can be achieved .
there is a possibility that active star - forming regions , such as w49 , where more than 100 o - type stars have been found in infrared observations @xcite , are hidden in the galactic plane due to heavy foreground extinction .
however , it is natural to think that regions with heavy foreground extinction are located distant from the sun ( 11.4 kpc in the case of w49 ) , and thus the @xmath0 car gmc still remains advantageous .
it is to be noted that the well - studied massive star - forming region , the orion region , is a much smaller system , containing only a single o4 - 6 star @xcite .
several observations have been made so far covering the entire @xmath0 car gmc with moderate spatial resolution .
@xcite observed the whole @xmath0 car gmc ( 3 @xmath12 2 deg@xmath40 ) in the @xmath1co ( @xmath4 = 10 ) line with an 8@xmath368 beam , and revealed the existence of a string of co clouds between @xmath41 = 284@xmath427 and 289@xmath43 with a total mass of 6.7 @xmath12 10@xmath44 @xmath14 . @xcite made sub - millimeter observations in @xmath1co ( @xmath4 = 43 ) and atomic carbon [ ci ] @xmath45@xmath46 with a 3@xmath32 beam , and found that these two emissions and @xmath1co ( @xmath4 = 10 ) are approximately coextensive , and that the excitation temperature is higher ( @xmath24 50 k near @xmath0 car and @xmath24 20 k in clouds farther from @xmath0 car ) than that in dark clouds .
high - resolution observations at a resolution of @xmath242040@xmath33 were made toward some selected clouds near @xmath0 car .
there are two main co emission regions near @xmath0 car ; northern and southern clouds @xcite .
the northern cloud lies surrounding the cluster tr 14 ( @xmath24 7 pc northwest of @xmath0 car ) .
@xcite observed the northern cloud and found that the cloud consists of 4 main co clouds with masses in the range 40500 @xmath14 , two of which are gravitationally bound with density exceeding 10@xmath44 @xmath17 .
they proposed that these clouds are exposed to a radiation field originating mainly from the most massive stars in tr 14 .
the southern cloud is located adjacent to the cluster tr 16 , which includes @xmath0 car .
@xcite found a gravitationally - bound cloud with a mass of 70 @xmath14 , and that a cold iras point source with luminosity @xmath24 10@xmath18 @xmath19 is associated with the cloud .
they concluded that star formation is on - going at the interface between the cloud and the cluster tr 16 , which is triggered by radiation driven shocks . in between the northern and the southern clouds ,
the distribution of the molecular gas is patchy due to dissociation by the strong uv radiation and stellar winds from nearby massive stars in tr 16 @xcite .
they also found that the typical mass of the clouds is @xmath24 10 @xmath14 , and that the clouds are not gravitationally bound .
recently , @xcite studied the region called the giant pillar @xcite , @xmath24 0@xmath425 south of @xmath0 car . since the distance from the clusters is larger than that of the northern and southern clouds , and the giant pillar
is located behind the southren cloud , the influence of the uv radiation and stellar winds from the clusters may be smaller and the photo - dissociation takes longer than the northern and southern clouds . from co @xmath4 = 21 observations toward four mid - ir sources identified by @xcite using mid - infrared data obtained by midcourse space experiment ( msx ) , they found these sources are also influenced by the radiation from clusters tr 14 and tr 16
. the mass of the associated clouds ranges from 1045 @xmath14 and they are probably not gravitationally bound .
they also found 12 candidates for massive young stellar objects and uchiis from the mid - ir data .
these studies revealed that star formation is still on - going in the @xmath0 car gmc under the strong influence of the most massive stars in clusters , in at least 1.5 @xmath12 2 deg@xmath40 regions in the vicinity of @xmath0 car .
here we report the results of an unbiased survey for dense cores in the entire @xmath0 car gmc .
the purpose of the present study is to obtain a sample of massive-_starless_-core candidates which are possible sites for future massive star formation . with moderate spatial resolution , 2@xmath367 , our observation covered the entire @xmath0 car gmc ( 3 @xmath12 2 deg@xmath40 ) including regions farther from @xmath0 car , such as the western part of the @xmath0 car gmc around the h ii region gum 31 and the southern most part of the @xmath0 car gmc , where no previous observations had been made except coarse large - scale surveys .
observations were made in c@xmath3o ( @xmath4 = 10 ) as well as in @xmath1co and @xmath2co with nanten .
we also made high - resolution observations in h@xmath2co@xmath20 ( @xmath4 = 10 ) with the mopra 22-meter millimeter telescope and searched for molecular outflows in @xmath1co ( @xmath4 = 21 ) toward some selected dense cores with the aste 10-meter sub - millimeter telescope .
out of the 15 c@xmath3o cores detected in this survey , two are on - going massive star forming cores associated with luminous infrared sources and another 4 are possibly less - massive star - forming cores , and thus as many as 913 cores which have not experienced massive star formation , are detected in this survey .
we present the physical properties of the dense cores and compare the star - formation activity with them .
we also compare the properties of the c@xmath3o cores with those in other massive star forming regions , such as orion , cepheus , vela , s35 , and centaurus , in order to study the characteristics of the @xmath0 car gmc .
observations of the @xmath4 = 10 transitions of @xmath1co , @xmath2co , and c@xmath3o were made with the nanten 4-meter telescope of nagoya university at las campanas observatory of the carnegie institution of washington @xcite .
the half - power beam width of the telescope was 2@xmath367 at 110 ghz and 2@xmath366 at 115 ghz , corresponding to @xmath242 pc for the distance of the @xmath0 car gmc , 2.5 kpc .
the 4 k cooled sis mixer receiver provided a typical system temperature of @xmath24140 k ( ssb ) at 110 ghz and @xmath24200 k ( ssb ) at 115 ghz , including the atmosphere toward the zenith @xcite .
the spectrometer was an acousto - optical spectrometer with a total bandwidth of 40 mhz divided into 2048 channels .
the effective spectral resolution was 40 khz , corresponding to a velocity resolution of 0.11 km s@xmath11 at 110 ghz .
the @xmath1co data were obtained from 1998 may to july , and in 1999 march .
most of the region was observed by using a position - switching technique with a grid spacing of 2@xmath32 , and the rest with 4@xmath32 grid . the total integration time and the typical rms noise of the data at a velocity resolution of 0.1 km
s@xmath11 were @xmath245 sec and @xmath47 1.0 k , respectively . in @xmath2co , observations were made from 1999 january to march with 4@xmath32 grid by using a position - switching technique .
the total integration time per point was @xmath245 sec and the typical rms noise of the data was @xmath47 0.4 k at a velocity resolution of 0.1 km s@xmath11 .
c@xmath3o observations were made in 3 periods from 1998 july to august , in 1999 march , and from 2003 august to october .
the data were obtained with a grid spacing of 2@xmath32 .
a frequency - switching technique was used with a switching interval of 13 mhz in 1998 and 1999 , whereas a position - switching technique was used in 2003 .
the total integration time per point was @xmath2490 sec and the typical rms noise of the data was @xmath47 0.1 k at a velocity resolution of 0.1 km s@xmath11 . for the calibration of the spectral line intensity ,
a room - temperature chopper wheel was employed .
the absolute - intensity calibration was made by observing ori kl ( @xmath48 = 5@xmath4932@xmath5047@xmath510 , @xmath52=@xmath535@xmath4324@xmath3221@xmath33 ) and @xmath35 oph east ( @xmath48 = 16@xmath4929@xmath5020@xmath519 , @xmath52 = @xmath5324@xmath4322@xmath3213@xmath33 ) . the peak radiation temperatures , @xmath54 , are assumed to be @xmath54 ( @xmath1co , ori kl ) = 65 k @xcite , @xmath54 ( @xmath2co , ori kl ) = 10 k @xcite , and @xmath54 ( c@xmath3o , @xmath35 oph east ) = 4.4 k. the last value is consistent with the assumption that @xmath54 ( c@xmath3o , tmc-1 [ @xmath48 = 4@xmath4938@xmath5042@xmath510 , @xmath52 = 25@xmath4335@xmath3245@xmath33 ] ) = 2.0 k @xcite , which has been established by the long - term monitoring of standard sources made at the nanten telescope . the pointing accuracy was measured to be better than 20@xmath33 , as checked by optical observations of stars with a ccd camera attached to the telescope as well as by radio observations of jupiter , venus , and the edge of the sun .
observations of the @xmath4 = 10 transition of h@xmath2co@xmath20 were made with mopra , the 22-meter radio telescope of the australia telescope national facility ( atnf ) , on 2003 july .
the half - power beam width of the telescope was @xmath2437@xmath33 at 86 ghz @xcite .
the front end was a dual - channel 4 k cooled sis mixer receiver , and the typical system temperature was 200 k ( ssb ) at 86 ghz , including the atmosphere toward the zenith .
two orthogonal polarizations were simultaneously observed , both tuned to the h@xmath2co@xmath20 ( @xmath4 = 10 ) frequency .
the spectrometer was a digital correlator with a bandwidth of 64 mhz and 2048 channels , which was divided into two 1024 channels in order to measure the two polarizations simultaneously .
the velocity coverage was @xmath24 110 km s@xmath11 at 86 ghz , and the effective spectral resolution was @xmath24 0.13 km s@xmath11 at 86 ghz , corresponding to @xmath24 1.2 times the channel increment . the data were obtained with a grid spacing of 40@xmath33 by the position - switching mode .
the total integration time per point was @xmath24 8 minutes , and the typical rms noise of the data was @xmath47 0.1 k at a velocity resolution of 0.1 km s@xmath11 .
the intensity calibration was made by using a room - temperature chopper - wheel .
the absolute intensity calibration was made by observing ori kl , by assuming @xmath54 = 0.55 k @xcite .
the pointing accuracy was measured to be within 15@xmath33 by observing the sio maser source rw vel at 86 ghz every two hours during the observations .
observations of the @xmath4 = 21 transition of @xmath1co were made with atacama submillimeter telescope experiment ( aste ) , the 10-meter radio telescope of nobeyama radio observatory at pampa la bola , chile , on 2003 november @xcite .
the half - power beamwidth of the telescope was 30@xmath33 at 230 ghz .
the front end was a 4 k cooled sis mixer receiver @xcite .
the typical system temperature was 300 k ( dsb ) at 230 ghz , including the atmosphere toward the zenith .
we used a digital correlator with a bandwidth of 128 mhz and 1024 channels @xcite .
the effective spectral resolution was 150 khz , corresponding to a velocity resolution of 0.16 km s@xmath11 at 230 ghz .
the data were obtained with a grid spacing of 30@xmath33 by using the position - switching mode .
the total integration time per point was @xmath24 1 minute and the typical rms noise of the data was @xmath47 0.8 k at a velocity resolution of 0.5 km s@xmath11 .
the intensity calibration was made by using a room - temperature chopper - wheel .
the absolute intensity calibration was made by observing ori kl by assuming @xmath54 to be 111 k @xcite .
the pointing accuracy was measured to be within 15@xmath33 as checked by optical observations of stars with a ccd camera attached to the telescope as well as by radio observations of jupiter and mars .
the @xmath4 = 10 emission lines of carbon monoxide ( @xmath1co ) and its isotopes ( @xmath2co and c@xmath3o ) have been used to study molecular clouds , since they are among the most abundant molecules in the dense interstellar medium .
although all these lines have nearly the same critical density for collisional excitation , it is found empirically that the typical densities traced by them are different : @xmath56 @xmath17 for @xmath1co , @xmath57 @xmath17 for @xmath2co , and @xmath38 @xmath17 for c@xmath3o ( e.g. , mizuno et al .
1995 , onishi et al . 1996 , 1998 for taurus ; mizuno et al . 1998 , 1999 for chamaeleon ;
dobashi et al .
1994 , 1996 , yonekura et al .
1997 , kawamura et al . 1998 , kim et al .
2004 for a series of @xmath2co survey in the milky way ; tachihara et al .
2002 and references therein for c@xmath3o cores in nearby star forming regions ) . this density differentiation can be understood as follows : one of the causes for this density differentiation is due to the optical depth .
the @xmath1co transition is optically thick in most molecular clouds , making it to probe the lower density enveloping layer of a molecular cloud due to the photon trapping .
the other two lines have smaller optical depths , allowing us to look more deeply into clouds .
however , because of the selective photo - dissosiation of the @xmath2co and c@xmath3o molecules , abundance ratio of the isotopes vary as a function of the visual extinction @xmath58 ( e.g. , * ? ? ?
. the abundance ratio @xmath59(@xmath2co / c@xmath3o ) reaches up to @xmath60 near the cloud surface at @xmath58 @xmath24 15 . as a result
, it is expected that the @xmath2co emission mainly comes from the intermediate part of the density ( @xmath57 @xmath17 ) and that the c@xmath3o emission comes from the inner part of the density ( @xmath38 @xmath17 ) .
figure [ fig01 ] shows the total - intensity distribution of @xmath1co ( @xmath4 = 10 ) in the velocity range from @xmath61 = @xmath5330 km s@xmath11 to @xmath5310 km s@xmath11 , overlaid on an optical image taken from the digitized sky survey .
we also show the distribution of ob stars overlaid on a @xmath1co map and an h@xmath62 image in figure [ fig02 ] .
the positions of the ob stars are taken from the simbad database .
we find @xmath24 80 o - type stars including five o3 stars , @xmath24400 b stars , and 7 wolf - rayet stars in the observed area .
the h@xmath62 image is reproduced from the southern h@xmath62 sky survey atlas ( shassa ; * ? ? ?
it is clearly seen that the overall distribution of the @xmath1co emission shows an anti - correlation with the optical objects such as stellar clusters and h ii regions .
most of the o - type stars are located near the @xmath0 car nebula at ( @xmath41 , @xmath63 ) @xmath24 ( 287.5 , @xmath530.7 ) , creating the huge h@xmath62 emission nebulosity , while small h ii regions are made by a few o stars such as ngc 3293 at ( 285.9 , 0.1 ) and gum 31 at ( 286.2 , @xmath530.2 ) . on the other hand
, there is a good correlation of the @xmath1co emission with the optical obscuration around the @xmath0 car nebula ; the northern cloud at ( 287.3 , @xmath530.7 ) , the southern cloud at ( 287.7 .
@xmath530.7 ) , and the giant pillar at ( 288.0 , @xmath531.1 ) . in the @xmath1co emission ,
a dozen intense peaks are distributed within more extended diffuse emission .
the total molecular mass traced in the @xmath1co emission is @xmath24 3.5 @xmath12 10@xmath44 @xmath14 , which is estimated in the following manner : we assumed an empirical relation that the molecular hydrogen column density , @xmath8(h@xmath9 ) is proportional to the @xmath1co integrated intensity , @xmath64(@xmath1co ) .
the conversion factor is called the @xmath65-factor [ @xmath65 = @xmath8(h@xmath9)/@xmath64(@xmath1co ) ] , and the value lies in the range ( 13 ) @xmath12 10@xmath66 @xmath16 ( k km s@xmath11)@xmath11 toward the clouds in the galactic disk , as estimated by the virial mass , @xmath67-ray emission , and so on ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
here we adopted @xmath65 = 1.6 @xmath12 10@xmath66 @xmath16 ( k km s@xmath11)@xmath11 as derived from observations with the energetic gamma - ray experiment telescope ( egret ; * ? ? ?
the total mass of the @xmath1co emission is estimated by using @xmath68 where @xmath37 is the distance , @xmath69 is the solid angle subtended by the effective beam size ( 2@xmath32 @xmath12 2@xmath32 or 4@xmath32 @xmath12 4@xmath32 ) , @xmath70 is the proton mass , and @xmath31 is the mean molecular weight .
the factor of 2.8 is introduced by taking into account the contribution of helium ( one he atom for every five h@xmath9 molecules ) , which is the same assumption as @xmath31 = 2.33 .
the summation is performed over the observed points within the 3-sigma contour level ( 5 k km s@xmath11 ) of the integrated intensity .
the distance of the @xmath0 car gmc is not well determined .
@xcite used @xmath37 = 2.7 kpc , which is estimated from the distances of associated star clusters .
@xcite summarized the photometric data of the open clusters distributed over the entire @xmath0 car gmc , and obtained the average value of 2.5 kpc . in the vicinity of @xmath0 car ,
values in the range 2.22.8 kpc are reported for the distances of tr 14 and tr 16 ( e.g. , * ? ? ?
* ) . in this paper
, we adopt the frequently quoted value of 2.5 kpc as the distance of the @xmath0 car gmc @xcite . the uncertainty in determining the total mass is estimated to be @xmath71 from the equation ( 1 ) , since the distance @xmath37 and the conversion factor @xmath65 include @xmath72 and @xmath73 uncertainty , respectively .
the total mass obtained in this manner is somewhat smaller than that derived in @xcite , 6.7 @xmath12 10@xmath44 @xmath14 ( cloud 11 [ [email protected] ] in their catalog ) , who observed the carina arm with a 1.2-meter telescope at cerro tololo .
the difference in the total mass basically originates from the different assumptions on the @xmath65-factor and the distance used in deriving the mass ; they used @xmath65 = 2.8 @xmath12 10@xmath66 @xmath16 ( k km s@xmath11)@xmath11 and @xmath37 = 2.7 kpc .
when we use the same assumption as in @xcite , 7.2 @xmath12 10@xmath44 @xmath14 is obtained , which is in good agreement with the value derived by @xcite .
clumpy structure is more clearly seen in the @xmath2co ( @xmath4 = 10 ) emission ( fig .
[ fig03 ] ) . from the @xmath2co distribution
, we can divide the @xmath0 car gmc into 7 regions , which are clearly separated at the 3-sigma contour level ( 1.35 k km s@xmath11 ) .
the boundaries of the regions are similar to those in a previous study by @xcite , which is based on the @xmath1co ( @xmath4 = 10 ) map of 7@xmath365 grid spacing with an 8@xmath368 beam , but our definition traces more detailed structure .
moreover we identified one more region ( region 7 ) .
this is mainly because the spatial resolution in our study is @xmath24 3 times higher than that in @xcite and because we used @xmath2co emission instead of @xmath1co .
the physical properties of the @xmath2co gas are estimated in the following manner assuming local thermodynamical equilibrium ( lte ) : the excitation temperature , @xmath74 , was estimated from @xmath54 ( @xmath1co ) at the peak position of each c@xmath3o core ( see 3.2 ) by using @xmath75 \ } } ( { \rm k}).\ ] ] we used the average value of the @xmath74 of each of the c@xmath3o cores within the region as the @xmath74 of the region . in order to derive the @xmath2co column density , we divide each spectrum into 0.1 km s@xmath11 bins , calculate the column density within each bin , and sum them up within an lsr velocity range from @xmath5330 km s@xmath11 to @xmath5310 km s@xmath11 . the optical depth in each bin , @xmath76 ( @xmath77 ) ,
is calculated by the following equation : @xmath78 - 0.164\}}\right),\ ] ] where @xmath79 ( @xmath77 ) is the average temperature of the @xmath2co spectrum in each bin in the unit of kelvin , and @xmath80[@xmath29(k ) ] = 1/exp[5.29/@xmath29(k ) ] @xmath53 1 .
the @xmath2co column density is estimated from @xmath81 } } ( { \rm cm}^{-2}).\ ] ] the ratio @xmath8(h@xmath9)/@xmath8(@xmath2co ) is assumed to be 5 @xmath12 10@xmath44 @xcite .
the mass is estimated in the same manner as for the @xmath1co .
the total molecular mass traced in the @xmath2co emission is @xmath24 1.3 @xmath12 10@xmath44 @xmath14 , which is 38 % of the @xmath1co mass . the spatial distribution of the c@xmath3o ( @xmath4 = 10 ) integrated intensity is shown in figure [ fig04 ] .
the distribution of the c@xmath3o gas is more compact than those of the @xmath1co and @xmath2co gas .
the total molecular mass traced in the c@xmath3o emission is @xmath24 5.8 @xmath12 10@xmath18 @xmath14 ( see 3.2 ) , which is 44 % of the @xmath2co mass .
the mass of each region traced in @xmath1co , @xmath2co , and c@xmath3o is summarized in table [ table01 ] .
the fraction of the @xmath2co mass to the @xmath1co mass ranges from 3668 % , with the largest value of 68 % in region 1 .
the fraction of the c@xmath3o mass to the @xmath2co mass ranges from 2172 % , with the largest value of 72 % again in regions 1 and 5 .
we note that co emission with @xmath61 @xmath27 @xmath5310 km s@xmath11 was detected toward some areas around ( @xmath41 , @xmath63 ) = ( 286.5 , @xmath530.2 ) , although these velocity components are not shown in the figures .
these velocity components come from distant clouds ( clouds on the far side of the carina arm at @xmath37 = 7.5 kpc , * ? ? ?
* ) , not associated with the @xmath0 car gmc .
we will discuss these clouds in a separate paper ( yonekura et al .
2005 in preparation ) . in order to investigate the physical properties of the dense regions , we define a c@xmath3o core in the same manner as that adopted by @xcite ;
( 1 ) find a peak - intensity position , ( 2 ) draw a contour at the half level of the peak intensity , ( 3 ) identify a core unless previously identified cores exist within the half - level contour , ( 4 ) find the next intensity peak outside the core , ( 5 ) repeat the procedure after ( 2 ) until the peak intensity falls down below 6-sigma level (= 1.0 k km s@xmath11 ) .
we finally identified 15 c@xmath3o cores . in figure [ fig05 ]
we show the boundaries of each core , and their observed properties are listed in table [ table02 ] .
the absolute peak temperature , @xmath54 ( c@xmath3o ) , the radial velocity of the c@xmath3o emission , @xmath61 , and the fwhm line width , @xmath23 , are derived from a single gaussian fitting for a line profile at the peak position of the core .
there is a variation in the ratio of the c@xmath3o integrated intensity , @xmath64(c@xmath3o ) , to the @xmath2co integrated intensity , @xmath64(@xmath2co ) ; three prominent c@xmath3o cores , whose c@xmath3o intensities at the peak position are larger than @xmath24 2 k km s@xmath11 ( nos . 5 , 6 , and 13 ) , possess only moderate @xmath2co intensity among the dozens of @xmath2co local peaks .
this indicates that @xmath2co observation is not necessarily suited to trace the high - density regions .
the physical properties of the c@xmath3o cores were estimated using nearly the same procedure as for @xmath2co .
differences between the two procedures are the following : the optical depth of the c@xmath3o line in each bin , @xmath82 , is calculated by the following equation : @xmath83 - 0.166\}}\right),\ ] ] where @xmath84 is the average temperature of the c@xmath3o spectrum in each bin in kelvin , and @xmath85[@xmath29(k ) ] = 1/exp[5.27/@xmath29(k ) ] @xmath53 1 .
the c@xmath3o column density is estimated from @xmath86 } } ( { \rm cm}^{-2}).\ ] ] the ratio @xmath8(h@xmath9)/@xmath8(c@xmath3o ) is assumed to be 6 @xmath12 10@xmath28 @xcite .
the mass of a c@xmath3o core is derived by summing over the observed points within the core .
the radius of a core , @xmath6 , was calculated from @xmath87 where @xmath88 is the area inside the core .
the radius of small cores is overestimated by the beam - size effect .
the relation between the observed radius , @xmath6 , and the deconvolved radius , @xmath89 , is approximated by the relation @xmath90 we also calculated the virial mass , @xmath22 , @xmath91 with @xmath5 defined as the fwhm line width of the composite profile derived by using a single gaussian fitting , where the composite profile was obtained by averaging all the spectra within the core .
the average h@xmath9 number density of the core , @xmath10(h@xmath9 ) , was derived by dividing @xmath7 by the volume of the core , assuming that the core is spherical with radius @xmath6 .
the physical properties of the c@xmath3o cores thus obtained are listed in table [ table03 ] along with the average and the median values .
here we compare the physical properties of the cores in the @xmath0 car gmc to those of massive - star - forming cores observed with the nanten and the 4-m radio telescopes of nagoya university : orion b @xcite , orion a @xcite , vela c @xcite , cepheus ob3 @xcite , s35/s37 @xcite , and centaurus @xcite .
all these studies assume lte in deriving physical properties and use the same assumption on abundance .
since the definition of a c@xmath3o core in orion a , vela c , and cepheus ob3 is different from the one in this work , only values at the peak position ( i.e. , @xmath23 and @xmath8(h@xmath9 ) ) are compared for these regions . in order to be consistent for comparison , we used ( @xmath22/@xmath14 ) = 209 ( @xmath6/pc ) ( @xmath23/km s@xmath11)@xmath40 instead of equation ( 9 ) , and we excluded cores with serious beam - size effect , i.e. , @xmath89/@xmath6 @xmath92 0.8 , in calculating the average and median values of @xmath6 , @xmath10(h@xmath9 ) , and @xmath22/@xmath7 according to @xcite .
hereafter , we call the cores with @xmath89/@xmath6 @xmath93 0.8 as resolved cores . the average and median values in each region
are listed in table [ table04 ] .
the line width , @xmath23 , of the cores in the @xmath0 car gmc ranges from 2.2 to 5.4 km s@xmath11 with an average value of 3.2 km s@xmath11 .
this is the largest among the sample .
the peak column density ranges from 0.63.5 @xmath12 10@xmath15 @xmath16 with an average of 1.3 @xmath12 10@xmath15 @xmath16 , which is a typical value for the gmcs listed here .
the core mass ranges from 450 to 6100 @xmath14 with an average of 2600 @xmath14 .
it is difficult to compare the absolute values of the mass , since there is a dependence on the distance .
the radius of the resolved cores ranges from 1.6 to 3.1 pc with an average of 2.2 pc .
the radius is also dependent on the distance .
the average number density , @xmath10(h@xmath9 ) , ranges from 0.53.4 @xmath12 10@xmath13 @xmath17 with an average of 1.2 @xmath12 10@xmath13 @xmath17 , which is about one order smaller than the expected value for a c@xmath3o core , @xmath24 10@xmath18 @xmath17 .
this is mainly due to the beam dilution ; the resolution of the telescope is @xmath242 pc at the distance of the @xmath0 car gmc , which is larger than the typical size of the c@xmath3o cores in orion , @xmath24 0.5 pc .
the ratio of the virial mass , @xmath22 , to the lte mass , @xmath7 , ranges from 0.5 to 2.9 with an average of 1.6 , which is the second largest value among the whole sample . to summarize ,
the cores in the @xmath0 car gmc are characterized by larger values of @xmath23 and @xmath22/@xmath7 on average than those in other massive - star - forming regions . in this subsection
, we investigate the star - formation activity in c@xmath3o cores in the @xmath0 car gmc .
the iras point source catalog is a useful database to search for young stellar objects ( ysos ) embedded in molecular clouds . in order to select candidates for
recently formed stars among the iras point sources , we searched for sources having cold infrared spectra defined at 25 and 60 @xmath31 m , i.e. , sources with higher flux densities at 60 @xmath31 m than at 25 @xmath31 m with a data quality better than or equal to 2 in these two bands .
we excluded objects from the selection that have already been identified as planetary nebulae or external galaxies in the iras point source catalog . among the 316 iras point sources in the observed region , 73
were found to satisfy the criteria , and only two sources are located inside the boundaries of c@xmath3o cores , i.e. , within the half - intensity contours : iras 10365@xmath535803 and iras 10361@xmath535830 are associated with c@xmath3o cores nos . 5 and 6 , respectively ( figs .
[ fig04 ] and [ fig05 ] ) .
the luminosities of these sources are @xmath2430000 @xmath19 ( iras 10365@xmath535803 ) and @xmath2421000 @xmath19 ( iras 10361@xmath535830 ) , indicating that these are candidates for massive stars or star clusters .
the properties of these iras point sources are summarized in table [ table05 ] .
we also searched for ysos from the msx point source catalog .
we searched for sources with a data quality better than or equal to 3 in band c ( @xmath94 m ) , d ( @xmath95 m ) , and e ( @xmath96 m ) . among the 2799 msx point sources in the observed region , 147
were found to satisfy the criteria , and 8 sources are associated with 5 c@xmath3o cores ( nos . 5 , 6 , 9 , 10 , and 12 ; fig .
[ fig06 ] ) .
the properties of these msx point sources are summarized in table [ table06 ] .
the detection limit of the iras survey toward the @xmath0 car gmc is estimated to be @xmath24 1000 @xmath19 , which we derived from the distribution of the luminosity of the above 73 yso - candidate iras point sources on the assumption that these sources are located at the same distance as the @xmath0 car gmc .
therefore , we can conclude that massive star formation has not occured in the cores without iras point sources , although we can not exclude the existence of sources at an evolutionary stage prior to the formation of a massive star , which may not have enough emission at infrared wavelengths ( e.g. , * ? ? ?
* ) . along with the result in the next subsection , 6 c@xmath3o cores ( nos
. 5 , 6 , 9 , 10 , 12 , and 15 ) out of 15 (= 40@xmath21 ) are experienced star formation , and at least 2 of 15 (= 13 @xmath21 ) are massive - star forming cores in the @xmath0 car gmc , although the proportion of star - forming cores may vary when high - spatial - resolution molecular line observations and more sensitive infrared and ( sub)millimeter observations are made .
it should be noted that as many as 13 out of the 15 c@xmath3o cores , i.e. , 87 % , are not associated with luminous iras point sources ( @xmath97 @xmath98 10@xmath13 @xmath19 ) .
the proportion , 13 % , is much lower than in other massive star forming regions at distances within a factor of two of the @xmath0 car gmc : 28 % in s35/s37 and 86 % in centaurus iii @xcite .
we will discuss the inactiveness in detail later in
4.3 in this subsection , we present the results from the search for high - density regions and molecular outflows , in order to reveal the physical properties and the evolutionary stages of dense cores , especially those without iras point sources , more precisely . in order to detect high - density regions within c@xmath3o cores ,
we have observed 14 c@xmath3o cores ( all except for c@xmath3o core no .
4 ) in h@xmath2co@xmath20 ( @xmath4 = 10 ) emission , which traces a typical density of @xmath24 10@xmath44 @xmath17(e.g .
* ; * ? ? ?
therefore , the c@xmath3o cores without iras point sources , which are detected in h@xmath2co@xmath20 emission , can be considered as good candidates for the sites of future massive star formation .
significant h@xmath2co@xmath20 emission above the 3-sigma noise level was detected toward 4 c@xmath3o cores ( nos . 5 , 6 , 9 , and 15 ) , two of which ( nos . 5 and 6 ) have associated iras point sources and three of which ( nos . 5 , 6 , and 9 ) with msx point sources ( see @xmath99 [ sec_iras ] )
. distributions of h@xmath2co@xmath20 emission are presented in figure [ fig07 ] . in order to study the physical properties , we define the h@xmath2co@xmath20 cores in the same manner as c@xmath3o cores . as a result
, we identified 3 cores .
physical properties of the h@xmath2co@xmath20 cores were estimated using the same method as for c@xmath3o : the h@xmath2co@xmath20 optical depth in each bin , @xmath100 , is calculated from @xmath101 } \right),\ ] ] where @xmath102 = 1 / [ exp(4.16/@xmath29 ) @xmath53 1 ] .
we tentatively estimate @xmath74 of each h@xmath2co@xmath20 core from the @xmath1co ( @xmath4 = 21 ) observations ( see 3.4.2 ) , since the beam sizes of both observations are similar . the h@xmath2co@xmath20 column density , @xmath8(h@xmath2co@xmath20 ) ,
is estimated by using the following equation : @xmath103 } } ( { \rm cm}^{-2}).\ ] ] the h@xmath9 column density , @xmath8(h@xmath9 ) , was derived from @xmath8(h@xmath2co@xmath20 ) by making the same assumption as in @xcite ; [ hco@xmath20]/[h@xmath9 ] = 4.0 @xmath12 10@xmath106 and @xmath1c/@xmath2c = 89 .
the assumption of @xmath74 may not be correct , since the density is perhaps smaller than the critical density of h@xmath2co@xmath20 ( @xmath4 = 10 ) , i.e. , @xmath4 = 10 line of h@xmath2co@xmath20 is perhaps sub - thermally excited .
however , the uncertainty in @xmath74 would not seriously affect the estimate of the column density @xmath8 : even if we changed the value of @xmath74 by a factor of 2 , the estimated value of @xmath8 changes within a factor of 1.9 , which is smaller than the uncertainty in the hco@xmath20 abundance ( e.g. , * ? ? ?
* ; * ? ? ?
the physical properties of the h@xmath2co@xmath20 cores are listed in table [ table07 ] .
the total mass of the molecular gas traced by h@xmath2co@xmath20 in the observed area above the 3-sigma level is estimated to be @xmath24 4300 @xmath14 , corresponding to @xmath24 7 % of the c@xmath3o mass .
it is to be noted that a significant difference is found in the behavior of the line width between cores with and without iras point sources : the line widths in the h@xmath2co@xmath20 cores are much smaller than those in the c@xmath3o cores for the cores without iras point sources ( nos . 9 and 15 ) , whereas the line widths in the h@xmath2co@xmath20 cores are nearly the same as or somewhat larger than those in c@xmath3o cores for cores with iras point sources ( nos . 5 and 6 ) . the present result can be compared to the multitracer , single - cloud line width - size relation , which is classified as @xmath107 in @xcite .
the relation @xmath23 @xmath108 @xmath109 with 0.2 @xmath26 @xmath110 @xmath26 0.7 is reported with the tendency that massive star - forming regions tend to have lower values of @xmath110 ( e.g. , * ? ? ?
the present result is consisitent with the above tendency in a sense that the value @xmath110 in massive star - forming cores tends to be smaller than in the others , although the trend is more extreme ; i.e. , @xmath110 @xmath26 0
. the trend might be indicating that line - width enhancement due to the input of turbulence by the activity of star - formation , such as jets and molecular outflows , is effective especially near the center of the massive star - forming cores .
however , we have to wait until statistically more accurate studies can be made based on a large sample .
this section can be summarized as follows : we detected h@xmath2co@xmath20 emission from 4 c@xmath3o cores , which indicates that high - density gas is presented in these cores and that these cores are capable of star formation .
indeed , two ( nos .
5 and 6 ) have already experienced massive star formation and another one ( no .
9 ) has experienced less - massive star formation .
the other c@xmath3o core ( no . 15 ) may have not yet experienced star formation , which makes the core the most probable site for future star formation , though we can not rule out the possibility that less - massive stars are forming inside ( see [ sec_outflow ] ) .
it is well known that molecular outflows are ubiquitous in low - mass star formation ( lada 1985 ; fukui et al . 1986 ; fukui 1989 ; fukui et al .
1993 and references therein ) , and they are believed to be related to the accretion process .
it is also suggested by these studies that molecular outflows are also common in massive star forming regions .
recent systematic surveys for outflows toward regions of massive - star formation showed a detection rate of molecular outflow phenomena as high as @xmath2490 % , confirming that outflows are also common in massive - star forming regions ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the formation process of massive stars is , however , still controversial ; it is not yet known whether massive stars are formed by direct accretion on a central protostar through an accretion disk ( e.g. , * ? ? ?
* ) , or by the coalescence of low- to intermediate - mass protostars @xcite .
at any rate , the detection of a molecular outflow is a direct indication of on - going star formation .
the age and the luminosity of the driving source can be estimated from the observations , and moreover , the kinematic information , such as the interaction with parent molecular clouds as well as the systemic velocity of the driving source , can be used to solve the distance ambiguity in cases where more than one molecular cloud is located in the same line of sight .
we searched for molecular outflows toward 7 c@xmath3o cores ( nos .
59 , 14 , and 15 ) .
we detected one outflow as well as three candidates for outflows .
figure [ fig08 ] displays the @xmath1co spectra around c@xmath3o core no .
15 . in the figure , both the blue - shifted ( @xmath5335 km s@xmath11 @xmath111 @xmath61 @xmath111 @xmath5325 km s@xmath11 ) and red - shifted wings ( @xmath5315 km s@xmath11 @xmath111 @xmath61 @xmath111 @xmath535 km s@xmath11 ) are clearly seen in some spectra , where the velocity ranges for wing components are determined at a 1-@xmath112 noise level in the spectra smoothed to the velocity resolution of 0.5 km s@xmath11 ( @xmath54 @xmath24 0.8 k ) : the blue - shifted component is prominent at the offset ( @xmath113 , @xmath114 ) = ( @xmath531@xmath32 , 0@xmath32 ) , while the red is most intense at ( @xmath530@xmath365 , 0@xmath32 ) .
the distribution of the @xmath1co line intensities integrated between @xmath61 = @xmath5335 and @xmath5325 km s@xmath11 ( blue component ) and between @xmath5315 and @xmath535 ( red component ) is shown in figure [ fig09 ] .
figure [ fig10 ] shows @xmath1co profiles of three outflow candidates .
although their spectra show asymmetry , or have broad wing features , we can not make a contour map of the wing components , due to the existence of multiple velocity components and/or contamination from the emission at off position .
physical parameters of the detected outflow associated with the c@xmath3o core no . 15 were estimated as follows : the excitation temperature of the high - velocity gas was assumed to be the same as that of the quiescent gas , and estimated to be 30 k from the @xmath1co ( @xmath4 = 21 ) observations by using the following equation @xmath115\ } } ( { \rm k}).\ ] ] the column densities of @xmath1co ( @xmath4 = 21 ) are calculated under the assumption of lte using the following equation : @xmath116 where the integration is made over the wing component ; @xmath117 is defined from the @xmath1co optical depth as @xmath117 = [ 1 @xmath53 exp(@xmath118 ) ] / @xmath119 .
we assumed @xmath119 = 2 for both the red and blue components , which is a typical value for outflows observed in the @xmath1co ( @xmath4 = 21 ) line @xcite .
the column density of h@xmath9 molecules was estimated using an [ h@xmath9]/[co ] ratio of 1 @xmath12 10@xmath18 @xcite .
the mass of each component is derived using equation ( 1 ) with an effective beam size of 30@xmath33 @xmath12 30@xmath33 .
the summation is performed over the observed points within the 3-sigma contour level ( 5 k km s@xmath11 ) of the intensity integrated over the wing component .
sizes of the lobes are defined at the 3-sigma contour level of the integrated intensity map of the outflow .
we tentatively defined the position of the driving source of this outflow as the center of the peak position of each lobe , because no driving sources , such as iras and msx point sources , have been identified around the outflow yet .
the radius is defined as the maximum separation between the position of the driving source and the 3-sigma contour of the wing emission .
using the mass and the radius of the outflow , we estimate the kinetic energy , mechanical luminosity , momentum , and dynamical timescale of the outflow in the same manner as in @xcite .
we calculated physical properties of the outflow for the velocity ranges from @xmath5335 to @xmath5325 km s@xmath11 for the blue - shifted wing and from @xmath5315 to @xmath535 km s@xmath11 for the red - shifted wing .
these values are summarized in table [ table08 ] .
the nature of the driving source is not clear because no sources have been identified yet around the outflow .
we estimate the upper limit to the luminosity from the iras sky survey atlas to be 3900 @xmath19 .
another estimate for the luminosity can be made by using the empirical relation between the mechanical luminosity of the outflow and the bolometric luminosity of the driving source @xcite ; 15 @xmath19 is obtained in this manner .
there are two possibilities for the nature of the driving source ; [ i ] a less - massive star with a luminosity below the iras and/or msx detection limit or [ ii ] a source at an evolutionary stage prior to the formation of a star without enough emission in the infrared ( e.g. , * ? ? ?
we can not rule out either of the above two possibilities at the moment , but the former might be more probable because the dynamical timescale ( @xmath24 3 @xmath12 10@xmath18 yrs ) , which will probably be overestimated because of the overestimation of size due to low angular resolution , is relatively long .
high - resolution co observations as well as sub - millimeter dust continuum observations will reveal the nature of the driving source .
we discuss here the relationship between the star - formation activity and the physical properties of c@xmath3o cores , which has become possible thanks to the detection of a large number of cores without any signs of star formation . in the study of c@xmath3o cores in orion b , @xcite found that star formation occurs preferentially in cores with larger @xmath23 , @xmath8 , and @xmath7 , and that the luminosity of the forming stars increases with @xmath8 . in centaurus , the luminosity of the forming stars increases with @xmath8 and @xmath7 as well as @xmath23 , and decreases with @xmath22/@xmath7 @xcite .
@xcite also found that the luminosity increases with @xmath23 by compiling the data in orion a , b , s35 , vela c , and centaurus , and proposed a possibility that the mass of forming stars is determined by @xmath23 , since high mass - accretion rates can be achieved in cores with large @xmath23 .
however , it is controversial whether large @xmath23 is a cause or effect of massive star formation .
we divided the c@xmath3o cores into 3 groups : ( a ) cores with iras point sources ( nos . 5 and 6 ) , ( b ) other star - forming cores ( nos .
9 , 10 , 12 and 15 ) , and ( c ) cores without any signs of star formation .
the cores in group a are thought to be sites of massive star formation ( see table [ table05 ] for the luminosities of the associated iras point sources ) .
there are two possibilities for the nature of the cores in group b ; these are either [ i ] less - massive stars with luminosities below the iras detection limit or [ ii ] sources at an evolutionary stage prior to the formation of stars without enough emission in the infrared ( see 3.4.2 ) . here
we tentatively treat them as less - massive star - forming cores , following the argument in 3.4.2 .
the cores in group c seem to be sites of less - active star formation than those in groups a and b , which is inferred from the absence of the high - density regions ( see 3.4.1 ) .
table [ table09 ] gives the average physical properties of these 3 groups as well as the average of all cores in the @xmath0 car gmc .
we also calculated the average properties of all the star - forming cores , i.e. , cores in groups a and b. figure [ fig11 ] shows a series of histograms of the core parameters , [ a ] line width @xmath5 , [ b ] h@xmath9 column density @xmath8(h@xmath9 ) , [ c ] lte mass @xmath7 , [ d ] radius @xmath6 , [ e ] h@xmath9 number density @xmath10(h@xmath9 ) , and [ f ] ratio @xmath22/@xmath7 .
note that only resolved cores are used to make the histograms of @xmath6 , @xmath10(h@xmath9 ) , and @xmath22/@xmath7 .
it is clearly seen that @xmath8(h@xmath9 ) , @xmath7 , and @xmath10(h@xmath9 ) are larger and the ratio @xmath22/@xmath7 is smaller in group a than in groups b and c , while the tendency is not clear in @xmath5 and @xmath6 ( trend 1 ) . between star - forming cores ( groups a+b ) and non - star - forming cores ( group c ) , @xmath8(h@xmath9 ) and @xmath10(h@xmath9 ) are larger in groups a+b than in group c ( trend 2 ) .
however , no significant difference between group b and group c is seen ( trend 3 ) .
these trends ( trends 13 ) might be indicating the differences in the physical properties among the cores as a function of the masses of the forming stars . here
we compare the present results to those in centaurus @xcite .
in centaurus , the luminosity of the forming stars increases with @xmath8 and @xmath7 as well as @xmath23 , and decreases with @xmath22/@xmath7 .
this trend is same as the trend 1 in @xmath0 car gmc except @xmath23 .
although we have obtained 127 c@xmath3o cores in massive star forming regions so far by using the same telescope including 15 in @xmath0 car gmc , we can not make direct comparison at present , because of the differences in the detection limit and/or the definition of c@xmath3o cores .
the results of detailed comparion will be published in a separate paper .
we can conclude that massive star formation occurs preferentially in cores with larger @xmath8 , @xmath7 , @xmath10 , and smaller @xmath22/@xmath7 in @xmath0 car gmc .
high spatial - resolution and high sensitivity observations in infrared ( e.g. , spitzer , astro - f ) as well as the observations in high density tracer such as cs ( @xmath4 = 76 ) , hco@xmath20 ( @xmath4 = 43 ) , hcn ( @xmath4 = 43 ) ( e.g. , aste ) , will help to reveal the properties of c@xmath3o cores .
it is of interest to compare the distribution of co and c@xmath3o ( @xmath4 = 10 ) emissions to that of co ( @xmath4 = 43 ) emission obtained with the antarctic submillimeter telescope and remote observatory ( ast / ro ; * ? ? ?
* ) , since different excitation conditions are required between co ( @xmath4 = 43 ) and co ( @xmath4 = 10 ) .
the @xmath4 = 4 level of the co molecule is 55 k above the ground level , and the critical density of the @xmath4 = 43 transition is @xmath120 @xmath17 for the optically thin case , whereas the @xmath4 = 1 level of the co molecule is 5.5 k above the ground level , and the critical density of the @xmath4 = 10 transition is @xmath57 @xmath17 . at kinetic temperature of 40 k ,
the @xmath4 = 43 emission of co is most intense at density of @xmath24 10@xmath44 @xmath17 while the @xmath4 = 10 intensity is peaked at around 10@xmath13 @xmath17 @xcite .
observationally , the density effectively emitting these emissions may correspond to somewhat lower than these values ( @xmath24 10@xmath40 @xmath17 for the @xmath4 = 10 emission ) , which can be understood if we take into account the optical depth effect .
thus , the co ( @xmath4 = 43 ) line requires higher temperature and density than the co ( @xmath4 = 10 ) line .
however , the overall distribution of the molecular gas traced in the co ( @xmath4 = 10 ) line and that of the co ( @xmath4 = 43 ) line is quite similar except the following two regions : \(1 ) c@xmath3o core no .
10 ( adjacent to tr 14 ) + the integrated intensities of co ( @xmath4 = 43 ) and co ( @xmath4 = 10 ) are the strongest in the observed region , although the integrated intensity of c@xmath3o ( @xmath4 = 10 ) shows only moderate values . a possible explanation for this is a high excitation temperature due to the star cluster tr 14 : because of the high excitation temperature , the optical depth of c@xmath3o ( @xmath4 = 10 ) remains very small in spite of its large column density , resulting in very low intensity in c@xmath3o ( @xmath4 = 10 ) emission .
the nanten data indicate the excitation temperature is @xmath24 25 k , and this value seems to be a lower limit , since higher resolution data obtained with a 24@xmath33 beam show the excitation temperature to be as high as @xmath24 45 k @xcite .
\(2 ) c@xmath3o core no . 5 + the intensity of co (
@xmath4 = 43 ) is very weak , whereas co and c@xmath3o ( @xmath4 = 10 ) emission is quite strong
. in general , two explanations are possible ; ( 1 ) the temperature is not high enough to excite the @xmath4 = 43 emission and/or ( 2 ) the density is lower than the typical value traced in co ( @xmath4 = 43 ) .
however , the excitation temperature derived from the co ( @xmath4 = 10 ) observations at nanten is 28 k , the highest value among the c@xmath3o cores in the observed region .
the number density , 2800 @xmath17 , is also the highest .
moreover , the detection of h@xmath2co@xmath20 ( @xmath4 = 10 ) emission toward the core indicates that a high - density region of @xmath98 10@xmath18 @xmath17 exists within the core .
thus , the above two hypotheses fail to explain the properties of the core .
high - resolution observations of both low-@xmath4 and mid-@xmath4 co lines may unveil the properties of the core .
in this subsection , we will describe the characteristics of the star formation in the @xmath0 car gmc .
the proportion of massive - star forming cores , i.e. , cores with iras point sources with @xmath97 @xmath98 10@xmath13 @xmath19 , 13 % (= 2/15 ) , is much lower than in massive star forming regions at distances within a factor of two of the @xmath0 car gmc ; 28 % in s35/s37 and 86 % in centaurus iii @xcite .
this is consistent with the fact that the ratio of @xmath22/@xmath7 is large ( see @xmath99 [ sec_core ] ) , i.e. , most of the cores are not gravitationally bound .
the large line widths of the cores ( @xmath24 3 km s@xmath11 ) compared to the other gmcs ( @xmath24 2 km s@xmath11 , see @xmath99 [ sec_core ] ) also support this .
a large amount of turbulence in the @xmath0 car gmc may prevent star formation at the present time .
let us then consider the origin of the large turbulence in the @xmath0 car gmc .
the kinetic energy due to the turbulence is roughly estimated from @xmath121 = @xmath122 @xmath7 @xmath123 . by using the @xmath1co mass @xmath7 = 3.5 @xmath12 10@xmath44 @xmath14 , and the average line width of the c@xmath3o cores @xmath124 = 3 km s@xmath11 , @xmath121 @xmath24 3.4 @xmath12 10@xmath125 ergs is obtained as the kinematic energy of the @xmath0 car gmc .
the momentum @xmath126 = @xmath7 @xmath124 is estimated to be 1.0 @xmath12 10@xmath28 @xmath14 km s@xmath11 . in the absence of any mechanisms to prevent it
, the turbulence would dissipate on the crossing time - scale of the cloud , @xmath127 @xmath24 @xmath128/@xmath129 ( see vzquez - semadeni et al .
2000 for a review ) , which corresponds to @xmath24 15 myr in the case of the @xmath0 car gmc .
thus the presence of a large amount of turbulence means either that the turbulence is somewhat constantly injected by some mechanisms , or that a large amount of turbulence is still retained although the turbulence is dissipating because the gmc is young and/or the initial amount of the turbulence is large . in the followings , we examine four possibilities : ( 1 ) turbulence injection by mass outflows from young stellar objects , ( 2 ) by supernova remnants , ( 3 ) by stellar winds from massive stars , and ( 4 ) turbulence is pre - existing .
\(1 ) turbulence injection by mass outflows from young stellar objects : the kinetic energy and the momentum of such outflows range from @xmath130 = 10@xmath13110@xmath132 ergs and from @xmath126 = 0.11000 @xmath14 km s@xmath11 , respectively @xcite .
thus , even if we assume that all outflows have @xmath130 = 10@xmath132 ergs and @xmath126 = 1000 @xmath14 km s@xmath11 and that all the energy and momentum of such outflows are converted into turbulence ( i.e. , conversion coefficient @xmath133 = 1 ) at least 1000 ysos are needed to supply the present turbulence of the @xmath0 car gmc . here we estimate the contribution of outflows from low - mass ysos , since most of the stars are formed as clusters in massive star forming regions .
observationally , it is known that the kinetic energy and the momentum of outflows are correlated with the bolometric luminosity of the driving source , @xmath134 @xcite .
least - squares fits to their data yield @xmath126 = 10@xmath135 @xmath136 and @xmath130 = 10@xmath137 @xmath138 .
if we assume that the relation between the mass and the luminosity of ysos as @xmath97 = @xmath139 with @xmath110 = 3.45 @xcite and that the initial mass function ( imf ) of ysos as scalo s imf ( d@xmath10/d@xmath7 = @xmath140 , @xmath63 = 2.7 , * ? ? ?
* ) with the cutoff at 1 @xmath14 , 1.6 @xmath12 10@xmath44 ysos are needed to supply the present turbulence of the @xmath0 car gmc .
the number of ysos largely depends on the cutoff mass of the imf and the indices @xmath110 and @xmath63 , however , 3 @xmath12 10@xmath13 are still needed for the imf with the cutoff at 3 @xmath14 with @xmath110 = 4 and @xmath63 = 2 .
although these estimate are quite rough , the number required to supply the present turbulence of the @xmath0 car gmc is extremely large compared to the number of ysos formed in a single gmc in any assumption .
therefore , we can conclude that the turbulence injection by mass outflows from ysos are not important .
\(2 ) turbulence injection by supernova remnants : the kinetic energy and the initial speed of ejecta of a single supernova remnant are typically @xmath121 = 10@xmath141 ergs and @xmath77 = 10@xmath18 km s@xmath11 .
if we assume the mass of the ejecta to be @xmath7 = 3 @xmath14 , @xmath126 = 3 @xmath12 10@xmath18 @xmath14 km s@xmath11 can be provided by a single supernova . thus , even if we assume @xmath133 @xmath24 1 , at least @xmath2430 supernovae are needed to supply the present turbulence of the @xmath0 car gmc , which is too large when we consider the ages of the associated clusters , 37 myr ( e.g. , * ? ? ? * ) . the fact that no supernova remnants are known to exist @xcite supports this , although possible non - thermal radio emission is reported @xcite .
@xcite suggested the possible influence of the carina flare supershell , which may have been formed by more than 20 supernovae located 100 pc away from the @xmath0 car gmc .
the momentum supplied to the @xmath0 car gmc from the carina flare supershell through the cross section of the @xmath0 car gmc is estimated in the following manner : the total momentum of the carina flare supershell is assumed to be 20 times the typical momentum of a single supernova .
we also assume the depth of the gmc to be the same as the projected extent of the gmc , @xmath24 130 pc .
the momentum supplied through the cross section of the @xmath0 car gmc is estimated to be @xmath126 = 3 @xmath12 10@xmath18 @xmath12 20 @xmath12 ( 130 @xmath12 130)/(4@xmath142 @xmath12 100@xmath40 ) = 8
@xmath12 10@xmath18 @xmath14 km s@xmath11 on the assumption of @xmath133 = 1 , and thus it fails to explain the origin of the turbulence .
\(3 ) turbulence injection by stellar winds from massive stars : the terminal velocity of the stellar wind and the mass - loss rate from @xmath0 car are @xmath77 = 700 km s@xmath11 and @xmath143
= 5 @xmath12 10@xmath144 @xmath14 yr@xmath11 , respectively @xcite . during the time scale of 10@xmath28 yr
, @xmath0 car will be able to provide @xmath121 = 2 @xmath12 10@xmath141 ergs and @xmath126 = 3 @xmath12 10@xmath44 @xmath14 km s@xmath11 , which is comparable to the energy and momentum needed if the conversion efficiency is @xmath24 100 % .
however , the effect of the stellar wind will be confined to a small region around @xmath0 car ; a radius of @xmath24 25 pc is derived from the models of @xcite and @xcite by taking the typical density of a molecular cloud , 100 @xmath17 , and thus injection from @xmath0 car alone is not sufficient to explain the observed turbulence .
there are 7 wr stars and @xmath2480 o - type stars around the @xmath0 car gmc ( see fig .
[ fig02 ] ) .
a typical value of the terminal velocity of the stellar wind and the mass - loss rate of a wr star is @xmath77 @xmath24 2000 km s@xmath11 and @xmath143 = 4 @xmath12 10@xmath145 @xmath14 yr@xmath11 @xcite .
the total energy and momentum from these wr stars in 10@xmath28 yr are estimated to be @xmath121 = 1.1 @xmath12 10@xmath146 ergs and @xmath126 = 6 @xmath12 10@xmath44 @xmath14 km s@xmath11 , respectively .
for an o - type star , typical values are @xmath77 @xmath24 15002500 km s@xmath11 and @xmath143 @xmath24 10@xmath14510@xmath147 @xmath14 yr@xmath11 ( e.g. , * ? ? ?
over a time scale of 10@xmath28 yr , one o - type star will be able to provide @xmath121 @xmath24 2.3 @xmath12 [email protected] @xmath12 10@xmath149 ergs and @xmath126 @xmath24 1.5 @xmath12 [email protected] @xmath12 10@xmath18 @xmath14 km s@xmath11 .
thus about one hundred o stars can provide enough @xmath121 and @xmath126 on the assumption of @xmath133 = 1 .
it may therefore be possible that somewhat uniformly distributed wr stars and o - type stars account for the large turbulence , although the estimate of @xmath133 is highly uncertain .
\(4 ) turbulence is pre - existing : @xcite argued that massive cores are magnetically supercritical .
this conclusion is extended to low - mass cores by @xcite .
magnetically supercritical cores are supported by turbulence against the pressure of the surrounding medium @xmath150 .
when @xmath150 exceeds a critical value @xmath151 , the core can not be in magnetohydrostatic equilibrium and collapses .
@xcite suggested that dissipation of turbulence is the most important process in reducing @xmath151 . the condition @xmath151
@xmath92 @xmath150 can , however , be achieved without dissipating turbulence if @xmath150 is large .
thus , star formation in molecular clouds with large turbulence is possible .
here we investigate the possibility that the @xmath0 car gmc still retains a large amount of turbulence although that turbulence is dissipating , i.e. , the @xmath0 car gmc is not a relaxed system .
there are two possibilities ; the @xmath0 car gmc is young and/or it has an initially large turbulence .
the former may be inconsistent with the fact that the ages of the associated clusters are not extremely small , 37 myr ( e.g. , * ? ? ?
* ) , and thus we concentrate the latter possibility . concerning molecular cloud formation , various theories have been proposed so far ; cloud formation by spontaneous instabilities , by the compression of diffuse interstellar medium , and by random coalescence of existing clouds ( see elmegreen 1993 for a review ) .
the most realistic scenario may be a combination of the above processes .
the origin of the turbulence in molecular clouds is not yet understood , but possible formation mechanisms of a molecular cloud with large turbulence have been proposed : for example , @xcite investigated the propagation of a shock wave into the interstellar medium and suggested that the postshock region collapses into a cold layer through thermal instability .
their recent two - dimensional calculation shows that the layer breaks up into small cloudlets , in which a large amount of molecules are formed , in several myr and that the velocity dispersion of the cloudlets is typically several km s@xmath11 @xcite .
the fragments then coalesce and become larger clouds with supersonic velocity dispersions . in the case of the @xmath0 car gmc
, multiple supernova explosions may have contributed to a certain degree ; @xcite suggested that @xmath2420 supernova explosions may have occurred @xmath24 100 pc away from the @xmath0 car gmc over the last 2 @xmath12 10@xmath25 yr . these supernova explosions may have triggered the formation of the @xmath0 car gmc as well as the formation of massive stars within the gmc .
although the above estimate is quite rough , we suggest the possibility that a large amount of turbulence was supplied when the gmc was formed , and is now dissipating . to identify the formation mechanism for the @xmath0 car gmc is beyond the scope of the paper , but compression due to supernova explosions in the carina flare supershell may have played an important role .
turbulent injection from stellar winds , supernova remnants , and outflows originating in the stars formed within the gmc may also contribute , supplying turbulence to some degree .
we carried out an unbiased survey for massive dense cores in the giant molecular cloud associated with @xmath0 carinae with the nanten telescope in @xmath1co , @xmath2co , and c@xmath3o @xmath4 = 10 emission lines .
the purpose of the present study is to obtain a sample of massive-_starless_-core candidates which are possible sites for future massive star formation . with moderate spatial resolution , 2@xmath367 ,
our observations covered the entire @xmath0 car gmc ( 3 @xmath12 2 deg@xmath40 ) including regions farther from @xmath0 car , where no previous observations had been made except coarse large - scale surveys .
we also made high - resolution observations in h@xmath2co@xmath20 ( @xmath4 = 10 ) with the mopra 22-m millimeter telescope and searched for molecular outflows in @xmath1co ( @xmath4 = 21 ) toward selected regions with the aste 10-m sub - millimeter telescope .
the main results are summarized as follows :
\(1 ) we identified 15 c@xmath3o cores , whose typical line width @xmath5 , radius @xmath6 , mass @xmath7 , column density @xmath8(h@xmath9 ) , and average number density @xmath10(h@xmath9 ) were 3.3 km s@xmath11 , 2.2 pc , 2.6@xmath1210@xmath13 @xmath14 , 1.3@xmath1210@xmath15 @xmath16 , and 1.2@xmath1210@xmath13 @xmath17 , respectively .
\(2 ) two of the 15 cores are associated with iras point sources whose luminosities are larger than 10@xmath18 @xmath19 , which indicates that massive star formation is occuring within these cores .
five cores including the two with iras sources are associated with msx point sources .
we detected h@xmath2co@xmath20 ( @xmath4 = 10 ) emission toward 4 c@xmath3o cores , two of which are associated with iras and msx point sources , another one is associated only with an msx point source , and the other is associated with neither iras nor msx point sources .
the core with neither iras nor msx point sources shows the presence of a bipolar molecular outflow in @xmath1co ( @xmath4 = 21 ) , which indicates that star formation is also occuring in the core , and the other three of the four h@xmath2co@xmath20 detections show wing - like emission . in total , six c@xmath3o cores out of 15 (= 40@xmath21 ) are experienced star formation , and at least 2 of 15 (= 13 @xmath21 ) are massive - star forming cores in the @xmath0 car gmc .
\(3 ) we found that massive star formation occurs preferentially in cores with larger @xmath8(h@xmath9 ) , @xmath7 , @xmath10(h@xmath9 ) , and smaller ratio of @xmath22/@xmath7 .
we also found that the cores in the @xmath0 car gmc are characterized by large @xmath23 and @xmath22/@xmath7 on average compared to the cores in other gmcs observed with the same telescope .
these properties of the cores may account for the fact that as much as 6087 % of the cores do not show any signs of massive star formation .
\(4 ) we investigated the origin of a large amount of turbulence in the @xmath0 car gmc .
we found that turbulence injection from stellar winds , molecular outflows , and supernova remnants which originated from stars formed within the gmc , are not enough to explain the existing turbulence .
we propose the possibility that the large turbulence was pre - existing when the gmc was formed , and is now dissipating .
mechanisms such as multiple supernova explosions in the carina flare supershell may have contributed to form a gmc with a large amount of turbulence . the nanten project ( southern 4-meter radio telescope )
is based on a mutual agreement between nagoya university and carnegie institution of washington .
we greatly appreciate the hospitality of all staff members of the las campanas observatory of the carnegie institution of washington .
we also acknowledge that this project could be realized by the contribution from many japanese public donators and companies .
aste ( atacama submillimeter telescope experiment ) is a joint project between japan and chile .
the telescope is operated by the aste team including naoj , university of tokyo , nagoya university , osaka prefecture university , and universidad de chile .
we are grateful to all the members of the aste team .
we are grateful to the atnf for thier hospitality , and to the staff of mopra observatory , especially stuart robertson , for his help in observations .
we are also grateful to peter barnes for his help in observations .
the australia telescope is funded by the commonwealth of australia for operation as a national facility managed by csiro .
this work made use of the southern h - alpha sky survey atlas ( shassa ) , which is supported by the national science foundation .
the digitized sky surveys were produced at the space telescope science institute under u.s .
government grant nagw-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 are also grateful to akiko kawamura , yoshiaki moriguchi , masanori nakagawa , and joanne dawson for their helpful comments .
this work was financially supported in part by grants - in - aid for scientific research ( kakenhi ) from the ministry of education , culture , sports , science , and technology of japan ( mext ) and japan society for the promotion of science ( jsps ) ; nos .
14102003 , 14403001 , 15071202 , and 15071205 .
allen , c. w. 1972 , astronomical quantities ( london : athlone ) alves , j. & homeier , n. 2003 , , 589 , l45 andr , p. , word - thompson , d. , & barsony , m. 1993 , 406 , 122 aoyama , h. , mizuno , n. , yamamoto , h. , onishi , t. , mizuno , a. , & fukui , y. 2001 , , 53 , 1053 bergin , e. a , goldsmith , p. f. , snell , r. l. , & langer , w. d. 1997 , , 482 , 285 bertoldi , f. & mckee , c. f. 1992 , , 395 , 140 bertsch , d. l. , dame , t. m. , fichtel , c. e. , hunter , s. d. , sreekumar , p. , stacy , j. g. , & thaddeus , p. 1993
, , 416 , 587 beuther , h. , schilke , p. , menten , k. m. , motte , f. , sridharan , t. k. , & wyrowski , f. 2002 , , 566 , 945 beuther , h. , schilke , p. , sridharan , t. k. , menten , c. m. walmsley , & wyrowski , f. 2002 , , 383 , 892 bloemen j. b. g. m. , et al .
1986 , , 154 , 25 van boekel , r. , et al .
2003 , , 410 , l37 bonnell , i. a. , bate , m. r. , & zinnecker , h. 1998 , , 298 , 93 bronfman , l. , nyman , l. - .
, & may , j. 1996 , , 115 , 81 brooks , k. 2000 , phd thesis , univ .
new south wales brooks , k. j. , cox , p. , schneider , n. , storey , j. w. v. , poglitsch , a. , geis , n. , & bronfman , l. 2003 , , 412 , 751 brooks , k. j. , whiteoak , j. b. , & storey , j. w. v. 1998 , , 15 , 202 cabrit , s. & bertout , c. 1992 , , 261 , 274 casseli , p. & myers , p. c. 1995 , , 446 , 665 castor , j. , mccray , r. , & weaver , r. 1975 , , 200 , 107 cox , p. & bronfman , l. 1995 , , 299 , 583 davidson , k. & humphreys , r. m. 1997 , , 35 , 1 de graauw , t. , lidholm , s. , fitton , b. , beckman , j. , israel , f. p. , nieuwenhuijzen , h. , & vermue , j. 1981 , , 102 , 257 dickman , r. l. 1978 , , 37 , 407 dobashi , k. , bernard , j. p. , & fukui , y. 1996 , , 466 , 282 dobashi , k. , bernard , j. p. , yonekura , y. , & fukui , y. 1994 , , 95 , 419 drissen , l. , moffat , a. f. j. , walborn , n. r. , & shara , m. m. 1995 , , 110 , 2235 elmegreen , b. g. 1993 , , in protostars and planets iii , ed . e. h. levy & j. i. lunine ( tucson : univ .
arizone press ) , 97 evans , n. j. , ii .
1999 , , 37 , 311 evans , n. j. , ii , shirley , y. l. , mueller , k. e. , & knez , c. 2002 , in asp conf .
267 , the earliest stages of massive star birth , ed .
crowther ( san francisco : asp ) , 17 ezawa , h. , kawabe , r. , kohno , k. , and yamamoto , s. 2004 , proc .
spie , 5489 , 763 fandez , s. , bronfman , l. , garay , g. , chini , r. , may , j. , & nyman , l. .
2004 , , in press feinstein , a. 1995 , , 2 , 57 forbrich , j. , schreyer , k. , posselt , b. , klein , r. , & henning , th .
2004 , , 602 , 843 frerking , m. a. , langer , w. d. , & wilson , r. w. 1982 , , 262 , 590 fukui , y. 1989 , proc .
eso workshop on low mass star formation and pre - main sequence objects , ed .
b. reipurth ( eso : garching ) , 95 fukui , y. , iwata , t. , mizuno , a. , bally , j. , & lane , a. p. 1993 , in protostars and planets iii , ed .
e. h. levy & j. i. lunine ( tucson : univ .
arizone press ) , 603 fukui , y. , ogawa , h. , kawabata , k. , mizuno , a. , & sugitani , k. 1991 , in iau symp .
148 , the magellanic clouds , ed .
r. haynes & d. mline ( dordrecht : kluwer ) , 105 fukui , y. , onishi , t. , abe , r. , kawamura , a. , tachihara , k. , yamaguchi , r. , mizuno , a. , & ogawa , h. 1999 , , 51 , 751 fukui , y. & sakakibara , o. 1992 , mitsubishi electric advance , 60 , 11 fukui , y. & yonekura , y. 1998 , in iau symp . 179 , new horizons from multi - wavelength sky surveys , ed . b. j. mclean , d. a. golombek , j. j. e. hayes , & h. e. payne ( dordrecht : kluwer ) , 165 garay , g. , fandez , s. , mardones , d. , bronfman , l. , chini , r. , & nyman , l. - .
2004 , , 610 , 313 gaustad , j. e. , mccullough , p. r. , rosing , w. , & van buren , d. 2001 , , 113 , 1326 goldsmith , p. f. & langer w. d. 1978 , , 222 , 881 goodman , a. a. , barranco , j. a. , wilner , d. j. , & heyer , m. h. 1998 , , 504 , 223 grabelsky , d. a. , cohen , r. s. , bronfman , l. , & thaddeus , p. 1988
, , 331 , 181 grabelsky , d. a. , cohen , r. s. , bronfman , l. , thaddeus , p. , & may , j. 1987 , , 315 , 122 hara , a. , tachihara , k. , mizuno , a. , onishi , t. , kawamura , a. , obayashi , a. , & fukui , y. 1999 , , 51 , 895 harju , j. , walmsley , c. m. , & wouterloot , j. g. a. 1993 , , 98 , 51 van der hucht , k. a. 2001 , , 45 , 135 hunter , s. d. , et al .
1997 , , 481 , 205 jijina , j. , myers , p. c. , & adams , f. c. 1999 , , 125 , 161 johansson , l. e. b. , andersson , c. , elldr , p. , friberg , p. , hjalmarson , .
, hglund , b. , irvine , w. m. , olofsson , h. , & rydbeck , g. 1984 , , 130 , 227 jones , b. b. 1973 , aust . j. phys . , 26 , 545 jrgensen , j. k. , schier , f. l. , & van dishoeck , e. f. 2004 , , 416 , 603 juvela , m. 1996 , , 118 , 191 kato , s. , mizuno , n. , asayama , s. , mizuno , a. , ogawa , h. , & fukui , y. 1999 , , 51 , 883 kawamura , a. , onishi , t. , yonekura , y. , dobashi , k. , mizuno , a. , ogawa , h. , & fukui y. 1998 , , 117 , 387 kim , b. g. , kawamura , a. , yonekura , y. , & fukui , y. 2004 , , 56 , 313 kohno , k. , et al .
2003 , in springer proceedings in physics 91 , the dense interstellar medium in galaxies , ed .
s. pfalzner , c. kramer , c. staubmeier , & a. heithausen ( berlin : springer ) , 349 koyama , h. , & inutsuka , s. 2000 , , 532 , 980 koyama , h. , & inutsuka , s. 2002 , , 564 , l97 kurtz , s. , cesaroni , r. , churchwell , e. , hofner , p. , & walmsley , c. m. 2000 , protostars and planets iv , ed . v. mannings , a. boss , & s. russel ( tucson : univ . of arizona press ) , 299 kutner , m. l. & ulich b. l. 1981 , , 250 , 341 lada , e. a. , bally , j. , & stark , a. a. 1991 , , 368 , 432 ladd , e. f. , myers , p. c. , & goodman , a. a. 1994 , , 433 , 117 ladd , n. , purcell , c. , wong , t. , & robertson , s. 2005 , , 22 , 62 levreault , r. m. 1988 , , 67 , 283 mckee , c. f. & tan , j. c. 2003 , , 585 , 850 maz - apellniz , j. , walborn , n. r. , galu , h. . , & wei , l. h. 2004 , , 151 , 103 markova , n. , puls , j. , repolust , t. , & markov , h. 2004 , , 413 , 693 megeath , s. t. , cox , p. , bronfman , l. , & roelfsema , p. r. 1996 , , 305 , 296 mizuno , a. , et al .
1998 , , 507 , l83 mizuno , a. , onishi , t. , hayashi , m. , ohashi , n. , sunada , k. , hasegawa , t. , & fukui , y. 1994 , , 368 , 719 mizuno , a. , onishi , t. , yonekura , y. , nagahama , t. , ogawa , h. , & fukui , y. 1995 , , 445 , l161 mizuno , a. , et al .
1999 , , 51 , 859 mueller , k. e. , shirley , y. l. , evans ii , n. j. , & jacobson , h. r. 2002 , , 143 , 469 myers , p. c. , ladd , e. f. , & fuller , g. a. 1991 , , 372 , l95 nagahama t. 1997 , phd thesis , nagoya university nakano , t. 1998 , , 494 , 587 nozawa , s. , mizuno , a. , teshima , y. , ogawa , h. , & fukui , y. 1991 , , 77 , 647 obayashi , a. , kun , m. , sato , f. , yonekura , y. , & fukui , y. 1998 , , 115 , 274 ogawa , h. , mizuno , a. , hoko , h. , ishikawa , h. , & fukui , y. 1990 , int .
j. infrared millimeter waves , 11 , 717 onishi , t. , mizuno , a. , kawamura , a. , ogawa , h. , & fukui , y. 1996 , , 465 , 815 onishi , t. , mizuno , a. , kawamura , a. , ogawa , h. , & fukui , y. 1998 , , 502 , 296 onishi , t. , mizuno , a. , kawamura , a. , tachihara , k. , & fukui , y. 2002 , , 575 , 950 onishi , t. , et al . 1999 , , 51 , 871 plume , r. , jaffe , d. t. , & evans ii , e. j. 1992 , , 78 , 505 plume , r. , jaffe , d. t. , evans ii , e. j. , martn - pintado , j. , & gmez - gonzlez , j. 1997 , , 476 , 730 rathborne , j. m. , brooks , k. j. , burton , m. g. , cohen , m. , & bontemps , s. 2004 , , 418 , 563 saito , h. , mizuno , n. , moriguchi , y. , matsunaga , k. , onishi , t. , mizuno , a. , & fukui , y. 2001 , , 53 , 1037 saito , h. , tachihara , k. , onishi , t. , yamaguchi , n. , mizuno , n. , mizuno , a. , ogawa , h. , & fukui , y. 1999 , , 51 , 819 scalo , j. m. 1986 , fund .
cosmic phys . , 11 , 1 sekimoto , y. , et al .
2001 , , 53 , 951 seward , f. d. , forman , w. r. , giacconi , r. , griffiths , r. e. , harnden jr .
, f. r. , jones , c. , & pye , j. p. 1979
, , 234 , l55 shepherd , d.s . &
churchwell , e. 1996 , , 472 , 225 shirley , y. l. , evans ii , n. j. , young , k. e. , knez , c. , & jaffe , d. t. 2003 , , 149 , 375 shu , f. h. , adams , f. c. , & lizano , s. 1987 , , 25 , 23 smith , n. , davidson , k. , gull , t. r. , ishibashi , k. , & hillier , d. j. 2003 , , 586 , 432 smith , n. , egan , m. p. , carey , s. , price , s. d. , morse , j. a. , & price , p. a. 2000 , , 532 , l145 solomon , p. m. , rivolo , a. r. , barrett , j. , & yahil , a. 1987 , , 319 , 730 sorai , k. , sunada , k. , okumura , s. k. , iwasa , t. , tanaka , a. , natori , n. , & onuki , h. 2000 , proc .
spie , 4015 , 86 sridharan , t. k. , beuther , h. , schilke , p. , menten , k. m. , & wyrowski , f. 2002 , , 566 , 931 stacy , j. g. , benson , p. j. , myers , p. c. , & goodman , a. a. 1988 , in interstellar matter : proc .
2d haystack obs .
meeting , ed .
j. m. moran & p. t. p. ho ( new york : gordon & breach ) , 179 strong , a. w. , et al .
1988 , , 207 , 1 sugitani , k. , fukui , y. , mizuno , a. , & ohashi , n. 1989 , , 342 , l87 sutton , e. c. , blake , g. a. , masson , c. r. , & phillips , t. g. 1985 , , 58 , 341 tachihara , k. , mizuno , a. , & fukui , y. 2000 , , 528 , 817 tachihara , k. , onishi , t. , mizuno , a. , & fukui , y. 2002 , , 385 , 909 tateyama , c. e. , strauss , f. m. , & kaufmann , p. 1991
, , 249 , 716 ulich , b. l. & haas , r. w. 1976 , , 30 , 247 vzquez - semadeni , e. , ostriker , e. c. , passot , t. , gammie , c. f. , & stone , j.m .
2000 , protostars and planets iv , ed . v. mannings , a. boss , & s. russel ( tucson : univ . of arizona press ) , 3 warin , s. , benayoun , j. j. , & viala , y. p. 1996
, , 308 , 535 weaver , r. , mccray , r. , castor , j. , shapiro , p. , & moore , r. 1977 , , s218 , 377 whiteoak , j. b. z. 1994 , , 429 , 225 wouterloot , j. g. a. , walmsley , c. m. , & henkel , c. 1988 , , 203 , 367 wu , y. , wei , y. , zhao , m. , shi , y. , yu , w. , qin , s. , & huang , m. 2004 , , 426 , 503 yamaguchi , n. , mizuno , n. , saito , h. , matsunaga , k. , mizuno , a. , ogawa , h. , & fukui , y. 1999a , , 51 , 775 yamaguchi , r. , saito , h. , mizuno , n. , mine , y. , mizuno , a. , ogawa , h. , & fukui , y. 1999b , , 51 , 791 yonekura , y. , dobashi , k. , hayashi , y. , sato , f. , ogawa , h. , & fukui , y. 1998 , , 115 , 2009 yonekura , y. , dobashi , k. , mizuno , a. , ogawa , h. , & fukui , y. 1997 , , 110 , 21 yonekura , y. , mizuno , n. , saito , h. , mizuno , a. , ogawa , h. , & fukui , y. 1999 , , 51 , 911 yu , z .- a . ,
nagahama , t. , & fukui , y. 1996 , , 471 , 867 zhang , q. , hunter , t.r . , brand , j. , sridharan , t.k . , molinari , s. , kramer , m.a .
, & cesaroni , r. 2001a , , 552 , l167 zhang , x. , lee , y. , bolatto , a. , & stark , a. a. 2001b , , 553 , 274 zinchenko , i. , mattila , k. , & toriseva , m. 1995 , , 111 , 95 zinchenko , i. , pirogov , l. , & toriseva , m. 1998 , , 133 , 337 rrrrrrrrccrll 1 & 287.800 & 288.400 & @xmath531.467 & @xmath530.933 & 2.2 & 1.5 & 1.07 & 0.68 & 0.72 & 20 & 13 , 15 & giant pillar + 2 & 287.600 & 287.900 & @xmath530.900 & @xmath530.400 & 1.7 & 0.8 & 0.28 & 0.44 & 0.37 & 20 & 12 & @xmath0 car , southern cloud + 3 & 286.867 & 287.567 & @xmath531.000 & 0.067 & 9.8 & 3.8 & 0.80 & 0.39 & 0.21 & 23 & 8 , 9 , 10 , 11 & @xmath0 car , northern cloud + 4 & 286.033 & 286.967 & @xmath530.800 & @xmath530.033 & 8.0 & 3.0 & 1.00 & 0.38 & 0.33 & 21 & 6 , 7 & gum 31 + 5 & 285.800 & 286.533 & @xmath530.200 & 0.333 & 6.6 & 3.2 & 2.29 & 0.48 & 0.72 & 21 & 2 , 3 , 4 , 5 & gum 31 , ngc 3293 + 6 & 285.700 & 286.000 & @xmath530.800 & @xmath530.300 & 2.0 & 0.7 & 0.27 & 0.36 & 0.38 & 16 & 1 & + 7 & 287.933 & 288.200 & @xmath530.900 & @xmath530.600 & 0.4 & 0.2 & 0.06 & 0.48 & 0.31 & 9 & 14 & + total & & & & & 34.6 & 13.2 & 5.78 & 0.38 & 0.44 & & & + rrrrrcccrrc 1 & 285.967 & @xmath530.500 & 10 34 23.6 & @xmath5358 46 56 & 0.6 & @xmath5316.6 & 2.4 & 1.6 & 9.5 & 0.17 + 2 & 286.100 & @xmath530.133 & 10 36 41.1 & @xmath5358 31 50 & 0.3 & @xmath5320.7 & 4.5 & 1.4 & 10.6 & 0.14 + 3 & 286.100 & 0.233 & 10 38 03.5 & @xmath5358 12 40 & 0.5 & @xmath5320.6 & 2.6 & 1.4 & 16.5 & 0.09 + 4 & 286.167 & 0.033 & 10 37 45.3 & @xmath5358 25 06 & 0.4 & @xmath5321.6 & 2.6 & 1.2 & 4.3 & 0.28 + 5 & 286.233 & 0.167 & 10 38 41.8 & @xmath5358 20 04 & 1.2 & @xmath5320.0 & 2.9 & 3.6 & 17.6 & 0.20 + 6 & 286.400 & @xmath530.233 & 10 38 19.0 & @xmath5358 45 54 & 0.5 & @xmath5324.1 & 3.1 & 1.9 & 9.3 & 0.20 + 7 & 286.433 & @xmath530.400 & 10 37 54.6 & @xmath5358 55 36 & 0.6 & @xmath5313.0 & 2.5 & 1.5 & 13.4 & 0.11 + 8 & 287.133 & @xmath530.833 & 10 41 03.7 & @xmath5359 38 42 & 0.3 & @xmath5319.9 & 3.8 & 1.2 & 12.2 & 0.10 + 9 & 287.233 & @xmath530.533 & 10 42 53.0 & @xmath5359 25 44 & 0.5 & @xmath5317.6 & 2.6 & 1.2 & 9.4 & 0.13 + 10 & 287.333 & @xmath530.600 & 10 43 19.7 & @xmath5359 32 05 & 0.3 & @xmath5318.7 & 4.0 & 1.2 & 12.7 & 0.10 + 11 & 287.333 & @xmath530.433 & 10 43 56.8 & @xmath5359 23 16 & 0.3 & @xmath5319.3 & 5.4 & 1.5 & 17.1 & 0.09 + 12 & 287.700 & @xmath530.733 & 10 45 24.0 & @xmath5359 49 28 & 0.3 & @xmath5325.8 & 3.9 & 1.1 & 14.1 & 0.08 + 13 & 287.967 & @xmath531.133 & 10 45 48.2 & @xmath5360 18 09 & 0.7 & @xmath5321.3 & 2.8 & 2.3 & 17.2 & 0.13 + 14 & 288.100 & @xmath530.700 & 10 48 21.3 & @xmath5359 58 44 & 0.5 & @xmath5312.1 & 2.2 & 1.1 & 3.4 & 0.32 + 15 & 288.233 & @xmath531.133 & 10 47 43.0 & @xmath5360 25 31 & 0.4 & @xmath5317.8 & 3.5 & 1.3 & 9.7 & 0.14 + rrccrrrrcrcccc 1 & 16 & 10.6 & 7.4 & 2.0 & 1.8 & 2100 & 900 & 2.7 & 3000 & 1.4 & & & + 2 & 15 & 9.2 & 6.9 & 3.1 & 2.9 & 4600 & 540 & 3.2 & 6500 & 1.4 & & & + 3 & 21 & 11.1 & 8.4 & 2.7 & 2.5 & 4400 & 750 & 3.3 & 6100 & 1.4 & & & + 4 & 20 & 9.1 & 7.3 & 1.4 & 1.0 & 1000 & 1200 & 2.6 & 1900 & 1.9 & & & + 5 & 28 & 34.8 & 25.7 & 1.8 & 1.6 & 6100 & 3400 & 2.8 & 2900 & 0.5 & 1 & 1 & 1 + 6 & 26 & 17.3 & 12.8 & 2.0 & 1.8 & 3700 & 1600 & 3.6 & 5300 & 1.5 & 1 & 3 & 2 + 7 & 17 & 10.7 & 8.6 & 1.6 & 1.3 & 1600 & 1300 & 3.1 & 3200 & 2.0 & & & + 8 & 24 & 10.1 & 7.0 & 2.7 & 2.5 & 3700 & 630 & 3.6 & 7500 & 2.0 & & & + 9 & 21 & 9.7 & 7.3 & 1.6 & 1.3 & 1400 & 1100 & 2.5 & 2200 & 1.6 & & 1 & 3 + 10 & 25 & 11.0 & 11.0 & 0.8 & & 520 & 3300 & 4.0 & 2700 & 5.2 & & 2 & + 11 & 22 & 11.8 & 10.1 & 1.4 & 1.0 & 1400 & 1700 & 4.5 & 6100 & 4.3 & & & + 12 & 20 & 8.8 & 6.5 & 2.3 & 2.1 & 2600 & 710 & 3.5 & 5900 & 2.3 & & 1 & + 13 & 20 & 17.4 & 12.8 & 2.2 & 1.9 & 4200 & 1400 & 4.9 & 11000 & 2.6 & & & + 14 & 9 & 5.9 & 4.8 & 1.2 & 0.6 & 450 & 1000 & 1.9 & 900 & 2.0 & & & + 15 & 19 & 9.9 & 7.0 & 2.0 & 1.8 & 2000 & 850 & 2.9 & 3500 & 1.8 & & & 4 + average & 20 & 12.5 & 9.6 & 1.9 & 1.7 & 2600 & 1400 & 3.3 & 4600 & 2.1 + median & 20 & 10.6 & 7.4 & 2.0 & 1.8 & 2100 & 1100 & 3.2 & 3500 & 1.9 + lccclcccccc orion b & 400 & 19 & 18 & 1.7 ( 1.7 ) & 2.1 ( 1.6 ) & 250 ( 160 ) & 0.5 ( 0.5 ) & 7.6 ( 5.7 ) & 2.1 ( 1.4 ) & 1 + orion a & 480 & 19 & 12 & 1.7 ( 1.6 ) & 2.4 ( 1.7 ) & & & & & 2 + vela c & 700 & 13 & 10 & 2.5 ( 2.3 ) & 2.8 ( 2.8 ) & & & & & 3 + cepheus ob3 & 730 & 8 & 8 & 2.0 ( 1.9 ) & 1.5 ( 1.4 ) & & & & & 4 + s35/37 & 1800 & 18 & 4 & 1.7 ( 1.6 ) & 1.7 ( 1.1 ) & 1100 ( 640 ) & 1.3 ( 1.3 ) & 11.0 ( 9.2 ) & 0.5 ( 0.6 ) & 5 + @xmath0 car gmc & 2500 & 15 & 11 & 3.2 ( 2.9 ) & 1.3 ( 1.1 ) & 2600 ( 2100 ) & 2.2 ( 2.0 ) & 1.2 ( 0.9 ) & 1.6 ( 1.3 ) & 6 + centaurus iii & 3500 & 14 & 10 & 2.3 ( 2.3 ) & 1.3 ( 1.3 ) & 5400 ( 4700 ) & 2.8 ( 2.8 ) & 1.1 ( 1.0 ) & 0.8 ( 0.6 ) & 7 + centaurus ii & 4500 & 5 & 5 & 3.1 ( 3.0 ) & 1.9 ( 1.9 ) & 20000 ( 24000 ) & 4.2 ( 3.8 ) & 1.1 ( 0.7 ) & 0.6 ( 0.5 ) & 7 + centaurus i & 5300 & 16 & 11 & 2.3 ( 2.2 ) & 1.3 ( 1.0 ) & 11000 ( 8600 ) & 4.0 ( 3.6 ) & 0.7 ( 0.7 ) & 0.5 ( 0.5 ) & 7 + rrrrrrrrrrrccr 10365@xmath535803 & 286.203 & 0.170 & 10 38 30.6 & @xmath5358 19 00 & & 7.242 & 85.64 & 1173 & 2782 & 30000 & 1333 & caba & 5 + 10361@xmath535830 & 286.375 & @xmath530.255 & 10 38 03.8 & @xmath5358 46 18 & & 12.38 & 38.38 & 625.7 & 2156 & 21000 & 1333 & dcbd & 6 + clrrrrrrrrrccr 1 & msx6c g286.2086 + 00.1694 & 286.2086 & 0.1694 & 10 38 32.5 & @xmath5358 19 12 & & 1.3532 & 2.8816 & 7.1817 & 40.572 & 4444 & 5 + 2 & msx6c g286.3579 - 00.2933 & 286.3579 & @xmath530.2933 & 10 37 48.4 & @xmath5358 47 48 & & 0.71259 & 1.8153 & 2.6774 & 6.0653 & 4344 & 6 + 3 & msx6c g286.3747 - 00.2630 & 286.3747 & @xmath530.2630 & 10 38 02.0 & @xmath5358 46 43 & & 3.5907 & 4.7559 & 2.4093 & 7.5775 & 4444 & 6 + 4 & msx6c g286.3773 - 00.2563 & 286.3773 & @xmath530.2563 & 10 38 04.6 & @xmath5358 46 26 & & 1.6278 & 2.9183 & 3.8554 & 12.095 & 4444 & 6 + 5 & msx6c g287.2238 - 00.5339 & 287.2238 & @xmath530.5339 & 10 42 48.9 & @xmath5359 25 29 & & 1.2040 & 1.6909 & 1.5151 & 4.3560 & 4344 & 9 + 6 & msx6c g287.3399 - 00.6008 & 287.3399 & @xmath530.6008 & 10 43 22.3 & @xmath5359 32 19 & & 0.80028 & 2.9386 & 5.9723 & 12.516 & 4444 & 10 + 7 & msx6c g287.3506 - 00.5900 & 287.3506 & @xmath530.5900 & 10 43 29.2 & @xmath5359 32 03 & & 0.55922 & 2.0313 & 3.9602 & 7.9096 & 4444 & 10 + 8 & msx6c g287.6393 - 00.7219 & 287.6393 & @xmath530.7219 & 10 45 00.9 & @xmath5359 47 10 & & 3.4863 & 4.2647 & 3.6021 & 17.835 & 4444 & 12 + cccccccccccrrccc 1 & 10 38 31.7 & @xmath5358 18 44 & 0.38 & @xmath5319.4 & 2.9 & 30 & 0.018 & 4.5 & 0.77 & 3.8 & 1400 & 2300 & 3.4 & 1.1 & 5 + 2 & 10 37 52.9 & @xmath5358 47 20 & 0.41 & @xmath5322.5 & 3.0 & 55 & 0.011 & 7.6 & 0.27 & 3.0 & 400 & 500 & 7.6 & 6.8 & 6 + 3 & 10 42 47.8 & @xmath5359 26 24 & 0.35 & @xmath5317.3 & 1.3 & 35 & 0.017 & 2.0 & & & & & & & 9 + 4 & 10 47 46.7 & @xmath5360 26 26 & 0.81 & @xmath5317.8 & 1.6 & 30 & 0.035 & 5.0 & 0.39 & 1.6 & 470 & 200 & 4.4 & 2.8 & 15 + lclcccccc cores with iras point sources & a & 5 , 6 & 3.2 ( 3.2 ) & 26.1 ( 26.1 ) & 4900 ( 4900 ) & 1.9 ( 1.9 ) & 2.5 ( 2.5 ) & 1.0 ( 1.0 ) + other star - forming cores & b & 9 , ( 10 ) , 12 , 15 & 3.2 ( 3.2 ) & 9.9 ( 9.8 ) & 1600 ( 1700 ) & 2.0 ( 2.0 ) & 0.9 ( 0.9 ) & 1.9 ( 1.8 ) + all star - forming cores & a+b & 5 , 6 , 9 , ( 10 ) , 12 , 15 & 3.2 ( 3.2 ) & 15.3 ( 10.5 ) & 2700 ( 2300 ) & 2.0 ( 2.0 ) & 1.5 ( 1.1 ) & 1.5 ( 1.6 ) + non - star - forming cores & c & 1 , 2 , 3 , ( 4 ) , 7 , 8 , ( 11 ) , 13 , ( 14 ) & 3.3 ( 3.2 ) & 10.7 ( 10.6 ) & 2600 ( 2100 ) & 2.4 ( 2.4 ) & 0.9 ( 0.8 ) & 1.8 ( 1.7 ) + all c@xmath3o cores & & all & 3.3 ( 3.2 ) & 12.5 ( 10.6 ) & 2600 ( 2100 ) & 2.2 ( 2.0 ) & 1.2 ( 0.9 ) & 1.7 ( 1.6 ) + | we carried out an unbiased survey for massive dense cores in the giant molecular cloud associated with @xmath0 carinae with the nanten telescope in @xmath1co , @xmath2co , and c@xmath3o @xmath4 = 10 emission lines .
we identified 15 c@xmath3o cores , whose typical line width @xmath5 , radius @xmath6 , mass @xmath7 , column density @xmath8(h@xmath9 ) , and average number density @xmath10(h@xmath9 ) were 3.3 km s@xmath11 , 2.2 pc , 2.6@xmath1210@xmath13 @xmath14 , 1.3@xmath1210@xmath15 @xmath16 , and 1.2@xmath1210@xmath13 @xmath17 , respectively .
two of the 15 cores are associated with iras point sources whose luminosities are larger than 10@xmath18 @xmath19 , which indicates that massive star formation is occuring within these cores .
five cores including the two with iras sources are associated with msx point sources .
we detected h@xmath2co@xmath20 ( @xmath4 = 10 ) emission toward 4 c@xmath3o cores , two of which are associated with iras and msx point sources , another one is associated only with an msx point source , and the other is associated with neither iras nor msx point sources .
the core with neither iras nor msx point sources shows the presence of a bipolar molecular outflow in @xmath1co ( @xmath4 = 21 ) , which indicates that star formation is also occuring in the core , and the other three of the four h@xmath2co@xmath20 detections show wing - like emission . in total , six c@xmath3o cores out of 15 (= 40@xmath21 ) are experienced star formation , and at least 2 of 15 (= 13 @xmath21 ) are massive - star forming cores in the @xmath0 car gmc .
we found that massive star formation occurs preferentially in cores with larger @xmath8(h@xmath9 ) , @xmath7 , @xmath10(h@xmath9 ) , and smaller ratio of @xmath22/@xmath7 .
we also found that the cores in the @xmath0 car gmc are characterized by large @xmath23 and @xmath22/@xmath7 on average compared to the cores in other gmcs observed with the same telescope .
these properties of the cores may account for the fact that as much as 6087 % of the cores do not show any signs of massive star formation .
we investigated the origin of a large amount of turbulence in the @xmath0 car gmc .
we found that turbulence injection from stellar winds , molecular outflows , and supernova remnants which originated from stars formed within the gmc , are not enough to explain the existing turbulence .
we propose the possibility that the large turbulence was pre - existing when the gmc was formed , and is now dissipating .
mechanisms such as multiple supernova explosions in the carina flare supershell may have contributed to form a gmc with a large amount of turbulence . |
coupling resonant rf radiation to the spins of atoms in inhomogeneous magnetic fields allows to locally address clouds of ultracold atoms with high spatial resolution .
this technique is commonly used in combination with atom chips , where it offers huge flexibility in the types of potentials that can be created @xcite .
since the rf fields can also be time dependent , this technique offers the possibility to control the atoms dynamically and has , for example , been used to split bose - einstein condensates ( becs ) @xcite . by today
significant work has been done in the modelling of rf systems to create highly non trivial trapping geometries @xcite , including annular traps @xcite , examining effects beyond the rotating wave approximation @xcite and investigating non - linear zeeman effects @xcite .
the use of multi - frequency fields has recently been suggested to create multi - well or even periodic potentials @xcite .
these can also be time dependent and in @xcite a method to coherently control the centre of mass motion of both , a single atom and a cloud of interacting atoms , was suggested .
while this work highlighted how versatile multi - frequency rf potentials are for experimentally realistic parameters , the theoretical model used could only describe limited tunnel coupling strengths between two potential wells , as it required all radio frequencies to be well separated . for large tunnel couplings , however , it is usually necessary to have potentials where the curvature can be controlled over short distances , which , using rf technology , requires two frequencies to be brought close . in such situations
the effects of both frequencies have to be considered simultaneously and a more advanced model is necessary . in this work
we address the question of how to accurately deal with near degenerate frequencies and take full advantage of all degrees of freedom offered by rf fields by using floquet theory and comparing the results to the currently most common model @xcite .
floquet theory allows to describe the interaction between a periodically oscillating field and a quantum system and its basic idea is to replace the semiclassical time - dependent hamiltonian with a time - independent hamiltonian of infinite dimensions @xcite .
we will show that this method allows to obtain highly accurate potentials .
floquet theory is applicable to single and many - mode systems @xcite and it has been extensively used in studying multi - photon processes for atoms in strong laser fields @xcite .
for example drese an holthaus @xcite used floquet theory to describe adiabatic potentials produced by short laser pulses in order to stimulate a stirap ( stimulated raman adiabatic passage ) process .
more recently this was extended to study multiphoton resonance dynamics driven by intense frequency - comb laser fields using many - mode floquet theory @xcite .
a comprehensive overview of the application of floquet theory to atomic and molecular multiphoton processes in intense laser fields can be found in @xcite .
the layout of this paper is as follows . in sec .
[ sec : genrf ] we briefly review the solutions of the hamiltonian describing a magnetically trapped atom , which is exposed to a single - frequency rf field . in sec .
[ sec : piecewiseresonance ] we discuss the piecewise resonance model for dealing with multiple rf fields , which was developed in @xcite , and show how it breaks down when two frequencies come too close . in sec .
[ sec : floquetmodel ] we show how floquet theory can be used to obtain a more accurate description and in sec .
[ sec : potentialgen ] we compare the two approaches and discuss the corrections and benefits of the floquet approach . to demonstrate the power of the approach , we detail in sec .
[ sec:2dpotential ] how an extension to higher dimensions can be done and show how controlling the orientation of the rf field can be used to produce non trivial trapping geometries . as an example
, we show how controlling the tunnel coupling in multidimensional rf traps can allow to create ring vortex solitons in bose - einstein condensates .
finally we conclude in sec .
[ sec : conclusion ] .
rf trapping is based on the idea of coupling magnetic sublevels in the presence of an inhomogeneous magnetic field .
consider a hyperfine atomic groundstate manifold with total spin @xmath0 . in the presence of a static magnetic field of modulus @xmath1 , the zeeman sublevels ,
@xmath2 undergo a linear splitting according to @xmath3 , as long as the field strength is not too large and second order splitting is negligible . here
, @xmath4 is the atomic g - factor and @xmath5 is the bohr magneton .
irradiating such a system with a radio frequency @xmath6 , which can excite atomic larmor precession , will couple the sublevels @xmath7 with spatial selectivity due to the inhomogeneity of the magnetic field and the spatially varying larmor frequency . for simplicity ,
let us first consider a one - dimensional situation with a static field pointing and increasing linearly along the @xmath8-direction , i.e. @xmath9 where g is the magnetic field gradient .
such a situation arises , for example , when an atom travels along an axis of a quadrupole field .
it is convenient to express the oscillating field in complex spherical components ( @xmath10 , @xmath11 , and @xmath12-polarizations ) , such that in cartesian coordinates @xmath13\ ; .\ ] ] the total hamiltonian for the magnetic interaction can then be written as @xmath14 where the angular momentum operator is expressed in units of @xmath15 .
transforming to a frame rotating about the @xmath8-axis at frequency @xmath16 using @xmath17 with @xmath18 leads to @xmath19 , where the effective field in cartesian coordinates is given by @xmath20+\frac{1}{\sqrt{2}}\text{re}[b_+]\\ \frac{1}{\sqrt{2}}\text{im}[b_-e^{-i2\omega t}]-\frac{1}{\sqrt{2}}\text{im}[b_+]\\ \text{re}[b_\pi e^{-i\omega t}]+b_s(z)-\frac{\hbar\omega}{\mu_b g_f } \end{array}\right).\ ] ] expressing this hamiltonian in the spherical basis @xmath21 then results in @xmath22+c.c.\nonumber\;,\end{aligned}\ ] ] with the usual definitions @xmath23 . if the rf field is purely @xmath10-polarized , the effective field and rotating frame hamiltonian are time - independent . in this case , the situation in the rotating frame is fully equivalent to an atom moving in an inhomogeneous field across a non - zero field minimum , defined by the resonance condition @xmath24 .
conversely , the widely used ioffe - pritchard type trap , where the field - zero of a 2d quadrupole field is lifted with an orthogonal applied field @xmath25 , can be viewed as a dressed state trap with zero coupling frequency .
the quasi - energies , i.e. , the eigenvalues of the hamiltonian in the rotating frame can be obtained straightforwardly as @xmath26 where the @xmath27 are eigenvalues of the spin component aligned with the effective field , i.e. , rotating in the laboratory frame .
if polarization components other than @xmath10 are present , one can apply the rotating wave approximation ( rwa ) by discarding the fast oscillating terms as long as @xmath28 @xcite .
this however , will neglect shifts in the energy spectrum , which are known as bloch - siegert shifts @xcite .
a second order correction is caused by a @xmath11 field component , which can be seen as equivalent to a @xmath10 polarized field at negative frequency , leading to off - resonant coupling between states with @xmath29 ( see fig . [
fig : scheme ] ) .
other higher order effects can also stem from @xmath12 polarised components .
( green lines ) and @xmath30 ( red lines ) polarised rf fields irradiating an atom in a linear static field @xmath31 .
the blue lines show the bare hamiltonian eigenvalues @xmath32 and @xmath33 .
the dashed black lines indicate @xmath34 and @xmath35 . ]
more control and complexity can be introduced into rf dressed potentials by using multiple frequencies @xmath36 . using the method outlined above
, one can in principle apply an iterative procedure by performing consecutive rotating wave approximations for each frequency .
however , it is then necessary at each step to transform all oscillating fields into the new rotating frame and determine new spherical polarization components with respect to the effective field of the previous step .
this procedure will lead to an accumulation of errors and can suffer from a certain arbitrariness as the order of eliminating individual frequencies is not defined .
however , if the individual frequencies are spaced sufficiently far apart and the coupling is small with respect to the detuning , one can approximate the dynamics locally by considering only the nearest resonance frequency , @xmath37 .
formally this means that @xmath38 is chosen such that @xmath39 is minimized at any position @xmath8 . in the language of the above consecutive approach ,
this is equivalent to approximating the effective field by its @xmath8-component before decomposing the polarisation components for the next step . in the following ,
we briefly revisit the method outlined in @xcite to analyse the resulting potentials . with the local frequency approximation , and after applying the rwa , the hamiltonian of eq .
( [ eq : sphericalhamiltonian ] ) for a system of total spin @xmath40 can be expressed as @xmath41 where @xmath42 is the rabi frequency of the @xmath43 rf field with @xmath10-polarisation and amplitude @xmath44 ( or @xmath11-polarisation for negative @xmath36 ) . for practical reasons ,
experiments are often performed with linearly polarised rf fields , orthogonal to the static field . in this case , the @xmath45-components are given by @xmath46 .
if only one linear polarisation is used , say @xmath47 , each frequency produces two equally strong resonances at @xmath48 or @xmath49 with rabi frequency @xmath50 . for large detuning the eigenenergies of the two - level rwa - hamiltonian can be written as @xmath51 where the second step is valid far from the resonance , @xmath52 , and the resulting second term can be viewed as a stark shift .
the effect of all other rf fields which are not closest to resonance can be approximated by creating an effective stark shift , which is the sum of the shifts of all rf fields of frequency @xmath53 ( @xmath54 ) @xcite , @xmath55 this leads to a correction of the energies , which are now given by @xmath56 from this , and considering that the couplings are strong enough to yield a landau - zener transition probability close to unity , the resulting adiabatic potential can be written as @xmath57 \mp \sum^{n-1}_{k=1}(-1)^k \hbar \omega_k \;.\ ] ] in fig .
[ fig : pwpotential ] we show the adiabatic rf potentials as seen by an atom in a static field gradient irradiated by two rf fields with frequencies , @xmath58 and @xmath59 , for three different values of @xmath60 . in order to set scales applicable to typical atom chip experiments
, all work presented here assumes a static field in the form of an ioffe - pritchard ( ip ) type trap with radial field profile @xmath61 where we chose a magnetic field gradient @xmath62 and an offset field @xmath63 .
while for well spaced frequencies the potential resulting from the above procedure is visually smooth ( see fig . [
fig : pwpotential ] ( a ) ) , a discontinuity appears for decreasing values of @xmath64 , which grows as the frequencies approach each other .
treating the rf fields independently from each other is therefore no longer appropriate . in order to describe these situations
we will explore the advantages a treatment using floquet theory offers .
situations in which this becomes important are , for example , potentials in which the curvature changes on a small spatial scale . for @xmath65 hz and rabi frequencies @xmath66
note that the axis for each plot changes so that the important features of each potential are resolved . in all cases @xmath67 and @xmath68 . ]
floquet theory provides a practical and powerful nonperturbative method for studying the interaction of a quantum system with an oscillating field , such as ionization and multiphoton excitation processes .
it allows to find the solutions of a schrdinger equation with a temporally periodic hamiltonian by representing it as an infinite matrix @xcite . by using a floquet approach to describe a system containing multiple rf fields
we can ensure that higher - order frequency shifts , which are neglected by the piecewise resonance model and other approaches which rely on the rwa , are accounted for .
as in the previous section , we aim to describe the state of an atom confined to a static magnetic field @xmath69 subjected to a radio - frequency field @xmath70 with n different frequencies .
however , unlike the previous section , we consider linearly polarised fields that can have arbitrary orientation relative to the static field by defining the @xmath71 and @xmath72 components of the rf field as @xmath73 and requiring the @xmath74 component to be real . according to eq .
( [ eq : rfbdef ] ) , the @xmath43 rf field can then be written as @xmath75 where @xmath76 is the rf field vector .
it can be decomposed into its components along each cartesian axis , @xmath77 , and controlling the angle @xmath78 allows to chose orientation of the rf field .
considering @xmath79 such fields then leads to the hamiltonian @xmath80 we now move to a frame in which the direction of @xmath69 always points along the @xmath8-axis by applying a rotation matrix @xmath81 to the hamiltonian .
while this can be done for any static field , here we assume that it is of ioffe - pitchard form ( [ eq : ip ] ) . in this case
, the rotation matrix @xmath81 can be written as , @xmath82 and we define @xmath83 as the rotated rf field vector such that @xmath84 in the limit @xmath85 the component that oscillates parallel to the static field , @xmath86 , can be neglected and only the orthogonal parts , @xmath87 and @xmath88 , contribute to the coupling between the atomic levels @xcite .
the hamiltonian ( [ eq : fhamiltonian ] ) therefore becomes @xmath89 \ ; .
\label{eq : hamiltonianfinal}\end{aligned}\ ] ] let us briefly demonstrate the floquet approach by first explicitly considering the situation where only a single frequency , @xmath58 , is present .
the hamiltonian is then periodic in time with a period of @xmath90 and the floquet theorem states that the time - dependent schrdinger equation has a complete set of quasi - periodic solutions , which acquire a phase @xmath91 when @xmath92 . here
the @xmath93 are quasi - eigenenergies and the phase factor defines @xmath93 mod @xmath58 . the first step towards finding the quasi - eigenenergies is to remove the time dependence of the hamiltonian by replacing it with an infinite matrix of the form @xmath94 where the _ floquet blocks _
@xmath95 ( with @xmath38 integer ) are given by @xmath96 this procedure can be interpreted as an expansion of the original hamiltonian in terms of the fourier components of @xmath58 . for the hamiltonian
the floquet blocks can be calculated by using the identity @xmath97 , where @xmath98 is the kronecker delta @xcite .
this leads to @xmath99 where the rabi frequency is given by @xmath100 and @xmath101 . in order to obtain the eigenvalues and eigenvectors of @xmath102
, it needs to be truncated to a finite size , and a consistent way of doing this is to fix the number of multiples of @xmath103 to include .
this corresponds to limiting the order of the photonic processes that can occur , i.e. if we limit the matrix to terms up to @xmath104 , maximally two photons can be absorbed and emitted . including higher orders will give a more accurate description , but usually these terms are quickly decreasing in magnitude . to first order , this then leads to @xmath105 and diagonalising this matrix gives the quasi - energy spectrum , which in turn allows the adiabatic rf potential to be calculated .
this process will be detailed in sec .
[ sec : potentialgen ] .
in this section we will describe the treatment of an rf field containing several radio frequencies and use the model of many mode floquet theory ( mmft ) @xcite .
we will carry out the calculations explicitly for two different frequencies , @xmath58 and @xmath59 , with the extension to any number being straightforward .
the hamiltonian follows from and is given by @xmath106 which leads to the basic form for the floquet matrix @xmath107 here @xmath108 is the identity matrix and @xmath109 as we want to work in the dressed state picture .
the elements of the @xmath110 matrix are matrices themselves and @xmath110 can therefore be thought of as a composite matrix of two single frequency floquet matrices .
the @xmath58 frequency terms and the @xmath111 rabi frequency terms are contained in matrix @xmath112 , which is the same matrix as @xmath102 in the previous section .
the off diagonal elements @xmath113 contain the rabi frequency of the second field , @xmath114 , and are of the form @xmath115 where @xmath116 by following the method outlined above for the single frequency case , one can then evaluate each of these matrices . however , the explicit form of @xmath110 is too large to reproduce here and we only show a small excerpt of it below to highlight the general structure .
the key points of the structure are that the bare trap eigenenergies are given along the diagonal and each floquet block differs by multiples of both @xmath58 and @xmath117 .
the off - diagonal terms correspond to the coupling strengths and contain terms for @xmath111 and @xmath118 @xmath119
in order to produce the adiabatic potential for the trapped atoms we need to extract the eigenvalues , @xmath120 , and eigenvectors , @xmath121 , from the floquet matrix for every position on the spatial numerical grid , @xmath122 .
we do this numerically as producing analytic expressions becomes quickly unfeasible .
the quasi - eigenenergies ( the eigenvalues of @xmath102 ) generated when applying a two frequency , linearly polarised , rf field orthogonal to the movement of an atom in a one - dimensional ip trap are shown in fig .
[ fig : spectrum ] .
the spectrum has a periodic structure , which stems from the fact that the quasi eigenenenergies are defined mod @xmath58 and mod @xmath117 , and shows the expected avoided crossings ( see fig .
[ fig : spectrum](b ) ) .
it should be noted that the rabi frequencies @xmath123 and @xmath124 in this geometry are constant and can be changed independently from each other , which allows to control the size of the avoided crossings associated with each frequency separately .
hz and @xmath125hz with rabi frequencies @xmath126hz .
@xmath67 and @xmath68 in all cases . ] the potential seen by the atoms can now be found by identifying the energy eigenvalue in the centre of the trap which is closest to the bare trap eigenenergy , @xmath127 ( normally the smallest positive quasi - eigenvalue ) , and following its evolution in space . to find the potential at the next numerical grid point @xmath128 ,
we take the eigenvector @xmath129 associated with the eigenvalue at @xmath127 and calculate @xmath130 , the dot - product between it and all eigenvectors corresponding to the eigenvalues @xmath131 .
the index of the eigenvector which maximises this inner product gives the value of the potential at @xmath128 .
iterating this process allows the construction of the potential along the whole length of the trap ( see red line in fig . [
fig : spectrum ] ) .
[ fig : comparison ] shows the adiabatic rf potentials generated using both the piecewise resonance and floquet methods for the same parameters as fig .
[ fig : pwpotential ] .
one can see that the floquet approach produces the same potential for ( a ) @xmath132 and avoids the discontinuity for ( b ) @xmath133 , which is a notable improvement over the piecewise resonant model . in the extreme case of @xmath134 shown in fig .
[ fig : comparison ] ( c ) the floquet model produces another pair of avoided crossings on either side of the original avoided crossings seen at larger frequency spacing .
these extra avoided crossings are the result of the frequencies no longer just coupling the local static field @xmath69 , where they are resonant . instead , at this very close frequency separation , each frequency
is applied to the effective field created by the static field and the other frequency , which leads to a shift in the resonance position and results in the additional avoided crossings .
as one can also see from fig .
[ fig : comparison ] ( c ) , these avoided crossings are not present in the piecewise resonance model , since the rotating wave approximation eliminates the higher order coupling terms . .
see text for discussion . ] from this example it is clear that the floquet approach to modelling adiabatic rf potentials produces highly accurate results with significant corrections to the piecewise resonance approach .
it is therefore well suited to accurately describe multi - frequency rf potentials created using closely spaced frequency . in the next section
we will show how the floquet approach can be extend to systems with multiple rf fields where the orientations of the fields are not necessarily perpendicular to the static field .
the well known flexibility of rf potentials stems from the large number of parameters ( number of frequencies , frequency spacing , field strength and orientation ) that can be controlled to modify the geometry of the potentials .
these parameters can also be time - dependent , which allows to use the resulting potentials to instigate adiabatic or non - adiabatic dynamics . in the following we consider a multi - dimensional setup and use the floquet approach to describe some of the features which appear when taking these additional parameters into account .
the main change necessary to extend the model to higher dimensions is to use a multi - dimensional static trapping field @xmath135 which then leads to coupling terms @xmath136 of the same , higher dimension .
this means that the floquet matrix contains elements that are multidimensional , but its basic structure is unchanged from the one - dimensional case .
the added dimension however means that the numerical processing power needed to diagonalise the floquet matrix is significantly higher . in the following
we will show an example of an interesting potential that can be created using just two frequencies .
the numerical approach is similar to the one detailed above , in that we construct the 2d potentials by splitting them into 1d slices and combine them to produce the full 2d potential . to ensure that our minimum rotation method produces 1d slices which are consistent with each other , we use the previously generated 1d slice for each consecutive 1d slice to define an initial quasi - eigenvector and eigenvalue pair .
two - dimensional rf potentials generated using one frequency have already been demonstrated to study rotating becs in annular geometries @xcite .
one of the simplest ways to create a more involved structure using two frequencies is shown in fig .
[ fig:2dpotential ] ( a ) where we consider a 2d ip trap and a constant rf field vector which contains a @xmath74 component only ( @xmath137 ) .
one can see that the geometry one can create consists of an outer ring and an inner potential , which is actually simply the result from the one - dimensional model of sec . [ sec : potentialgen ] rotated around the axis along z=0 .
however , by controlling @xmath58 and @xmath59 one can easily change the depth and radius of the inner harmonic well and the outer ring , as well as the size and height of the barrier between them . performing such a change in a time - dependent way one
can then engineer a situation where the tunnel coupling between the ring and inner trap can be increased or decreased with high accuracy .
hz and @xmath138hz for ( a ) @xmath137 , @xmath139 ( b ) @xmath140 , @xmath141 , @xmath142 and ( c ) @xmath143 , @xmath144 , @xmath142 . in all cases
@xmath67 and @xmath68 .
] furthermore , having control over the orientation of the rf fields allows one to break the rotational symmetry and create potentials that are no longer isotropic . in fig .
[ fig:2dpotential ] ( b ) and ( c ) we show two examples of a 2d potential generated by using constant rf field vectors for ( b ) @xmath140 , @xmath141 , @xmath142 and ( c ) @xmath143 , @xmath144 , @xmath142 .
the resulting potentials display two minima and two maxima on the rings of resonance , which can be understood by considering that the orientation of the @xmath145 and @xmath146 components of the rf field with respect to the static field is not constant in space .
this leads to the variable coupling .
furthermore , controlling the angle @xmath78 allows to create rf fields which point in any direction , which in turn allow the creation of minima and maxima at any position and even in a time dependent fashion . as an example of how useful the time - dependent control of the tunnelling interaction between the outer ring and the inner harmonic trap can be
, we demonstrate in this section how the potential discussed above can be used to create so - called ring vortex solitions ( rvss ) in gaseous bose - einstein condensates .
these states consist of multiple concentric density - wave rings with a non - zero winding number and they were recently extensively studied by li _
_ @xcite . to create them in a laboratory
, one can consider starting out with a single - frequency rf field in a two - dimensional setting , and creating a condensate carrying a persistent current in the resulting annular potential .
adding the second frequency would allow one to create the central potential and by adjusting all frequencies and intensities appropriately , the current would start to tunnel into the inner potential . to simulate the creation of rvss in such a way , the potential must allow large tunnelling strength and therefore the model must be able to deal with situations where two frequencies come close .
this makes the floquet approach the most suitable option . in fig .
[ fig : rvs ] we show a simulations of this tunnelling process using a gross - pitaevskii model for the condensate and a potential calculated using the floquet approach outlined above .
the density and phase distribution for the bec initially confined to the outer ring with @xmath147 can be seen in figs . [
fig : rvs ] ( a ) and ( b ) , and the density and phase after transfer into the harmonic trap is shown in figs .
[ fig : rvs ] ( c ) and ( d ) .
clearly the resulting state can be identified as a ring vortex soliton .
hz and @xmath148hz with @xmath149 for both fields .
( a ) shows the density of the bec of winding number n=1 initially confined to the outer ring and ( b ) is its corresponding phase plot .
( c ) shows the density of the bec at @xmath150 and ( d ) is corresponding phase plot . ] while we have explicitly shown a 2d adiabatic rf potential generated using two frequencies , this can easily be extended to include more frequencies to create potentials that consist of multiple concentric rings . as in the previous example of just two frequencies , control over the frequency spacing
would allow one to adjust the tunnel coupling between the individual rings and can be accurately determined by using a floquet approach .
such systems are highly desirable as they offer experimentally realistic systems in which one could observe rotational fluxons ( josephson vortices ) @xcite or study the kibble - zurek mechanism in becs @xcite .
we have shown that using a non - perturbative floquet approach to calculate radio - frequency potentials for atoms on atom chips leads to good results for multi - frequency situations in one and two dimensions .
this approach offers significant corrections to a piecewise - resonant model and can be used to accurately describe situations where two frequencies come close ( for example when trying to engineer large tunnel couplings ) .
we have also shown that a floquet approach can deal with higher dimensions and rf fields of arbitrary field orientation .
we have demonstrated the flexibility of the floquet approach by considering the creation of ring vortex solitons in a 2d adiabatic rf potential which requires that the tunnelling dynamics be determined accurately .
such potentials can be used in a variety of ways , and for example , allow the creation of rotational fluxons and non standard trapping potentials that can connect different topological geometries .
this will be the focus of future work .
the authors would like to thank igor lesanovsky for valuable discussions .
this work was supported by the european science foundation ( esf ) within the framework of the common perspectives for cold atoms , semiconductor polaritons and nanoscience activity and science foundation ireland under project numbers 05/in / i852 and 10/in.1/i2979 .
d. t. pegg , j. phys .
b * 6 * 241 ( 1974 ) j. li , d.s .
wang , z.y .
wu , y.m . yu and w.m . liu , phys . rev . a * 86 * 023628 ( 2012 )
j. brand , t. j. haigh and u. zlicke , phys .
a * 80 * 011602 ( 2009 ) shih - wei su , shih - chuan gou , a. bradley , o. fialko and j. brand , phys .
lett . * 110 * 215302 ( 2013 ) | adiabatic radio frequency ( rf ) potentials are powerful tools for creating advanced trapping geometries for ultra - cold atoms . while the basic theory of rf trapping is well understood , studies of more complicated setups involving multiple resonant frequencies in the limit where their effects can not be treated independently are rare . here
we present an approach based on floquet theory and show that it offers significant corrections to existing models when two rf frequencies are near degenerate .
furthermore it has no restrictions on the dimension , the number of frequencies or the orientation of the rf fields .
we show that the added degrees of freedom can , for example , be used to create a potential that allows for easy creation of ring vortex solitons . |
cosmological observations have started to provide valuable upper limits on absolute neutrino masses ( see , e.g. , the reviews @xcite ) , competitive with those from laboratory experiments .
in particular , the combined analysis of high - precision data from cosmic microwave background ( cmb ) anisotropies and large scale structures ( lss ) has already reached a sensitivity of @xmath0 ( see , e.g. , @xcite ) for the sum of the neutrino masses @xmath1 , @xmath2 we recall that the total neutrino energy density in our universe , @xmath3 ( where @xmath4 is the hubble constant normalized to @xmath5 km s@xmath6 mpc@xmath6 ) is related to @xmath1 by the well - known relation @xmath7 @xcite , and plays an essential role in theories of structure formation .
it can thus leave key signatures in lss data ( see , eg.,@xcite ) and , to a lesser extent , in cmb data ( see , e.g.,@xcite ) . very recently
, it has also been shown that accurate lyman-@xmath8 ( ly@xmath8 ) forest data @xcite , taken at face value , can improve the current cmb+lss constraints on @xmath1 by a factor of @xmath9 , with important consequences on absolute neutrino mass scenarios@xcite .
on the other hand , atmospheric , solar , reactor and accelerator neutrino experiments have convincingly established that neutrinos are massive and mixed .
world neutrino data are consistent with a three - flavor mixing framework ( see @xcite and references therein ) , parameterized in terms of three neutrino masses @xmath10 and of three mixing angles @xmath11 , plus a possible cp violating phase @xmath12 .
neutrino oscillation experiments are sensitive to two independent squared mass difference , @xmath13 and @xmath14 ( with @xmath15 ) , hereafter defined as @xcite @xmath16 where @xmath17 fixes the absolute neutrino mass scale , while the cases @xmath18 and @xmath19 identify the so - called normal and inverted neutrino mass hierarchies , respectively .
neutrino oscillation data indicate that @xmath20 ev@xmath21 and @xmath22 ev@xmath21 .
they also indicate that @xmath23 , @xmath24 , and @xmath25 .
however , they are currently unable to determine the mass hierarchy ( @xmath26 ) and the phase @xmath12 , and are insensitive to the absolute mass parameter @xmath17 in eq .
( [ deltadef ] ) .
the absolute neutrino mass scale can also be probed by non - oscillatory neutrino experiments .
the most sensitive laboratory experiments to date have been focussed on tritium beta decay and on neutrinoless double beta decay .
beta decay experiments probe the so - called effective electron neutrino mass @xmath27 @xcite , @xmath28^\frac{1}{2}\ , \ ] ] where @xmath29 and @xmath30 .
current experiments ( mainz @xcite and troitsk @xcite ) provide upper limits in the range @xmath31 ev @xcite .
neutrinoless double beta decay ( @xmath32 ) experiments are instead sensitive to the so - called effective majorana mass @xmath33 ( if neutrinos are majorana fermions ) , @xmath34 where @xmath35 and @xmath36 parameterize relative ( and unknown ) majorana neutrino phases @xcite .
all @xmath32 experiments place only upper bounds on @xmath33 ( the most sensitive being in the ev range , with the exception of the heidelberg - moscow experiment @xcite , which claims a positive ( but highly debated ) @xmath32 signal @xmath37 ev at @xmath38 c.l . and corresponding to @xmath33 in the sub - ev range at best fit @xcite . in these proceedings
, we will briefly illustrate the impact of the cosmological constraints on the sum of neutrino masses on the three - flavor mixing theoretical and observational scenario .
the neutrino contribution to the overall energy density of the universe can play a relevant role in large scale structure formation and leave key signatures in several cosmological data sets . more specifically , neutrinos suppress the growth of fluctuations on scales below the horizon when they become non relativistic .
a massive neutrinos of a fraction of ev would therefore produce a significant suppression in the clustering on small cosmological scales ( namely , for comoving wavenumber @xmath39 ) . to constrain @xmath1 from cosmological data ,
we perform a likelihood analysis comparing the recent observations with a set of models with cosmological parameters sampled as follows : cold dark matter ( cdm ) density @xmath40 $ ] in steps of @xmath41 ; baryon density @xmath42 $ ] ( motivated by big bang nucleosynthesis ) in steps of @xmath43 ; a cosmological constant @xmath44 $ ] in steps of @xmath45 ; and neutrino density @xmath46 $ ] in steps of @xmath47 .
we restrict our analysis to _ flat _ @xmath48-cdm models , @xmath49 , and we add a conservative external prior on the age of the universe , @xmath50 gyrs .
the value of the hubble constant in our database is not an independent parameter , since it is determined through the flatness condition .
we adopt the conservative top - hat bound @xmath51 and we also consider the @xmath52 constraint on the hubble parameter , @xmath53 , obtained from hubble space telescope ( hst ) measurements @xcite .
we allow for a reionization of the intergalactic medium by varying the cmb photon optical depth @xmath54 in the range @xmath55 $ ] in steps of @xmath45 .
we restrict the analysis to adiabatic inflationary models with a negligible contribution of gravity waves .
we let vary the spectral index @xmath56 of scalar primordial fluctuations in the range @xmath57 $ ] and its running @xmath58 $ ] assuming pivot scales at @xmath59 and @xmath60 .
we rescale the fluctuation amplitude by a prefactor @xmath61 , in units of the value @xmath62 measured by the wilkinsin microwave anisotropy probe ( wmap ) satellite .
finally , concerning the neutrino parameters , we fix the number of neutrino species to @xmath63 , all with the same mass ( the effect of mass differences compatible with neutrino oscillation being negligible in the current cosmological data @xcite ) . an higher number of neutrino species can weakly affect both cmb and lss data ( see , e.g. , @xcite ) but is highly constrained by standard big bang nucleosynthesis and is not considered in this work , where we focus on @xmath64 mixing .
the cosmological data we considered comes from observation of cmb anisotropies and polarization , galaxy redshift surveys and luminosity distances of type ia supernovae . for the cmb data we use the recent temperature and cross polarization results from the wmap satellite @xcite using the method explained in @xcite and the publicly available code .
given a theoretical temperature anisotropy and polarization angular power spectrum in our database , we can therefore associate a @xmath65 to the corresponding theoretical model .
we further include the latest results from other cmb datasets .
the cmb data analysis methods have been already described in @xcite and will not be reported here .
in addition to the cmb data we also consider the real - space power spectrum of galaxies from either the 2 degrees fields ( 2df ) galaxy redshifts survey or the sloan digital sky survey ( sdss ) , using the data and window functions of the analysis of @xcite and @xcite .
we restrict the analysis to a range of scales over which the fluctuations are assumed to be in the linear regime ( @xmath66 ) . when combining with the cmb data , we marginalize over a bias @xmath67 for each data set considered as an additional free parameter .
we also include information from the ly@xmath8 forest in the sdss , using the results of the analysis of @xcite and @xcite , which probe the amplitude of linear fluctuations at very small scales .
for this data set , small - scale power spectra are computed at high redshifts and compared with the values presented in @xcite . as in @xcite
, we do not consider running .
we finally incorporate constraints obtained from the sn - ia luminosity measurements of @xcite using the so - called gold data set .
luminosity distances at sn - ia redshifts are computed for each model in our database and compared with the observed apparent bolometric sn - ia luminosities . in fig . 1 we plot the likelihood distribution for @xmath1 from our joint analysis of cmb + sn - ia + hst + lss data , transformed into an equivalent @xmath68 function , which allows to derive bounds on @xmath1 at any fixed confidence level .
we take lss data either from the sdss or the 2df survey ( dashed and solid curves , respectively ) .
as we can see , these curves do not show evidence for a neutrino mass ( the best fit being at @xmath69 ) and provide the @xmath70 bound @xmath71 ev .
such bound is in good agreement with previous results in similar analyses @xcite .
also plotted in fig .
1 is the @xmath68 function from a joint analysis of cmb + sn - ia + hst + 2df + ly@xmath8 .
no running is assumed in this analysis , and we find a @xmath70 bound @xmath72 ev , in very good agreement ( despite the more approximate method we used ) with the analysis already presented in @xcite . as shown in fig . 1 and already discussed in @xcite ,
the inclusion of the ly@xmath8 data from the sdss set greatly improves the constraints on @xmath1 .
upper bounds on the sum of neutrino masses @xmath1 from our @xmath64 analysis of cosmological data , given in terms of the @xmath68 function .
the solid and dashed curves refer to the combination of cmb and lss data ( cmb+2df and cmb+sdss , respectively ) .
the two cmb+lss fits provide comparable results and , for definiteness , the cmb+2df one is adopted .
in addition , we consider also the case where the recent ly@xmath8 data from the sdss are included , providing significantly stronger constraints on @xmath1 ( dotted curve ) .
see @xcite for details.,width=226 ] global @xmath64 analysis in the @xmath73 parameter space , using oscillation data plus laboratory data and cosmological data .
this figure implements also upper limits ( shown as dashed lines at @xmath70 level ) on @xmath27 from mainz+troitsk data , on @xmath33 from @xmath32 data , and on @xmath1 from cmb+2df+ly@xmath8 data . in combination with oscillation parameter bounds , the cosmological upper limit on @xmath1 dominates over the laboratory upper limits on @xmath27 and @xmath33 .
see @xcite for details.,width=226 ]
here we consider confidence regions obtained from analysis of neutrino oscillation data , of @xmath27 and @xmath33 data and cosmological cmb+lss data ( see @xcite for more details ) .
figure 2 shows such regions projected in the three coordinate planes .
separate laboratory and cosmological upper bounds at the @xmath70 level are shown as dashed lines , while the regions allowed by the combination of laboratory , cosmological , and oscillation data are shown as thick solid curves for normal hierarchy and as thin solid curves for inverted hierarchy .
it can be seen that the upper bounds on the @xmath74 observables are dominated by the cosmological upper bound on @xmath1 .
this bound , via the @xmath75 and @xmath76 correlations induced by oscillation data , provides upper limits also on @xmath33 and @xmath27 , which happen to be stronger than the current laboratory limits by a factor @xmath77 . since significant improvements on laboratory limits for @xmath33 and @xmath27 will require new experiments and several years of data taking @xcite , cosmological determinations of @xmath1 , although indirect , will continue to provide , in the next future , the most sensitive upper limits ( and hopefully a signal ) for absolute neutrino mass observables . in fig .
2 , the tension ( at @xmath70 ) between the limits from cosmology and the lower limit on @xmath78 ev claimed by the heidelberg - moscow experiment is a clear symptom of possible problems , either in some data sets or in their theoretical interpretation , which definitely prevent any global combination of data
. it would be premature to conclude that , e.g. , the @xmath32 claim is``ruled out '' by cosmological data but it is anyway exciting that global neutrino data analyses have already reached a point where fundamental questions may start to arise . _ acknowledgements _ a.m. would like to thank the organizers of the _ now-2004 _ workshop .
the work of a.m. is supported by the italian ministero dellistruzione , universit e ricerca ( miur ) through the `` gemini '' research project and istituto nazionale di fisica nucleare ( infn ) through the `` astroparticle physics '' research project . the work of g.l.f . ,
e.l . , a.m.@xmath79 , and a.p .
is supported by the italian ministero dellistruzione , universit e ricerca ( miur ) and istituto nazionale di fisica nucleare ( infn ) through the `` astroparticle physics '' research project .
mckellar , phys .
b 97 , 93 ( 1980 ) ; f. vissani , nucl .
b ( proc . suppl . ) 100 , 273 ( 2001 ) ; j. studnik and m. zralek , hep - ph/0110232 .
see also the discussion in y. farzan and a.yu .
smirnov , phys .
b 557 , 224 ( 2003 ) . | we derive upper limits on the sum of neutrino masses from an updated combination of data from cosmic microwave background experiments and galaxy redshifts surveys .
the results are discussed in the context of three - flavor neutrino mixing and compared with neutrino oscillation data , with upper limits on the effective neutrino mass in tritium beta decay from the mainz and troitsk experiments and with the claimed lower bound on the effective majorana neutrino mass in neutrinoless double beta decay from the heidelberg - moscow experiment . |
the totally asymmetric simple exclusion process ( tasep ) is a paradigmatic model for driven diffusive systems @xcite .
tasep is particularly used for modeling transport in a wide variety of nonequilibrium systems , _
e.g. _ motion of molecular motors on microtubules @xcite , filamentous fungal growth @xcite and flow of traffic on a road @xcite .
the model is defined on a one dimensional lattice on which hardcore particles perform biased diffusion . despite its simplicity , this model shows interesting properties like the existence of a phase transition in one dimension , which are hallmarks of nonequilibrium systems @xcite .
one widely - studied variant of the tasep is defined on an open lattice where in addition to performing biased diffusion , the system can also exchange particles with the reservoirs at the boundaries . in this open system , particles are injected at one end and extracted at the other end of the lattice and by tuning these injection and extraction rates one finds that the system shows a boundary induced phase transition @xcite .
the resulting phase diagram was derived using exact calculations and an alternative simpler derivation was also presented using mean field theory @xcite .
the phase diagram consists of a high density phase , a low density phase and a maximal current phase .
as the names suggest , the high density ( low density ) phase is characterized by a large ( small ) bulk density and the maximal current phase by the presence of largest admissible current in the system .
the open chain tasep is widely used to model transport in complex biological systems @xcite .
owing to the inherent complexity of a biological system , transport processes are generally coupled with other biochemical reactions . at the entry and exit terminals other reactions can take place whose outcome may influence the injection and exit processes strongly .
such a situation was considered in @xcite where the species of particles undergoing transport were assumed to take part in other reactions before or after the transport _
i.e. _ entrance or exit are possible only when certain reactions take place .
the system was modeled as an open - chain tasep with stochastically mediated entrance or exit , facilitated by gates present at the boundaries .
the opening and closing of the gate at the exit ( entrance ) terminal is determined by the binding and unbinding of certain receptor ( initiator ) to the gate . when the receptor ( initiator ) is bound to the gate
, the particles can pass through the gate and exit ( enter ) the channel . when the gate at the exit ( entrance ) end is not bound to any receptor ( initiator ) , it is assumed to be in a closed state when no particle can pass through .
one important component in this model is the two - way coupling between the gate dynamics and the particle flow .
when a particle passes through an open gate , the receptor ( initiator ) immediately unbinds , _
i.e. _ the particles ` close the gate behind them ' .
the new phase diagram for this model was studied using monte carlo simulations and mean - field theory .
one physical system , where stochastic gates are explicitly present and control cellular transport , can be found in ion - channels , which are embedded on cell membrane and exchange of ions between cells takes place through these channels @xcite .
an ion - channel stochastically switches between ` open ' and ` closed ' conformations and while the open conformation allows fast passage of ions through the channel resulting in a large ion - flux , no transport is possible in the closed state @xcite .
the closed state is also known as the high - affinity state , when the ions strongly bind to the channel and hence gets trapped . in the open conformation ,
the channel is in the low - affinity state and the ions immediately dissociate from the channel and move to the cell cytosol . in this paper , we explicitly model the gating dynamics , following the modeling strategy used to describe ion - channel gating @xcite , and study its effect on the transport .
we consider a generalization of an open chain tasep , where the injection and extraction processes at the two boundaries are controlled by the presence of gates or special boundary sites , which can stochastically switch between high- and low - affinity states , as described above . while in the high - affinity state ,
a particle gets trapped in the gate , in the low - affinity state it immediately dissociates from the gate and moves towards right .
our monte carlo simulations show that the inclusion of the above gating mechanism produces interesting and non - trivial modifications in the phase diagram .
the residence time of the gates in the open state are important parameters here and depending on their value all three phases may or may not be observed . for example , unlike in ordinary tasep , sufficiently large values of the injection and exit rates do not always yield an mc phase the residence time of the gates in the open state must exceed certain critical values for mc phase to be observed . to explain these results we perform analytical calculations within mean - field theory .
a simple mean - field theory where all correlations in the system are ignored , shows qualitative agreement with the simulation results but there is significant quantitative mismatch . to improve the quantitative agreement
, we use a refined mean - field theory , developed in @xcite , where important correlations at the boundaries are taken into account and this approach yields better agreement with the simulation data . in the next section
we briefly summarize earlier results on the phase diagram for an ordinary open chain tasep and then introduce our stochastic gated model in detail . in section
[ sec : sim ] , we present the results of monte carlo simulation of our model .
our analytical calculations within simple and refined mean - field theory are presented in sections [ sec : mft ] and [ sec : rmft ] , respectively .
discussion and conclusions are presented in section [ sec : discussion ] .
the usual open - chain tasep is defined on a one dimensional open lattice of @xmath0 sites . at each site
there can be at most one particle which can hop towards right , if the right neighboring site is empty .
particles are injected and extracted at the two ends of the lattice .
if site @xmath1 is empty , then a particle is injected into this site with probability @xmath2 .
if a particle is present at site @xmath0 then it exits the system with probability @xmath3 . by varying @xmath2 and @xmath3 ,
the following phase diagram is obtained in steady state . for @xmath4 and @xmath5 ,
the system is in a low - density ( ld ) phase with a bulk density @xmath2 and current @xmath6 . for @xmath7 and @xmath8
the system is in a high density ( hd ) phase when the bulk properties are controlled by the exit conditions . in this phase one
has bulk density @xmath9 and current @xmath10 . for @xmath11 ,
the system is in a coexistence phase when part of the system has a bulk density @xmath2 and the remaining part has bulk density @xmath12 .
these two parts are connected by a domain wall or shock .
the @xmath13 line is known as the coexistence line which terminates at @xmath14 . for @xmath15 ,
the system is in a maximal current ( mc ) phase with bulk density @xmath16 , and current @xmath17 .
this phase diagram was derived exactly in @xcite .
remains trapped at site @xmath1 .
( b ) : left gate is empty and in the open state , particle injected with probability @xmath2 immediately detaches from site @xmath1 and hops to site @xmath18 .
( c ) : left gate is closed and occupied , switches to an open state with probability @xmath19 and the particle is immediately untrapped and hops to site @xmath18 .
( d ) : right gate is closed and empty , particle hopping from site @xmath20 to @xmath0 , gets trapped at site @xmath0 .
( e ) : right gate is open and empty , particle hops from site @xmath20 to @xmath0 and after hopping immediately dissociates from site @xmath0 and exits the system with probability @xmath3 .
( f ) : right gate is closed and occupied , switches to an open state with probability @xmath21 when the trapped particle immediately dissociates and exits the system with probability @xmath3 . ] in this paper , we consider a variant of the above system , where two stochastic gates are present at the two boundary sites @xmath1 and @xmath0 .
the gating dynamics is modeled so as to mimic ion - channel gating
these gates can stochastically switch between high - affinity ( closed ) and low - affinity ( open ) states .
the configuration of the system is specified by specifying the occupancy of each site and the state of the gates at the boundaries .
if the gate at site @xmath1 is in the closed or high - affinity state , then the injected particle strongly binds to the site and therefore remains trapped in that site .
if the gate is in the open or low - affinity state , then the particle immediately dissociates from site @xmath1 and hops to the next site if it is empty . similarly ,
if the gate at site @xmath0 is closed , then a particle arriving at that site gets trapped .
but if the gate is open then the particle exits the system with probability @xmath3 . because of the hard core constraint , no particle can enter the gate if it is already occupied .
these moves are shown schematically in fig .
[ fig : model ] . the switching rate of the gates are assumed to follow local detailed balance . if @xmath19 is the probability that the left gate is open , then the ratio of the forward and reverse transition rates between the open and closed states is @xmath22 .
for the right gate the probability to be in open state is denoted by @xmath21 and the switching rates are chosen following similar local detailed balance .
we are interested in how the @xmath23 phase diagram is modified for different choices of @xmath19 and @xmath21 .
one important point to note here is that even in the limit @xmath24 , _
i.e. _ when the gates are always in the open state , our model does not map onto regular open - chain tasep .
since the gating dynamics in our system is motivated by the models for ion - channel gating @xcite , an open state or low - affinity state of the gate implies that any particle arriving at the gate immediately detaches and goes to the next site , as shown in fig .
[ fig : model]b @xcite .
this effect is not present for a regular tasep where the waiting time at the boundary sites is decided according to the rules of usual stochastic update .
however , we have verified that all our qualitative conclusions remain valid when we consider a variant of our model which maps onto regular tasep in @xmath24 limit . in this modified version ,
a particle gets trapped in a closed gate as before , and it is allowed to leave while the gate is open , but it does not immediately dissociate from an open gate .
the waiting time of a particle at the open gate is same as that at any other site .
we have performed monte carlo simulations and analytical calculations within mean - field theory in this model and find similar results .
see appendix [ sec : passive ] for more details .
we perform monte carlo simulations of the above model . starting from a random initial configuration , we evolve the system in time following random sequential updates where we choose a site at random and update it following the steps outlined in appendix [ sec : algo ] .
each monte carlo step consists of @xmath0 such update trials .
since we are mainly interested in the phase diagram , we use a large system - size @xmath25 in our simulations so as to avoid errors due to finite size effects which may cause significant difference close to the phase boundaries . the current and density profiles which characterize the phase of the system
are measured after evolving for a long enough time to ensure that the system reaches steady state .
vs exit probability @xmath3 for fixed values of @xmath2 and @xmath26 .
discrete points show simulation results , solid ( dotted ) continuous lines show analytical predictions from mean - field ( refined mean - field ) theory .
top panel : @xmath27 , @xmath28 .
( a ) : hd - ld transition at @xmath29 , @xmath30 ( simulation ) and @xmath31 ( refined mean - field ) .
mean - field predicts hd - mc transition at @xmath32 .
( b ) : simulation shows that for @xmath33 ( @xmath34 ) , hd - ld ( hd - mc ) transition at @xmath35 ( @xmath36 ) . for both these @xmath2 values , mean - field ( refined mean - field ) theory predicts hd - mc transition at @xmath32 ( @xmath37 ) .
middle panel : @xmath38 , @xmath39 .
( c ) : hd - ld transition for @xmath40 with @xmath41 ( simulation ) , @xmath42 ( mean - field ) and @xmath43 ( refined mean - field ) .
( d ) : hd - ld transition at @xmath44 and @xmath45 ( simulation ) , @xmath46 ( mean - field ) , @xmath47 ( refined mean - field ) . for @xmath48 simulations
show an hd - mc transition at @xmath49 .
but mean - field and refined mean - field predict hd - ld transition at @xmath50 and @xmath51 , respectively .
bottom panel : @xmath52 , @xmath53 .
( e ) : hd - ld transition at @xmath54 , @xmath55 ( simulation ) , @xmath56 ( mean - field ) , @xmath57 ( refined mean - field ) .
( f ) : hd - ld transition at @xmath58 , @xmath59 ( simulation ) , @xmath60 ( mean - field ) and @xmath61 ( refined mean - field ) . ] to map out the phase diagram , we study the variation of the steady state current @xmath62 as a function of @xmath3 for fixed @xmath2 .
the current is obtained by measuring the average flux across a bond per unit time , averaged over a time - period of @xmath63 monte carlo steps .
the measured current has an error - bar @xmath64 caused by statistical fluctuations .
we find that for small @xmath3 when the system is in hd phase , @xmath62 increases with @xmath3 and then saturates beyond a particular @xmath65 . if the saturation value of @xmath62 depends on @xmath2 , then @xmath65 marks the transition from hd to ld phase . if the saturation current takes an @xmath2-independent value @xmath66 then @xmath65 marks the transition between hd and mc phase . using this method
, we calculate the phase - diagram for various @xmath19 and @xmath21 , from the plot of @xmath62 as a function of @xmath3 for fixed @xmath2 . in fig .
[ fig : jb ] we show the data for @xmath62 vs @xmath3 for few representative values of @xmath2 and @xmath19 , @xmath21 . from the @xmath62 vs
@xmath3 data we obtain @xmath2 vs @xmath65 coexistence line for each given set of @xmath19 and @xmath21 . in fig .
[ fig : coex ] we plot these coexistence lines .
the critical values of @xmath2 and @xmath3 that describe this phase boundary have an error - bar of @xmath67 which comes from the resolution of our measurement . for @xmath68 and @xmath69
the coexistence line terminates at a particular point and the mc phase begins .
but for @xmath70 and @xmath71 , there is no mc phase , and for any @xmath2 in the entire range @xmath72 , one can find a @xmath65 that marks the hd - ld phase transition .
hence for this case , the coexistence line extends all the way upto @xmath73 . in the following section , we derive these results analytically using a mean - field theoretic approach . plane for different values of @xmath19 and @xmath21 .
continuous lines and dotted lines show analytical predictions from simple mean - field theory and refined mean - field theory , respectively .
the simulation results are shown by discrete points .
the refined mean - field approach significantly improves the quantitative agreement with simulation . ]
let @xmath74 be the occupancy of the @xmath75-th site in a particular configuration . within simple mean - field theory
we assume all correlations of the type @xmath76 are factorized and can be written as @xmath77 , where @xmath78 is the average density at the @xmath75-th site .
the time - evolution equations for these densities are @xmath79 here , the first two terms on the right - hand - side of eq .
[ eq : rho1 ] represent the possible mechanisms that lead to the increase of the density at the first site
the first term corresponds to the case when the left gate is closed and an injected particle gets trapped in site @xmath1 ( as shown in fig .
[ fig : model]a ) , the second term stands for the case when the left gate is open but the injected particle can not immediately detach from site @xmath1 and hop to site @xmath18 because site @xmath18 is already occupied .
the last term in that equation describes the situation shown in fig .
[ fig : model]c , when the left gate switches from closed to open state and the trapped particle is released and goes to site @xmath18 . in this way , all the terms in the above set of equations can be explained .
although we have a total of four control parameters @xmath80 and @xmath21 , from the time - evolution equations it follows that @xmath3 and @xmath21 always appear in a product .
so we treat @xmath81 as a single independent parameter .
note that , as mentioned earlier , even when @xmath82 , these equations are not identical to those in regular open - chain tasep .
consider the example shown in fig .
[ fig : model]b , the site @xmath18 can directly receive a particle from the reservoir when the left gate is open .
thus in the time - evolution equation for @xmath83 there is a term that couples to @xmath2 .
no such term is present for regular open - chain tasep .
similarly , at the right boundary , the time - evolution equation for @xmath84 has in general a different form even when @xmath85 .
but here one can retrieve the limit for regular tasep by redefining the exit probability as @xmath86 . in the steady state ,
all time derivatives on the left - hand - side of the equations ( 1)-(4 ) become zero and a constant current @xmath62 flows through every bond in the lattice . at the left end of the system ,
the injection current , _
i.e. _ the rate of influx from the reservoir to site @xmath1 is @xmath87 .
depending on the state of the gate at site @xmath1 and on the occupancy of site @xmath18 , a fraction of the injected particles stays at site @xmath1 and the remaining fraction immediately dissociates from site @xmath1 and hops to site @xmath18 .
thus , the current flowing through the bond between site @xmath1 and @xmath18 is given by @xmath88 .
we can now construct the following set of steady state equations , by recognizing that the current @xmath62 has the same value through all the bonds of the system in the steady state .
@xmath89 note that for @xmath90 , the local density @xmath91 satisfies the recursive relation in eq .
[ eq : rec ] .
let the two fixed points of the recursive map be given by @xmath92 , defined as @xmath93 .
it is easy to see that @xmath94 is the stable fixed point and @xmath95 is the unstable fixed point of the map . also , @xmath96 and @xmath97 .
below we identify different phases and the phase - boundaries following the same steps as in @xcite . in this phase ,
the bulk density is infinitesimally close to the unstable fixed point @xmath95 and deviates from this value close to the right boundary , while satisfying @xmath98 .
note that due to presence of the stochastic gate at site @xmath1 , the starting point of the recursive map is @xmath83 , instead of @xmath99 , unlike ordinary tasep .
putting @xmath100 yields @xmath101 using the relation between @xmath94 and @xmath95 , one gets a quadratic equation for @xmath95 : @xmath102 this has the solution @xmath103^{1/2 } \label{eq : ld}\ ] ] now , by definition , @xmath104 , and since the second term in the above equation is always positive for @xmath105 , one must take the negative root and in addition one must satisfy @xmath106^{1/2 } \end{aligned}\ ] ] which after simplification gives the condition @xmath107 since @xmath108 , above inequality is satisfied if @xmath109 for @xmath110 , _ i.e. _ for @xmath111 , the inequality eq .
( [ eq : ldpa ] ) holds for all @xmath2 .
as shown later , this implies that for @xmath111 there is no mc phase in the system . in this phase , the density at the bulk and the right end is infinitesimally close to the stable fixed point @xmath94 and close to the left boundary there is deviation from this value , with @xmath112 .
using @xmath113 gives @xmath114 the condition @xmath115 implies @xmath116 in this phase , @xmath117 and @xmath118 .
the current @xmath62 takes the maximal value @xmath119 and the bulk density is @xmath120 . from conditions specified by eqs .
( [ eq : rho2 ] ) and ( [ eq : rhob ] ) this phase can be observed if @xmath121 since both @xmath2 and @xmath3 must be less than unity , mc phase can be observed if and only if @xmath122 and @xmath123 . for @xmath19 or @xmath21 lower than these limiting values
mc phase does not exist .
note that this takes care of the condition @xmath124 derived in eq .
[ eq : am ] . on the critical line separating hd and ld phase
, we match the expression for @xmath125 obtained from eqs .
( [ eq : ld ] ) and ( [ eq : hd ] ) for the two phases : @xmath126^{1/2 } = \frac{\beta p_b}{1-\beta p_b}\ ] ] this is the equation for the coexistence line which terminates at @xmath127 and @xmath128 and mc phase starts from this point .
it follows from this equation that when the gates are always open , _
i.e. _ @xmath129 , then the coexistence line is given by @xmath13 , as in regular tasep , but it terminates at @xmath130 , instead of @xmath131 .
however , for general values of @xmath19 and @xmath21 , the coexistence line has a curvature .
we compare the coexistence line obtained from mean - field theory ( continuous lines ) with that from simulation ( discrete closed symbols ) in fig .
[ fig : coex ] .
we find that there is good qualitative agreement between the two but there is a systematic quantitative deviation .
for example , at @xmath52 , @xmath71 , our simulation shows there is no mc phase and this is consistent with the criterion derived in section [ sec : mc ] .
however , mean - field systematically under - estimates @xmath65 values for all @xmath2 and the mismatch becomes more prominent for large @xmath2 values .
this also affects the mc phase boundaries . for @xmath68 , mean - field
predicts that the coexistence line should terminate at @xmath132 and @xmath133 .
but our simulation gives @xmath134 and @xmath135 .
similarly , for @xmath69 , mean - field prediction is @xmath136 and @xmath137 , whereas our simulation yields @xmath138 and @xmath139 .
thus there is significant mismatch between the mean - field and simulation results . to improve the quantitative agreement with simulations , we develop a refined mean - field theory in the next section
unlike the simple mean field theory described above , where all correlations between the sites are ignored , in this refined mean - field approach , we take into account certain correlations in the system , which are expected to be more important than others , in a self - consistent way , as was first demonstrated in a model for a dynamically extending exclusion process @xcite
. in the bulk of the system one still expects the correlations to be weak .
but close to the boundary this assumption breaks down and strong correlations may be present between the sites . for example , in the limit of small @xmath19 when the residence time of the left gate in the closed state is large , an injected particle remains trapped at site @xmath1 for long enough such that the particle at site @xmath18 ( if any ) has time to drift away towards the bulk of the system before the gate opens .
when the gate opens finally , the trapped particle hops to site @xmath18 . therefore , for small values of @xmath19 , a strong anti - correlation develops between sites @xmath1 and @xmath18 . for general values of @xmath140 and @xmath3 , depending on the interplay of the time - scales associated with boundary injection , exit and switching process of the left gate ,
the correlation between the sites @xmath1 and @xmath18 can become positive or negative and vary in strength .
similarly , at the right boundary a correlation develops between sites @xmath20 and @xmath0 .
let @xmath141 and @xmath142 be the correlation between the sites near the left and right boundary , respectively .
unlike simple mean - field theory , we do not replace these correlations by simple product , rather we treat @xmath143 and @xmath144 as additional variables .
all other correlations in the system are assumed to be factorized .
therefore , the equations of motion are : @xmath145 in steady state , when all time - derivatives vanish , we express @xmath143 , @xmath99 , @xmath146 in eqs [ eq : r1 ] and [ eq : ca ] in terms of @xmath83 and @xmath62 to obtain the following equation @xmath147 = 0 \label{eq : r2j}\ ] ] for a given @xmath62 the above quadratic equation can be solved for @xmath83 : @xmath148 } \label{eq : r2q}\ ] ] from here , one can work out the conditions for the mc phase , where @xmath149 and @xmath117 .
the last criterion implies that in eq .
[ eq : r2q ] only positive root should be taken and one must have @xmath150 \ge ( \alpha j + p_a j)^2\ ] ] for @xmath119 in mc phase this condition gives @xmath151 this is the condition that must be satisfied at the left boundary of the chain to observe mc phase . to work out the condition for the right boundary , we consider eqs .
[ eq : rl-1 ] , [ eq : rl ] and [ eq : cb ] from where we replace @xmath84 , @xmath152 and @xmath144 in terms of @xmath153 and @xmath62 to obtain the following equation .
@xmath154 for a given @xmath62 this equation can be solved for @xmath153 @xmath155 from which negative root must be taken since in mc phase @xmath156 .
in addition , one must satisfy @xmath157 and for @xmath119 this gives @xmath158 eqs .
[ eq : mca ] and [ eq : mcb ] define the conditions that must be satisfied in order for the system to support an mc phase
. the hd - ld coexistence line can be obtained from matching the solutions of eqs .
[ eq : r2j ] and [ eq : rlj ] .
for example , in eq .
[ eq : rlj ] we use the conditions for hd phase and substitute @xmath159 and @xmath160 to obtain @xmath161 which has a solution @xmath162 and another trivial solution @xmath163 which we neglect . since @xmath164
, one must have @xmath165 for hd phase , which is consistent with the criterion obtained in eq .
[ eq : mcb ] . for ld phase we use eq .
[ eq : r2j ] and substitute @xmath166 and @xmath167 to obtain the following cubic equation @xmath168 the solution for @xmath95 obtained from this equation must match that obtained in eq .
[ eq : rp ] on the hd - ld coexistence line .
therefore , the equation for coexistence line is obtained by substituting @xmath169 in eq .
[ eq : cube ] . , @xmath19 and @xmath81 .
the coexistence surface is the one demarcated by the thick blue lines .
the region enclosed by the thick red lines indicates the maximal current phase .
the region above the coexistence surface is the low - density phase , while the region below the coexistence surface is the high - density phase . ] in fig .
[ fig : jb ] we show variation of @xmath62 with @xmath3 ( dotted lines ) for different @xmath26 values .
the critical point @xmath65 is the value of @xmath3 for which the current @xmath62 reaches a saturation .
as can be seen from the figure , refined mean field theory consistently predicts a more accurate value for @xmath65 as compared to mean field theory .
the actual value of the current is also predicted better by the refined mean field theory .
the remaining discrepancy between the theory and simulation results may be attributed to the presence of non - zero correlations extending into the bulk , beyond the first and last site correlations taken into account in our refined mean - field approach . in fig .
[ fig : coex ] we plot the coexistence line ( dotted line ) from refined mean - field theory and compare with the simulation results . even in this case
we find that compared to simple mean - field , the refined mean - field prediction shows a better quantitative agreement with simulation results . in fig .
[ fig:3d ] we present the complete phase diagram in three dimension with the three independent parameters , @xmath2 , @xmath19 and @xmath81 as axes .
the coexistence region ( eq . [ eq : cube ] ) is a surface in this three dimensional space , as shown in fig .
[ fig:3d ] .
in this paper , we have studied the phase diagram of an open - chain tasep whose entrance and exit at the boundaries are controlled by two gates , whose dynamics are chosen so as to mimic ion - channel gates in cell membrane . in our model , these gates can switch between a high - affinity state , when it strongly binds to the particles thereby trapping them , and a low - affinity state , when particle immediately dissociates from the gates and move to the next site .
this immediate dissociation implies even in the limit when the gates are always open , our model does not map onto a regular open - chain tasep .
however , all our results and conclusions remain valid in a slightly different version of the model , where there is no immediate dissociation in the open state ( see appendix [ sec : passive ] ) and usual tasep limit is recovered , when the gates are always open .
the effect of stochastic gates on the phase diagram was studied earlier where a two - way coupling between the gating dynamics and particle flow was considered @xcite .
while entrance of exit of particles can take place only through an open gate , immediately after passage the gate shuts close . on the other hand , in our model ,
no such two - way coupling is assumed and the gates can switch between two states , completely independent of the particle flow .
even this uncoupled , independent gating dynamics yields important modification in the phase diagram which can not be reproduced by mere rescaling of entrance and exit rates .
stochastic gating , as investigated in our model , allows systems to actively control the phase diagram , and hence the current flow through the lattice , independent of the entry and exit rates .
the entry and exit rates often depend on the concentration of particles near the boundary sites , and gating then allows a mechanism for systems to prevent flow even in the presence of high concentrations ( or equivalently , high entry rate ) and vice versa .
this opens new possibilities for mechanisms for the regulation of particle current , especially in biological systems .
it might be of interest to include more biologically relevant features in our model and check how generic our present results are .
mkm would like to acknowledge support from the ramanujan fellowship ( rjn-09/2012 ) , department of science and technology , india and the ircc seed grant from iit - bombay .
we evolve the system following random sequential update where a randomly chosen site is updated according to the steps outlined below .
\(a ) choose an integer @xmath170 within @xmath171 $ ] .
\(b ) if @xmath172 , then a particle hops from site @xmath170 to @xmath173 , if the @xmath170-th site is occupied and @xmath174-th site is empty .
\(c ) if @xmath175 , and if @xmath176-th site is occupied and @xmath0-th site is empty , a particle hops from @xmath20 to @xmath0 . after the hopping takes place , the particle attempts to exit the system with probability @xmath3 , if the gate at site @xmath0 is open .
if the hopping attempt is unsuccessful , the particle stays at site @xmath0 .
\(d ) if @xmath177 , then if site @xmath1 is empty , a particle is injected into site @xmath1 with probability @xmath2 . if the gate at site @xmath1 is open , the injected particle immediately hops to site @xmath18 , if site @xmath18 is empty .
\(e ) if k=1 , and if the gate at site @xmath1 is open , close it with probability @xmath178 . if the gate is closed , open it with probability @xmath19 and if there is a particle , it hops to site @xmath18 , if site @xmath18 is empty .
\(f ) if @xmath179 and if the gate at site @xmath0 is open , close it with probability @xmath180 . if the gate is closed , open it with probability @xmath21 and after opening the gate , the particle at site @xmath0 , if any , attempts to exit the system with probability @xmath3 .
in this section , we discuss a variant of our model that maps onto the regular open - chain tasep in the limit when the gates are always open .
the only difference between this model and the one discussed in the main part of the paper is that in the open state of the gate there is no immediate dissociation . in other words ,
the hopping rate of a particle from an open gate is same as the hopping rate from any bulk site ( as opposed to an infinite hopping rate that stands for immediate dissociation ) .
the mean - field equations for this system read @xmath181 it is easy to see that in the limit @xmath24 this model maps onto regular open - chain tasep . following the method illustrated in section [ sec : mft ] these equations
can be solved to derive the following equation for the coexistence line @xmath182^{1/2 } = 2 \beta p_b\ ] ] which terminates at @xmath183 and @xmath184 and mc phase starts from here .
the coexistence line obtained from simulation ( data not shown ) matches qualitatively with this calculation but there is a quantitative mismatch .
this is taken care of when refined mean - field theory is used .
following the steps outlined in section [ sec : rmft ] , it is easy to show that within refined mean - field theory the equation for the coexistence line becomes @xmath185 = 0\ ] ] this line shows much better quantitative agreement with the simulation ( data not presented here ) . | abstract : we study an open - chain totally asymmetric exclusion process ( tasep ) with stochastic gates present at the two boundaries .
the gating dynamics has been modeled keeping the physical system of ion - channel gating in mind .
these gates can randomly switch between an open state and a closed state . in the open state ,
the gates are highly permeable such that any particle arriving at the gate immediately passes through . in the closed state ,
a particle gets trapped at the gate and can not pass through until the gate switches open again .
we calculate the phase - diagram of the system and find important and non - trivial differences with the phase - diagram of a regular open - chain tasep . in particular , depending on switching rates of the two gates , the system may or may not admit a maximal current phase .
our analytic calculation within mean - field theory captures the main qualitative features of our monte carlo simulation results .
we also perform a refined mean - field calculation where the correlations at the boundaries are taken into account .
this theory shows significantly better quantitative agreement with our simulation results . |
one of the most promising avenues for physics beyond the standard model ( sm ) is supersymmetry ( susy ) @xcite , since it can provide a fundamental understanding of electroweak symmetry breaking ( ewsb ) and it allows unification of the electroweak and strong interactions at a grand unified scale @xcite .
because of its great theoretical attraction , extensive phenomenological work continues to explore the ways for discovery and precision study of supersymmetric particles at present and future colliders .
most of these investigations are directed to the minimal supersymmetric standard model ( mssm ) , which has the minimal new particle content @xcite .
the mssm contains two higgs doublets which develop vacuum expectation values @xmath10 and @xmath11 that break the @xmath12 gauge symmetry spontaneously @xcite .
there are 5 physical higgs bosons in the mssm : two cp - even states @xmath13 and @xmath1 , a cp - odd state @xmath2 , and two charged states @xmath14 ; the lightest higgs boson is @xmath13 .
the ratio @xmath15 is a critical parameter of the mssm : it characterizes the relative fraction that the two higgs doublets contribute to the ewsb , and it enters all sectors of the theory .
the interactions of both the susy particles and the higgs bosons depend on @xmath3 , and the relations of susy particle masses to the soft symmetry breaking parameters of supersymmetry involve @xmath3@xcite .
a measurement of @xmath3 from one sector will thereby allow predictions or tests in other sectors .
the renormalization group evolution of the yukawa couplings from the unification scale to the electroweak scale are sensitive to the value of @xmath3 .
the large top quark mass can naturally be explained with @xmath16 unification as a quasi - infrared fixed point of the top yukawa coupling if @xmath17 or @xmath18 @xcite .
the possibility of so(10 ) yukawa unification requires the high @xmath3 solution @xcite .
the predicted mass of the lightest susy higgs boson also depends on @xmath3 , with @xmath19gev at @xmath17 and @xmath20 gev at @xmath21 @xcite . because of the significance of @xmath3 for the theory and phenomenology of the mssm , it is important to find processes in which @xmath3 can be best determined .
some regions of the mssm parameter space have been excluded at lep2 @xcite due to the lower bound on the lightest higgs boson mass ( @xmath22 ) , particularly at low @xmath3 near 1 .
much of the parameter space remains to be explored at the upgraded tevatron @xcite , the lhc @xcite , the future linear colliders @xcite and muon colliders @xcite .
the @xmath3 constraints that may be obtained from @xmath22 via radiative corrections @xcite , or from precision electroweak measurements @xcite , or from susy particle production usually depend also on other susy parameters .
furthermore , measurements of @xmath23 or @xmath24 via other susy processes without directly involving higgs bosons do not accurately determine large @xmath3 values @xcite .
for general susy higgs phenomenology , we refer the readers to reviews @xcite .
the higgs couplings of @xmath25 to heavy quarks are given by @xmath26 , \label{equ : htb}\end{aligned}\ ] ] where the decoupling limit @xmath27 has been assumed for the approximate forms of @xmath28 . in this limit
, the lightest higgs boson @xmath13 becomes sm - like and its couplings are insensitive to susy parameters . for the @xmath25 higgs bosons ,
@xmath3 is essentially the unique parameter for higgs - heavy quark couplings .
this suggests that studies of the associated production of the higgs bosons and heavy quarks may effectively probe the @xmath3 parameter .
heavy higgs boson production at future @xmath0 colliders was discussed in ref . @xcite . in a recent study feng and moroi@xcite evaluated the prospects for determining @xmath3 in @xmath0 collisions with @xmath29 and 1 tev c.m .
energy via the processes @xmath30 the primary channel in their study involves the @xmath31 coupling .
they found that the strong dependence of heavy higgs branching fractions on @xmath3 allows stringent constraints to be placed for moderate @xmath3@xcite in the mssm . in the present paper ,
we report results of a complementary study of the associated production of a neutral higgs boson and heavy quarks @xmath32 these processes involve @xmath33 and @xmath34 production separately and are thereby expected to be complementary for low and high values of @xmath3 .
we study the sensitivity to probe @xmath3 in two scenarios : the minimal supergravity model ( msugra ) and the mssm .
the paper is organized as follows : we present the higgs decay branching fractions and the cross sections for the associated production of the higgs bosons and heavy quarks in sec .
we analyze the sensitivity to determine the value of @xmath3 at future linear colliders in sec .
we discuss our results , make some general remarks and conclude in sec .
the mass matrix of the cp - even higgs bosons of the mssm is given by @xmath35 where @xmath36 represents the radiative corrections . at tree level ,
the input parameters are @xmath37 and @xmath3 , and the phenomenology is relatively easy to analyze
. the radiative corrections may be substantial in this cp - even sector and they are dependent on other susy parameters , especially on the masses and couplings / mixings of the top quark and heavy scalar quarks ( squarks ) @xcite . in a general mssm analysis
the parameters required as input are @xmath38 where @xmath39 is the soft susy breaking mass parameter of left - handed stop ( where only the heavy third generation parameters are relevant ) , @xmath40 the susy breaking mass parameter of right - handed stop ( sbottom ) , @xmath41 the gaugino masses , @xmath42 the stop trilinear soft breaking term , and @xmath43 the higgs mixing parameter .
the large parameter space involved makes phenomenological studies difficult . on the other hand ,
once a precision measurement is made in the higgs sector in future collider experiments , we would expect to learn more about the susy sector due to the radiative relations among the physical susy masses . instead of exploring the large space of the mssm soft parameters , we focus on the following two scenarios for illustration .
* * motivated by the msugra model and the requirements of radiatively generated electroweak symmetry breaking ( ewsb ) , we relate the soft susy breaking parameters to the common scalar , fermion and trilinear parameters @xmath44 at the grand unified scale . for specific choices of @xmath3
the results depend on the sign of @xmath43 .
the @xmath45 sign is less constrained by @xmath46 decay @xcite and we adopt this convention in our analysis . we make use of the isajet package @xcite to determine the susy masses and couplings from the gut scale input parameters .
the higgs mass eigenvalues are among the outputs of this program .
these values agree with the corresponding results from the code of ref .
@xcite to a precision @xmath47 .
the soft - supersymmetry - breaking parameters are evolved according to renormalization group ( rg ) equations @xcite . for our illustrations
we make the parameter choice @xmath48 along with the positive sign of @xmath43 . the magnitude of @xmath43 is fixed in terms of @xmath49 through the radiately generated ewsb . for three representative @xmath3 values ,
the mass eigenvalues of higgs bosons and susy soft - breaking terms are listed in table [ para ] . for charginos and neutralinos
, we list only the masses of the lightest ones .
in fact , our choice of the above parameters is somewhat conservative in exploring the susy higgs sector . a large @xmath50 results in heavy @xmath51 .
consequently it leads to the `` decoupling limit '' @xcite so that the lightest higgs boson @xmath13 becomes sm - like and thus insensitive to the susy parameters .
.isajet output parameters based on the msugra input in eq .
( [ msugra ] ) . [ para ] [ cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] 0.2 cm -1 cm
for the msugra scenario , @xmath3 is essentially the only variable after fixing the other soft susy breaking parameters as in eq .
( [ msugra ] ) .
the masses of the @xmath52 higgs bosons decrease as @xmath3 increases .
thus the corresponding higgs branching fractions and the production cross sections at a given energy increase with @xmath3 , especially for large values .
this leads to possible accurate determinations of @xmath3 in msugra ( fig .
[ fig : const ] ) , for high values in particular .
in contrast , for the mssm scenario the masses of @xmath2 and @xmath1 are independent of @xmath3 , and are kept fixed in the analyses . at large @xmath3 values the decay branching fractions and the production cross sections of @xmath2 or @xmath1 with @xmath53 reach a plateau in the mssm . consequently the determination of @xmath3 in that range is less effective .
there are other processes by which @xmath3 may also be constrained : ( i ) chargino pair production in @xmath0 collisions can provide good measurements on @xmath3 for low @xmath3 values @xcite ; ( ii ) @xmath54 mixing can be a sensitive probe of @xmath3 @xcite ; ( iii ) gaugino production in @xmath55 collisions may provide information on @xmath3 @xcite ; ( iv ) kinematical distributions from the decay products of susy particles can be used to determine the @xmath3 value @xcite ; ( v ) the magnetic dipole moment of the muon may be useful for @xmath56 if slepton masses @xmath57 gev @xcite ;
( vi ) the branching fractions of @xmath58 may be useful to set a lower bound @xmath5 @xcite . the alternative methods in ( i)-(iv ) probe either @xmath23 or @xmath24 ;
thus the sensitivity to @xmath3 is degraded at high values of @xmath3 . on the other hand , methods ( v ) and ( vi )
are only effective for high values of @xmath3 .
in contrast , higgs boson production processes under consideration and the @xmath59 process discussed in ref.@xcite are direct probes of @xmath3 with the complementary constraints from @xmath60 and @xmath61 final states at low and high values of @xmath3 , respectively . in summary
, we studied heavy neutral higgs boson production in the minimal supersymmetric theories at a linear collider with @xmath4 tev with the expected integrated luminosities of @xmath62 .
the cross sections have a strong dependence on the fundamental supersymmetry parameter @xmath3 , and thus provide a good way to determine it .
we considered the @xmath7 and @xmath8 final states which are sensitive and complementary in determining @xmath3 . in the supergravity scenario ,
the sensitivity is particularly good for @xmath5 in comparison with other methods , reaching a @xmath6 or better determination in a @xmath63 c.l .
cross section measurement . in the general mssm scenario ,
the interplay between the @xmath7 and @xmath8 channels results in a good determination for @xmath9 ( see table [ result ] ) . for higher values of @xmath3
the sensitivity is weakened .
the accuracy of @xmath3 determination is generally sufficient to distinguish theories with a low value ( @xmath64 ) from a high value @xmath65 and thus to provide information on testing certain guts scenarios
. for recent reviews see , _ e.g. _ , h. haber , scipp-92/33 , hep - ph/9306207 ; r. arnowitt and p. nath , in _
perspectives on supersymmetry _ , p.442 , ed . by g.l .
kane ; m. drees , hep - ph/9611409 ; s. martin , hep - ph/9709356 ; x. tata , hep - ph/9807526 .
j. ellis , s. kelley and d.v .
nanopoulos , phys .
b249 * , 441 ( 1990 ) ; j. ellis , s. kelley and d.v .
nanopoulos , phys .
b260 * , 131 ( 1991 ) ; u. amaldi , w. de boer and h. furstenau , phys . lett . *
b260 * , 447 ( 1991 ) ; p. langacker and m. luo , phys . rev . *
d44 * , 817 ( 1991 ) .
h. haber and r. hempfling , phys .
lett . * 66 * , 1815 - 1818 ( 1991 ) ; m. carena , p.h .
chankowski , s. porkorski and c. wagner , phys . lett .
* b411 * , 205 ( 1998 ) ; s. heinemeyer , w. hollik and g. weiglein , phys . rev . *
d58 * , 091701 ( 1998 ) ; phys .
* b440 * , 296 ( 1998 ) ; r .- j .
zhang , phys .
b447 * , 89 ( 1998 ) ; j.r . espinosa and r
zhang , hep - ph/0003246 .
v. ruhlmann - kleider , talk at _ xix international symposium on lepton and photon interactions at high energies _ , stanford university , aug 9 - 14 ( 1999 ) ; a. quadt , talk at the _ xxxv rencontres de moriond @xmath66 electroweak and unified theories _
, les arcs , savoie , france , march 11 - 14 ( 2000 ) .
m. carena , s. mrenna and c.e.m .
wagner , phys . rev .
* d60 * , 075010 ( 1999 ) ; c. balazs , j.l .
diaz - cruz , h.j .
he , t. tait and c.p .
yuan , phys . rev .
* d59 * , 055016 ( 1999 ) ; see also in physics at run ii workshop : higgs boson working group report , http://fnth37.fnal.gov//higgs.html .
haber , t. han , f.s .
merritt and j. womersley , in _ proceedings of the 1996 dpf / dpb summer study on new directions for high energy physics _ , hep - ph/9703391 ; j.f .
gunion , l. poggioli and r. van kooten , in _ proceedings of the 1996 dpf / dpb summer study on new directions for high energy physics _ , hep - ph/9703330 .
p. nath and r. arnowitt , phys
b336 * , 395 ( 1994 ) ; phys . rev . *
d54 * , 2374 ( 1996 ) ; v. barger , m.s .
berger , p. ohmann and r.j.n .
phillips , phys .
rev . * d51
* , 2938 ( 1995 ) ; h. baer and m. brhlik , phys . rev . * d55 * , 3201 - 3208 ( 1997 ) ; h. baer , m. brhlik , d castano and x. tata , phys . rev .
* d58 * , 015007 ( 1998 ) ; m. ciuchini , g. degrassi , p. gambino and g.f . giudice nucl . phys . *
b534 * , 21 ( 1998 ) .
chamseddine , r. arnowitt and p. nath , phys .
* 49 * , 970 ( 1982 ) ; l. ibaez and g. ross , phys .
* b110 * , 215 ( 1982 ) ; j. ellis , d. nanopoulos and k. tamvakis , phys . lett . *
b121 * , 123 ( 1983 ) ; l.j .
hall , j. lykken and s. weinberg , phys .
* d27 * , 2359 ( 1983 ) ; l. alvarez - gaum , j. polchinski and m. wise , nucl .
* b121 * , 495 ( 1983 ) . | we study the production at future @xmath0 linear colliders of the heavy neutral higgs bosons @xmath1 and @xmath2 of the minimal supersymmetric standard model in association with top and bottom quarks .
the cross sections have a strong dependence on the parameter @xmath3 , and thus provide a good way to determine it . at a linear collider with @xmath4 tev and expected integrated luminosities , we find significant sensitivities for determining @xmath3 . in the supergravity scenario ,
the sensitivity is particularly strong for @xmath5 , reaching a @xmath6 or better measurement . in the general mssm scenario , the interplay between the @xmath7 and @xmath8 channels results in a good determination for @xmath9 , while the sensitivity is weakened for higher values of @xmath3 . |
the study of isospin asymmetric nuclear matter is closely related to a few issues which are presently the subjects of ongoing theoretical and experimental researches . at first
, the isospin dependence of the nuclear equation of state ( eos ) shows up in the density dependence of the symmetry energy . a quantity which drives the structure of neutron rich nuclei and at supra - high densities the physics of supernovae explosions and the structure of compact neutron stars @xcite .
for instance , the stiffness @xcite of the symmetry energy is directly related to the proton - fraction in @xmath1-equilibrated neutron stars , which is limited by the onset of the urca - process @xcite . however , the symmetry energy is empirically known up to saturation densities , where it can be determined in nuclear structure studies from giant and pigmy resonances and from neutron skin thickness @xcite . in the laboratory
the symmetry energy can be explored in heavy ion collisions , where the knowledge of the high density eos is mandatory @xcite for the description of the collision dynamics @xcite within transport theory @xcite .
different heavy ion observables , _
e.g. _ , collective transverse ( in - plane ) and elliptic ( out - of - plane ) flows @xcite of particles and/or light fragments , as well as subthreshold strangeness production @xcite are sensitive to the stiffness of the eos .
furthermore , the isospin tracing @xcite , isoscaling @xcite , collective isospin flows @xcite and isospin ratios of produced mesons @xcite represent plausible tools to constrain the isovector eos at high densities and energies .
moreover , heavy ion collisions allow to probe not only the density of created matter , but also the energy dependence of the nuclear potential @xcite .
theoretically , the density and energy dependence of the isovector eos beyond the saturation density is still controversial . in non - relativistic skyrme models
both possibilities of a soft and a stiff symmetry energies @xcite can be obtained . in the rmf models
the symmetry energy is determined by the lorentz - structure of the isovector potential @xcite , which is described by the vector - isovector @xmath0-field , or by @xmath0- and an attractive scalar @xmath2-fields @xcite . in both cases ,
the symmetry energy rises with increasing baryon density leading to a stiff isovector eos at high densities .
the models with density dependent ( dd ) couplings @xcite predict a soft symmetry energy @xcite , while the dirac - brueckner - hartree - fock ( dbhf ) calculations @xcite and other rmf models with complementary exchange ( fock ) terms ( density - dependent relativistic hartree - fock ( rhf ) @xcite ) favour again a stiff one . concerning the energy dependence ,
the schrdinger equivalent optical potential in symmetric matter is well known phenomenologically @xcite .
however , different empirical analyses give different predictions for the energy dependence of isovector lane - potential @xcite .
the microscopic dbhf @xcite and bhf @xcite models predict a rather constant optical potential at low momenta with a strong decrease at high energies . in skyrme approaches both cases of a strong increase and a strong decrease of the lane - potential as a function of energy
can be realized @xcite .
the dd models do not contain any explicit energy dependence , except if one modifies the original dd formalism @xcite with additional interaction terms @xcite , however , with the cost of additional parameters .
note that , the energy dependence of the optical potential has been so far a crucial problem in standard rmf , where the energy dependence appears only due to relativistic effects .
another question concerns the quasi - particle properties of nucleons in the nuclear matter .
the isospin - dependent forces may result in a splitting between the in - medium masses of protons and neutrons which can strongly influence the dynamics in heavy - ion collisions @xcite .
however , it is still under debate , whether @xmath3 or @xmath4 holds in neutron rich matter . in non - relativistic approaches
the in - medium particle mass is described through the energy dependence of the real part of the phenomenological mean - field @xcite .
again , the skyrme - like interactions predict both possibilities @xcite for the non - relativistic in - medium mass @xcite . in relativistic models
the in - medium dependence of the particle mass arises from the lorentz - scalar character of the relativistic mean - field , i.e. , from the scalar selfenergy @xcite .
it is often denoted as `` effective mass '' or `` dirac mass '' , and these definitions we will use in this work . as in the non - relativistic case
, same situation holds for the dbhf models where some calculations predict for the dirac mass @xmath3 @xcite and others @xmath4 @xcite . in rmf models
the mass splitting is merely fixed by the lorentz structure of the isovector mean - field . due to the lorentz - vector character of the @xmath5-interaction the minimal @xmath6-models
do not generate any in - medium mass splitting , except if one in addition introduces the lorentz - scalar @xmath2-field in the isovector channel @xcite , or if one includes explicitly exchange ( fock ) terms in the relativistic formalism @xcite .
including the @xmath2-field one gets @xmath3 , while in the rhf models both cases @xmath7 are possible .
so far these were the only possibilities to generate the mass splitting in the relativistic mean field theory @xcite . in this work we study the isovector interactions in asymmetric nuclear matter .
our studies are based on the non - linear derivative ( nld ) model to rmf proposed in @xcite .
this approach describes remarkably well the bulk properties of nuclear matter and , in particular , the density dependence of the nuclear eos and the energy dependence of the proton - nucleus @xcite as well as the antiproton - nucleus schrdinger equivalent optical potentials @xcite .
the present paper is organized as follows . in section 2
we briefly outline the nld model . in section
3 the approach is applied to isospin symmetric and isospin asymmetric nuclear matter . the numerical realization and details of the calculations
are described in section 4 .
the results and discussions are presented in section 5 .
we finally end up with the summary in section 6 .
the nld approach is based on the lagrangian density of relativistic hadrodynamics ( rhd ) @xcite and describes the in - medium interaction of nucleons through the exchange of auxiliary scalar - isoscalar @xmath8 , vector - isoscalar @xmath9 and vector - isovector @xmath10 meson fields .
the lagrangian density is given by @xmath11 - \overline{\psi}\psi m \nonumber\\ & - & \frac{1}{2}m^{2}_{\sigma}\sigma^{2 } + \frac{1}{2}\partial_{\mu}\sigma\partial^{\mu}\sigma + \frac{1}{2}m^{2}_{\omega}\omega_{\mu } \omega^{\mu } -\frac{1}{4}f_{\mu\nu}f^{\mu\nu } + \frac{1}{2}m^{2}_{\rho}\vec{\rho\,}_{\mu}\vec{\rho\,}^{\mu } -\frac{1}{4}\vec{g\,}_{\mu\nu}\vec{g\,}^{\mu\nu } \nonumber\\ & + & { \cal l}_{int } \quad , \label{ndc - free}\end{aligned}\ ] ] where the first two terms describe the ( symmetrized ) lagrangian for the nucleon field @xmath12 with bare mass @xmath13 and the second line contains the standard lagrangian densities for @xmath8 , @xmath9 and @xmath10 fields with masses @xmath14 , @xmath15 and @xmath16 , respectively . the strength tensors are defined as @xmath17 and @xmath18 for the vector - isoscalar and vector - isovector fields , correspondingly . in eq .
( [ ndc - free ] ) the last term @xmath19 contains the interaction between the meson and nucleon fields . in the conventional rhd
@xcite the meson fields couple to the nucleons via the corresponding lorentz structures @xmath20 , -@xmath21 and -@xmath22 .
such minimal walecka - type interactions describe successfully the static properties of nuclear matter around saturation density . however , they are not sufficient for the description of dynamical situations which occur in nucleon - nucleus and heavy - ion collisions , where this minimal model results in a wrong energy dependence of the mean - field .
this has been a crucial problem of the rmf models which already attracted much attention in the past @xcite .
a possible solution to this problem has been suggested in @xcite , where momentum - dependent form factors which suppress the high momentum components of fields were introduced .
this idea has been followed in the nld model @xcite where it was generalized in a manifestly covariant way .
the nld interaction lagrangian introduces the non - linear derivative terms into the conventional rhd meson - nucleon vertices .
in particular , the lagrangian density depends on higher - order partial derivatives of the nucleon field @xmath23 @xmath24 with the symmetrized meson - nucleon interactions given by ( @xmath25 denotes the isospin operator ) @xmath26 \nonumber\\ & - \frac{g_{\omega}}{2 } \left [ \overline{\psi } \ , \stackrel{\leftarrow}{{\cal d } } \gamma^{\mu}\psi\omega_{\mu } + \omega_{\mu}\overline{\psi}\gamma^{\mu } \ , \stackrel{\rightarrow}{{\cal d } } \psi \right ] \nonumber\\ & - \frac{g_{\rho}}{2 } \left [ \overline{\psi } \ , \stackrel{\leftarrow}{{\cal d } } \gamma^{\mu}\vec{\tau}\psi\vec{\rho}_{\mu } + \vec{\rho}_{\mu}\overline{\psi}\vec{\tau}\gamma^{\mu } \ , \stackrel{\rightarrow}{{\cal d } } \psi \right ] \quad , \label{ndc}\end{aligned}\ ] ] with obvious notations for the fields and their couplings .
the new operator @xmath27 acts on the nucleon fields and is the generic non - linear function of partial derivatives . in the model of ref .
@xcite a simple exponential cut - off function has been assumed @xmath28 @xmath29 in eqs .
( [ ope ] ) and ( [ ope2 ] ) @xmath30 denotes an auxiliary dimensionless @xmath31-vector and @xmath32 stands for the cut - off parameter . a way to realize the meaning of the cut - off function is the presence of the momentum ( or energy ) dependent form factors in the one - boson - exchange models , which cut the high energy behaviour of the meson - exchange vertices @xcite .
furthermore , in the limiting case of @xmath33 the standard rhd ( or walecka ) model is retained .
the euler - lagrange equations and the noether currents are derived from variational principles which are applied to the nld lagrangian containing higher order derivatives of the spinor fields @xmath23 and @xmath34 . the generalized euler - lagrange equations read @xcite ( @xmath35 @xmath36 the invariance of the nld lagrangian under global phase transformations results in the following noether current @xmath37 @xcite @xmath38\phi \nonumber\\ + & \left [ \hspace*{2.25 cm } \frac{\partial{\cal l}}{\partial(\partial_{\mu}\partial_{\beta}\phi ) } - \hspace*{0.4 cm } \partial_{\gamma } \frac{\partial{\cal l}}{\partial(\partial_{\mu}\partial_{\beta}\partial_{\gamma}\phi ) } \pm \cdots \hspace*{3.6 cm } \right]\partial_{\beta}\phi \nonumber\\ + & \left [ \hspace*{5.25 cm } \frac{\partial{\cal l}}{\partial(\partial_{\mu}\partial_{\beta}\partial_{\gamma}\phi ) } - \partial_{\delta } \frac{\partial{\cal l}}{\partial(\partial_{\mu}\partial_{\beta}\partial_{\gamma}\partial_{\delta}\phi ) } \pm \cdots \hspace*{0.25 cm } \right]\partial_{\beta}\partial_{\gamma}\phi \nonumber\\ + & \cdots \quad .
\label{noether - current}\end{aligned}\ ] ] furthermore , the invariance under space - time translations gives the energy - momentum tensor @xmath39 @xcite @xmath40\partial^{\nu}\phi \nonumber\\ + & \left [ \hspace*{2.25 cm } \frac{\partial{\cal l}}{\partial(\partial_{\mu}\partial_{\beta}\phi ) } - \hspace*{0.4 cm } \partial_{\gamma } \frac{\partial{\cal l}}{\partial(\partial_{\mu}\partial_{\beta}\partial_{\gamma}\phi ) } \pm \cdots \hspace*{3.4 cm } \right]\partial_{\beta}\partial^{\nu}\phi \nonumber\\ + & \left [ \hspace*{5.25 cm } \frac{\partial{\cal l}}{\partial(\partial_{\mu}\partial_{\beta}\partial_{\xi}\phi ) } - \partial_{\gamma } \frac{\partial{\cal l}}{\partial(\partial_{\mu}\partial_{\gamma}\partial_{\beta}\partial_{\xi}\phi ) } \pm \cdots
\right]\partial_{\beta}\partial_{\xi}\partial^{\nu}\phi \nonumber\\ + & \cdots \nonumber\\ - & g^{\mu\nu}{\cal l } \quad . \label{noether}\end{aligned}\ ] ] the euler - lagrange equations ( [ euler ] ) and the noether theorems ( [ noether - current ] ) and ( [ noether ] ) contain now an infinite series of higher order derivatives of the nucleon field .
it was shown in @xcite that in mean - field approximation ( up to derivatives of the meson fields which in any case vanish in nuclear matter ) these infinite series can be resumed exactly .
indeed , the application of the generalized euler - lagrange equations ( [ euler ] ) to the full lagrangian density ( [ ndc ] ) with respect to the field @xmath34 leads to the following dirac equation @xcite @xmath41\psi = 0 \ ; , \label{dirac}\ ] ] with the selfenergies @xmath42 and @xmath43 given by @xmath44 both lorentz - components of the selfenergy , @xmath42 and @xmath43 , show a linear dependence with respect to the meson fields @xmath10 , @xmath9 and @xmath8 , as in the standard rmf .
however , they contain an additional non - linear dependence on the partial derivatives .
the following proca and klein - gordon equations for the meson fields are obtained @xmath45 , \label{omega_meson } \\
\partial_{\mu}\vec{g}^{\mu\nu } + m_{\rho}^{2}\vec{\rho\,}^{\nu } & = & \frac{1}{2}g_{\rho } \left [ \overline{\psi}{e^{\frac{i\stackrel{\leftarrow}{\partial}_{\alpha}v^{\alpha}+m}{\lambda}}}\gamma^{\nu}\vec{\tau}\psi + \overline{\psi}\vec{\tau}\gamma^{\nu}{e^{\frac{-v^{\alpha}i\stackrel{\rightarrow}{\partial}_{\alpha}+m}{\lambda}}}\psi \right ] , \label{rho_meson } \\ \partial_{\mu}\partial^{\mu}\sigma + m_{\sigma}^{2}\sigma & = & \frac{1}{2}g_{\sigma } \left [ \overline{\psi}{e^{\frac{i\stackrel{\leftarrow}{\partial}_{\alpha}v^{\alpha}+m}{\lambda}}}\psi + \overline{\psi}{e^{\frac{-v^{\alpha}i\stackrel{\rightarrow}{\partial}_{\alpha}+m}{\lambda}}}\psi \right ] \ ; , \label{sigma_meson}\end{aligned}\ ] ] where eqs .
( [ omega_meson ] , [ rho_meson ] , [ sigma_meson ] ) show a similar form as in conventional rmf ( @xmath46 ) , except of the highly non - linear structure in the source terms .
we apply now the nld model to isospin - asymmetric nuclear matter . in the rmf approximation
all meson fields are treated as classical fields . in this approximation
the spatial components of the vector meson fields vanish .
moreover , only the third component of the @xmath0-meson field in isospin space survives . to simplify the notation , the time - like component of the @xmath47 field
will be labelled as @xmath48 .
similar notation holds also for the time - like and third components in lorentz- and isospin - space , respectively , of the vector - isovector @xmath49 field , which will be denoted as @xmath0 .
the baryon density will be labelled as @xmath50 .
for the auxiliary vector we choose @xmath51 .
note that , additional rearrangement terms can appear for different choices of @xmath52 in the nld formalism .
however , they exactly vanish in the mean - field approximation to infinite nuclear matter @xcite . the dirac equation , see eq . ( [ dirac ] ) ,
can be solved using the following _ ansatz _ @xmath53 where @xmath54 and @xmath55 stand , respectively , for the spin and @xmath31-momentum of the proton ( @xmath56 ) or neutron ( @xmath57 ) , @xmath58 denotes the space - time coordinate , and @xmath59 is the dirac spinor for positive energy states .
the dirac equation ( [ dirac ] ) maintains its form in infinite nuclear matter with selfenergies given by @xmath60 the upper ( lower ) sign in eq .
( [ selfenvnm ] ) refers from now on always to protons , @xmath61 , ( neutrons , @xmath62 ) .
the dirac equation is solved inserting eqs .
( [ plane_wave ] ) , ( [ selfenvnm ] ) and ( [ selfensnm ] ) into eq . ( [ dirac ] ) for protons ( @xmath61 ) or neutrons ( @xmath62 )
@xmath63 with the in - medium energy and mass given by @xmath64 and @xmath65 these quantities are related to each other through the dispersion relation @xmath66 which , because of the @xmath0-field , is different for protons and neutrons .
the solutions of the dirac equations for protons and neutrons are found in the usual way @xmath67 with the spin eigenfunctions @xmath68 .
the factor @xmath69 is defined as @xmath70 and normalizes the spinors as follows @xmath71 and @xmath72 .
in nuclear matter the meson field equations of motion read @xmath73 where the vector - isoscalar @xmath74 , vector - isovector @xmath75 and scalar - isoscalar @xmath76 source terms take the forms ( @xmath77 is the spin degeneracy factor ) @xmath78 contrary to the walecka model , in the nld approach the vector - density @xmath74 is not a conserved nucleon density .
latter one has to be derived from the generalized noether - theorem , see eq .
( [ noether - current ] ) .
following ref .
@xcite the conserved nucleon density after the resummation procedure takes the form @xmath79 \ , , \nonumber\end{aligned}\ ] ] where the expectation value @xmath80 is just the usual density of the walecka model @xcite .
the equation of state ( eos ) is obtained from the @xmath81-component of the energy - momentum tensor @xmath39 .
the resummation procedure described in @xcite results in @xmath82 with the additional expectation values @xmath83 the pressure @xmath84 is given as usual by @xmath85 however , since in the nld approach rearrangement terms do not appear @xcite , thermodynamic consistency is always fulfilled .
then , the pressure can be obtained from the energy density by using the thermodynamic relation @xmath86 note that in eq .
( [ press ] ) the conserved baryon density @xmath50 from eq .
( [ rhobar ] ) appears , and not the standard walecka density .
two important effects appear in nld model , which are absent in the minimal @xmath6-models . at first , the selfenergies depend not only on density , but also explicitly on single - particle energy , as discussed in detail in ref .
these feature is retained also in isospin - asymmetric nuclear matter . as an additional novel feature of the nld model , not only the lorentz - vector component of the nucleon selfenergy , but
also its lorentz - scalar part depends implicitly ( through the dispersion relation ) on the isovector @xmath0-meson field .
this feature arises from the isospin dependence of the single - particle energy , and it will generate an in - medium mass splitting between protons and neutrons in isospin asymmetric matter . note that the @xmath8 field is also affected by isospin effects , but the residual impact of the dispersion relation is negligibly small in this case .
the meson - field equations depend explicitly on the baryon density , but only implicitly on the single - particle energy .
latter quantity is integrated out in the source terms .
therefore , it is sufficient to solve selfconsistently only the coupled set of meson - field equations ( [ mesonsnm ] ) at a given density , and then determine the proton or neutron single - particle energy by solving numerically the dispersion relation ( [ mass - shel ] ) for a given proton or neutron momentum relative to the nuclear matter at rest . for the density dependence
we proceed as follows : the input parameters are the fermi - momentum and the isospin - asymmetry .
the isospin - asymmetry parameter @xmath87 is determined according to the standard definition @xmath88 the set of the meson field equations ( [ mesonsnm ] ) is solved then selfconsistently until convergence is achieved .
for each iteration step the integrals in eq .
( [ dens-1 ] ) are evaluated numerically , where at each momentum step the dispersion relation , see eq .
( [ mass - shel ] ) , has to be calculated also numerically .
the whole procedure provides the meson - fields , with which we calculate the corresponding conserved baryon density according eq .
( [ rhobar ] ) and the isospin - asymmetry parameter according eq .
( [ alpha ] ) .
note again , that the conserved baryon density @xmath50 differs from the walecka one @xmath89 .
finally , the single - particle energies , which appear explicitly in the selfenergies , are evaluated at the corresponding proton or neutron fermi - momentum . for the energy dependence at a fixed baryon density
the procedure is similar as outlined above .
we first calculate the meson - fields at a given density and isospin - asymmetry .
then we solve numerically the dispersion relation for protons ( neutrons ) .
in the first case the additional input parameter is the proton ( neutron ) momentum by keeping the momentum of the other isospin state at the corresponding fixed fermi - momentum .
the in - medium energy of a proton ( @xmath61 ) or neutron ( @xmath62 ) is defined then as @xmath90 where the proton or neutron lorentz - vector selfenergy is given by eq .
( [ selfenvnm ] ) . in principle
, the nld model contains no parameters .
the original isoscalar @xmath91 and @xmath92 couplings can be taken from any linear walecka model , e.g. , @xcite , as it has been done in ref .
@xcite and in this work .
the @xmath5 coupling @xmath93 close to its universal value is used @xcite which is also well determined by the one - boson - exchange model @xcite .
a small ( big ) numerical value of @xmath32 provides a stronger ( weaker ) density and energy dependence of the mean - field , which influences the density dependence of the equation of state for densities at and above saturation and , at the same time , strongly affects the energy dependence of the schrdinger - equivalent optical potential at saturation density .
the simplest way followed here is to fix @xmath32 just from the empirically well - known energy dependence of the schrdinger equivalent optical potential .
the cut - off parameter @xmath32 is set to @xmath94 mev @xcite .
a different numerical value for the cut - off parameter is of course possible if one uses a different walecka - like parametrization than the original one or if one adopts a different functional form for the non - linear operator than the simpler exponential one .
( 18.,8.0 ) ( 0.25,0.0 ) before presenting the calculations for isospin asymmetric nuclear matter we discuss briefly the nld results for the special cases of symmetric matter ( @xmath95 ) and pure neutron matter ( @xmath96 ) . in fig .
[ fig1 ] we present the eos for @xmath95 and @xmath96 in terms of the pressure density as function of the baryon density .
here we compare the empirical heavy - ion data @xcite , see the shaded bands in fig .
[ fig1 ] , with nld and walecka models .
the inserted panels in fig .
[ fig1 ] shows the nuclear eos in terms of the binding energy per nucleon .
the conventional walecka model ( dashed curves ) leads to a very stiff eos for symmetric nuclear matter and also pure neutron matter with a high compression modulus at saturation , because of the well - known linear divergent behaviour of the repulsive @xmath48 field . on the other hand , the lorentz - vector mean - field
is suppressed to large extend at high baryon densities in the nld model .
see further discussions in ref .
this effect , which is retained also for the extreme case of pure neutron matter , weakens considerably the stiffness of the high - density eos , see the insert in fig .
[ fig1 ] , and the empirical region in the pressure - density phase diagram is reproduced fairly well by the nld approach ( solid curve )
. however , not only the density dependence of the nuclear mean - field , but also its energy dependence matters for nuclear reactions .
this can be demonstrated using the in - medium optical potential which is defined as follows @xmath97 ( 18.,8.0 ) ( 2.0,0.0 ) as shown in fig .
[ fig2 ] , the empirical data from dirac phenomenology ( open diamonds ) predict a saturation of the in - medium nucleon optical potential @xcite at intermediate energies .
as well known , standard walecka - type rmf models ( rhd , dashed curve ) are not able to reproduce this empirical saturation , and strongly overestimate the data at high energies .
this disagreement remains also in other rmf models ( @xmath98 , dot - dot - dashed curve ) , where the meson - nucleon couplings depend explicitly on the baryon density .
a better description can be achieved here by the consideration of additional energy dependent terms in the interaction lagrangian ( @xmath99 , dot - dashed curve ) . however , with the cost of many additional parameters @xcite .
microscopic dbhf models ( dbhf , dot - dashed - dashed curve ) reproduce the optical potential in average up to energies close to pion production threshold @xcite . on the other hand , the nld model ( nld , solid curve )
not only softens the eos at high densities , but simultaneously generates the correct energy behaviour of the nucleon - nucleus optical potential at high energies .
note that , the same nld approach describes the puzzling energy dependence of the antiproton - nucleus potential , see ref .
@xcite .
( 18.,8.0 ) ( 2.0,0.0 ) in fig . [ fig3a ]
we present the density dependence of the nucleon selfenergies for various asymmetry parameters @xmath87 . in symmetric nuclear matter ( @xmath95 )
, the nld vector @xmath100 and scalar @xmath101 selfenergies ( solid curves ) saturate at high densities .
as one can see , they essentially follow the density dependence of the microscopic dbhf calculations @xcite ( filled squares ) . as discussed in detail in @xcite , this saturation of the selfenergies with increasing density , in particular , of the repulsive vector part , is responsible for a softening of the eos at high compressions .
( 18.,8.0 ) ( 2.0,0.0 ) the saturation in density remains in the general case of isospin - asymmetric nuclear matter , see the dashed @xmath102 and dot - dashed @xmath103 curves in fig .
[ fig3a ] . as a novel feature , not only the vector , but also the scalar component of the nucleon selfenergy
gets isospin dependent .
the isospin dependence of the scalar selfenergy , observed here , arises from the @xmath0-field induced isospin splitting between the in - medium proton and neutron single particle energies .
this effect will be discussed in more details later on .
the saturation of the selfenergies in density affects considerably the density dependence of the symmetry energy .
the symmetry energy is extracted from the empirical parabolic low and defined as the second derivative of the energy per nucleon versus the asymmetry parameter @xmath87 , or as the energy difference between pure neutron matter and symmetric nuclear matter .
[ fig3 ] ( left panel ) shows the density dependence of the symmetry energy within the nld ( solid curve ) and the walecka ( dashed curve ) models .
the symmetry energy @xmath104 in the standard rmf grows almost linearly with baryon density . here
the symmetry potential @xmath105 is proportional to the @xmath0 field , which increases linearly with density , @xmath50 .
the moderate saturation at very high densities arises from the suppression of the kinetic part @xmath106 ( with @xmath107 being the in - medium fermi - energy ) .
the nld model gives a different behaviour at high baryon densities with respect to the conventional rmf .
a considerable softening of the high density eos is obtained , but now for the isovector sector .
available data for the symmetry energy exist around saturation density only @xcite .
the theoretical models describe the experimental data fairly well , however , they give different predictions for the high - density symmetry energy , see for reviews refs .
@xcite . in non - relativistic models
the adjustment of the model parameters to the empirical saturation point does not necessarily constraint the stiffness of the symmetry energy .
in fact , a soft and a stiff symmetry energy can be obtained at high densities @xcite . in the relativistic walecka - type models ,
the high density dependence of the symmetry energy is always stiff @xcite .
similar results are obtained in the dbhf approach and within density - dependent relativistic hartree - fock ( rhf ) approaches @xcite , see the dashed and dot - dot - dashed curves in the right panel of fig.[fig3 ] , where the symmetry energy becomes stiffer with increasing density @xcite .
it is interesting to note that the rhf models contains also an energy dependence , apart the density dependent coupling functions . on the other hand , the dd models @xcite ,
see the dot - dashed ( dd ) and dot ( dd - me@xmath2 ) curves in the right panel of fig.[fig3 ] , predict a rather soft symmetry energy , which is similar to the nld results ( solid curve )
. direct experimental access to the high - density region of the symmetry energy is still incompletely known . a plausible way to constraint the stiffness of the symmetry energy at supra - normal densities consists in studies of compact neutron stars @xcite .
in fact , the symmetry energy must saturate or , at least , show a rather soft behaviour at high baryon densities in order to respect the direct urca - limit @xcite .
this constraint is satisfied in the nld model .
( 18.,8.0 ) ( 2.0,0.0 ) the density dependence of the nucleon effective or dirac mass in the nld model @xcite is in remarkable agreement with the dbhf calculations .
the insert in fig .
[ fig4 ] describes the dirac mass of the nucleon in symmetric nuclear matter . here
the solid curve corresponds to the nld results and the square symbols describe the dbhf calculations of ref .
@xcite .
interestingly , in the nld model an effective mass splitting between protons and neutrons is generated in asymmetric matter by taking into account at the hartree - level the @xmath0-field only .
this is demonstrated in fig .
[ fig4 ] for two asymmetry parameters @xmath108 ( dashed curves ) and @xmath96 ( dot - dashed curves ) .
in particular , a mass splitting @xmath4 is obtained for neutron - rich matter , which becomes pronounced with increasing density and isospin asymmetry parameter .
the in - medium mass splitting in the nld model is due to the isospin dependence of the single particle energy which appears in the selfenergies .
note that , any explicit energy dependence of the scalar selfenergy will result in a @xmath0-induced mass splitting of nucleons in isospin asymmetric nuclear matter .
as discussed in the introduction , an effective mass splitting between protons and neutrons do not occur in the walecka - type @xmath6-models solved at the hartree - level .
the introduction of the scalar - isovector @xmath2-meson in the nuclear mean - field is necessary for this purpose @xcite .
however , one gets more parameters to fit the bulk asymmetry parameter at saturation density and , in particular , one needs to re - adjust again the @xmath5 coupling .
this procedure can be ambiguous , since the @xmath109 coupling is not fixed neither by nuclear structure @xcite nor by the one - boson - exchange models @xcite .
on the other hand , as it has been investigated intensively in the past , the @xmath2-meson implies interesting effects if one goes beyond the saturation density of ground state nuclear matter and study its influence in compressed matter created in heavy - ion collisions @xcite .
unfortunately , the available experimental data do not allow still decisive conclusions @xcite .
the @xmath110-models predict a mass splitting @xmath3 for neutron - rich matter @xcite .
however , just from the energy dependence of the scalar selfenergy the nld model predicts an opposite splitting @xmath4 for neutron - rich matter . this trend is unique and
can not be reversed in the nld model .
we remind here , that rmf models treated on the hartree - fock level ( rhf ) @xcite , in which the selfenergies depend also explicitly on the single - particle energy due to the fock - terms , predict for the non - relativistic effective mass a mass splitting @xmath4 up to densities @xmath111 , but this trend is reversed for densities @xmath112 @xcite . therefore
, the property @xmath3 is not a genuine feature of the rmf models , as has been believed so far .
( 18.,8.0 ) ( 2.0,0.0 ) ( 18.,8.0 ) ( 2.0,0.0 ) the energy dependence of the isovector nuclear potential is described by the schrdinger - equivalent optical potential , see eq .
( [ v_opt ] ) , but now separated between protons and neutrons .
[ fig5 ] shows these potentials as function of the corresponding particle momentum and different asymmetry parameters @xmath87 .
the observed isospin dependence of the proton ( right panel ) and neutron ( left panel ) optical potentials results from cancellation effects in eq .
( [ v_opt ] ) between different isospin dependent components of the selfenergies .
it is remarkable that the microscopic ( d)bhf calculations @xcite predict very similar behaviour as the nld model .
this is shown in fig .
[ fig6 ] , where the proton and neutron optical potentials are shown as a function of the asymmetry parameter @xmath87 at fixed density @xmath113 and three different momenta , as indicated in fig .
indeed , at low momenta up to the fermi momentum ( left and middle panels ) the isospin splitting between the proton and neutron optical potentials in the nld model ( solid and dashed curves ) follows closely the microscopic calculations ( filled and open squares for dbhf , filled and open triangles for bhf ) . at higher momenta ( right panel )
the nld isospin splitting is similar to the ( d)bhf calculations , up to an absolute value .
this is related to a stronger energy dependence in the nld model relative to these microscopic approaches .
recall , that this strong energy dependence of the nld potential is essential for the description of the schrdinger - equivalent optical potential shown in fig .
[ fig2 ] .
( 18.,8.0 ) ( 2.0,0.0 ) the energy dependence of the isovector optical potential is conventionally described by the lane - type potential .
it is defined as follows @xcite @xmath114 where @xmath115 denotes the proton / neutron schrdinger - equivalent optical potentials .
[ fig7 ] shows the momentum dependence of the lane - potential in the nld approach ( solid curve ) . also
the results of linear walecka model denoted as rhd ( dashed curve ) and different microscopic approaches are shown .
the linear walecka model predicts an almost linear behaviour in energy , as in the isoscalar case , and it diverges for @xmath116 .
the lane potential in the nld model shows an opposite behaviour with respect to rhd .
it decreases with increasing particle momentum .
in contrast to the symmetric proton- and antiproton - nucleus optical potentials @xcite , the empirical situation in the isovector case is still not well known .
different empirical studies predict an opposite behaviour of @xmath117 versus momentum .
in fact , the analysis of madland _ et al . _
@xcite predicts an increase of the isovector potential with increasing momentum , while in the analysis of lane _
et al . _
@xcite a decrease of @xmath117 with rising momentum has been obtained . nevertheless , as shown in fig .
[ fig7 ] , the nld calculations agree with the calculations from the microscopic brueckner theory which also show the decrease of the lane potential as a function of the momentum .
in summary , the nld formalism which incorporates on a mean - field level the energy dependence of the nucleon selfenergies has been applied to isospin - asymmetric nuclear matter . the model describes the bulk properties of symmetric matter and compares well with the results from microscopic brueckner calculations .
due to the explicit energy dependence of the nuclear mean - field , the nld model reproduces very well the empirical proton and antiproton optical potentials @xcite .
same energy dependence is responsible for the softening of the eos at high densities . in the present model
the isospin effects are generated by using the conventional @xmath0-meson exchange .
such a minimal extension of the nld formalism describes the symmetry energy around saturation and predict its softening ( saturation ) at high baryon densities .
this feature is an important element in the description of dense nuclear systems created in heavy ion collisions and/or in the interior of neutron stars .
however , such a soft symmetry energy is at variance with standard rhd models , where it rises linearly with increasing density . as a novel feature
we find that the energy dependent nld interactions generate a mass splitting between protons and neutrons already within the conventional @xmath0-exchange scheme .
in particular , a dirac mass splitting @xmath4 is obtained for neutron - rich matter .
however , we point out that the @xmath0-induced nucleon mass splitting is a rather general feature of energy dependent interactions and should show up in any model which incorporates them .
we further discussed the energy dependence of the isovector optical potentials .
we found that the nld model leads to a decreasing lane - type potential with increasing particle momentum .
this disagrees with the conventional rmf models , but is consistent with the results from microscopic brueckner calculations .
also the comparison of the isovector splitting in the proton / neutron optical potentials agrees well with the results from dbhf and bhf microscopic approaches .
these suggest that the nld @xmath6 mean - field model which accommodates the energy dependence of nucleon selfenergies , is in qualitative agreement with microscopic ( dirac ) brueckner calculations . since the nld model describes also asymmetric matter features fairly well and agrees with microscopic dbhf models , it could be interesting to apply the nld approach to finite nuclei
. however , due to the energy dependence of the fields , the main advantage of the nld approach would be its application to the description of heavy - ion reactions within the transport theory .
this work was supported by hic for fair , dfg through tr16 and by bmbf .
57 v. baran , m. colonna , v. greco and m. di toro , phys .
* 410 * ( 2005 ) 335 .
b. a. li , l. w. chen and c. m. ko , phys .
* 464 * ( 2008 ) 113 . with the term _ stiffness _ of
a quantity one usually refers to its slope at rather high densities in the range @xmath118 , where @xmath119 is the density of ground state nuclear matter .
a soft ( stiff ) behaviour of a quantity means then a small ( big ) slope parameter or a moderate ( strong ) density dependence of a quantity at supra - normal densities .
d. vretenar , t. nik@xmath120ic , p. ring , phys .
c * 68 * ( 2003 ) 024310 ; + a. carbone _ et al .
_ , phys .
c * 81 * ( 2010 ) 041301(r ) ; + j. zenihiro _ et al .
_ , phys . rev .
c * 82 * ( 2010 ) 044611 ; + m. centelles _ et al .
_ , phys .
* 102 * ( 2009 ) 122502 ; + m. warda _ et al .
_ , phys .
c * 80 * ( 2009 ) 024316 .
w. reisdorf and h. g. ritter , ann .
nucl . part .
* 47 * ( 1997 ) 663 .
n. herrmann , j. p. wessels and t. wienold , ann .
nucl . part .
* 49 * ( 1999 ) 581 .
p. danielewicz , annals phys .
* 152 * ( 1984 ) 305 .
p. danielewicz , annals phys .
* 152 * ( 1984 ) 239 .
w. botermans and r. malfliet , phys .
* 198 * ( 1990 ) 115 .
a. andronic _ et al . _
[ fopi collaboration ] , phys .
b * 612 * ( 2005 ) 173 . c. fuchs , prog . part .
* 56 * ( 2006 ) 1 .
l. w. chen , c. m. ko and b. a. li , phys .
* 94 * ( 2005 ) 032701 .
d. v. shetty , s. j. yennello and g. a. souliotis , phys .
c * 75 * ( 2007 ) 034602 .
j. rizzo , m. colonna , m. di toro and v. greco , nucl .
a * 732 * ( 2004 ) 202 .
v. giordano _ et al .
_ , phys .
c * 81 * ( 2010 ) 044611 .
b. a. li , phys .
c * 69 * ( 2004 ) 064602 .
b. a. li , phys .
* 85 * ( 2000 ) 4221 .
t. gaitanos _ et al .
_ , nucl .
a * 732 * ( 2004 ) 24 .
g. ferini _ et al .
_ , phys . rev . lett .
* 97 * ( 2006 ) 202301 .
v. prassa _ et al .
_ , nucl .
phys . a * 832 * ( 2010 ) 88 .
q. f. li , z. x. li , e. g. zhao and r. k. gupta , phys .
c * 71 * ( 2005 ) 054907 .
k. weber _ et al .
_ , nucl .
a * 539 * ( 1992 ) 713 .
p. danielewicz , nucl .
a * 673 * ( 2000 ) 375 .
k. sahu _ et al . _ , nucl .
a * 640 * ( 1998 ) 493 .
s. kubis and m. kutschera , phys .
b * 399 * ( 1997 ) 191 .
b. liu _ et al .
_ , phys .
c * 65 * ( 2002 ) 045201 . c. fuchs , h. lenske and h. h. wolter , phys .
c * 52 * ( 1995 ) 3043 .
s. typel and h. h. wolter , nucl .
a * 656 * ( 1999 ) 331 . t. niksic _ et al . _ ,
c * 66 * ( 2002 ) 024306 ; + g.a .
et al . _ ,
c * 71 * ( 2005 ) 024312 ; + w .- h .
long , n. van giai , s .-
zhou , phys .
c * 69 * ( 2004 ) 034319 .
s. typel , phys .
c * 71 * ( 2005 ) 064301 .
e. n. e. van dalen , c. fuchs and a. faessler , phys .
c * 72 * ( 2005 ) 065803 .
e. n. e. van dalen , c. fuchs and a. faessler , eur .
j. a * 31 * ( 2007 ) 29 .
e. d. cooper , s. hama , b. c. clark and r. l. mercer , phys .
c * 47 * ( 1993 ) 297 .
a. lane , nucl .
* 35 * ( 1962 ) 676 .
j. p. jeukenne , a. lejeune and c. mahaux , phys .
c * 10 * ( 1974 ) 1391 .
a. j. koning and j. p. delaroche , nucl .
a * 713 * ( 2003 ) 231 .
r. kozack and d. g. madland , phys . rev .
c * 39 * ( 1989 ) 1461 .
r. kozack and d. g. madland , nucl .
a * 509 * ( 1990 ) 664 .
d. alonso and f. sammarruca , phys .
c * 67 * ( 2003 ) 054301 .
f. sammarruca , w. barredo and p. krastev , phys .
c * 71 * ( 2005 ) 064306 .
t. gaitanos , m. kaskulov and u. mosel , nucl .
a * 828 * ( 2009 ) 9 [ arxiv:0904.1130 [ nucl - th ] ] . t. gaitanos , m. kaskulov and h. lenske , phys .
b * 703 * ( 2011 ) 193 [ arxiv:1105.4450 [ nucl - th ] ] . h. p. duerr , phys . rev . *
103 * ( 1956 ) 469 . b. d. serot and j. d. walecka , int .
j. mod .
e * 6 * ( 1997 ) 515 .
t. maruyama , et al .
phys . a * 573 * ( 1994 ) 653 .
r. machleidt , k. holinde and c. elster , phys .
* 149 * ( 1987 ) 1 .
b. ter haar and r. malfliet , phys .
rept . * 149 * ( 1987 ) 207 . m. m. kaskulov , u. mosel , [ arxiv:1103.2097 [ nucl - th ] ] ; m. m. kaskulov , u. mosel , phys . rev .
* c81 * ( 2010 ) 045202 .
l. sehn , c. fuchs and a. faessler , phys .
c * 56 * ( 1997 ) 216 .
r. brockmann and r. machleidt , phys .
c * 42 * ( 1990 ) 1965 .
d. v. shetty , s. j. yennello and g. a. souliotis , phys .
c * 76 * ( 2007 ) 024606 [ erratum - ibid .
c * 76 * ( 2007 ) 039902 ] .
x. lopez _ et al .
_ [ fopi collaboration ] , phys .
c * 75 * ( 2007 ) 011901 . | the lagrangian density of relativistic mean - field ( rmf ) theory with non - linear derivative ( nld ) interactions is applied to isospin asymmetric nuclear matter .
we study the symmetry energy and the density and energy dependences of nucleon selfenergies . at high baryon densities
a soft symmetry energy is obtained .
the energy dependence of the isovector selfenergy suppresses the lane - type optical potential with increasing energy and predicts a @xmath0-meson induced mass splitting between protons and neutrons in isospin asymmetric matter .
relativistic hadrodynamics , non - linear derivative model , isospin asymmetric matter , equation of state , symmetry energy , optical potential .
+ pacs numbers : 21.65.cd , 21.65.mn , 25.70.-z |
frustrated systems can present a wide range of physical phenomena due to more configuration options than the ordered systems , and , as such have been paid particular attention in recent past and these systems have emerged as a topic of contemporary research in condensed matter physics .
@xcite spin - glass systems are characterized by a random frozen spin orientation below a characteristic freezing temperature and present a nice example of frustrated magnetism .
in contrast to a unique ground state found in conventional collective systems , spin - glass systems possess a multitude of possible disordered ground states and freeze into one of them .
investigations into spin - glass systems have shown that frustration and randomness are the key ingredients for spin - glass behavior .
the crystallographic disorder in a magnetic system may cause frustration of the magnetic moments leading to spin - glass behavior .
however , for a well - ordered crystal structure there is no obvious source of frustration to magnetic moments and , therefore , one would not expect spin - glass behavior in crystallographically ordered compounds ( except for the specific class of geometrically frustrated lattice , i.e. due to antiferromagnetic coupling among the magnetic moments in a triangular lattice ) . for this reason research into spin - glass behavior in magnetic systems
has been generally focused on geometrically frustrated kagome ( two dimensional ) and pyrochlore ( three - dimensional ) lattices .
@xcite however , the recent observations of spin - glass behavior in the well - ordered stoichiometric intermetallic compounds urh@xmath3ge@xmath3 @xcite and prau@xmath3si@xmath3 @xcite have brought further scope and insight into the mechanism of spin - glass behavior . the spin - glass behavior in urh@xmath3ge@xmath3 was explained in terms of the site disorder on the rhodium and germanium sublattices and , extended annealing was found to remove the disorder resulting in an antiferromagnetically ordered state .
@xcite in contrast , the spin - glass behavior in prau@xmath3si@xmath3 was found to persist despite extended annealing .
this suggests that the spin - glass behavior in prau@xmath3si@xmath3 is not driven by the crystallographic disorder
. a novel mechanism due to the dynamic fluctuations of the crystal field levels has been proposed to be responsible for the frustration of the magnetic moments in prau@xmath3si@xmath3.@xcite in this paper we present another example of well - ordered stoichiometric intermetallic system prrusi@xmath0 that exhibit the signatures of spin - glass behavior most likely through the dynamic fluctuations of the crystal field levels .
further , the spin - glass state in prrusi@xmath0 is found to compete with the spontaneously induced magnetic order with a cef singlet ground state .
its ce - analog , cerusi@xmath0 is reported to exhibit no magnetic order down to 2 k both for polycrystalline and single crystal samples .
@xcite a broad maximum is observed in the magnetic susceptibility of cerusi@xmath0 near 150 k which indicates the presence of strong hybridization between the 4@xmath6 and conduction electrons .
polycrystalline samples of prrusi@xmath0 and its nonmagnetic analog larusi@xmath0 were prepared using standard arc melting technique on a water cooled copper hearth under the titanium gettered inert argon atmosphere using the high purity elements ( pr , la : 99.9% , ru : 99.99% , si : 99.999% ) in stoichiometric ratio . during the melting process the samples were flipped and remelted several times to achieve homogeneity .
further , to improve homogeneity and phase formation , and reduce disorder ( if any present ) the as - cast samples were wrapped in tantalum foil and annealed for a week at 900 @xmath7c under the dynamic vacuum . the crystal structure and phase purity
were checked by the powder x - ray diffraction ( xrd ) and scanning electron microscopy ( sem ) .
the stoichiometry was checked by the energy dispersive x - ray ( edax ) composition analysis .
a commercial squid magnetometer ( mpms , quantum - design ) was employed for magnetization measurements .
specific heat was measured by the relaxation method in a physical properties measurement system ( ppms , quantum - design ) .
electrical resistivity was measured by standard four - probe ac technique using the ppms .
the @xmath1sr measurements were performed in longitudinal geometry using the musr spectrometer at isis facility of the rutherford appleton laboratory , didcot , uk . since silver gives a non - relaxing muon signal we used a silver holder ( purity 4n ) for mounting the sample .
the stray fields at the sample position were cancelled to within 1 @xmath1 t by using correction coils .
the neutron scattering experiment was performed on the powdered prrusi@xmath0 sample in the temperature range 4 to 50 k using the mari time of flight spectrometer at isis .
the powdered sample was mounted by wrapping the sample in a thin al - foil inside a thin walled al - can .
the low temperature was achieved using he - exchange gas with a top - loading closed cycle refrigerator .
the data were collected using the neutron incident energy e@xmath8 = 6 , 20 and 40 mev .
the neutron diffraction measurements were performed on the powdered sample using the g4.1 diffractometer at llb , saclay .
the neutron diffraction data were collected at 1.8 and 25 k using the neutrons of wavelength 2.423 .
( colour online ) powder x - ray diffraction pattern of prrusi@xmath0 recorded at room temperature .
the solid line through the experimental points is the two - phase rietveld refinement profile calculated for banisn@xmath0-type tetragonal ( space group _
i4 mm _ ) and thsi@xmath9-type tetragonal ( space group _ i4@xmath10/amd _ ) structural model .
the vertical bars indicate the bragg positions of both the phases .
the lowermost curve represents the difference between the experimental data and calculated results.,width=340 ] .[tab : table1 ] crystallographic parameters for prrusi@xmath0 determined from the full structure refinement of powder x - ray diffraction data using fullprof software .
[ cols="<,<,^,^,^,^ " , ] the analysis of powder x - ray diffraction data reveals that prrusi@xmath0 crystallizes in the banisn@xmath0-type tetragonal structure ( space group _
i4 mm _ ) .
1 shows the xrd pattern together with the two - phase rietveld refinement profile fit , which indicates a well crystallographic ordered structure and the single phase nature of sample , the impurity phase identified as prsi@xmath9 is very small ( 1.54 % in volume fraction , equivalent to 2.19 % mole fraction ) .
the lattice parameters were found to be @xmath11 = 4.213(1 ) and @xmath12 = 9.923(1 ) .
the crystallographic parameters obtained from the rietveld refinement of prrusi@xmath0 are listed in table i. larusi@xmath0 also crystallizes in similar banisn@xmath0-type tetragonal structure ( space group _
i4 mm _ ) with lattice parameters @xmath11 = 4.263(1 ) and @xmath12 = 9.944(1 )
the best fits the least squares refinement gave the value of @xmath13 = 1.84 and 1.47 respectively for prrusi@xmath0 and larusi@xmath0 .
the single phase nature of prrusi@xmath0 and larusi@xmath0 samples were further confirmed by the high resolution sem ( scanning electron microscopy ) images , and the edax composition analysis confirmed the desired 1:1:3 stoichiometry .
the impurity phase prsi@xmath9 which forms in thsi@xmath9-type tetragonal ( space group _ i4@xmath10/amd _ ) at room temperature is reported to exhibit ferromagnetic order below 11 k followed by another transition near 7 k , and undergoes a structural phase transition to gdsi@xmath9-type orthorhombic structure ( space group _ imma _ ) below 153 k. @xcite coincidentally the dc magnetic susceptibility of our compound prrusi@xmath2 also exhibits two sharp anomalies at 8.6 k and 4.9 k which are very close to what have been observed in prsi@xmath9 .
however , we do not observe any signature of ferromagnetic order in low temperature neutron diffraction data which suggests that the effect of impurity phase prsi@xmath9 , if any , is very small .
in fact from a systematic study of ac and dc magnetic susceptibility , specific heat , electrical resistivity , inelastic neutron scattering and @xmath1sr data we have found evidence of an induced moment spin - glass type behavior in this compound . the reproducibility of results have been checked by the magnetic susceptibility and specific heat measurements on two different batches of samples ( having different impurity phases ) and we believe that the small impurity content does not influence the results and conclusions derived on the bulk properties of prrusi@xmath2 which we present in this paper .
( colour online ) the zero field cooled dc magnetic susceptibility @xmath14 of prrusi@xmath0 as a function of temperature in the temperature range 2 300 k measured in a field of 0.1 t. the solid line represents the fit to curie - weiss behavior .
the upper inset shows the expanded view of low temperature susceptibility measured at different applied magnetic fields .
the lower inset shows the zero field cooled ( zfc ) and field cooled ( fc ) susceptibility data at 2.5 and 10 mt.,width=321 ] ( colour online ) the dc isothermal magnetization @xmath15 as a function of magnetic field measured at constant temperatures of 2 , 6 , 12 and 20 k.,width=321 ] results obtained from the dc magnetization measurements are shown in fig . 2 and fig .
3 . at low fields ( e.g. , at 0.01 t )
the low temperature magnetic susceptibility data clearly show two well defined anomalies at 8.6 k and 4.9 k ( insets in fig .
2 ) . while with an increase in applied magnetic field the 4.9 k anomaly becomes more pronounced , the anomaly at 8.6 k gets smoothened , however , no change is observed in the temperature at which this anomaly occurs .
the nature of these transitions at 4.9 k and 8.6 k is not clear from the temperature dependent dc magnetic susceptibility data , however the ac susceptibility and specific heat measurements as discussed later hint for a signature of spin - glass type behavior in this compound .
further , the behavior of magnetic anomaly at 4.9 k suggests that an increase in the magnetic field causes the spin alignment and the 4.9 k transition might be related to spin reorientation . on further increase in applied magnetic field , at an applied field of 0.5 t both the anomalies in the susceptibility are hardly detectable and susceptibility tends to saturate . in the lower inset of fig .
2 we have plotted the zero field cooled ( zfc ) and field cooled ( fc ) dc susceptibility data as a function of temperature at an applied field of 2.5 mt and 10 mt .
the splitting of fc and zfc susceptibility data can be taken as the first indication for the spin - glass behavior in this compound .
the splitting of fc and zfc susceptibility sets in at @xmath16 = 9.8 k defining the quasi - static freezing temperature ( @xmath17 ) for the spin - glass transition .
the high temperature magnetic susceptibility data are consistent with the curie - weiss behavior , @xmath18 ) . from the fit to the inverse susceptibility data above 50 k ( solid line in fig .
2 ) , we obtain an effective moment @xmath19 = 3.41 @xmath20 and curie - weiss temperature @xmath21 = @xmath2210.0 k. the value of effective moment obtained is very close to the theoretically expected value of 3.58 @xmath20 for pr@xmath23 free ions . fig .
3 shows the isothermal magnetization data as a function of magnetic field measured at constant temperatures of 2 , 6 , 12 and 20 k. the isotherms exhibit almost a linear field dependence and attain a value of @xmath24 0.9 @xmath20 at 7 t at 2 k which is very small compared to the theoretical saturation magnetization of 3.2 @xmath20 , which indicates the presence of the crystal field effect that will split the 4@xmath6 multiplet ground state of the pr - ion ( @xmath25 ) .
( colour online ) the time dependence of thermo - remnant magnetization ( trm ) , @xmath26 recorded at 2 k after switching off the cooling magnetic field of 0.05 t. the solid line is the fit to the superposition of a stretched exponential and a constant term .
the inset shows the semi - log plot of the thermo - remnant magnetization .
the solid lines in the inset are the guide to eyes.,width=321 ] in fig .
4 we have shown the time dependence of the thermo - remnant magnetization ( trm ) , @xmath26 which we recorded at 2 k , the sample was cooled in a magnetic field of 0.05 t from 50 k ( well above @xmath17 ) to 2 k and the field - cooled isothermal remanence magnetization was measured after switching off the magnetic field .
it is interesting to note that the magnetization does not become zero immediately after switching off the applied field as one expects for a normal antiferromagnetic order system , but it decays slowly with time as observed for many spin - glass systems .
@xcite the non zero value of the magnetization after 70000 s , indicates frozen spin dynamics which transforms from one spin configuration to another spin configuration with time .
many models have been proposed to describe the time dependence of magnetization relaxation in spin - glasses,@xcite the logarithmic relaxation decay , @xmath27 and the stretched exponential decay @xmath28 $ ] are the most common in literature .
the inset of fig .
4 shows the semi - log plot of trm data , the data significantly deviates from the ideal linear behavior , probably due to the aging time effects .
@xcite this suggests that a logarithmic relaxation decay is not a good choice for our data and the relaxation of thermo - remnant magnetization in prrusi@xmath0 is much slower .
a superposition of a stretched exponential and a constant term is found to represent the experimentally observed trm data very well , @xmath29\ ] ] where the additional constant term is interpreted as the longitudinal spontaneous magnetization coexisting with the frozen transverse spin component .
@xcite the solid line in fig .
4 represents the fit to this expression , the best fit parameters are @xmath30 = 53.03 emu / mole , @xmath31 = 2.53 emu / mole , mean relaxation time @xmath32 = 13357 s and @xmath33 = 0.62 .
( colour online ) the temperature dependence of the real part of ac magnetic susceptibility of prrusi@xmath0 measured at different frequencies.,width=321 ] ( colour online ) the temperature dependence of the imaginary part of ac magnetic susceptibility of prrusi@xmath0 measured at different frequencies.,width=321 ] in order to get further information about the magnetic behavior of prrusi@xmath0 we also measured the ac magnetic susceptibility at different frequencies .
the real and imaginary parts of ac magnetic susceptibility are shown in fig . 5 and fig . 6 , respectively .
at 11 hz the real part of ac susceptibility @xmath34 shows two sharp peaks at 5.7 k and 9.8 k. while there is no change in the peak position of the 5.7 k anomaly with frequency up to 10 khz , the 9.8 k ( 11 hz ) peak shifts to 10.0 k at 919 hz and remains unchanged up to 10 khz .
this frequency dependence is reminiscent of a spin glass .
the imaginary component of ac susceptibility @xmath35 shows only one peak near 9.8 k which is independent of frequency and there is almost no noticeable anomaly ( except a weak change in slope ) near 5.7 k. the maximum of @xmath35 can be taken as the freezing temperature ( @xmath17 ) of spin - glass behavior .
however , the extremely weak ( almost negligible ) frequency dependence of ac susceptibility maximum suggests that the prrusi@xmath0 is a non - canonical spin - glass system . for a canonical spin - glass system one expects an increase in transition temperature with increasing frequency , characterized by @xmath36 , for most of the canonical spin - glass system @xmath37 is found to lie in between 0.0045 and 0.06 .
@xcite the 5.7 k anomaly in @xmath34 can be attributed to the transition to a further frozen spin - glass state .
( colour online ) the temperature dependence of specific heat data of larusi@xmath0 and prrusi@xmath0 measured in different magnetic fields.,width=321 ] the specific heat data of prrusi@xmath0 and larusi@xmath0 are shown in fig .
the specific heat of prrusi@xmath0 shows a very broad anomaly with a maximum near 10 k which we believe is due to crystal field effects .
we do not see any sharp anomaly near 4.9 k or 8.6 k which would suggest a magnetic phase transition corresponding to the peaks in dc magnetic susceptibility data . with the application of magnetic field , the specific heat anomaly becomes broader and the position of maximum increases in temperature .
the fact that the specific heat anomaly persists up to the investigated field of 9 t clearly rules out the possibility of the anomaly being related to a magnetic phase transition .
the absence of the magnetic phase transition in specific heat is consistent with the spin - glass type behavior .
the specific heat data of larusi@xmath0 does not show any anomaly down to 2 k and the low temperature data have a temperature dependence of @xmath38 with the sommerfeld coefficient @xmath39 @xmath24 5 mj / mole k@xmath40 .
in contrast , the specific heat of prrusi@xmath0 does not obey the @xmath38 temperature dependence , the sommerfeld coefficient @xmath39 of prrusi@xmath0 seems enhanced due to the crystal field effect with a c / t value of @xmath24 93 mj / mole k@xmath40 at 2 k. an extrapolation of low temperature specific heat data gives a rough estimate of @xmath39 @xmath24 58 mj / mole k@xmath40 for prrusi@xmath0 .
the magnetic contributions to the specific heat and entropy of prrusi@xmath0 are shown in fig .
the magnetic contribution to specific heat was obtained by subtracting the lattice contribution from the specific heat of prrusi@xmath0 which we took roughly equal to the specific heat of larusi@xmath0 .
the magnetic entropy was obtained by integrating the @xmath41 _ vs. _ @xmath16 plot .
the broad schottky - type anomaly observed in @xmath42 is reasonably reproduced with the crystal field analysis . the solid line in fig .
8 represents the fit to @xmath42 data with the crystal field scheme obtained from the inelastic neutron scattering data discussed later .
the ground state is a singlet and the first excited state is also a singlet at @xmath24 22 k , the second excited state being a doublet at @xmath24 28 k. the temperature dependence of magnetic entropy is consistent with the singlet ground state and attains a value of @xmath43 near 12 k and @xmath44 near 22 k which supports the proposed crystal field scheme ( the excess entropy being the contribution from higher excited states ) . the enhancement of sommerfeld coefficient @xmath39 can thus be understood in terms of excitonic mass enhancement due to low lying crystal field excitations as suggested in references 28 and 29 for a system with cef - split singlet ground state .
( colour online ) the temperature dependence of magnetic contribution to the specific heat and entropy of prrusi@xmath0 at zero field .
the solid line shows the fit based on the crystal field model obtained from the inelastic neutron scattering data.,width=321 ] ( colour online ) the temperature dependence of electrical resistivity of prrusi@xmath0 measured in zero magnetic field .
the inset shows the expanded view below 50 k for different applied magnetic fields.,width=321 ] figure 9 shows the electrical resistivity data of prrusi@xmath0 measured in different applied magnetic fields . above 50
k the electrical resistivity exhibit metallic behavior and almost a linear temperature dependence with a residual resistivity of @xmath24 16 @xmath45 cm ( at 1.85 k ) and residual resistivity ratio of @xmath24 5 .
the low temperature resistivity data show a broad curvature with a maximum near 15 k which can be attributed to the crystal field effect as revealed by the specific heat data discussed above .
an extremely weak effect of magnetic field is observed on the electrical resistivity anomaly up to the investigated field of 9 t. ( colour online ) the @xmath1sr spectra of prrusi@xmath0 collected at various temperatures ( @xmath46 1.2 k , @xmath47 8 k , @xmath48 12 k , @xmath49 20 k , @xmath50 30 k ) .
the solid lines are the fits to the data as discussed in the text.,width=321 ] ( colour online ) the temperature dependence of ( a ) the initial asymmetries @xmath51 and @xmath52 , ( b ) the precession frequency @xmath53 , ( c ) the depolarization rate @xmath54 , and ( d ) the depolarization rate @xmath55 .
the solid line in ( b ) is a guide to the eye.,width=321 ] in order to further probe the nature of the ground state of prrusi@xmath0 , spin - glass behavior versus long range antiferromagnetic order , we have carried out zero - field @xmath1sr study .
the @xmath1sr spectra were collected while warming the sample in zero field .
10 shows @xmath1sr spectra of prrusi@xmath0 at various temperatures in the temperature range 1.2 30 k. a marked change is observed in the muon depolarization rate for the temperatures above and below 12 k ( fig .
10 ) , the temperature close to 9 - 10 k at which we have observed the anomaly in the dc and ac magnetic susceptibility . above 12 k , the @xmath1sr spectra can be described by a simple exponential decay , indicating that the muons are sensing paramagnetic fluctuations . however , below 12 k the @xmath1sr spectra are best described by a heavily damped oscillating function , namely , @xmath56 where @xmath57 and @xmath52 are the initial asymmetries , @xmath53 is the precession frequency , @xmath58 is the phase , @xmath55 and @xmath54 are the depolarization rates , and c is the background .
the temperature dependencies of these parameters are shown in fig .
11(a ) clearly shows that at 12 k there is a loss of asymmetry in a@xmath59 to 1/3 the high temperature value . at the same temperature an increase in @xmath51 is observed together with an increasing frequency ( see fig .
11(b ) ) , indicating the presence of a long range ordered state in prrusi@xmath0 which contrasts the specific heat observation of no long range magnetic ordering . the solid line in fig .
11(b ) is a guide to the eye and emphasizes the salient feature , namely that at the points of interest in the magnetization , t = 5 and 9 k , the frequency shows plateaus which indicates a spin - reorientation .
figures 11(c ) and 11(d ) show the temperature dependence of the muon depolarization rates .
it is interesting to note that @xmath54 continues to increase as the temperature is reduced , which is contrary to the expected temperature dependence for a static long range magnetically ordered system .
this may suggest that there may be a degree of frustration in prrusi@xmath2 in line with the magnetization and specific heat data .
furthermore , the support of the spin - glass behavior in a homogenous system arises from the well - defined crystal field excitations in this compound .
as expected the temperature dependence of @xmath55 shows a peak at the ordering temperature and also continues to rise as the temperature is reduced , again indicating slow down of spin dynamics .
( colour online ) ( a ) neutron diffraction pattern of prrusi@xmath0 recorded at 25 k. the solid line through the experimental points is the rietveld refinement profile calculated for banisn@xmath0-type tetragonal ( space group _
i4 mm _ ) structural model .
the vertical bars indicate the bragg positions .
the lowermost curve represents the difference between the experimental data and calculated results . neutron diffraction pattern of prrusi@xmath0 recorded at 1.8 and 25 k together with their difference ( b ) .
no magnetic bragg peaks are observed in the difference plot.,width=340 ] in order to explore the nature of long range order observed in @xmath1sr study we performed a low temperature neutron diffraction .
the neutron diffraction patterns of prrusi@xmath0 recorded at 1.8 and 25 k are shown in fig .
a comparison of 1.8 and 25 k data ( above and below the transition temperature ) reveals no magnetic bragg peak ( fig .
12(b ) ) , the very small peaks observed in difference plot are due to the imperfect subtraction of data sets .
this indicates that the long range ordered state moment of the pr ion is below 0.2(1 ) @xmath60 . a structural rietveld refinement of 25 k neutron diffraction pattern ( fig .
12(a ) ) reveals single phase nature of sample with crystallographic parameters very close to what have been extracted from the room temperature x - ray diffraction pattern . at 25
k the lattice parameters are @xmath11 = 4.2162(5 ) and @xmath12 = 9.9339(14 ) , and that at 2 k are @xmath11 = 4.2161(5 ) and @xmath12 = 9.9340(14 ) .
( colour online ) inelastic neutron scattering response , a color - coded map of the intensity , energy transfer ( @xmath61 ) versus momentum transfer ( @xmath62 ) of prrusi@xmath0 measured with the incident energy @xmath63 = 6 mev ( top ) and 20 mev ( bottom ) at 4.7 k and 20 k.,width=283 ] as discussed above the @xmath1sr study shows an evidence for the presence of long range magnetic order , however the specific heat and neutron diffraction results suggest that there is no long range magnetic ordering in prrusi@xmath0 , it seems that the moment is weak which is sensed by muons but not manifested in the specific heat and neutron diffraction measurements .
a rough estimate of magnetic moment from @xmath1sr suggests a moment of @xmath24 0.05 @xmath60 in prrusi@xmath2 .
further , considering that the ground state is a singlet , the observation of weak moment in @xmath1sr measurements suggest an induced magnetic moment behavior in prrusi@xmath0 due to low lying crystal field excitations . therefore in order to understand the origin of spin - glass and/or
induced moment magnetism in prrusi@xmath0 we have performed the inelastic neutron scattering ( ins ) study to investigate crystal field excitations and their energy scheme .
neutrons with incident energy @xmath63 = 6 , 20 and 40 mev were used to record the ins spectra at 4.7 , 20 and 50 k for scattering angles between 3@xmath7 and 135@xmath7 .
the spectra have been corrected for the background signal and the absolute normalized response ( in units of mbr / sr / mev ) , using vanadium standard , is presented in fig .
13 as a color - coded map of the intensity , energy transfer versus momentum transfer , for scattering from prrusi@xmath0 measured with 6 mev and 20 mev incident neutron energies at 4.7 k and 20 k. the presence of sharp inelastic excitations near 2.4 mev and 14.7 mev are clearly observed in the ins spectra at 4.7 k. further we observe that the peak position of both 2.4 mev and 14.7 mev excitations remain nearly invariant from 4.7 to 50 k , which suggest that these excitations are the result of the crystal field effects , indirectly reflecting an absence of spin wave contributions , which one would expect below the magnetic ordering temperature .
consistent with neutron diffraction data we did not find any magnetic bragg peak in @xmath62-cuts ( integrated over the elastic line from @xmath221 to 1 mev ) at 4.7 k , which suggests that long range order if present in this compound has very small magnetic moment , which can not be detected in the present data . at 20 k we can see a weak intensity near 12.3 mev , which is due to excited transition from the ground state to excited state doublet ( fig .
13 , right bottom ) .
the @xmath62-dependent integrated intensity between 2 and 3 mev at 4.7 k follows the square of pr@xmath23 magnetic form factor [ @xmath64 ( fig .
14 ) , which suggests that the inelastic excitations result mainly from single - ion cef - transitions .
the @xmath65 behavior of the @xmath62-dependent integrated intensity indicates a very small phonon contribution for @xmath66 @xmath67 3 which we have used for crystal field analysis . a similar behavior was also found for 14.7 mev peak .
we have therefore not corrected the spectra for phonon contribution and we presumed that the presence of weak phonon part , if it is present would not affect our analysis for the low @xmath66 data based on the cef model .
( colour online ) the @xmath62 dependence of total intensity integrated between 2 and 3 mev at 4.7 k for incident energy @xmath63 = 20 mev .
the solid line represents the square of the pr@xmath23 magnetic form factor , scaled to 200 at @xmath62 = 0.,width=321 ] ( colour online ) the inelastic neutron scattering spectra of prrusi@xmath0 measured with incident energy @xmath63 = 6 mev for momentum transfer @xmath66 less than 3 @xmath68 at different temperatures .
the solid lines represent the crystal electric field fits .
different dotted , dashed , dashed - dotted lines correspond to the different components of the fit.,width=283 ] ( colour online ) the inelastic neutron scattering spectra of prrusi@xmath0 measured with incident energy @xmath63 = 20 and 40 mev for momentum transfer @xmath66 less than 3 @xmath68 at different temperatures .
the solid lines represent the crystal electric field fits .
different dotted , dashed , dashed - dotted lines correspond to the different components of the fit.,width=283 ] ( colour online ) crystal electric field level scheme of the pr@xmath23 ions in prrusi@xmath0 deduced from the inelastic neutron scattering experiment .
the transitions from the ground state to the excited states that contribute to the observed excitations are shown by arrows.,width=321 ] the ins spectra were analyzed further to obtain the full information about crystal field levels scheme .
the crystal field hamiltonian for the pr ion having a tetragonal point group ( @xmath69 ) symmetry in prrusi@xmath0 is given by @xmath70 where @xmath71 are the stevens operator and @xmath72 are the phenomenological crystal field parameters that is determined from the experimental results of inelastic neutron scattering .
the action of crystal electric field tends to remove the degeneracy of the 4__f _ _ ground multiplet . for tetragonal symmetry
the nine fold - degenerate ground state of pr@xmath23 ( @xmath73 ) splits in five singlets ( @xmath74 , @xmath75 , @xmath76 , @xmath77 , @xmath5 ) and two doublets ( @xmath78 , @xmath79 ) , the @xmath80 s are the irreducible representations of the point group . the solid lines in fig .
15 and fig .
16 represent the fit to the cef model for simultaneous refinement of all six data sets for 6 , 20 and 40 mev incident energies ( from 4.7 , 20 and 50 k ) .
the phenomenological crystal field parameters @xmath72 obtained from the best fit are @xmath81 = 0.4268 ( @xmath82 0.0025 ) mev , @xmath83= 0.1031 ( @xmath82 0.0009 ) @xmath84 10@xmath85 mev , @xmath86 = 0.3151 ( @xmath82 0.0018 ) @xmath84 10@xmath87 mev , @xmath88 = -0.1996 ( @xmath82 0.0016)@xmath89 10@xmath68 mev , and @xmath90 = 0.1563 ( @xmath82 0.0008 ) @xmath84 10@xmath85 mev .
the crystal field level scheme corresponding to these @xmath72 parameters has a singlet ( @xmath5 ) ground state followed by a singlet ( @xmath74 ) at 1.87 mev , a doublet ( @xmath78 ) at 2.45 mev , another doublet ( @xmath79 ) at 13.41 mev , and three singlets @xmath77 , @xmath76 and @xmath75 respectively at 15.06 mev , 23.73 mev and 24.40 mev .
the crystal field level scheme of the pr@xmath23 ions in prrusi@xmath2 obtained this way is shown in fig .
17 . to compare
, spin - glass system prau@xmath9si@xmath9 also has a singlet @xmath5 ground state , lying below a doublet first excited state @xmath78 at 0.72 mev and second excited singlet @xmath74 at 7.18 mev . @xcite
it is worth to mention here that though the first excited state is a singlet at 1.87 mev , there is no transition from the ground state to the first excited state , the matrix element is zero , the nonzero matrix elements for the transitions from ground state to the excited states are obtained only for the transitions shown by arrows in fig .
thus at temperatures close to the freezing temperature , which is well below 155 k ( 13.41 mev ) , prrusi@xmath2 is effectively a two - level system . the specific heat calculated with the crystal field scheme presented in fig .
17 is shown by solid line in fig .
a good agreement between the experimentally observed specific heat data and the calculated one validates the obtained cef level scheme .
( colour online ) the mean - field calculated @xmath91 as a function of the exchange energy , @xmath4 , for the singlet - doublet transition , @xmath92 = 2.45 mev .
the vertical dotted line shows @xmath4 = 0.32 mev for prrusi@xmath0 , which gives ordering temperature @xmath24 11.5 k that is very close to peak in the susceptibility at 9.8 k.,width=321 ] now let us discuss the @xmath1sr finding of weak moment taking into consideration that the prrusi@xmath0 has a singlet ground state as deduced from inelastic neutron scattering and specific heat measurements . for a singlet ground state system
the existence of magnetic order critically depends on the ratio @xmath93 , where @xmath4 is the heisenberg exchange interaction and @xmath92 is the crystal field splitting energy between the ground state and the excited state coupled by the matrix element @xmath94 .
@xcite above a critical value of @xmath93 system undergoes a self - induced spontaneous ordering . the mean - field critical temperature below which self - induced moment forms spontaneously in a two - level system is given by @xcite @xmath95 \right\}^{-1}\ ] ] where @xmath33 is the degeneracy of the excited state and @xmath94 is the the matrix element between the ground state singlet and excited state doublet . a plot of the @xmath91 calculated from this expression as a function of @xmath4 for @xmath33 = 2 , @xmath96 = 9.58 and @xmath92 = 2.45 mev is shown in fig .
18 . from fig .
18 it follows that the exchange energy must be @xmath4 = 0.30 mev for prrusi@xmath2 to have an induced moment ordering below @xmath91 = 9.8 k. a rough estimate of @xmath4 following the approach used to estimate the @xmath4 for pr in reference 29 , i.e. using @xmath97 and @xmath98 , yields @xmath4 = 0.32 for prrusi@xmath2 which is very close to the mean - field required value of @xmath4 for an induced moment magnetic ordering below 9.8 k. therefore this clearly suggests that prrusi@xmath0 undergoes an induced moment ordering below 9.8 k. the value of @xmath99 = 0.122 for prrusi@xmath0 is comparable but little higher than the corresponding @xmath99 = 0.085 for prau@xmath3si@xmath3 .
@xcite furthermore , as in the case of prau@xmath3si@xmath3 , if the exchange coupling is very close to the critical value required to induce the magnetic order , the dynamic fluctuations of crystal field levels can destabilize the induced moment long range magnetic order resulting in a frustrated ground state , i.e. spin - glass type behavior .
the dynamic fluctuations of crystal field levels is the leading mechanism for spin - glass behavior in prau@xmath3si@xmath3 . @xcite for our system prrusi@xmath0 it seems that the j@xmath100/@xmath92 is higher than the critical value , resulting in an induced magnetic order as sensed by @xmath1sr , however the value of @xmath99 is not high enough to be unaffected by the dynamic fluctuations of crystal field levels and a frustrated ground state is realized together with induced moment magnetism .
the two contradictory ground states , spin - glass and magnetic order , can be explained if the dynamic fluctuation time of the induced moment is smaller than the muon probing time window . in this case , the muon will observe the presence of an internal field , whereas the magnetization will observe the spin - glass behavior .
we have investigated the magnetic and transport properties of prrusi@xmath0 using various experimental techniques .
our dc and ac magnetization studies reveal the signatures of non - canonical spin - glass behavior in this compound with the freezing temperature @xmath17 @xmath24 9.8 k. the thermo - remnant magnetization in spin - glass state is found to relax with a very high mean relaxation time @xmath32 = 13357 s. the magnetic and thermal properties are strongly influenced by the crystal field effect that manifest as a broad anomaly in the specific heat and electric resistivity .
the @xmath1sr data show the presence of long range magnetic ordering below 12 k with very small internal field at the muon site .
the inelastic neutron scattering confirms a singlet cef ground state and a possibility of the induced moment magnetism arising from the excited cef doublet at 2.45 mev in the presence of strong exchange in prrusi@xmath0 .
the cef level scheme of prrusi@xmath0 is very similar to that of prau@xmath3si@xmath3 in which the spin - glass behavior arising from the dynamic fluctuations of the crystal field levels has been observed .
this leads us to believe that the origin of spin - glass behavior in prrusi@xmath2 lies in the dynamic fluctuations of crystal field levels .
considering the two contradictory ground states of prrusi@xmath0 , spin - glass behavior seen in the magnetization and specific heat study and long range ordering observed in the @xmath1sr study , we suspect that the time of dynamic fluctuations of the induced moments is smaller than the muon probing time window , therefore muon observes the presence of an internal field and hence the ordered state , whereas the magnetization observes the spin - glass behavior .
further , the value of the transition temperature of prsusi@xmath0 increases with frequency of the probe , i.e. , the ac susceptibility shows the peak at 10 k ( at 10 khz ) , while muon frequency in mhz shows the highest ordering temperature at 12 k. thus the overall behavior of prrusi@xmath0 leads us to characterize it as an induced moment spin - glass system , however further investigations preferably on single crystals are highly desired to understand better the mechanism of spin - glass behavior and induced moment magnetism in this well crystallographically ordered compound . | we have investigated the magnetic and transport properties of a ternary intermetallic compound prrusi@xmath0 using dc magnetization , ac susceptibility , specific heat , electrical resistivity , neutron diffraction , inelastic neutron scattering and @xmath1sr measurements .
the magnetic susceptibility and specific heat data reveal the signatures of spin - glass behavior in prrusi@xmath0 with a freezing temperature of 9.8 k. at low magnetic fields , we observe two sharp anomalies ( at 4.9 and 8.6 k ) in magnetic susceptibility data .
in contrast , the specific heat data show only a broad schottky - type anomaly centered around 10 k. however , @xmath1sr reveals very low frequency coherent oscillations at 1.8 k with an onset of fast relaxation below 12 k indicating a long range magnetically ordered ground state with very small moment . on the other hand ,
no magnetic bragg peaks are observed in low temperature neutron diffraction data at 1.8 k. these two contradictory ground states , spin - glass versus magnetic order , can be explained if the spin - glass behavior in prrusi@xmath2 is considered due to the dynamic fluctuations of the crystal field levels as has been proposed for spin - glass behavior in prau@xmath3si@xmath3 .
two sharp inelastic excitations near 2.4 mev and 14.7 mev are observed in the inelastic neutron scattering ( ins ) spectra between 4 k and 50 k. further , exchange coupling @xmath4 obtained from the analysis of ins data with cef model provides evidence for the spontaneously induced magnetic order with a cef - split singlet ( @xmath5 ) ground state .
however , the exchange coupling seems to be close to the critical value for the induced moment magnetism , therefore we tend to believe that the dynamic fluctuations between the ground state singlet and excited doublet cef levels is responsible for spin - glass behavior in prrusi@xmath2 . |
axisymmetric spacetimes has been studied mainly for two reasons .
the first one is that they often appear in astrophysical models like rotating stars and black holes .
the second is because in the presence of any symmetry einstein equations simplify considerable and hence these spacetimes are useful as intermediate step to understand more complex problems . in particular ,
axially symmetric gravitational waves in vacuum do not carry angular momentum , this represents an important simplification in the dynamics .
also , axial symmetry is the only symmetry compatible with asymptotic flatness and non - trivial gravitational radiation @xcite . from this perspective , axially symmetric gravitational waves are the simplest possible waves emitted from isolated sources .
and hence they represent the natural candidates to study the strong field dynamics of gravitational waves in einstein equations .
however , axial symmetry presents a major difficulty . to take advantage of the symmetry
an adapted coordinate system should be used in order to reduce the field equations to a lower - dimensional system ( there is a well known procedure to do this for any symmetry in a geometrical way @xcite , we review this result in sec . [
sec : symmetry - reduction ] ) .
the problem is that the norm of the axial killing vector vanishes at the axis , and hence the reduced equations are formally singular there .
this difficulty is so severe that until recently axially symmetric spacetimes have not been studied in detail even using numerical techniques ( see chapter 10.4 in @xcite and references therein ) . in a number of recent articles @xcite , @xcite , @xcite , @xcite ,
@xcite this kind of singular behavior has been successfully implemented numerically .
there is however no analytical study of axial symmetry in the dynamical regime ( see the review article @xcite for results for other kind of symmetries ) .
in fact , it can be argued that this singular behavior near the axis is so complicated that the axially symmetric case is as hard as the full general case from the analytical point of view .
there exists however a new ingredient that makes , in our opinion , the problem worth studying . in the article @xcite
it has been proved that there exists a gauge in axial symmetry such that the total mass of the spacetime can be written as a positive definite volume integral over the spacelike slices of the foliation . moreover ,
this integral is conserved along the evolution . this conserved integral control the norm of the fields along the whole evolution .
this is certainly a very desirable property of this gauge which is not present in the general , non - symmetric , case .
also , this mass integral formula appears to be connected with stability properties of black holes in axial symmetry @xcite .
the gauge mentioned above is a combination of the well known maximal condition for the lapse and the choice of isothermal coordinates ( also called quasi isotropical ) for the shift .
the later condition is only possible in axial symmetry .
we call it the maximal - isothermal gauge .
this gauge has been known for long time ( see @xcite and @xcite ) but without noticing this property of the mass .
it is also important to emphasize that this gauge is the one used in most of the recent numerical computations @xcite @xcite @xcite@xcite ( examples of other gauge choices in axial symmetry are given in @xcite @xcite ) . that is , this gauge has not only desirable analytical properties but it is also useful for numerical studies .
the very basic question of well - posedness of the equations in this gauge is open .
this question is rather subtle because of the singular behavior mentioned above .
the standard theory in partial differential equations does not seems to apply in a direct way .
this is the problem we want to study in this article . in order to do this
, the first step is to study the linearization of the equation around fixed solutions .
we chose minkowski as a background for simplicity .
as we describe in the next section , we obtain a remarkable simple system of linear equation together with a conserved quantity which corresponds to the mass of the spacetime up to second - order corrections .
this system allows us to formulate the problem of well - posedness in a simplified setting which is nevertheless relevant and physically interesting . remarkably enough , even for this linear system the well - posedness appears to be a nontrivial problem . in order to get insight into this problem we numerically evolve these equations to provide evidences that the system is in fact well - posed and that the solutions have the expected behavior . if the local existence problem is so complicated in this gauge one can wonder what can be said about the global behavior of the evolution , which is , of course , the ultimate goal . however , many of the main complications of this gauge are already present in the well - posedness problem because they are related with the local behavior of the fields at the symmetry axis .
if one can solve them at the linearized level in a satisfactory way there is a good chance that the mass integral formula can be used to control the global evolution in some way .
also , the well - posedness of the linear equations are relevant by themselves for the following two reasons .
first , the mass formula at the linear level can in principle be used to prove linear stability in axial symmetry of a background solution like a black hole .
second , the well - posedness of the linear equations and the mass formula give insight on appropriate boundary conditions on a bounded domain .
in particular , the mass formula allows us to calculate the gravitational waves that leave or enter a bounded domain .
the plan of the article is the following . in sec .
[ sec : main - results ] we summarize our mains results . in sec .
[ sec : axisymm - vacu - einst ] we review the axially symmetric , vacuum , einstein equations .
although this is well known , the way we process to obtain the final equations in the maximal - isothermal gauge is slightly different than the standard one used in the numerical works mentioned above . in sec .
[ sec : linearized - equations ] we derive our main linear equations and in sec .
[ sec : prop - line - equat ] we describe their main properties .
in particular , in this section we discuss the mass conservation and boundary conditions on a bounded domain . in sec .
[ sec : numerical - techniques ] we describe the numerical techniques used to evolve these equations . and in sec .
[ sec : numerical - results ] we present the numerical results . finally , in sec .
[ sec : final - comments ] we conclude with a discussion of the relevant open problems .
this article has two main results .
the first one is to prove that the linearized einstein vacuum equations in the maximal - isothermal gauge reduce to a very simple set of equations together with a conserved quantity .
this conserved quantity is the mass up to second - order corrections and it is written as a positive definite integral over a spacelike surface , which has a similar form as the energy of the wave equation .
this property of the mass , which only holds in this gauge , is of course what distinguished this system of equations from any other linearization .
the second result is the numerical study of these equations , together with the analysis of appropriate boundary conditions on a finite grid .
let us describe the first result . in axial symmetry ,
the dynamical degrees of freedom of the vacuum gravitational field are prescribed by two functions , which can be chosen to be the norm and twist potential of the axial killing vector ( see sec .
[ sec : axisymm - vacu - einst ] ) .
we make , for simplicity , the extra assumption that the twist is zero ( although we discuss the full non - linear equations with twist in sec .
[ sec : axisymm - vacu - einst ] ) .
this assumption simplify the equations but it is by no means essential . in the maximal - isothermal gauge , the linearized einstein equations with respect to a minkowski background
reduce to the following two equations for the functions @xmath0 and @xmath1 ( the reason for the notation for the last function is that it represents the @xmath2 component of the shift vector as we will see below ) @xmath3 these equations are deduced in sec .
[ sec : linearized - equations ] .
we have chosen cylindrical coordinates @xmath4 .
the relevant domain for these equations is the half plane @xmath5 , @xmath6 denoted by @xmath7 .
a dot denotes time derivative and @xmath8 is the flat laplacian in 2-dimensions @xmath9 the boundary condition for eqs . and arise from the regularity of the spacetime metric at the axis and the standard asymptotically flat fall - off behavior at infinity .
we discuss this in detail in sec .
[ sec : bound - cond - axial ] and [ sec : prop - line - equat ] .
let us present here a summary .
is an elliptic equation for @xmath1 , we need to prescribe boundary conditions on @xmath7 . on the axis
@xmath10 we require @xmath11 and at infinity we impose @xmath12 where @xmath13 . with these boundary conditions , eq .
has a unique solution .
. is a wave equation for @xmath0 , we need to prescribe initial conditions , which are functions @xmath14 and @xmath15 such that @xmath16 the axis represents a timelike boundary for the wave equation , and hence we need to prescribe also boundary conditions there .
this is the delicate part , because the equations are singular at the axis and hence we are not free to chose arbitrary boundary conditions there . from the axial regularity of the spacetime metric
we deduce that the initial data @xmath17 and @xmath18 should vanish at the axis , namely @xmath19 in sec .
[ sec : prop - line - equat ] , using series expansions , we prove that conditions on the initial data imply that @xmath20 for all times .
moreover , solutions @xmath0 and @xmath1 of equations and satisfy a parity conditions , namely @xmath0 is an even function of @xmath2 and @xmath1 is an odd function of @xmath2 .
these parity conditions imply that the spacetime metric is smooth at the axis .
it is important to emphasize that these conditions are consequences of equations and alone , without any extra requirement . in the numerical implementation , eqs .
are used as boundary conditions at the axis .
there are various ways to re - express and in order write conditions as proper timelike boundary conditions ( e.g. dirichlet or neumann ) .
for example , following @xcite , in sec .
[ sec : numerical - techniques ] we write them in terms of the rescaled variable @xmath21 .
we are interested in asymptotically flat solutions of and .
we will argue in sec .
[ sec : prop - line - equat ] , that the typical fall off behavior as @xmath22 for this kind of solutions is @xmath23 that is , if we chose initial data @xmath17 and @xmath18 which satisfy then @xmath0 will satisfy for all times .
all the other components of the linear perturbation can be calculated in terms of @xmath0 and @xmath1 as follows . in our gauge
the four dimensional coordinates are given by @xmath24 .
a general twist free linear perturbation is written as follows @xmath25 where the functions @xmath26 , @xmath27 , @xmath1 and @xmath28 depends only on @xmath4 .
the function @xmath1 is given by ( [ eq:120b ] ) , the other functions are calculated in terms of @xmath0 as follows .
the functions @xmath27 is a time derivative of @xmath0 @xmath29 the function @xmath26 is determined by the following elliptic equation @xmath30 where @xmath31 is defined as @xmath32 this operator , which appears frequently in the rest of the article , is the flat laplace operator in 3-dimensions written in cylindrical coordinates and acting on axially symmetric functions .
the boundary condition for equation ( [ eq:86 ] ) at the axis is given by @xmath33 and at infinity we impose @xmath34 eq .
can be also viewed as an equation in @xmath35 . in this case
we do not need to prescribe any boundary condition at the axis .
condition will be automatically satisfied for any regular solution . finally , the other component of the shift vector is determined by the following equation @xmath36 with boundary condition at the axis @xmath37 and decay condition at infinity @xmath38 the total mass of the system is given by the following integral @xmath39 note that in order to compute the mass we need the function @xmath26 , which satisfies eq . .
this equation is uncoupled with equations and .
the integral is conserved .
that is , for every solution of and which satisfy the boundary conditions , , and decay at infinity like we have @xmath40 the conservation law is deduced from a local conservation formula which involves the integrand of the mass formula .
this local conservation law can be also used to compute the gravitational waves entering or leaving a bounded domain .
we discuss this in sec .
[ sec : prop - line - equat ]
. the second main result of this article is the numerical study of the system and .
we describe this in detail in sec .
[ sec : numerical - results ] .
let us briefly summarize these results .
the system and appears , from the numerical evidences , to be well posed and numerically stable .
in particular , this imply that the functions @xmath0 and @xmath1 remain bounded for all times by a constant that depends only on the initial data .
this is consistent with the linear stability of minkowski spacetime .
the numerical calculations are , of course , performed on a finite grid .
hence , we need to prescribe boundary conditions on a bounded domain . these conditions should be compatible with asymptotically flatness in the following sense .
assume we have a sequence of bounded domains such that in the limit they cover the half plane @xmath7 .
if we solved the equations for this sequence of domains we should recover in the limit the asymptotically flat solution described above .
there exists many different boundary conditions that have this property .
in particular , homogeneous dirichlet conditions for @xmath1 and @xmath0 . for each bounded domain the mass is not conserved .
however , as the size of the domain increase we expect that the mass approach a time independent constant .
this is precisely what we observe in our numerical calculations . for our present goal
, this kind of asymptotically flat boundary conditions is all what we need
. there is , however , an interesting extra point here . to model an isolated system on a finite grid it is important to prescribe boundary conditions such that the gravitational radiation leaves the domain . in general
, this is a very difficult problem since it is not even clear what we mean by gravitational radiation at a finite distance .
however , as we mention above , in our gauge the mass formula allow us to compute gravitational radiation on a bounded domain .
although it appears not to be possible to prescribe boundary conditions such that the gravitational waves always leave the domain , the mass formula suggests a particular kind of boundary conditions that has this behavior in our numerical calculations .
that is , under these boundary conditions , the mass on a bounded domain is monotonically decreasing with time for the particular kind of initial data used in the computations .
we emphasize however that we have not been able to prove this analytically .
we explore this in detail in sec .
[ sec : prop - line - equat ] and [ sec : numerical - results ] .
the purpose of this section is to write the vacuum einstein equations for axially symmetric spacetimes in the maximal - isothermal gauge .
this involves three clearly distinguished steps . in the first one ,
described in sec .
[ sec : symmetry - reduction ] , we perform a symmetry reduction of einstein equations to obtain a set of geometrical equations in the 3-dimensional quotient manifold .
these equations can be viewed as 3-dimensional einstein equations coupled with effective matter sources . in the second step ( sec .
[ sec:2 + 1-decomposition ] ) we chose an arbitrary spacelike foliation in the quotient manifold and split the equations in time plus space . in sec .
[ sec : gauge ] we fixes the foliation and the coordinate system .
we also write the mass formula in this gauge . finally , in sec .
[ sec : bound - cond - axial ] we discuss boundary conditions at the axis and at infinity . in this section we perform the symmetry reduction of the field equations .
we follow @xcite and @xcite .
see also @xcite @xcite . consider a vacuum solution of einstein s equations ,
i.e. , a four dimensional manifold @xmath41 with metric @xmath42 ( with signature @xmath43 ) such that the corresponding ricci tensor vanishes @xmath44 suppose , in addition , that the metric @xmath42 admits a killing field @xmath45 , that is @xmath45 satisfies the equation @xmath46 where @xmath47 is the connection with respect to @xmath42 .
greek indices @xmath48 denote four dimensional indices .
we define the square of the norm and the twist of @xmath45 , respectively , by @xmath49 using the field eq .
it is possible to prove that @xmath50}=0,\ ] ] and hence @xmath51 is locally the gradient of a scalar field @xmath52
@xmath53 let @xmath54 denote the collection of all trajectories of @xmath45 , and assume that it is a differential 3-manifold .
we define the metric @xmath55 on @xmath54 by @xmath56 the vacuum field equations can be written in the following form on @xmath54 @xmath57 where @xmath58 and @xmath59 are the connexion and the ricci tensor of @xmath60 , we have defined @xmath61 and latin indices @xmath62 denote three dimensional indices on @xmath54 . note
that the definition of the metric involves a conformal rescaling with respect to the canonical metric @xmath63 defined by @xmath64 that is , we have @xmath65 this rescaling simplify considerably the field equations .
in particular , on the right hand side of eq .
there are no second derivatives of the fields @xmath66 and @xmath52 ( compare , for example , with equation ( 20 ) in @xcite ) .
finally , we note that eq .
( [ eq:4 ] ) can be written in the following form @xmath67 where we have defined @xmath68 up to this point , the only assumption we have made is that the spacetime admits a killing vector field @xmath45 and that @xmath45 is not null , otherwise the metric @xmath60 is not defined .
if the killing field is timelike ( @xmath69 ) then the metric @xmath60 is riemannian and the equations are the stationary einstein vacuum equations . on the other hand , when the killing vector is spacelike ( @xmath70 ) , the metric @xmath60 is a is a 3-dimensional lorenzian metric ( we chose the signature @xmath71 ) . in axially symmetry ,
the killing vector @xmath45 is spacelike and its norm vanishes at the axis of symmetry .
hence , the equations are formally singular at the axis .
this singular behavior at the axis represents the main difficulty to handle these equations . in the lorenzian case , eq . has
the form of einstein equations in three dimensions , with effective matter sources produces by @xmath66 and @xmath52 .
the effective matter eqs . imply that the energy - momentum tensor defined in terms of @xmath66 and @xmath52 by @xmath72 is divergence free , i.e. @xmath73 . a particularly relevant special case is when @xmath74 . in that case eqs .
simplify considerable @xmath75 we have pointed out that the rescaling simplifies the equations and allow us to write them in a more geometric form .
this is the reason why this scaling is used in the case of @xmath76 cosmologies where the equations are locally the same but the norm @xmath66 never vanishes ( see @xcite @xcite and the review article @xcite ) . in our case the conformal scaling is singular at the axis .
however , since the behavior of @xmath66 at the axis , as we will see in the next sections , is controlled a priori this singular scaling does not seems to introduce any extra difficulty in the equations .
we also remark that in all the numerical works mentioned above this conformal rescaling was not used , the equations are written in terms of the metric @xmath77 defined by .
( [ eq:4])([eq:4c ] ) are purely geometric with respect to the metric @xmath60 . to solve these equations we need to prescribe some gauge for the metric @xmath60 .
this will be done in the next two sections .
in order to formulate an initial value problem , we will perform an standard @xmath78 decomposition of eqs .
( [ eq:4])([eq:4c ] ) .
note that this is completely analogous to the @xmath79 decomposition of einstein equations , in fact all the formulas are formally identical because the dimension do not appears explicitly in them ( see , for example , @xcite , @xcite ) . consider a foliation of spacelike , 2-dimensional slices @xmath80 of the metric @xmath60 .
let @xmath81 be an associated time function and let @xmath82 be the unit normal vector orthogonal to @xmath80 with respect to the metric @xmath60 .
the intrinsic metric on @xmath80 is denoted by @xmath83 and is given by @xmath84 define the density @xmath85 by @xmath86 and the current @xmath87 by @xmath88 where @xmath89 denotes the trace of @xmath59 .
then , using eq .
we obtain @xmath90 where @xmath91 is the connexion with respect to @xmath92 .
the prime denotes directional derivative with respect to @xmath82 , that is @xmath93 where @xmath94 is the lapse and @xmath95 is the shift vector of the foliation .
the indices @xmath96 denotes two dimensional indices on @xmath80 .
the constraints equations corresponding to ( [ eq:4c ] ) are given by @xmath97 where @xmath98 is the ricci scalar of @xmath92 , @xmath99 is the second fundamental form of @xmath80 and @xmath100 its trace @xmath101 we use the following sign convention for the definition of @xmath99 @xmath102 where @xmath103 denotes lie derivative .
the evolution equations are given by @xmath104 where @xmath105 and @xmath106 the evolution equations and the constraint equations
constitute a complete @xmath78 decomposition of the 3-dimensional einstein eq . .
it remains to decompose the effective matter eqs .
. this can easily be obtained using the decomposition formula for the wave operator @xmath107 and the definition of the metric @xmath83 given by .
the result is the following @xmath108 where instead of we have use , and @xmath109 is the laplacian with respect to @xmath92 , i.e. @xmath110 . finally , we mention that the line element of the metric @xmath60 takes the standard form @xmath111 in this section we describe the maximal - isothermal gauge .
in particular we review the mass formula for this gauge ( see @xcite for details ) . for the lapse ,
we impose the maximal condition on the 2-surfaces @xmath112 note that we are not imposing that the surfaces are maximal in the 3-dimensional picture as in @xcite .
the later condition is the one generally used @xcite @xcite , but the difference is only minor . in particular the mass formula is positive definite for both conditions as we will see .
the one used here appears to be natural with respect to the rescaled metric @xmath60 .
. implies the following well known equation for the lapse @xmath113 where @xmath114 the maximal gauge can be , of course , imposed in any dimensions and it is not related at all with axial symmetry . in contrast , the condition for the shift is peculiar for two space dimensions .
the shift vector is fixed by the requirement that the intrinsic metric @xmath92 has the following form @xmath115 where @xmath116 is a fixed ( i.e. @xmath117 ) flat metric in two dimensions .
then , using , we obtain that the trace free part of is given by @xmath118 where @xmath119 is the conformal killing operator in two dimensions with respect to the metric @xmath92 defined in eq . .
equation ( [ eq:42 ] ) is an elliptic first order system of equations for @xmath95 .
the elliptic eqs .
( [ eq:37 ] ) and ( [ eq:42 ] ) determine lapse and shift for the metric @xmath60 and hence fixes completely the gauge freedom in eqs .
. this gauge has associate a natural cylindrical coordinate system @xmath120 for which the metric @xmath116 is given @xmath121 and the axis of symmetry is given by @xmath10 .
the slices @xmath80 are the half planes @xmath7 . for the analysis of the equations it is of course important to write them explicitly as partial differential equations in these coordinates .
we will do this in the remainder of this section .
in general , due to the complexity of einstein equations , the partial differential equations obtained in a particular gauge can be quite involved . in our case , however , the geometric nature of the gauge plus the symmetry reductions will provide a relative simple set of equations .
we first present some useful definitions .
we need to subtract from @xmath66 the part that vanishes at the axis .
we define the function @xmath26 by @xmath122 due to the rescaling , the lapse @xmath94 vanishes also at the axis , hence we define the normalized lapse @xmath123 by @xmath124 from the regularity conditions presented in the next section we will see that it is useful to define the function @xmath27 defined by @xmath125 we now proceed to write the equations .
we begin with the evolution equations for @xmath26 and @xmath52 .
the evolution equation for @xmath26 is given by .
using the definition and the conformal rescaling expression for the laplacian we obtain @xmath126 in the same way , from we get @xmath127 where @xmath128 denotes partial derivatives with respect to @xmath2 and @xmath129 and all the indices are moved with respect to the flat metric @xmath116 . in these equations the lapse @xmath94 and the shift @xmath95 appear trough the prime operator defined in .
the momentum constraint ( [ eq:8b ] ) is given by @xmath130 where @xmath131 , that is we have @xmath132 to obtain we have used the conformal rescaling of the divergence in 2-dimensions given by .
the indices in eq . and in the rest of the article , are moved with the flat metric @xmath116 . to avoid confusion ,
it is useful to introduce the following notation @xmath133 that is , we want to distinguish between , say , the covector @xmath134 used in the previous section and @xmath135 ( see the discussion after eq . in the appendix ) . the hamiltonian constraint , eq .
( [ eq:8 ] ) , is given by @xmath136 where @xmath137 let us consider the evolution equations for @xmath92 and @xmath99 .
the evolution equation for the metric @xmath92 reduces to @xmath138 and the evolution equation for the second fundamental form @xmath99 is given by @xmath139 where @xmath140 denotes the trace free part ( with respect to @xmath116 ) of @xmath141 . using eq . )
we obtain @xmath142 and @xmath143 denotes the trace free part of @xmath144 , namely @xmath145 where @xmath144 is given by .
the equation for the lapse is given by @xmath146 and for the shift we have @xmath147 where @xmath148 is the flat conformal killing operator defined by . using the identity
, we can transform the the first order system of eqs . for the shift and for the momentum constraint in a pair of second - order uncoupled equations . for the shift ,
we take a divergence to eq . to obtain @xmath149 for eq .
we define the vector @xmath150 by @xmath151 and hence eq .
transform to @xmath152 the total adm mass of the spacetime can be calculated as a volume integral on the half plane @xmath7 of the positive definite effective energy density ( see @xcite ) @xmath153 finally , we mention that for the twist free case ( @xmath74 ) the four dimensional spacetime metric @xmath42 has a simple expression in these coordinates , namely @xmath154 the boundary conditions at the axis in axial symmetry have been extensible analyzed in the literature @xcite , @xcite , @xcite , @xcite .
they involve parity conditions in the @xmath2 dependence of the different fields .
that it , the relevant functions are either even or odd functions of @xmath2 . in order to use these results in our setting ,
it is useful to write the relations of the quantities with respect to the rescaled metric @xmath60 and the canonical metric @xmath77 , since all the above mentioned articles work with the metric @xmath77 .
using relation we obtain for the 2-dimensional metric @xmath155 and for the second fundamental form @xmath156 where quantities with a tilde are written with respect to the metric @xmath157 .
we also have @xmath158 using these relations and the results mentioned above it is straightforward to obtain the following behavior of the relevant variables @xmath159 and @xmath160 note that odd functions vanishes at the axis and the @xmath2 derivative of even functions vanishes at the axis .
it follows that one can impose homogeneous dirichlet boundary conditions at the axis for odd functions and homogeneous neumann boundary conditions for even functions .
in addition , we have that the function @xmath27 defined by should vanished at the axis @xmath161 since @xmath27 is an even functions , from we deduce that @xmath162 near the axis .
finally , there is an important regularity condition which comes from the axial regularity of the 3-dimensional extrinsic curvature .
let us define the following quantity @xmath163 then it follows that @xmath164 near the axis .
this is the equivalent of the regularity condition given in equation ( 50 ) in @xcite adapted to our conformally rescaled metric .
see also @xcite , @xcite .
the fall off conditions at infinity are the standard asymptotically flat ones .
in particular we have @xmath165 and @xmath166 as @xmath167 .
in this section we make a linear expansion around minkowski of the einstein equations in the maximal - isothermal gauge described in the previous section . note that for minkowski we have @xmath168 and hence , due to the rescaling ( [ eq:17 ] ) , the background metric @xmath60 , given in coordinates by , is non - flat @xmath169 the other background quantities are given by @xmath170 and @xmath171
the hamiltonian constraint and the equation for the lapse are non - trivial for the metric ( [ eq:55 ] ) , namely @xmath172 let us proceed with the linearization . for simplicity we will consider only the case @xmath74 .
the first step is to compute the lapse function . the right hand side of eq .
( [ eq:49 ] ) is second - order , then , using the boundary condition we obtain @xmath173 that is , the maximal condition for the lapse is trivial at the linearized level . on the contrary , as we will see , the equation for the shift plays a crucial role .
the next step is to compute the linearization of the wave eq .
( [ eq:137 ] ) for @xmath26 , we obtain @xmath174 where dot means partial derivative with respect to @xmath81 and we have defined @xmath175 in order to close the system we need an equation for @xmath1 . using equation ( [ eq:65 ] ) and for the momentum we obtain @xmath176 with @xmath177 we define the vector field @xmath150 by eq .
( [ eq:46 ] ) and then by eq .
( [ eq:47 ] ) we obtain @xmath178 from ( [ eq:76 ] ) we deduce @xmath179 and hence we get @xmath180 by the fall off condition , we obtain @xmath181 in the following , to simplify the notation we set @xmath182 eq . ( [ eq:93 ] ) reads @xmath183 using ( [ eq:46 ] ) we also obtain @xmath184 for the shift we have the equation @xmath185 taking a divergence to this equation ( or linearizing ( [ eq:50 ] ) ) we obtain @xmath186 note that in ( [ eq:94 ] ) we get an equation for @xmath1 decoupled from @xmath28 . using eq .
( [ eq:76 ] ) , ( [ eq:77 ] ) from this equation we get @xmath187 eq . , together with ( [ eq:135 ] ) and ( [ eq:73 ] ) form a complete system for the variables @xmath0 , @xmath26 and @xmath1 .
alternative , using eq .
and we can eliminate @xmath99 and hence also @xmath0 .
we get the following equation for @xmath95 @xmath188 eq . , together with ( [ eq:135 ] ) and ( [ eq:73 ] ) form a complete system for the variables @xmath26 , @xmath1 and @xmath28 .
there is however an important difficulty .
the linearization of the regularity condition is given by @xmath189 were we have used eq .
( [ eq:56 ] ) . from the set of equations presented above , it is difficult to ensure that this condition will be satisfied . to enforce this condition we will write the equations in terms of different variables .
in order to do that , we need first to compute the remaining equations , namely the hamiltonian constraint and the evolution equation for the metric and second fundamental form .
since @xmath190 defined in is second - order , the hamiltonian constraint is given by @xmath191 the evolution equation for @xmath27 is obtained from ( [ eq:164 ] ) @xmath192 the evolution equation for @xmath99 is obtained linearizing ( [ eq:140 ] ) @xmath193 we can also write the evolution eqs .
( [ eq:79 ] ) in components @xmath194 using eqs .
( [ eq:56 ] ) we deduce the important relation @xmath195 which only holds in the twist free case .
this relation simplify the equations considerable . from eq .
( [ eq:79 ] ) we also deduce @xmath196 with these equations we can compute the time derivative of @xmath197 @xmath198 and hence the evolution equation for @xmath27 is given by @xmath199 as a consequence , @xmath27 satisfies the following wave equation @xmath200 we also have @xmath201 eqs . and form a complete systems for the variables @xmath27 and @xmath1 .
a similar choice of variable was used in by @xcite @xcite@xcite . however , in our particular case ( i.e. linear equation without twist ) it is possible a further simplification , namely to use eq . and hence replace @xmath27 by @xmath0 in these equations and then integrate in time . in this way
we obtain our main eqs . and .
the advantage of using @xmath0 as a variable is that the mass integral has a simple expression in terms of @xmath0 given by .
this formula for the mass is obtained expanding up to second - order the energy density , using equations to replace @xmath99 and eq . to replace @xmath202 .
we discuss this in more detail in the next section . the boundary conditions and for @xmath1 arise from the axial regularity condition and the asymptotically flat fall off .
conditions arise from the axial regularity conditions for @xmath27 given by and . the main advantage of eqs . and
is that they have build in the regularity condition as we will see in the next section .
let us mention that the non - trivial regularity condition is written in terms of @xmath0 as follows @xmath203 the component @xmath28 , which does not appears in eqs . and , can be calculated using @xmath204 or , alternative , using eq .
which is is obtained taking a derivative to eqs .
. finally , the four dimensional perturbation is obtained using the line element and the background values , , .
in this section we analyze some properties of our linear equations and .
we begin with the symmetries of these equations .
the first symmetry is given by translation in @xmath129 .
this is to be expected since the gauge fixes the axis ( and hence there is no translation freedom in @xmath2 ) , but we still have the freedom to chose the origin in the @xmath129 coordinate .
then , if we have a solution @xmath205 ; the derivative @xmath206 is also a solution , since @xmath207 commute with all the differential operators because their coefficients depend only on @xmath2 .
the same argument applies to time translations , which is the second symmetry of the equations .
the third symmetry is scaling .
let @xmath208 a positive real number .
for a given solution @xmath209 we define the rescaled function as @xmath210 where @xmath211 and the same for @xmath1 .
then , @xmath212 define also a solution in terms of the rescaled coordinates .
the mass rescales like @xmath213 in order to understand the equations in a simpler situation , let us first consider eqs . and on a bounded domain @xmath214 which does not contain the axis .
on @xmath214 the coefficient of eqs . and
are smooth .
is an elliptic equation for @xmath1 if we consider @xmath0 as a given function .
hence , in order to solve this equation we need to prescribe elliptic boundary conditions for @xmath1 on @xmath215 .
for example , dirichlet or neumann boundary conditions .
eq . is a wave equation for @xmath0 if we consider @xmath1 as a given function . to solve this wave equation we need to prescribe initial data for @xmath0 and @xmath216 at @xmath217 together with compatible boundary conditions for @xmath0 at @xmath218 for example dirichlet , neumann or sommerfeld boundary conditions for @xmath0 at @xmath215 .
the equations are of course coupled , so it is not obvious that the above procedure of fixing boundary conditions is correct since @xmath0 and @xmath1 are not `` given functions '' .
however , it is possible to prove that this is procedure is in fact correct . consider the following iteration scheme @xmath219 in this iteration , the equations are not coupled and hence the boundary conditions mentioned above ( which are kept fixed ) are correct .
following similar arguments to the one presented in @xcite ( see also @xcite ) it is not difficult to see that this iteration converges for some small time interval .
and hence we get well - posedness for the linear system and under these boundary conditions on the domain @xmath214 .
the reason why the iteration and converges is the following . from eq . , using standard elliptic , estimates we obtain that @xmath1 is equivalent ( in number of derivatives ) to @xmath0 .
hence , the term containing @xmath1 in eq . is equivalent to a first order derivative of @xmath0 and then it is not in the principal part of the wave equation .
this rough argument suggests that the combination of elliptic estimates and energy estimates for the wave equations will close and hence the iteration will converge .
this is basically the argument presented in @xcite and @xcite .
if the domain @xmath214 is not bounded , this argument will still work if we add appropriate fall - off conditions at infinity .
however , the situation change drastically when @xmath214 includes the axis .
let us analyze that case . since the axis is a singular boundary for the equations , we are not free to chose arbitrary boundary condition there .
in fact @xmath1 and @xmath220 should vanishes at the axis , otherwise the equations become singular . if we use lhpital rule ,
we conclude that the term with @xmath1 in contain in fact two derivatives with respect to @xmath2 at the axis .
that is , due to the lhopital limit , to divide by @xmath2 is equivalent as to take a derivative with respect to @xmath2 at the axis .
but then , using eq . , we conclude that this term is equivalent to second derivatives of @xmath0 and hence it is in the principal part of the wave equation .
we can not conclude that the iteration scheme converges if we include the axis in the domain .
this is , roughly speaking , the main difficulty to prove the well - posedness of the linear system and .
it appears to be difficult to identify the principal part of the system at the axis and to construct an appropriate iteration scheme .
let us discuss in detail the boundary conditions at the axis .
we are interested in solutions @xmath0 which vanish at the axis , this comes from the regularity condition .
moreover , we have seen in section [ sec : bound - cond - axial ] , the smoothness of the spacetime metric at the axis implies that the functions @xmath1 and @xmath0 should satisfies the parity conditions and . let us see heuristically , how these conditions are automatically implied by the equations provided we impose the following standard boundary conditions . at the axis
we impose @xmath221 for @xmath0 we prescribe initial data @xmath222 such that @xmath223 note that we are not imposing any condition on @xmath0 at the axis for @xmath224 .
we make a formal series expansion , namely let as assume that our solution is smooth at the axis and has the form @xmath225 substituting these expansions in eqs .
and we obtain the following recurrence relation for the coefficients @xmath226 and @xmath227 these expressions are valid for all integer @xmath228 , with the convention that the coefficients @xmath229 and @xmath230 vanished for @xmath231 . the first non - trivial @xmath228 in eq .
is @xmath232 , which gives the relation @xmath233 the term @xmath234 is given by @xmath235 this is a wave equation in 1-dimension . from the boundary conditions we obtain @xmath236
hence we deduce from that @xmath237 from the initial data conditions we have that @xmath238 these provides trivial initial data for the wave eq .
and hence we deduce @xmath239 that is , we have deduced the behavior @xmath240 only from the boundary conditions and the condition on the initial data .
we want to prove now that and imply that all @xmath230 with @xmath228 odd and all @xmath229 with @xmath228 even are zero .
we prove this by induction .
let us assume that for some @xmath228 ( with @xmath241 ) we have that @xmath242 using eq .
we deduce @xmath243 and from we have @xmath244 and then we have @xmath245 . since is valid for @xmath246 we have proved the desired result .
that is , the solutions @xmath0 and @xmath1 satisfy the parity conditions and respectively . using that @xmath0 is an even function of @xmath2 and that @xmath240 it is straightforward to deduce that the regularity condition holds for all times .
we analyze the fall off behavior of the solution @xmath0 .
this behavior is completely determined by the initial data @xmath17 and @xmath18 .
let us assume that the initial data has compact support . in the case of the wave equation , the signal will propagate with finite speed and hence the solution will always have compact support for any finite time . in our case , however , the coupling with the elliptic eq . produce a non - local behavior .
even if we start with compactly supported data , the function @xmath1 will instantaneously spread to all space .
let us perform a formal expansion in @xmath247 to see the typical behavior of @xmath0 .
we have that @xmath248 for all times , this is prescribed by the boundary conditions . in @xcite it has been proved that this implies that @xmath249 , and hence the terms containing @xmath1 in is @xmath250 .
then , at @xmath217 we obtain that @xmath251 .
if we take time derivatives of the equations and repeat this argument , we get that all time derivatives of @xmath0 are @xmath250 .
then , we conclude that the typical fall - off behavior for asymptotically flat solutions is given by , in the sense that we can not expect a faster decay in general . instead of compactly
supported data we can begin with initial data for @xmath0 such that they are @xmath250 at infinity .
let us discuss now the most important property of equations and namely the mass conservation . as usual
, the mass appears as a second - order quantity that can be calculated in terms of squares of first order quantities .
the density ( [ eq:102 ] ) up to this order is given @xmath252 the total mass is calculated by the integral ( [ eq:103 ] ) .
the mass integral is conserved for the full nonlinear equations in this gauge ( see @xcite ) and hence it is conserved at the linearized level .
it is however important to compute explicitly this conservation formula using only the linear eqs .
( [ eq:119b ] ) and ( [ eq:120b ] ) .
note that the function @xmath26 appears in the mass and this function should be calculated from @xmath0 using eq . .
to compute the time derivative of @xmath253 we need first to calculate the time derivative of @xmath26 .
using the evolution eq .
( [ eq:119b ] ) only , we compute @xmath254,\ ] ] where we have defined @xmath255 then , using and the time derivative of we get @xmath256 if we are solving in the whole half plane @xmath7 then , by the fall - off conditions , we deduce that the only possible solution of this equation is the trivial one , and hence @xmath257 we have proved that eqs .
( [ eq:119b ] ) and ( [ eq:120b ] ) together with ( [ eq:86 ] ) imply eq .
( [ eq:190 ] ) .
we can also formulate the system in a different way .
we can take ( [ eq:119b ] ) and ( [ eq:120b ] ) and eq .
( [ eq:190 ] ) , instead of ( [ eq:86 ] ) , as an evolution equation for @xmath26 . if we take the laplacian @xmath31 to both sides of eq .
( [ eq:190 ] ) and use the identity together with eq .
( [ eq:120b ] ) we obtain @xmath258 hence , if we chose initial condition for @xmath26 such that @xmath259 eq . implies ( [ eq:86 ] ) .
this two different ways of calculating @xmath26 correspond to a constrained system and a free system ( using the terminology defined in @xcite ) .
the previous calculation is nothing but the propagation of the hamiltonian constraint at the linearized level . for the full einstein equations ,
the difference of constrained and free evolution schemes involves different set of evolutions equations . in our linear system the evolution equations are the same ( namely , ( [ eq:119b ] ) and ( [ eq:120b ] ) ) , the difference is the way the function @xmath26 ( and hence the mass ) is calculated .
these two ways are of course completely equivalent when the domain is the whole half plane @xmath7 , however , as we will see , they are not equivalent for a bounded domain . using eqs .
( [ eq:190 ] ) , ( [ eq:86 ] ) , ( [ eq:119b ] ) and ( [ eq:120b ] ) we obtain the following local conservation law for the density @xmath190 defined by @xmath260 where @xmath261 the vector @xmath262 can be interpreted as the energy flow of the gravitational field .
if we integrate eq .
( [ eq:89 ] ) in @xmath7 we have that the boundary terms vanishes both at the axis ( by the axial regularity ) and at infinity ( by the fall off conditions ) .
then we have @xmath263 we can also integrate eq .
( [ eq:89 ] ) on a bounded domain @xmath214 , namely we define the mass contained in @xmath214 by @xmath264 and then we have @xmath265 where @xmath266 is the unit normal of @xmath215 . the quantity @xmath267 measure how much energy is leaving or entering the domain
. the local conservation formula ( [ eq:89 ] ) can be generalized for the non - linear equations @xcite . using the conservation of the mass ( [ eq:41 ] ) we can prove uniqueness of solutions of the system .
let us say we have two different solutions with the same initial data .
we take the difference between the two solutions .
the difference satisfies the same equation with zero initial data .
in particular @xmath26 on the initial surface is zero . and
hence the mass is zero .
since it is conserved the mass is zero for all times , which implies that the solution is zero . in the case of hyperbolic equations ( the wave equation for example )
the conservation of the energy gives also local properties of the solution , namely finite speed propagation of signals .
however this is not the case here ; the elliptic equation implies a non - local behavior of the solution .
the discussion above applies for the domain @xmath7 which is the relevant domain for the equations .
however , in numerical computation we need to solve the equations on a finite grid and hence it is necessary to impose boundary conditions on a bounded domain .
a typical domain for the numerics is shown in fig .
[ fig:1 ] . as we mention in section [ sec : main - results ] , for our present purpose we only need to prescribe some boundary conditions compatible with asymptotic flatness .
for example , homogeneous dirichlet boundary conditions for @xmath0 and @xmath1 .
however , the mass formula rise an interesting point here . on a bounded domain , to calculate @xmath26 we have two possibilities .
first , we can determine @xmath26 as the unique solution of the elliptic equation ( [ eq:86 ] ) with some boundary conditions .
if we do so , then we again deduce eq .
( [ eq:189 ] ) . however , from this equation we can not deduce ( [ eq:190 ] ) . in effect , we have @xmath268 where @xmath269 satisfies @xmath270 we can not conclude that @xmath269 is zero from this equation , because @xmath269 will have non - trivial boundary condition .
namely , let us assume the we prescribe some boundary condition for @xmath26 .
we can not control the boundary value of @xmath271 , and hence we can not ensure that @xmath269 vanishes at the boundary .
in fact , the function @xmath269 is fixed as the unique solution of with boundary values @xmath272 then , if we compute the time derivative of the density @xmath190 we get @xmath273 that is , we do not get a conservation law , there is a volume term given by @xmath269 .
there seems to be no boundary conditions for @xmath26 that can ensure @xmath269 to vanishes .
the other possibility is to compute @xmath26 using the evolution equation ( [ eq:190 ] ) with initial condition ( [ eq:62 ] ) . from ( [ eq:190 ] )
, in the same way as we mentioned above we deduce , since in this deduction the boundary conditions play no role . using the initial data condition ( [ eq:62 ] ) , from we deduce .
that is , we are in the same situation as the whole domain .
hence , in this case we recover ( [ eq:89 ] ) , where @xmath262 is given by the same expression . from this point of view , this evolution scheme appears to be better than the previous one . in this scheme
, we are free to chose any elliptic boundary condition for @xmath1 and any boundary condition for @xmath0 compatible with the wave equation
. for @xmath26 we do not have any freedom , and hence we can not prescribe the boundary value of this function . a natural choice of boundary conditions would be to force the boundary integral in to have a definite sign .
these conditions would have the interpretation of radiative boundary conditions , in the sense that the energy is leaving the domain . to prescribe such conditions
seems not to be possible ( at least for generic data ) since we do not have any control on the term with @xmath26 .
however , we can do something intermediate .
namely , if we impose sommerfeld boundary condition for @xmath0 @xmath274 and homogeneous dirichlet conditions for @xmath1 we have that the first term in has negative sign , the third term is zero . for the second and fourth term we have no control a priori .
but we can expect that the influence of these term is small at least for some class of initial data .
if this is true , then we get @xmath275 this is what we observe in our numerical simulations described in the next sections . for the numerical evolution [ fig:1],width=151 ]
in this section we want to study numerically the initial - boundary value problem ( ibvp ) for the eqs .
( [ eq:119b ] ) and ( [ eq:120b ] ) . in this problem
the symmetry axis , @xmath276 becomes a boundary of our domain .
notice then , that working with the variable @xmath0 poses an inconvenient as regards the boundary condition at @xmath10 since , according to , this function satisfies both , homogeneous dirichlet boundary condition and homogeneous neumann boundary condition .
it is then convenient to rewrite the equations in terms of a new variable for which the smoothness properties at the symmetry axis defines a unique , equivalent , boundary condition .
we define @xmath277 this new variable vanishes linearly with @xmath2 and the correct boundary condition is simply homogeneous dirichlet at @xmath278 the equation for @xmath279 with @xmath280,~z\in[0,l],$ ] and @xmath281 is @xmath282 where @xmath283 is determined by the elliptic equation @xmath284 with homogeneous dirichlet boundary conditions , @xmath285,\\ { \beta^\rho}(\rho,0,t)&={\beta^\rho}(\rho , l , t)=0,\quad \rho\in[0,r ] .
\end{split}\ ] ] for all @xmath286 the boundary condition for @xmath287 at the symmetry axis is @xmath288 while at the outer boundaries we study two possibilities , homogeneous dirichlet , @xmath289 or sommerfeld ( outgoing waves ) @xmath290 the initial data are @xmath291 where @xmath292 and @xmath293 are @xmath294 functions with compact support in @xmath295 so that the compatibility of the boundary and initial data is not an issue . the eqs .
( [ eq : hyp])([idata ] ) constitute the ibvp we approximate with our finite difference scheme .
we want to emphasize here an important difference between our numerical approach with the usual approaches in the area ( see for example @xcite ) .
we solve the ibvp for ( [ eq : hyp ] ) as a second - order equation just it is written above , i.e. we do not reduce ( [ eq : hyp ] ) to a first order system of equations .
the treatment of evolution equations as second - order equations as opposed to first order systems of equations has several advantages .
for example , the number of dynamical fields , and then the number of equations , is not increased .
this facilitate the treatment of the boundary conditions .
there are also numerical accuracy advantages . in the context of general relativity , this has been stressed in @xcite . in particular the simplest proofs of well - posedness for general initial - boundary value problems for einstein s equations have been found recently using second - order systems of equations @xcite,@xcite .
[ [ the - implementation . ] ] the implementation .
+ + + + + + + + + + + + + + + + + + + in our numerical experiments we always consider square domains , i.e. , @xmath296 to define the numerical grid let @xmath297 be a positive integer and @xmath298 the space stepsize .
we define our grid to be half a stepsize displaced from all the boundaries .
we think of our grid as a uniformly distributed set of points each of which is at the center of one of the @xmath299 square cells covering the domain .
the coordinates of the gridpoint at the site @xmath300 are then @xmath301 the sites @xmath300 with @xmath302 are within the domain , while the sites with @xmath303 and @xmath304 are `` ghost points '' used to ease the implementation of the boundary conditions @xcite .
time is discretized as @xmath305 we use capital latin letters to denote the grid functions associated to the dynamical variables .
also , we use sub - indices to denote the space - site indices and a super - index to denote the time step .
this is , @xmath306 besides the uniform grid we introduce the extra gridpoints placed at the physical boundary @xmath307 and denote the values of @xmath287 at these points as @xmath308 respectively . in our difference scheme
we approximate space derivatives by the standard fourth order accurate centered difference operators given by @xcite @xmath309 and add a sub - index @xmath2 or @xmath129 to indicate what coordinate the operator is acting on .
for example @xmath310 is approximated by @xmath311 at every time step we need to solve the elliptic eq .
( [ eq : ell ] ) which we approximate by @xmath312 we solve this difference equation iteratively using the gauss - seidel iteration scheme , and stop the iteration when the difference between both sides in ( [ eq : discrete - ell ] ) is smaller , in maximum norm , than a given small tolerance @xmath313 we then extend the solution to the ghost points so that the homogeneous boundary condition is satisfied as follows @xmath314 we now describe how we approximate ( [ eq : hyp ] ) using fourth - order accurate difference approximations in space ; to use second - order accurate approximations instead , we just need to change @xmath315 and @xmath316 in what follows by @xmath317 and @xmath318 respectively
. we set @xmath319 and extend the solution to vanish at all ghost points and boundary points since the initial data has compact support
. then we compute @xmath320 by solving ( [ eq : discrete - ell ] ) as explained above .
( first step ) .
we do , for @xmath321 @xmath322 now , if working with boundary condition ( [ bc : axis]),([bc : dir ] ) , we define the solution at the boundary points to vanish @xmath323 while if working with boundary condition ( [ bc : axis]),([bc : som ] ) we evolve the boundary points by integrating the boundary condition using explicit euler scheme . for example for the boundary @xmath324 @xmath325 where @xmath326 is a fourth - order accurate approximation of the normal first derivative at the border @xmath327 we now extend the solution to the ghost points as @xmath328 finally , we compute @xmath329 as explained above . with @xmath330
we evolve the solution with the two step method @xmath331 for @xmath332 then we impose the boundary conditions exactly as done in the first step . finally we compute @xmath333 as explained above .
we notice that the second derivative in time is approximated by @xmath318 which is second - order accurate .
the time step we use in all our runs is @xmath334 .
the ratio @xmath335 satisfies the courant condition and we see from our runs that the whole method turns out to be numerically stable . besides the solution @xmath336 and
@xmath337 an essential quantity we want to compute is the mass @xmath338 defined by ( [ eq:100 ] ) and ( [ eq : mass_omega ] ) , during the whole evolution . to this end
we need to compute @xmath339 on the physical domain at all times . given the approximations of @xmath287 and @xmath340 we compute @xmath339 by integrating ( [ eq:190])rewritten in terms of @xmath341 as an ode at each gridpoint .
the initial data for these odes is computed by solving the elliptic eq .
( [ eq:62 ] ) , also rewritten in terms of @xmath341 only once at initial time with homogeneous dirichlet boundary conditions and using the same technique we use to compute @xmath342 the first time step to integrate ( [ eq:190 ] ) is carried out with explicit euler method , and from there on with the two - step , second - order accurate , leap - frog method .
we evaluate the integral in ( [ eq : mass_omega ] ) with the midpoint rule .
the numerical calculations we carry out in this work pursue two main objectives .
the first objective is to make plausible that the initial - boundary value problem for ( [ eq : hyp]),([eq : ell ] ) is well - posed . if we were simulating an ibvp
that is not well - posed , the expectation would be that almost any consistent numerical simulation of the problem would fail to pass convergence tests , numerical stability tests , or both .
we show below that both kind of numerical tests are passed satisfactorily by our numerical approximation .
the second objective is to study the behavior of the mass in these initial - boundary value problems .
in particular , we will show that for fixed initial data , the larger the domain used in our calculation is , the longer and better the mass approaches a constant value .
we use in our runs two kinds of initial data which are smooth and strongly decaying outside a small region ( gaussian functions ) .
the first is @xmath343 which decays very fast as @xmath344 get away from @xmath345 and so approximate very well a compact support data on the domains we use .
the second is the same kind of function but for @xmath346 instead of @xmath347 namely @xmath348 linearity of the problem tells us that the runs with @xmath349 or @xmath350 can be performed separately .
a solution with general initial data is the superposition of two solutions , one with each kind of data .
[ [ elliptic - solver - tolerance . ] ] elliptic solver tolerance .
+ + + + + + + + + + + + + + + + + + + + + + + + + + we need to determine the value of @xmath351 to use in our runs . to this end
we perform runs for six different values of @xmath351 with all other parameters fixed to typical values in our runs . in these tests
runs we use initial data given by ( [ typical_idata ] ) and sommerfeld boundary conditions ( [ bc : axis]),([bc : som ] ) .
we then analyze the different values of the mass obtained for the six solutions . by comparing the variations of @xmath352 with respect to the initial value of @xmath353 we see from our runs
show that the evolution is not very sensitive to the tolerance @xmath313 fig .
[ fig : epsilon_full ] shows that the plot for the different computed masses superimpose when plotted in the full mass scale .
the detail in the figure shows convergence of @xmath354 as @xmath355 the upper plot shows the six runs in full mass scale . in this scale
the six curves look superimposed .
the lower plot shows a detail of the initial `` flat '' region in amplified scale.,title="fig:",width=302 ] + the upper plot shows the six runs in full mass scale . in this scale
the six curves look superimposed .
the lower plot shows a detail of the initial `` flat '' region in amplified scale.,title="fig:",width=302 ] in table [ table : epsilon ] we show the maximum absolute difference between the computed masses with respect to the most accurate one ( corresponding to @xmath356 ) and the time of occurrence .
.values of @xmath357 and time of occurrence for different values of the tolerance @xmath313 initial mass is @xmath358 [ cols="^,^,^,^",options="header " , ] [ [ stability - tests . ] ] stability tests .
+ + + + + + + + + + + + + + + + numerical stability means that the solution to the ibvp stays bounded during time evolution .
typical signs of instability are the appearance of artifacts in the plot of the solution as a consequence of evolution and in most cases , after a while , the complete break - down of the solution .
if an instability has its root on the ill - posedness of the analytic problem underneath , the expectation is that some high frequency modes of the solution explode exponentially fast and are detected at very short times of the numerical evolution .
for some more benign ill - posed problems ( like weakly - hyperbolic problems ) the growing of instabilities is only polynomial and it may take longer to detect them .
we performed several series of runs using both kinds of boundary conditions ( [ bc : axis]),([bc : dir ] ) or ( [ bc : axis]),([bc : som ] ) and both kinds of initial data ( [ typical_idata ] ) or ( [ typical_idata_2 ] ) on different domains and during several time intervals .
we studied the plots of the solutions in all cases and they always look smooth , agreement with the boundary conditions imposed and never showed any sort of strange artifact .
typical plots for @xmath359 are shown in fig .
[ fig : sol ] . both plots of solutions computed on a grid with @xmath360 gridpoints and initial data given by ( [ typical_idata ] ) .
upper plot is the solution with homogeneous dirichlet boundary conditions at time @xmath361 while lower plot is the solution with sommerfeld boundary conditions at time @xmath362,title="fig:",width=302 ] + both plots of solutions computed on a grid with @xmath360 gridpoints and initial data given by ( [ typical_idata ] ) .
upper plot is the solution with homogeneous dirichlet boundary conditions at time @xmath361 while lower plot is the solution with sommerfeld boundary conditions at time @xmath362,title="fig:",width=302 ] we have also studied the plots of @xmath363 in these runs and no sign of instability showed .
a second , physically meaningful , test for stability is provided by the study of the mass @xmath354 which in this problem is a sort of incomplete @xmath364 sobolev norm of the solution . as explained in sec .
[ sec : prop - line - equat ] the mass is conserved for the cauchy problem in the whole space . on bounded domains this is no longer true , but we expect that it stays bounded when using homogeneous dirichlet boundary conditions , and that it goes to zero when using sommerfeld boundary conditions .
we analyze the behavior of the mass below .
[ [ behavior - of - the - mass . ] ] behavior of the mass .
+ + + + + + + + + + + + + + + + + + + + + as explained before the mass , defined by ( [ eq:100 ] ) and ( [ eq : mass_omega ] ) , is a conserved quantity when the cauchy problem is considered in the whole space ( i.e. , @xmath214 is @xmath365 ) . in our numerical tests
we solve the initial boundary value problem on compact domains where no known boundary conditions imply mass conservation .
however , an interesting study for the mass evolution can be done as follows .
we solve the ibvp on domains of different size but use , in all runs , the same initial data , at the same distance from the symmetry axis .
the initial data are chosen to decay exponentially fast outside a region which is small compared to the smallest of the domains we use . clearly , the expectation is that the larger the domain is the closest to constant the mass stays during evolution .
we do series of runs for homogeneous dirichlet boundary conditions and for sommerfeld ( outgoing waves ) boundary conditions .
the plots for the dirichlet case are shown in fig .
[ fig : dir_domains ] .
observe that the plot is not on full mass scale .
the three curves show an almost constant initial region and then variations of small relative amplitude . after an initial peak immediately after the constant region the amplitude of the variations is , roughly speaking , @xmath366 for the @xmath367 domain , @xmath368 for the @xmath369 domain and @xmath370 for the @xmath371 domain .
the amplitude clearly diminishes when the domain increases size . as a function of time for three solutions with homogeneous dirichlet boundary conditions and the same initial data but on domains of different size . in the upper right corner
the each curve is associated to the corresponding domain.,width=302 ] for the case of sommerfeld boundary conditions , the plots of the mass evolution can be seen in fig .
[ fig : som_domains ] .
this series of three runs is totally analogous to the previous case .
the only change is the boundary condition used . as a function of time for three solutions with sommerfeld boundary condition , the same initial data but on domains of different size . in the upper right corner
the each curve is associated to the corresponding domain .
the lower plot shows in amplified scale that the `` flat '' region presents very small variations of around @xmath372,title="fig:",width=302 ] as a function of time for three solutions with sommerfeld boundary condition , the same initial data but on domains of different size . in the upper right corner
the each curve is associated to the corresponding domain .
the lower plot shows in amplified scale that the `` flat '' region presents very small variations of around @xmath372,title="fig:",width=302 ] as can be inferred from the plot in full mass scale , the energy leaks though the boundary as expected .
in this article we have deduced the linear system and and we have analyzed some of its properties . among them ,
the most relevant are the mass conservation and the numerical stability .
the main open problem is to prove that this system is well - posed .
remarkable enought , it seems to be not much literature on this class of linear systems which are singular at the axis .
once the well - posedness problem is solved , we believe that the future research on the subject can be divided in two paralel but complementary roads . the first one is to extend the well - posedness from the linear system to the full einstein equations in the maximal - isothermal gauge .
the non - linear lower order terms introduce extra difficulties ( see @xcite ) .
there are many possible evolutions schemes ( see the discussion in @xcite ) .
it is very likely that few of them ( or may be only one ) are well - posed .
if this is the case , the resolution of the well - posedness question will lead us to select ( or even discover ) the correct evolution scheme .
after the local problem is solved , the next step is to use the global conservation of the mass to control the full non - linear evolution in this gauge .
a natural first example would be to recover the non - linear stability of minkowski @xcite in this gauge .
the expectation is that the mass formula will provide a simpler ( and different ) kind of approach to this problem ; although , of course , always resticted to axial symmetry .
the ultimate and difficult goal is to say something , in this gauge , about the non - linear stability of a black hole in axial symmetry .
the second road is the study axially symmetric perturbation but with a black hole as background solution .
linear stability of the kerr black hole is a relevant open problem which is currently intensively studied ( see the review articles @xcite , @xcite and references therein ) .
the expectation is that the mass formula can help to prove linear stability under axially symmetric perturbation of the kerr black hole .
s. d. thanks piotr chruciel and helmut friedrich for useful discussions .
these discussions took place at the mathematisches forschungsinstitut oberwolfach during the workshop `` mathematical aspects of general relativity '' , october 11th october 17th , 2009 .
s. d. thanks andrs acea for useful discussions that took place at the max - planck - institut fr gravitationsphysik ( albert - einstein - institut ) during the conference `` space , time and beyond '' , october 19th
october 21th , 2009 .
s. d. thanks the organizers of these events for the invitation and the hospitality and support of the mathematisches forschungsinstitut oberwolfach and the max - planck - institut fr gravitationsphysik ( albert - einstein - institut ) .
s. d. is supported by conicet ( argentina ) .
this work was supported in part by grant pip 6354/05 of conicet ( argentina ) , grant 05/b415 secyt - unc ( argentina ) and the partner group grant of the max planck institute for gravitational physics , albert - einstein - institute ( germany ) .
we collect in this appendix some useful formula that are used in the main part of this article .
the conformal killing operator in 2-dimensions with respect to the metric @xmath92 is defined by @xmath373 for the particular case of a flat metric @xmath116 this definition reduce to @xmath374 for this operator we have the following identity often used in the article @xmath375 under the conformal rescaling the diferential operators relevant in this article transform as follows @xmath378 where we have defined @xmath379 we follow the convention that the indices for hat quantities are moved with the flat metric @xmath116 and inidices of non - hat quantities with the metric @xmath92 .
then , we have @xmath380 that is why we suppress the hat notation for the tensors @xmath381 and @xmath382 in the main part of this article . | for axially symmetric solutions of einstein equations there exists a gauge which has the remarkable property that the total mass can be written as a conserved , positive definite , integral on the spacelike slices .
the mass integral provides a nonlinear control of the variables along the whole evolution . in this gauge ,
einstein equations reduce to a coupled hyperbolic - elliptic system which is formally singular at the axis . as a first step in analyzing this system of equations we study linear perturbations on flat background .
we prove that the linear equations reduce to a very simple system of equations which provide , thought the mass formula , useful insight into the structure of the full system .
however , the singular behavior of the coefficients at the axis makes the study of this linear system difficult from the analytical point of view . in order to understand the behavior of the solutions , we study the numerical evolution of them . we provide strong numerical evidence that the system is well - posed and that its solutions have the expected behavior .
finally , this linear system allows us to formulate a model problem which is physically interesting by itself , since it is connected with the linear stability of black holes solutions in axial symmetry .
this model can contribute significantly to solve the nonlinear problem and at the same time it appears to be tractable . |
tools for enclosing the image set of factorable functions are the basis for many reliable numerical computing methods including global optimization algorithms @xcite , robust and semi - infinite optimization algorithms @xcite , as well as validated integration algorithms @xcite . here , factorable functions @xcite are functions that can be represented as a finite recursive composition of atom operations from a ( finite ) library @xmath0 this library typically includes binary sums , binary products , and a number of univariate atom functions such as trigonometric functions , exponential functions , or logarithms . in practice ,
factorable functions over a given library @xmath1 are represented in the form of a computational graph , which can be obtained conveniently in most object oriented programming languages by using operator overloading or source code transformation @xcite .
existing methods for computing enclosures of factorable functions can be divided into three categories : traditional interval arithmetics and its variants , arithmetics using other convex sets such as ellipsoids or zonotopes , as well as non - convex set arithmetics . in the following , advantages and limitations of these existing methods are reviewed .
[ [ interval - arithmetic . ] ] interval arithmetic .
+ + + + + + + + + + + + + + + + + + + + interval arithmetic is one of the oldest and most basic tools for computing enclosures of the image set of a factorable function on a given compact domain @xcite . throughout this paper
, we use the notation @xmath2 \subseteq r \ , \mid \ , a , b \in \mathbb r , \ , a \ , \leq \ , b \ , \right\}\ ] ] to denote real valued intervals .
interval arithmetic proceeds by defining bounding rules for binary sums and products as well as for all univariate atom operations in the given library @xmath1 .
for example , for two intervals @xmath3 \in \mathbb i$ ] and @xmath4 \in \mathbb i$ ] , their sum , product , and exponential are given by @xmath5 + [ c , d ] & = & [ a+b , c+d ] \ ; , \\[0.16 cm ] \label{eq::intervalproduct } [ a , b]*[c , d ] & = & \left [ \ , \min \ { ac , ad , bc , bd \ } , \ , \max \ { ac , ad , bc , bd \ } \
, \right ] \ ; , \\[0.16 cm ] \text{and } \quad e^{[a , b ] } & = & \left [ e^a , e^b \right ] \ ; .\end{aligned}\ ] ] the derivation of other univariate composition rules is straightforward for most commonly used univariate atom operations @xcite . additionally , in order to ensure that an interval arithmetic is compatible with the addition and multiplication with scalars @xmath6 , the notation @xmath7 with @xmath8 $ ] is used to denote the shifted interval @xmath9 $ ] .
similarly , @xmath10 denotes the scaled interval @xmath11 $ ] if @xmath12 and @xmath13 $ ] if @xmath14 .
unfortunately , one of the main limitations of interval arithmetics is that the computed interval enclosures are often much wider than the exact range of the given factorable function .
this overestimation effect is mainly caused by the so called _
dependency problem_. the dependency problem can already be observed for very simple functions , such as @xmath15 , @xmath16 $ ] . here , the exact image set is @xmath17 $ ] , but a naive application of standard interval arithmetics gives the interval bound @xmath18 $ ] . on the other hand ,
an advantage of interval arithmetics is its favorable computational complexity : storing an interval in floating point format amounts to storing two floating point numbers of the same format . in terms of computation time ,
evaluating a product of intervals is about @xmath19-@xmath20 times more expensive than evaluating a product of real numbers .
the above rules for the binary sum and evaluation of the exponential functions are only @xmath21-times more expensive than a corresponding function evaluation with real numbers .
another advantage of interval arithmetics is that the computational and storage complexity of its multivariate extension scales linearly with the number of variables of the factorable function .
this is due to the fact that the complexity of storing an @xmath22-dimensional interval vector @xmath23 is of order @xmath24 . [
[ convex - set - arithmetics . ] ] convex set arithmetics .
+ + + + + + + + + + + + + + + + + + + + + + + during the last decades , there have been many attempts to overcome the limitations of interval arithmetics by using more advanced set propagation techniques . here
, one way to generalize interval arithmetics is to replace the intervals ( or interval vectors ) with more general computer representable convex sets .
for example , relaxations propagate convex lower and concave upper bounds rather than standard intervals @xcite .
arithmetic has the same computational complexity as interval arithmetics with respect to the number of variables @xmath22 , namely @xmath24 . in practice
, it sometimes yields tighter bounds , but it is also slightly more expensive than interval arithmetics despite the fact that it has the same complexity , as illustrated in @xcite .
another class of convex set based enclosure tools is the so - called ellipsoidal calculus @xcite , where multi - dimensional ellipsoids rather than interval vectors are used in order to represent the set enclosures . in terms of storage requirements ,
@xmath22-dimensional ellipsoids can be represented by a vector in @xmath25 , representing the center , and a symmetric and positive semi - definite shape matrix in @xmath26 .
thus , the complexity of storing ellipsoids is of order @xmath27 , while intervals can be stored with complexity @xmath24 .
unfortunately , the computational complexity of ellipsoidal calculus is typically of order @xmath28 , if dense matrix - matrix multiplications are not avoided . concerning applications in the field of validated integration and control ,
ellipsoidal calculus has been found to be a suitable tool for mitigating the inherent dependency problem of set arithmetics @xcite .
for example ,
in contrast to interval arithmetics , ellipsoid arithmetics can be used to construct stable validated integration algorithms @xcite . in this sense
, it has been established that , at least for particular applications , the higher computational effort of storing ellipsoids rather than intervals pays out in terms of the accuracy of the enclosure set .
other convex enclosure methods use polyhedral sets , which are in general even more expensive to store than ellipsoids , but they can be used to represent convex sets with arbitrary precisions by controlling the number of facets .
polyhedral relaxations have become popular in the field of global optimization and are for example used in the software ` baron ` @xcite , which is based on linear programming relaxations .
another example for an enclosure algorithms based on polyhedral sets is the so - called `` affine arithmetics '' @xcite .
affine arithmetic is based on zonotopes , a particular class of point - symmetric polytopes .
[ [ non - convex - set - arithmetics . ] ] non - convex set arithmetics .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + a rather apparent disadvantage of all convex set based arithmetics is that they can , in the best case , represent the convex hull of the image of a given factorable function .
consequently , if the exact image set of a factorable function is non - convex , the benefit of investing into more accurate convex set representations , such as zonotopes or even general polytopes with many facets , is limited .
one way to overcome this limitation is by working with non - convex sets , which are in practice often represented by using polynomials .
the use of interval polynomials or polynomials with interval remainder terms have been developed since a long time starting in the 1960s @xcite and 1980s @xcite .
these early works have been the basis for the popular taylor model arithmetic , which has been developed by berz and coworkers @xcite .
nowadays , there exist mature tools , for example the software ` mc++ ` @xcite , implementing taylor model arithmetics with arbitrary order .
the favorable convergence properties of taylor models on variable domains with small diameter have been analyzed thoroughly @xcite .
however , the convergence properties of taylor series on wider domains are often less favorable .
for example , the taylor series of the function @xmath29 at @xmath30 is divergent outside the interval @xmath18 $ ] , as the analytic continuation of this function has poles at @xmath31 , @xmath32 .
a promising direction towards overcoming this limitation of taylor models is the ongoing research on so - called chebychev models . for functions with one or two variables
chebychev models can be constructed by the software ` chebfun ` as developed by trefethen and coworkers @xcite .
chebychev models for functions with more than two variables are the focus of recent research @xcite . while computing bounds on convex sets is computationally tractable , finding tight bounds of a multivariate polynomial is itself a complex task . here , one way to compute bounds on such polynomials is to use linear matrix inequalities @xcite .
other heuristics for computing range bounders for multivariate polynomials can be found in @xcite .
a principal goal of this paper is to develop a non - convex set arithmetic for factorable function that exploits _ global algebraic structures _ rather than attempting to borrow methods from local and numerical analysis such as variational analysis , taylor expansions , or other polynomial approximation techniques .
the following paragraphs outline two global algebraic structures that are exploited by the proposed interval superposition arithmetic , namely , _ addition theorems _ @xcite and _ additive separability _ @xcite .
[ [ addition - theorems . ] ] addition theorems .
+ + + + + + + + + + + + + + + + + + a rational addition theorem for a univariate atom operation @xmath33 with domain @xmath34 is a formula that expresses @xmath35 as a rational function of @xmath36 and @xmath37 , @xmath38 with @xmath39 being a bivariate rational function .
an important example is the addition theorem for the exponential function , @xmath40 , which holds globally on the domain @xmath41 .
notice that the right - hand side of the addition theorem for the sine function , @xmath42 , is an exception in the sense that its right - hand side depends not only @xmath43 and @xmath44 but also @xmath45 and @xmath46 .
however , if we allow complex arguments , this addition theorem can alternatively be regarded as a special case of the addition theorem for the exponential function such that it can be written in the form of .
a similar statement holds for the cosine function .
moreover , in analogy to the univariate case , a rational addition theorem for a bivariate atom operation @xmath33 with domain @xmath47 , @xmath48 is a formula of the form @xmath49 for all @xmath50 with @xmath51 and all @xmath52 with @xmath53 . here , @xmath54 is a rational function with @xmath19 arguments .
an incomplete list of examples for addition theorems , which are either in this form or in generalized versions relevant for the derivations in this paper , are collected in table [ tab::additiontheorems ] .
.a list of rational addition theorems for common atom operations .
the addition theorems for the sine and cosine functions can alternatively be regarded as special cases of the addition theorem for the exponential function , if complex arguments are allowed.[tab::additiontheorems ] [ cols="^,^,^ " , ] the goal of this section is to compare the performance of first order interval superposition models versus taylor models on wider domains .
let @xmath55 denote a non - convex factorable function of the form @xmath56 on the two - dimensional domain @xmath57 \times [ 0 , \overline x_2 ] \subseteq \mathbb r^2 $ ] . here , @xmath58 and @xmath59 are parameters that can be used to control the diameter of the domain @xmath60 . the upper left plot in figure [ fig::result1 ] shows a @xmath61-dimensional visualization of the function @xmath62 on the interval domain @xmath63 \times [ 0,20]$ ] , i.e. , for @xmath64 and @xmath65 .
the upper right plot in figure [ fig::result1 ] shows the overestimation of five different enclosure methods for bounding @xmath62 for @xmath66 as a function of the domain parameter @xmath67 $ ] : the red solid and red dotted lines show the overestimation of taylor models of order @xmath68 and @xmath21 , respectively .
the black solid , black dotted , and black dashed lines correspond to the overestimation of the enclosures that are obtained by using first order interval superposition models with @xmath69 , @xmath70 , and @xmath71 . here , the overestimation is measured in terms of the hausdorff distance between the exact image set and the five different enclosure sets .
the lower left plot in figure [ fig::result1 ] shows the overestimation of the five mentioned methods for a fixed @xmath72 as a function of @xmath67 $ ] .
similarly , the lower right plot in figure [ fig::result1 ] depicts the corresponding results for @xmath64 , again as a function of @xmath67 $ ] . here , the results for the taylor models is not shown , as the overestimation error is larger that @xmath73 , i.e. , taylor models do not yield reasonable enclosures on this rather large domain . in order to avoid misunderstanding at this point , notice that the width of the exact image set @xmath74 is monotonically increasing with respect to the parameter @xmath75 . however , the hausdorff difference between @xmath74 and an enclosure @xmath76 is not necessarily monotonous in @xmath75 .
in fact , also the overestimation of standard taylor models decreases in sections , if the domain @xmath60 is increased , although one might argue that , overall , a rough trend is that the overestimation error of the enclosure methods increases when increasing the domain @xmath60 .
one aspect that is not shown in the figure [ fig::result1 ] is that taylor models do outperform interval superposition models on very small domains , i.e. , if we would zoom in and analyze the overestimation for @xmath77 , we could see that taylor models are the better choice on such small domains .
however , as mentioned in previous section , the local convergence properties of the proposed superposition could be improved by constructing , e.g. , superpositions of first order taylor models rather than intervals , but this is not implemented here as the motivation of this paper is to develop a tool that yields reasonable bounds on wider domains . notice that on the domain @xmath63 \times [ 0,20]$ ] the overestimation of the first order interval superposition method with @xmath71 yields an enclosure that is approximately @xmath78 times larger than the width of the exact range , i.e. , the relative over - approximation is approximately @xmath79 .
this is in contrast to taylor models , which yield bounds that are more than @xmath73 times larger than the exact image set .
the performance of taylor models of order larger than @xmath21 is not shown in the figure , as they perform even worse than the taylor models of order @xmath68 and @xmath21 on the analyzed , particularly large domains .
here again , of course , if we would zoom in on smaller domains @xmath60 , we could see that increasing the taylor model order does improve the accuracy for such smaller @xmath60 , but this effect is well - known @xcite and therefore not analyzed further at this point .
moreover , as the above example considers a function with two variables , second or other higher order interval superposition models are equivalent to exhaustive branching , i.e. , this example is not suited for making a fair comparison of such higher order methods .
therefore , we refer to section [ sec::recursion ] , where first and second order interval superposition models are analyzed for a more challenging case study . in order to illustrate how the proposed interval superposition arithmetics performs for a more challenging example ,
we introduce the function @xmath80 \frac{1}{4 } x_2 + \frac{1}{4 } x_2 x_3 \\[0.16 cm ] -x_3 ^ 2 + x_3 ^ 4 + \frac{4}{10 } x_1 x_3 + \frac{1}{10 } x_2 ^ 2 \end{array } \right ) \ ; . \label{eq::f1}\end{aligned}\ ] ] notice that @xmath81 is a multivariate polynomial of order @xmath19 . in the next step
we define the functions @xmath82 recursively . of course , the function @xmath83 are `` simple '' polynomial functions in principle , but the actual challenge is that the order of these polynomials grows exponentially with @xmath84 .
more precisely , @xmath83 is a polynomial function of order @xmath85 .
the goal of this section is to find enclosure sets of the exact image sets @xmath86 on the rather large interval domain @xmath87 ^ 3 \subseteq \mathbb r^3 $ ] .
it can be checked easily that exact image sets satisfy a convergence rate condition of the form @xmath88
i.e. , the exact image set contracts to @xmath89 with linear convergence rate , @xmath90 for sufficiently large @xmath84 . the hausdorff distance between the exact image sets @xmath86 and their computed enclosure sets @xmath91 in dependence on the running index @xmath84 .
the corresponding results for standard interval arithmetic are labeled as `` ia '' .
the other results are obtained by using interval superposition models of order @xmath68 and @xmath21 with @xmath71 labeled as `` isa1 '' and `` isa2 '' , respectively . ]
figure [ fig::recursion ] shows the hausdorff distance between the exact image set @xmath86 and the enclosure sets that are obtained by applying standard interval arithmetic , first order interval superposition arithmetic with @xmath92 , as well as second order interval superposition arithmetic also with @xmath92 .
all results are shown in dependence on @xmath84 .
notice that the first order interval superposition models yield enclosure sets that are much less conservative than the enclosures that are obtained by standard interval arithmetic , but both enclosure methods are divergent for @xmath93 .
this is in contrast to the second order interval superposition arithmetic , which yields convergent enclosures for @xmath93 .
the results for first and second order taylor models are not shown in the figure , as they perform even worse than standard interval arithmetic for this particular example .
if we use a taylor model of order @xmath19 , the remainder interval is equal to @xmath89 when bounding the @xmath19-th order polynomial @xmath81 .
however , the enclosure set that is obtained by applying a @xmath19-th order taylor model for bounding the function @xmath94 is again worse than the corresponding enclosure obtained by standard interval calculus .
the authors also tried to use a taylor model of order @xmath95 , which does lead to a remainder interval of width @xmath89 when applied to the functions @xmath81 and @xmath94 recalling that @xmath94 is a multivariate polynomial of order @xmath95 .
however , if we apply a taylor of order @xmath95 for bounding the function @xmath96 the overestimation error is again larger than @xmath97 , which corresponds to about the same accuracy that is obtained by applying the much cheaper standard interval calculus .
thus , in summary , taylor models fail to yield reasonable bounds for this example on the particularly wide domain @xmath87 ^ 3 $ ] .
of course , this conclusion changes if the functions @xmath83 are bounded on smaller interval domains , where taylor models tend to yield more accurate enclosures . because the exact image sets converge linearly and because theorem [ thm::higherorderlocal ] predicts that the overestimation error of second order interval superpositions converge has the form @xmath98^{3 } \right ) } \ ; , \ ] ] with respect to the diameter of the input set @xmath99 , one would expect that the overestimation error converges cubically for larger input domains @xmath99 and then linearly , once term the linearly convergent term becomes dominant .
the convergence plot of the second order interval superposition models in figure [ fig::recursion ] indicates indeed such a behavior : for the range @xmath100 the hausdorff distance between the second order interval superposition enclosure and the exact set converges superlinearly , although one might argue that it is hard to tell from these numerical results whether this range corresponds to a cubic convergence phase .
however , for @xmath101 the convergence rate switches and becomes linear , since in this phase the terms with linear convergence order , @xmath102 , dominate recalling that these terms did not dominate early because we chose a rather large number of branches , @xmath92 , in this example .
consequently , the numerical convergence behavior of the second order interval superposition models for @xmath103 can be explained by the theoretical analysis results from theorem [ thm::higherorderlocal ] .
this paper has introduced an interval superposition arithmetic and illustrated the advantages of this new arithmetic compared to existing enclosure methods for factorable functions on wider domains .
the construction of interval superposition models is based on derivative - free composition rules which exploit global algebraic properties such as rational addition theorems , inverse rational addition theorems , and partially separable sub - structures of the computational graph of factorable functions .
the corresponding technical developments are based on dependency expansions , which can be viewed as a derivative - free , algebraic abstraction of taylor expansions .
interval superposition arithmetic has polynomial run - time and storage complexity of order @xmath104 , which depends on the number @xmath22 of variables of the factorable functions and the branching accuracy @xmath105 .
the order @xmath106 of this polynomial run - time bound is equal to the order of the dependency expansion .
moreover , this paper has established local and global convergence estimates of the proposed arithmetic in dependence on the order @xmath106 of the interval superposition and in dependence on the coupling degree of the given factorable function . from a practical perspective
, the main advantage of interval superposition arithmetics compared to other enclosure methods is that it yields reasonably accurate bounds of the image set of factorable functions on wider interval domains , for which existing methods often yield divergent or very conservative bounds .
this advantage has been illustrated by analyzing two numerical case studies .
many algorithms of current interest such as semi - infinite programming algorithms , validated integrators , as well as branch & bound methods for global optimization rely on the availability of set arithmetics for factorable functions . as interval superposition models have major advantages compared to existing set computation methods on wider domains , this novel arithmetic should be of practical relevance to everyone working on or using such algorithms . moreover , since interval superposition arithmetic is , on the one hand , a bounding tool , but , on the other hand , exploits the concept of coordinate aligned branching while enforcing polynomial run - time and storage requirements , this arithmetic might lead to new types of global optimization algorithms , where the branching and bounding operations are not considered as separate routines anymore . however , a deeper analysis of the corresponding interplay between the presence of global algebraic structures in factorable functions and the complexity of their associated optimization problems is beyond the scope of this paper and shall be part of future research
. 99 y. abe .
a statement of weierstrass on meromorphic functions which admit an algebraic addition theorem . j. math .
japan 57(3):709723 , 2005 . s.a .
algebraic addition theorems . adv .
math . 13:2030 , 1974 .
z. battles , l.n .
an extension of matlab to continuous functions and operators .
siam j. sci .
comput . 25:17431770 , 2004 .
m. berz . from taylor
series to taylor models . in nonlinear problems in accelerator physics , american institute of physics cp405 , pp.:127 , 1997 .
m. berz , g. hoffsttter . computation and application of taylor polynomials with remainder bounds .
comput . 4:8397 , 1998 .
a. bompadre , a. mitsos , b. chachuat .
convergence analysis of taylor and mccormick - taylor models .
journal of global optimization 57(1):75114 , 2013 .
b. buchberger , r. loos . algebraic simplification . in : computer algebra - symbolic and algebraic computation .
b. buchberger , g. e. collins , r. loos ( eds . ) , pp . 1143 , 1982 .
b. chachuat , b. houska , r. paulen , n. peric , j. rajyaguru , m.e .
set theoretic approaches in analysis , estimation and control of nonlinear systems .
ifac - papersonline volume 48(8 ) , pp:981995 , 2015 .
conn , n.i.m .
gould , ph.l .
an introduction to the structure of large scale nonlinear optimization problems and the lancelot project . in : glowinski , r. ,
lichnewsky , a. ( eds . ) computing methods in applied sciences and engineering , pp .
4251 . siam , philadelphia , 1990 .
conn , a.r .
, gould , n.i.m . ,
toint , ph.l . : improving the decomposition of partially separable functions in the context of large - scale optimization : a first approach . in : w.w .
hager , d.w .
hearn , p.m. pardalos ( eds . ) .
large scale optimization : state of the art , pp . 8294 .
kluwer academic publishers , amsterdam , 1994 .
du , r.b .
the cluster problem in multivariate global optimization .
journal of global optimization 5(3):253265 , 1994 .
eckmann , h. koch , p. wittwer . a computer - assisted proof of universality in area - preserving maps .
memoirs of the ams 47:289 , 1984 .
de figueiredo , j. stolfi .
affine arithmetic : concepts and applications .
numerical algorithms 37(1 - 4):147158 , 2004 .
floudas and o. stein .
the adaptative convexification algorithm : a feasible point method for semi - infinite programming .
siam journal on optimization , 18(4):11871208 , 2007 .
c.a . floudas .
deterministic global optimization : theory , methods and applications .
springer science & business media , vol .
37 , 2013 .
a. griewank and a. walther . .
siam , 2008 .
goemans and d.p .
improved approximation algorithms for maximum cut and satisfiability problems using semidefinite programming .
, 42:11151145 , 1995 .
d. henrion , s. tarbouriech , and d. arzelier .
pproximations for the radius of the intersection of ellipsoids : a survey .
, 108(1):128 , 2001 .
b. houska , f. logist , j. van impe , m. diehl .
robust optimization of nonlinear dynamic systems with application to a jacketed tubular reactor .
journal of process control , volume 22(6 ) , pp . 11521160 , 2012 .
b. houska , m.e .
villanueva , b. chachuat .
stable set - valued integration of nonlinear dynamic systems using affine set parameterizations .
siam journal on numerical analysis , 53(5 ) , pp:23072328 , 2015 .
a. kurzhanski , i. valyi .
ellipsoidal calculus for estimation and control .
series in systems & control : foundations & applications , birkhuser , 1997 .
moments , positive polynomials and their applications .
imperial college press , 2009 .
q. lin , j.g .
methods for bounding the range of a polynomial .
j. comput appl math 58:193199 , 1995 .
m. neher , k.r .
jackson , n.s .
on taylor model based integration of odes .
siam journal on numerical analysis 45:236262 , 2007 .
a. nemirovski , c. roos , and t. terlaky . on maximization of quadratic form over intersection of ellipsoids with common center .
, 86(3):463473 , 1999 .
y. nesterov .
semidefinite relaxation and non - convex quadratic optimization . , 12:120 , 1997 .
a. neumaier .
complete search in continuous global optimization and constraint satisfaction .
acta numer . 13:271369 , 2004 .
k. makino , m. berz .
efficient control of the dependency problem based on taylor model methods .
comput . 5(1):312 , 1999 .
computability of global solutions to factorable nonconvex programs : part i convex underestimating problems .
mathematical programing 10:147175 , 1976 .
r. misener , c.a .
glomiqo : global mixed - integer quadratic optimizer .
journal of global optimization , 57(1):350 , 2013 .
a. mitsos , p. lemonidis , p.i .
global solution of bilevel programs with a nonconvex inner program .
journal of global optimization 42.4:475513 , 2008 .
a. mitsos , b. chachuat , p.i .
mccormick - based relaxations of algorithms .
siam journal on optimization 20(2):573601 , 2009 .
moore . interval analysis .
prentice - hall , englewood cliffs , nj , 2966 .
moore , r.b .
kearfott , m.j .
introduction to interval analysis .
siam , philadelphia , pa , 2009 .
p.m. pardalos , s.a .
quadratic programming with one negative eigenvalue is np - hard .
journal of global optimization 1 ( 1 ) : 1522 , 1991 .
j. rajyaguru , m.e .
villanueva , b. houksa , b. chachuat .
higher - order inclusions of factorable functions by chebyshev models .
journal of global optimization , 2016 .
( submitted ) h. ratschek , j. rokne .
computer methods for the range of functions .
series in mathematics and its applications , ellis horwood ltd , mathematics and its applications , chichester , uk , 1984 .
a general purpose global optimization software package .
journal of global optimization , 8(2):201205 , 1996 .
a.m. sahlodin , b. chachuat .
convex / concave relaxations of parametric odes using taylor models .
computers and chemical engineering 35(5):844857 , 2011 .
m. tawarmalani , n.v .
sahinidis . a polyhedral branch - and - cut approach to global optimization .
mathematical programming , 103(2):225249 , 2005 .
computing numerically with functions instead of numbers .
sci . 1:919 , 2007 .
a. townsend , l.n .
an extension of chebfun to two dimensions .
siam j. sci .
comput 35(6):c495c498 , 2013 .
villanueva , j. rajyaguru , b. houska , b. chachuat .
ellipsoidal arithmetic for multivariate systems .
comput . aided chem .
eng . 37:767772 , 2015 .
villanueva , b. houska , b. chachuat .
unified framework for the propagation of continuous - time enclosures for parametric nonlinear odes .
j. of global optim 62(3 ) , pp:575613 , 2015 .
villarino . algebraic addition theorems . retrieved february 12 , 2016 , from the arxiv database , arxiv:1212.6471 , 2013 .
a. wechsung , s.d .
schaber , p.i .
the cluster problem revisited .
journal of global optimization 58(3):429438 , 2014 .
this section briefly discusses how to derive remainder bounds for the proposed first order interval superposition arithmetic .
these remainder bounds are needed in the composition rule of the first order interval superposition arithmetic ( algorithm 1 ) requiring that the inequality @xmath107 is satisfied for all @xmath108 with @xmath109 . in the sections below
we verify this inequality for all rows of table [ tab::bounds ] except for the remainder bounds for the univariate minus and univariate square in the first two rows in table [ tab::bounds ] , which are easy to check and left as an exercise for the reader . for the atom function @xmath110 we have to bound the expression @xmath111 \notag\end{aligned}\ ] ] for all @xmath112 with @xmath113 .
in the latter equation , we follow the concept of strategy 1 ; that is , we apply the addition theorem for the exponential function , @xmath114 it is convenient to introduce the auxiliary variables @xmath115 such that @xmath116 & = & e^{\omega } \left [ \sum_{i=1}^n t_i + 1 - \prod_{i=1}^n ( 1+t_i ) \right ] \ ; .\end{aligned}\ ] ] the absolute value of this expression can be bounded as @xmath117 this motivates to choose the central points @xmath118 such that @xmath119 takes the smallest possible value , given by @xmath120 in summary , we have shown that @xmath121 this is the remainder bound from that is listed in the fourth row of table [ tab::bounds ] . the aim of this section is to find a remainder bound for the atom function @xmath122 on the positive domain @xmath123 .
this is sufficient , as bounds on the domain @xmath124 can be found analogously .
if an interval contains @xmath89 , the bounds are set to @xmath125 $ ] .
the main idea is to exploit the addition theorem for the inverse function and to apply further standard manipulations for rational functions in order to simplify the result .
we start with the equation @xmath126 & = & \frac{1}{\omega } \left ( \sum_{i=1}^n \frac{-\delta_i}{\omega+\delta_i } + \frac{\sum_{i=1}^n \delta_i}{\omega + \sum_{i=1}^n \delta_{i } } \right ) \notag \\[0.16 cm ] & = & \frac{1}{\omega } \ , \frac{1}{\omega + \sum_{i=1}^n \delta_{i } } \ , \left ( \sum_{i=1}^n \frac { \delta_i ( \delta_i - \sum_{k=1}^n \delta_k)}{\omega+\delta_i } \right ) \ ; . \notag\end{aligned}\ ] ] next , we bound the terms in the last equation separately under the assumption that @xmath127 ( if we do nt have @xmath127 , we can not ensure that @xmath89 is not in the interval ) , @xmath128 \left| \frac{\delta_i}{\omega + \delta_{i } } \right| & \leq & \max \left\ { \ , \frac{a_i - l(a_i ) } { \omega - a_i + l(a_i ) } \ , , \ ,
\frac{u(a_i ) - a_i } { \omega - a_i + u(a_i ) } \ , \right\ } \ ; = \ ; s_i \ ; , \notag \\[0.16 cm ] \text{and } \quad \left| \delta_i - \sum_{k=1}^n \delta_k \right| & \leq & \mu(a ) - \omega - ( u(a_i)-a_i ) \ ; . \notag\end{aligned}\ ] ] substituting these inequalities yields @xmath129 this is the remainder bound that is listed in the third row of table [ tab::bounds ] .
the aim of this section is to find a remainder bound for the atom function @xmath130 on the positive domain @xmath123 .
we start by exploiting the inverse addition theorem for the logarithm , which leads to the equation @xmath131 & = & \log \left ( \frac{\prod_{i=1}^n \left ( \omega+\delta_i \right ) } { \omega^{n-1 } \left ( \omega + \sum_{i=1}^n \delta_{i } \right ) } \right ) \notag \\[0.16 cm ] & = & \log \left ( 1 + \frac{\prod_{i=1}^n \left ( \omega+\delta_i \right ) - \omega^{n-1 } \left ( \omega + \sum_{i=1}^n \delta_{i } \right ) } { \omega^{n-1 } \left ( \omega + \sum_{i=1}^n \delta_{i } \right ) } \right ) \ ; . \notag\end{aligned}\ ] ] next we bound the absolute value of this term by choosing the central points @xmath132 such that @xmath133 and @xmath134 this is the remainder bound that is listed in the fifth row of table [ tab::bounds ] .
the real - valued sine and cosine functions are similar to the exponential function in the sense that an addition theorem is available and can be used for constructing remainder terms . in order to exploit this relation systematically , we use euler s formula , @xmath135 with @xmath32 .
the derivation passes through the following steps . in the first step ,
we derive for all @xmath136 the bound @xmath137 & = & 2 \left| \sin \left ( \pm \frac{\delta_k}{2 } \right ) \right| \notag \\[0.16 cm ] & \leq & 2 \left| \sin \left ( \left [ -\frac{u(a_k)-l(a_k)}{4},\frac{u(a_k)-l(a_k)}{4 } \right ] \right ) \right| = s_k \ ; .\end{aligned}\ ] ] here , the expression for the scalars @xmath138 is evaluated by using standard interval arithmetic , i.e. , we define @xmath139 \right ) \right| = \left\ { \begin{array}{ll } 2 \sin \left ( \frac{u(a_k)-l(a_k)}{4 } \right ) \ ; \ ; & \text{if } \ ; \ ; \frac{u(a_k)-l(a_k)}{4 } \leq \frac{\pi}{2 } \\[0.16 cm ] 2 & \text{otherwise } \end{array } \right\ } \ ; .\ ] ] the auxiliary inequalities from step 2 , in turn , can be used to establish the inequalities @xmath141 & \leq & \prod_{k=1}^n ( 1+s_k ) - \sum_{k=1}^n s_k - 1\end{aligned}\ ] ] as well as @xmath142 using an analogous argument . for the sine function @xmath143 , the estimate from step 3
can be used to find the remainder bound @xmath144 & = & \left| \sum_{k=1}^n \sin \left ( \omega+\delta_k \right ) - \sin \left ( \omega + \sum_{k=1}^n \delta_{k } \right ) - ( n-1 ) \sin \left ( \omega \right ) \right| \notag \\[0.16 cm ] & = & \left| \sin(\omega ) \left ( \sum_{k=1}^n \cos \left ( \delta_k \right ) - \cos \left ( \sum_{k=1}^n \delta_{k } \right ) - ( n-1 ) \right ) + \cos(\omega ) \left ( \sum_{k=1}^n \sin \left ( \delta_k \right ) - \sin \left ( \sum_{k=1}^n \delta_{k } \right ) \right ) \right| \notag \\[0.16 cm ] & \leq & \left ( |\sin(\omega)| + |\cos(\omega)| \right ) \left ( \prod_{k=1}^n ( 1+s_k ) -
\sum_{k=1}^n s_k - 1 \right ) = r_g(a ) \ ; . \notag\end{aligned}\ ] ] the corresponding bound for the atom function @xmath145 is given by @xmath146 & = & \left| \sum_{k=1}^n \cos \left ( \omega+\delta_k \right ) - \cos \left ( \omega + \sum_{k=1}^n \delta_{k } \right ) - ( n-1 ) \cos \left ( \omega \right ) \right| \notag \\[0.16 cm ] & = & \left| \cos(\omega ) \left ( \sum_{k=1}^n \cos \left ( \delta_k \right ) - \cos \left ( \sum_{k=1}^n \delta_{k } \right ) - ( n-1 ) \right ) - \sin(\omega ) \left ( \sum_{k=1}^n \sin \left ( \delta_k \right ) - \sin \left ( \sum_{k=1}^n \delta_{k } \right ) \right ) \right| \notag \\[0.16 cm ] & \leq & \left ( |\sin(\omega)| + |\cos(\omega)| \right ) \left ( \prod_{k=1}^n ( 1+s_k ) - \sum_{k=1}^n s_k - 1 \right ) = r_g(a ) \ ; .
\notag\end{aligned}\ ] ] the above remainder bounds are listed in table [ tab::bounds ] . in order to construct a remainder bound for the function @xmath147 on the open domain
@xmath148 it is helpful to notice that the addition theorem for the tangent function , given by @xmath149 can alternatively be written in the difference form @xmath150 the correctness of this equation can be verified by multiplying the addition theorem for the tangent function by @xmath151 on both sides and by re - bracketing terms .
next , a generalization of the difference formula for general sums is given by the eqaution @xmath152 this equation can be proven by induction .
firstly , is true for @xmath153 , as this case reduces to .
secondly , if is true for a given @xmath22 , we have @xmath154 & \overset{\eqref{eq::tangentdifference}}{= } & \tan \left ( \sum_{i=1}^{n+1 } \delta_i \right ) \tan \left ( \sum_{i=1}^{n } \delta_i \right ) \tan(\delta_{n+1 } ) + \left [ \tan \left ( \sum_{i=1}^{n } \delta_i \right ) - \sum_{i=1}^{n } \tan ( \delta_i ) \right ] \notag \\[0.16 cm ] & \overset{\eqref{eq::gentandiff}}{= } & \sum_{i=1}^{n } \tan(\delta_{i+1 } ) \tan \left ( \sum_{k=1}^i \delta_k \right)\tan \left ( \sum_{k=1}^{i+1 } \delta_k \right ) \ ; , \end{aligned}\ ] ] which completes the induction step .
thus , the difference formula holds for all integers @xmath22 . in order to generalize the above formula further for the case @xmath155 ,
the following algebraic manipulations are made @xmath156 & = & \left [ \tan \left ( \omega + \sum_{i=1}^n \delta_i \right ) - \tan(\omega ) \right ] - \sum_{i=1}^n \left [ \tan ( \omega + \delta_i ) - \tan(\omega ) \right ] \notag \\[0.16 cm ] & \overset{\eqref{eq::tangentdifference}}{= } & \tan \left ( \sum_{i=1}^n \delta_i \right ) \left [ 1 + \tan \left ( \omega + \sum_{i=1}^n \delta_i \right ) \tan(\omega ) \right ] - \sum_{i=1}^n \tan ( \delta_i ) \left [ 1 + \tan ( \omega + \delta_i ) \tan(\omega ) \right ] \notag \\[0.16 cm ] & = & \left ( \tan \left ( \sum_{i=1}^{n } \delta_i \right ) - \sum_{i=1}^{n } \tan ( \delta_i ) \right ) + \tan(\omega ) \left ( \tan \left ( \sum_{i=1}^n \delta_i \right)\tan \left ( \omega + \sum_{i=1}^n \delta_i \right ) - \sum_{i=1}^n \tan ( \delta_i ) \tan ( \omega + \delta_i ) \right)\notag \\[0.16 cm ] & \overset{\eqref{eq::gentandiff}}{= } & \rho_0(\delta ) + \tan(\omega ) \left ( \rho_0(\delta ) \tan \left ( \omega + \sum_{i=1}^n \delta_i \right ) + \sum_{i=1}^n \tan \left ( \delta_i \right ) \left [ \tan \left ( \omega + \sum_{i=1}^n \delta_i \right ) - \tan ( \omega + \delta_i ) \right ] \right ) \notag \\[0.16 cm ] & = & \rho_0(\delta ) \left[1 + \tan(\omega ) \tan \left(\omega + \sum_{i=1}^n \delta_i \right ) \right ] \notag \\[0.16 cm ] & & + \sum_{i=1}^n \tan(\omega ) \tan \left ( \delta_i \right ) \tan \left ( \sum_{k \neq i } \delta_k \right ) \left [ 1 + \tan(\omega+\delta_i)\tan \left ( \sum_{k \neq i } \delta_k \right ) \tan \left(\omega + \sum_{i=1}^n \delta_i \right ) \right ] \ ; . \notag\end{aligned}\ ] ] now , we can bound the right - hand expression by using standard interval arithmetic .
this leads to a remainder bound of the form @xmath157 \right .
\notag \\[0.16 cm ] & & \left .
+ \sum_{i=1}^n \tan(\omega ) \tan \left ( s_i \right ) \tan \left ( t_i \right ) \left [ 1 + \tan ( \omega+s_i)\tan \left ( t_i \right ) \tan \left ( \omega + \sigma \right ) \right ] \right| \ ; , \notag\end{aligned}\ ] ] where we have introduced the auxiliary variables @xmath158 \quad \text{and } \quad \sigma = \sum_{i=1}^n s_i \ ; , \ ; \ ; \sigma = [ -\sigma , \sigma ] \ ; , \ ; \ ; t_i = [ -\sigma + s_i , \sigma - s_i ] \ ; .\ ] ] this is the remainder bound for the tangent function that is listed in the last row of table [ tab::bounds ] . | this paper is about a novel set based computing method , called interval superposition arithmetic , for enclosing the image set of multivariate factorable functions on a given domain . in order to represent enclosure sets , the proposed arithmetic stores a matrix of intervals .
every point in the domain is associated with a sequence of interval valued components of this matrix and the superposition , i.e. , the minkowski sum , over these components is the actual enclosure of the function at this point .
interval superposition arithmetic has polynomial runtime complexity with respect to the number of variables of the factorable function .
it is capable of representing highly complex enclosure sets , because the number of choices for picking a sequence of components from a given interval matrix grows exponentially with respect to the size of the matrix .
the composition rules that are associated with interval superposition arithmetic exploit algebraic addition theorems of atom operations as well as partially separable sub - structures of the computational graph of factorable functions .
besides analyzing the accuracy and favorable convergence properties of interval superposition arithmetic , this paper illustrates the advantages of the proposed method compared to existing set arithmetics by studying numerical examples . |
.8 cm the observation of the @xmath2-meson by the cdf collaboration in the channel @xmath3 , with ground state mass @xmath4 gev , and lifetime @xmath5@xmath6 ps [ 1 ] , has stimulated up the investigation of the properties of @xmath2-mesons theoretically , and experimentally on a new footing .
the particular interest of this observation is related to the fact that , the meson ground - state with @xmath7 can decay only weakly , thus providing the rather unique opportunity of investigating weak decays in a heavy quarkonium system . moreover , studying this meson could offer an unique probe to check the perturbative qcd predictions more precisely , and one can get essential new information about the confinement scale inside hadrons . the theoretical study of the pure - leptonic decays of @xmath2-meson , such as @xmath8 ( @xmath9 ) can be used to determine the leptonic decay constant @xmath10 [ 2 ] , as well as the fundamental parameters in the standard model ( sm ) , such as the cabibbo- kobayashi - maskawa ( ckm ) matrix elements which are poorly known at present .
nevertheless , the well - known helicity suppression effect make an experimental difficulty in the measurement of purely leptonic decays of @xmath2 . although , the @xmath11 channel is free of the helicity suppression , the observation of this decay is experimentally difficult due to the efficiency problem for detecting the @xmath12 lepton .
recently , the radiative leptonic @xmath13 decays ( @xmath9 ) , received considerable attention as a testing ground of sm and new physics
, where no helicity suppression exists any more [ 3 - 5 ] . among various radiative
leptonic decays , the @xmath14 decay provokes special interest , since the sm predictions has been exploited to establish a bound on the branching ratio of the above mentioned decay of order @xmath15 [ 5,6 ] , and therefore can be potentially measurable in the up coming lhc b - factories , where the number of @xmath2-mesons that will be produced are estimated to be @xmath16 [ 7,8 ]
. this will provide an alternative way to determine the decay constant @xmath10 and the ckm matrix elements .
the decay @xmath14 receive two types of contributions : internal bremsstrahlung ( ib ) , and structure - dependent ( sd ) parts .
the ib contributions are still helicity suppressed , while the sd ones contain the electromagnetic coupling constant @xmath17 but they are free of helicity suppression . therefore , the radiative decay rates of @xmath13 could have an enhancement with respect to the purely leptonic modes of @xmath8 due to the sd contributions , thus it enable to establish `` new physics '' beyond the standard model . in this work , we will study the radiative leptonic @xmath14 decay in the framework of the two - higgs doublet model ( 2hdm ) [ 9 - 11 ] at large @xmath1 . the so - called model i and model ii are considered , which are differ only in the couplings of the charged higgs bosons to fermions .
subsequently , this paper is organized as follows : in section 2 , the theoretical formalism relevance for the @xmath14 decay in 2hdm is presented .
section 3 , is devoted to the numerical analysis and the discussion of the results .
in the standard model ( sm ) , the decay @xmath14 can be studied to a very good approximation in terms of four - fermion interactions .
the effective hamiltonian relevant to the process @xmath11 is : @xmath18 where @xmath19 is the fermi constant , @xmath20 is the ckm mixing element , @xmath21 is a qcd corrected factor , which is equal @xmath22 .
however , in the next discussion we will put @xmath23 the emission of a real photon in leptonic decays of heavy mesons @xmath14 can proceed via the two mechanisms mentioned in sec . i. for the ib amplitude , the charged @xmath2-meson emits leptons via the axial - vector current , and the photon is radiated from the external charged particles . on the other hand
the sd amplitude is governed by the vector and axial vector form factors , in which the photon is emitted from intermediate states .
gauge invariance leaves only two form factors @xmath24 undetermined in the sd part .
the possible diagrams for this two mechanisms are shown in fig .
1 . following this framework , the general form of the gauge invariant amplitude corresponding to fig .
1 can be written as [ 6 ] : @xmath25 where @xmath26 and @xmath27 represent the contributions of inner bremsstrahlung
( ib ) , and structure - dependent ( sd ) parts , given by : m_1 & = & ie v_cb f_b_c m _ _ |u(p_1)\ { - } 1 - _ 5 v(p_2 ) , m_2 & = & ie v_cb _ \ { _ p_q _ + & + & } |u(p_1 ) _ 1 -_5 v(p_2 ) , where @xmath28 is the photon polarization vector , @xmath29 , @xmath30 , and q are the four momenta of @xmath31 , @xmath32 , and @xmath33 , respectively .
@xmath10 is the @xmath2- meson leptonic decay constant , @xmath24 corresponding to parity conserving and a parity violating formfactors , @xmath34 is the momentum transfer to lepton pair .
the necessary matrix elements related to the @xmath10 , and to the hadron transition form factors @xmath24 are defined as follows [ 4 ] : @xmath35 @xmath36 @xmath37.\end{aligned}\ ] ] we want now to consider the @xmath38 decay in the context of a 2hdm with no flavor changing neutral currents ( fcnc ) allowed at the tree level , i.e. model i and model ii .
the interaction lagrangian of fermions with the charged higgs fields in both models is given by [ 12 ] : l&= & \ { v_ijm_u_ix |u_i(1-_5)d_j+v_ijm_d_jy & + & m_z|(1+_5 ) } h^+h.c .
, where @xmath39 is the weak coupling constant , @xmath40 is the @xmath41- boson mass , @xmath42 is the charged physical field , and @xmath43 is the relevant elements of ckm matrix . in model
i , @xmath44 , @xmath45 and in model ii , @xmath44 , and @xmath46 .
the decay @xmath47 in 2hdm proceeds through the same feynman diagrams ( which are displayed in fig .
1 ) that mediate the process in sm , except the @xmath41-boson is replaced by the charged scalar higgs boson @xmath42 , i.e. @xmath48 . fig . 1
the matrix element corresponding to the diagram @xmath48 where the photon is radiated from @xmath31 lepton , ( see fig.1 ) is : @xmath49 where @xmath10 is the leptonic decay constant of @xmath2 meson , defined as : @xmath50 while the contribution of the structure dependent part to the @xmath51 decay , i.e. , when photon is radiated from initial quark lines , due to the charged higgs exchange can be obtained by considering the following correlation function : @xmath52 j_{\alpha}^{el } ( x ) \bigg \ } \mid b_{c } > \times \nonumber \\ & & \bar u(p_1 ) \gamma_\beta \ga 1 -\gamma_5 \dr v(p_2),\end{aligned}\ ] ] where @xmath53 is the electromagnetic current for @xmath54 or @xmath55 quarks .
the hadronic matrix elements involving the scalar and pseudoscalar currents in eq.(11 ) are parameterized such that : @xmath56 @xmath57 the parameterization of the hadronic matrix elements given in eqs .
( 12,13 ) are particularly well suited for our purposes since in the 2hdm , the vertex @xmath58 or @xmath59 , hence , the vector part of this correlator is zero , and the active part of this correlator is the axial - part given by : @xmath60 and the total matrix element in 2hdm becomes : m_2 ^ 2hdm(b_c| _ ) & = & ie v_cb f_b_c _ + & & |u(p_1)\ { - } 1 - _ 5 v(p_2 ) .
at this accuracy it is easy to check that the modified total amplitude for the radiative leptonic b - decays of @xmath47 is gauge invariant : @xmath61 where m^new_1 & = & ie v_cb f_b_c m _
_ c^2hdm|u(p_1)\ { - } 1 - _ 5 v(p_2 ) .
therefore , in this model the charged higgs contribution modifies only the so - called @xmath26 part of the sm , and it does not induce any new contribution to the so - called @xmath27 ( see eq.3 ) : c^2hdm=\ { + 1 } .
the 2hdm is sensitive to two basic free parameters , namely @xmath1 , and the charged higgs mass @xmath62 . if we formally set @xmath63 in eq .
( 18 ) , the resulting expression is expected to coincide with the @xmath64 decay , which was investigated in the framework of sm [ 6 ] .
after lengthy , but straightforward calculation for the squared matrix element , we get : @xmath65 + { \mid m_2 \mid}^2~,\end{aligned}\ ] ] where @xmath66 \nonumber \\ & + & \!\ ! ( p q ) ( p_1 q ) \left [ ( p p_2 ) ( p_1 q ) - ( p p_1 ) ( 4 p_1 p_2 + p_2 q ) \right ] \bigg\}~ , \\ \nonumber \\
\nonumber \\ 2 \,\mbox{\rm re}\left [ m^{new}_1 m_2^\dagger \right ] & = & \frac{g^2}{2 } { \mid v_{cb } \mid } ^2 e^2 ( - 16 f_{b_c } m_\tau^2)\,c^{2hdm } \frac{1}{(p_1 q ) ( p q)}\nonumber \\
& \times & \!\ ! \bigg\ { \frac { f_2(p^2)}{m_{b_c}^2 } \ , p^2 ( p_1 q ) ( p_2 q ) + ( p q)^2 \left [ \frac { f_2(p^2)}{m_{b_c}^2 } \,(p_1 p_2 + p_2 q ) - \frac { f_1(p^2)}{m_{b_c}^2 } \ , ( p_2 q ) \right ] \nonumber \\ & - & \!\ ! \frac { f_2(p^2)}{m_{b_c}^2 } \left [ ( p p_2 ) ( p_1 q ) + ( p p_1 ) ( p_2 q ) \right ] \bigg\}~ , \\ \nonumber \\
{ \mid m_2 \mid}^2 & = & \frac{g^2}{2 } { \mid v_{cb } \mid}^2 e^2 16\left [ \frac{{\mid f_1(p^2 ) \mid}^2}{m_{b_c}^4 } + \frac{{\mid f_2(p^2 ) \mid}^2}{m_{b_c}^4 } \right ] \nonumber \\ & \times & \!\ ! \left\ { ( p p_2 ) ( p q ) ( p_1 q ) + ( p_2 q ) \left [ ( p p_1 ) ( p q ) - p^2 ( p_1 q ) \right ] \right\}~.\end{aligned}\ ] ] all calculations have been performed in the rest frame of the @xmath67 meson .
the dot products of the four vectors are defined if the photon and neutrino ( or electron ) energies are specified .
the dalitz boundary for the photon energy @xmath68 and neutrino energy @xmath69 are defined as follows : @xmath70 it is now straightforward to work out the expression for the differential decay rate in the lepton and photon energies : @xmath71 the differential @xmath72 and total decay width are singular at the lower limit of the photon energy , and this singularity which is present only in the @xmath73 contribution is due to the soft photon emission from charged lepton line . on the other hand , @xmath74 and @xmath75 $ ] terms are free from this singularity . in this limit
the @xmath47 decay can not be distinguished from the @xmath76 decay .
therefore , in order to obtain a finite result for the decay width , we must consider both decays together .
the infrared singularity arising in the @xmath77 contribution must be canceled with @xmath78 virtual correction to the @xmath76 decay . in the sm
this cancellation explicitly was shown in [ 13 ] . in this work ,
the @xmath47 process is not considered as a @xmath78 correction to the @xmath76 decay , but rather a separate decay channel with hard photon radiation .
therefore we impose a cut off value on the photon energy , which will set an experimental limit on the minimum detectable photon energy .
we consider the case for which the photon energy threshold is larger than @xmath79 , mev i.e. , @xmath80 , where @xmath81 .
an integration over all the possible values of the lepton energy @xmath82 gives the total decay width as a function of the photon energy : + + @xmath83\nonumber \\ & - & \!\ ! 4 f_{b_c } \
, c^{2hdm}\int_\delta^{1-r } dx \,\frac{r x}{1-x } \bigg [ ( 1-r - x ) \big ( f_1(x ) x + f_2(x ) ( 1+r-2 x ) \big ) \nonumber \\ & - & \!\ ! ( 1-x ) \big ( f_1(x ) x + f_2(x ) ( 2r - x ) \big ) \ell n ( \frac{1-x}{r } ) \bigg]\nonumber \\ & + & \!\!\frac{1}{3 } \int_\delta^{1-r } dx \left [ { \mid f_1(x ) \mid}^2 + { \mid f_2(x ) \mid}^2 \right ] \frac{1}{(1-x)^2 } \ , x^3 ( 2 + r - 2 x ) ( 1-r - x)^2 \bigg\},\end{aligned}\ ] ] where @xmath84is the dimensionless photon energy , @xmath85 and @xmath86 .
to calculate the decay width , explicit forms of the form factors @xmath87 and @xmath88 are needed .
these form factors were calculated in the framework of the light - front quark model in [ 14 ] , and in the light - cone qcd sum rules [ 3 ] , where in [ 3 ] it was found that , the best agreement is achieved by the following pole forms for the form factors : f_1(p^2 ) = , f_2(p^2 ) = , where f_1(0 ) = 0.44 0.04 , m_1 ^ 2 = 43.1 ^2 , + f_2(0 ) = 0.21 0.02 , m_2 ^ 2 = 48.0 ^2 .
on the other hand , in evaluating the decay width , we have used the following set of parameters : @xmath89 , @xmath90 , @xmath91 , @xmath92 , @xmath93 , @xmath94 , @xmath95gev [ 15 ] , @xmath96 [ 16 ] and , @xmath97 [ 1 ] . in this
regard we should also recall that the free parameters of the 2hdm model namely @xmath1 , and @xmath62 are not arbitrary , but there are some semiquantitative restrictions on them using the existing experimental data .
the most direct bound on the charged higgs boson mass comes from top quark decays , which yield the bound @xmath98 for large @xmath1 [ 17 ] . for pure type - ii
2hdm s one finds @xmath99 , coming from the virtual higgs boson contributions to @xmath100 [ 18 ] .
furthermore , there are no experimental upper bounds on the mass of the charged higgs boson , but one generally expects to have @xmath101 in order that perturbation theory remain valid [ 19 ] . for large @xmath1
the most stringent constraints on @xmath1 and @xmath62 are actually on their ratio , @xmath102 .
the current limits come from the measured branching ratio for the inclusive decay @xmath103 , giving @xmath104 [ 20 ] , and from the upper limit on the branching ratio for @xmath105 , giving @xmath106 [ 21 ] . for illustrative purposes we consider four values of @xmath1 , namely @xmath1075 , 10 , 30 and 50 and let @xmath108 .
then we consider two values of @xmath109 , namely @xmath110=200 , 400 gev , and we allow @xmath1 to range between 0 to 60 .
the results of this numerical analysis are graphically shown in figures 2 - 4 . in these figures
the differences between the 2hdm s and the sm are shown for two different fixed cut off values , i.e. , @xmath111 and @xmath112 both for model i and model ii .
the results for the differential decay branching ratio dbr(@xmath113 as a function of @xmath114 for different values of @xmath1 , @xmath115 are presented in fig.2 , while the branching ratio ( br ) for @xmath116 decay is shown in figure 3 - 4 as a function of @xmath62 for various values of @xmath1 , and as a function of @xmath1 for different values of @xmath62 .
results are shown for model i , and mode li .
it is observed that model ii gives both a bigger differential decay branching ratio , and a bigger branching ratio than the sm rates of ( up to three orders of magnitude [ @xmath115 ] ) for large values of @xmath118 , while for small values of @xmath119 results approaches its sm value .
+ in model i the situation is somewhat totally different .
curves overlap with the sm results all the way .
this behavior obviously reflects the @xmath120 fermion couplings , which are proportional to @xmath121 in this model . in conclusion
, this study shows that the branching ratios for @xmath122 could be at the level of @xmath123 in the 2hdm , which may be detectable at the ongoing lhc .
when enough @xmath2 events are collected , this decay will be able to provide alternative channel to extract new restrictions for the free parameters @xmath1 and @xmath124 of the 2hdm model .
+ figure 1 .
: the relevant feynman diagrams , responsible for @xmath122 decay . + figure 2 . :
the dependence of the differential branching ratio of dbr(@xmath113 as a function of the photon energy @xmath114 for both models model i , and model ii . + figure 3 . :
the dependence of the branching ratio on the charged higgs boson mass at different values of @xmath1 for both models model i , and model ii . + figure 4 . : the dependence of the branching ratio on @xmath1 at different values of the charged higgs boson mass for both models model i , and model ii .
+ 99 f. abe _ et .
_ , the cdf collaboration , phys .
d * 58 * , 112004 ( 1998 ) . c. h. chang , y. q. chen , phys . rev .
d * 49 * , 3399 ( 1994 ) .
t. m. aliev , m. savci , phys .
b * 434 * , 358 ( 1998 ) .
g. eilam , i. halperin , r. r. mendel , phys .
b * 361 * , 137 ( 1995 ) . c. h. chang , j. p. cheng c. d. l , phys .
b * 425 * , 166 ( 1998 ) .
t. m. aliev , m. savci , j. phys .
g * 25 * , 1205 ( 1999 ) . c. h. chang , y. q. chen , phys .
d * 48 * , 4086 ( 1993 ) .
k. cheung , phys .
lett . * 71 * , 3413 ( 1993 ) . t. barakat , nuovo cimento a110 , 631 ( 1997 ) .
t. barakat , j. phys .
g24 , 1903 ( 1998 ) .
t. barakat , nuovo cimento a112 , 697 ( 1999 ) .
b. grinstein , m. j. savage , and m. b. wise , nucl .
b * 319 * , 271 ( 1989 ) . c. h. chang , et .
al , phys .
d * 60 * , 114013 ( 1999 ) .
c. q. geng , c. c. lih , and wei - min zhang , phys . rev .
d * 57 * , 5697 ( 1998 ) .
p. colangelo , g. nardulli , n. paver , z. phys .
c * 57 * , 43 ( 1993 ) , t. m. aliev , o. yilmaz , nuovo cimento a105 , 827 ( 1992 ) .
particle data group , r. m. barnett _
et al . _ ,
d * 54 * , 1 ( 1996 ) .
f. abe _ et .
_ , the cdf collaboration , phys .
* 79 * , 357 ( 1997 ) .
j. l. hewett , and j. d. wells , phys .
d * 55 * , 5549 ( 1997 ) .
m. veltman , phys .
b * 70 * , 253 ( 1977 ) .
aleph collaboration , in ichep96 , proceedings of the 28th international conference on high energy physics , warsaw , poland , edit by z. ajdeck and a. wroblewski ( world scientific , singapore , 1997 ) .
l3 collaboration , m. acciarri et al .
b * 396 * , 327 ( 1997 ) . | .8 cm the radiative leptonic @xmath0 decay is analysed in context of 2hdm .
it is shown that with large values of @xmath1 , the contributions of model ii to the decay rate exceeds considerably the standard model ones , while the contributions of model i overlap with the standard model predictions .
16.3 true cm 23.0 true cm -0.1 true in 0.00 true in |
decades after the experimental detection of the neutrino @xcite , it was generally accepted that the neutrino mass @xmath4 was rigorously zero . the crucial experiments with the 50 kton neutrino detector super - kamiokande found strong evidence for oscillations ( and hence - mass ) in the atmospheric neutrinos @xcite .
the direct neutrino measurements allowed to bound the neutrino mass .
the upper limit for the mass of the lightest neutrino flavor @xmath0 was obtained from experiments for measurement of the high - energy part of the tritium @xmath5-spectrum and recent experiments yielded @xmath6 upper limit @xcite . as a result of the recent experiments , the upper mass limits of @xmath7and @xmath8are @xmath9 kev @xcite and @xmath10 @xcite , respectively .
the solar neutrino experiments ( _ sne _ ) and atmospheric neutrino experiments ( _ ane _ ) allow to find the square mass difference @xmath11 and @xmath12 , but not the absolute value of neutrino masses
. the astrophysical constraint of the neutrino mass is @xmath13 @xcite .
the recent extensions of the standard model lead to non - zero neutrino masses , which are within the large range of @xmath14 . in the classical @xmath15 model
the mass relations between charged leptons and down quark masses are simply identities : @xmath16 @xmath17 and @xmath18 .
the mass relations of georgi - jarlskog @xcite ensue from the @xmath19 model and relate charged leptons and down quark masses : @xmath20 @xmath21 and @xmath18 .
however , these mass relations deviate several times , compared to experimental data .
moreover , similar mass relations are unsuited for neutral leptons ( neutrino ) masses .
the seesaw mechanism naturally generates small majorana neutrino mass @xmath22 from reasonable dirac mass @xmath23 and very heavy majorana sterile neutrino mass @xmath24 , namely @xmath25 .
but there are many seesaw models that differ in the scale @xmath24 and dirac mass .
the grand unified theories ( _ gut _ ) are the main candidates for seesaw models , with @xmath24 at or a few orders of magnitude below _
gut _ scale . successful _ gut _
models should essentially generate cabibbo - kobayashi - maskawa ( _ ckm _ ) quark mixing matrix @xcite and maki - nakagawa - sakata ( _ mns _ ) lepton mixing matrix @xcite and predict results compatible with the data from _ sne _ and _ ane_. yet , it is admitted that the predictions of the quark - lepton mass spectrum are the least successful aspect of the unified gauge theory @xcite .
the purpose of this paper is to find simple and reliable quark - lepton mass relations , based on experimental data and estimations for quark and lepton masses .
the next step is to estimate neutrino masses by means of these mass relations and data from _ sne _ and _
according to the standard model , the fundamental constituents of the matter are 6 quarks and 6 leptons .
the fundamental fermions group in three generations , having similar properties and increasing masses .
the three generations of the fundamental fermions and their masses are presented in table [ table 1 ] .
the estimations of quark masses are taken from @xcite and the upper mass limits of the neutrino flavors are taken from @xcite . [ c]lllllllfermions & & & + up quarks & @xmath26 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 + down quarks & @xmath32 & @xmath33 & @xmath34 & @xmath35 & @xmath36 & @xmath37 + charged leptons & @xmath38 & @xmath39 & @xmath40 & @xmath41 & @xmath42 & @xmath43 + neutral leptons & @xmath0 & @xmath44 & @xmath1 & @xmath45 & @xmath2 & @xmath46 + [ table 1 ] a clear feature of the quark and charged lepton mass spectrum is the hierarchy of masses belonging to different generations : @xmath47 most likely , a similar hierarchy of masses of the neutral leptons ( neutrinos ) could be anticipate @xmath48 .
based on the experimental data , we search for a simple relation between the masses of charged leptons ( @xmath49 ) and the respective up quarks ( @xmath50 ) by the least squares .
although the linear regression @xmath51 @xmath3 shows close correlation , it yields electron mass many times lower than the experimental value .
after examination of other simple approximations ( logarithmic , exponential and power law ) we found out that the power law fits best experimental data : @xmath52 where @xmath53 and @xmath54 despite the large uncertainty of the @xmath55 quark mass ( from @xmath56 to @xmath57 @xmath58 ) and @xmath59 quark mass ( from @xmath60 to @xmath61 ) , the slope remains within the narrow interval from 0.683 to 0.782 .
although only three points make the approximation , the correlation coefficient is very high ( @xmath62 ) and the maximal ratio of the predicted mass in relation to the respective experimental value is @xmath63 @xmath64 ( for muon ) .
the predicted masses of the electron and tau lepton differ from the respective experimental values with less than 40% .
therefore , the mass relation ( [ eqn2 ] ) could be accepted satisfactory .
we suggest that a mass relation similar to ( [ eqn2 ] ) connects the masses of the neutral leptons ( @xmath65 ) and the respective down quarks ( @xmath66 ) : @xmath67 where @xmath68 and @xmath69 is an unknown constant . for @xmath70 formula ( [ eqn3 ] ) yields
@xmath71 @xmath58 , @xmath72 and @xmath73 @xmath58 .
these values are several orders of magnitude bigger than the experimental upper limits of the neutrino masses ( see table 1 ) , therefore @xmath74 .
astrophysical constraints allow to limit more closely the value of _ k _ since they give @xmath75 @xmath3 .
thus , from equation ( [ eqn3 ] ) we obtain : @xmath76 _ ane _ @xcite determine the squared mass difference : @xmath77 relation ( [ eqn3 ] ) yields : @xmath78 @xmath79 solving system ( [ eqn5 ] ) - ( [ eqn7 ] ) we obtain @xmath80 , @xmath81 and @xmath82 .
these results support the normal hierarchy of neutrino masses .
on the other hand , the large mixing angle ( _ lma _ ) of mikheyev - smirnov - wolfenstein ( _ msw _ ) solution for _ sne _ @xcite yields : @xmath83 this equation , together ( [ eqn6 ] ) and ( [ eqn7 ] ) , yield @xmath84 , @xmath85 and @xmath86 @xmath3 .
these values are almost three times bigger than the values obtained by super kamiokande data , therefore , they do not fit well with the latter . however , the small mixing angle ( _ sma _ ) _ msw _ solution for _ sne _ @xcite yields : @xmath87 this equation , together ( [ eqn6 ] ) and ( [ eqn7 ] ) , yield @xmath88 , @xmath89 and @xmath90 .
these values differ by less than 25% from the results obtained above by super kamiokande data , which show that according to the suggested approach , _ sma msw _ solution fits better with the _ ane _ than _ lma msw_. thus , the obtained quark - lepton mass relations and the results of the solar and atmospheric neutrino experiments provide to estimate the masses of @xmath0 , @xmath1 and @xmath2 of @xmath91 , @xmath92 and @xmath93 , respectively .
these values are close to the neutrino masses ( @xmath94 , @xmath95 and @xmath96 ) found in @xcite by the mass relation , connecting the masses of four stable particles and the coupling constants of the fundamental interactions .
we could calculate constant @xmath69 using the most trustworthy data for neutrinos and down quarks masses , namely @xmath2 and @xmath36-quark masses : @xmath97 [ t ] fig1.png the obtained mass relations ( [ eqn2 ] ) and ( [ eqn3 ] ) are shown in fig .
[ figure1 ] .
it shows that the neutrino masses estimated by _
sma msw _ are close to the neutrino masses estimated by _ ane _ , i.e. two sets of estimations are compatible .
the attempt to relate the masses of the charged leptons with the masses of the down quarks and the masses of neutral leptons with the masses of up quarks did not yield satisfactory results since the data from _ sne _ and _ ane _ did not fit within the framework of the suggested approach . besides , the respective mass relations predict the muon mass which is nearly three times less than the experimental value ( see table [ table 2 ] ) and an electron neutrino mass less than @xmath98 . [ c]lllllllmodel & _ su(5 ) _ & _ so(10 ) _ & power law & linear & power law & exp .
+ & & & ( doun & ( up & ( up & data + & & & quarks ) & quarks ) & quarks ) & + electron & 6 & 2 & 0.905 & 0.031 & * 0.663 * & * 0.511 * + muon & 122 & 366 & 36.9 & 12.8 & * 60.7 * & * 105.7 * + tau & 4200 & 4200 & 2877 & 1775 & * 2449 * & * 1777 * + @xmath99 & 11.74 & 3.91 & 2.86 & 16.48 & * 1.74 * & * 1 * + [ table 2 ] table [ table 2 ] shows the masses of charged leptons calculated by different approaches and experimental values .
the last row of the table shows the maximal ratio ( deviation ) @xmath100 of the masses predicted by the respective approach in relation to the experimental values @xmath101 .
clearly , the power law relating to the masses of charged leptons and up quarks fits best experimental data .
based on the experimental data and estimations of charged leptons and quarks masses , a power law with exponent @xmath102 has been found , connecting charged leptons masses and up quarks masses . it has been shown that this approximation is considerably better than any known approach .
a similar mass relation has been suggested for neutral leptons and down quarks .
the latter mass relation and the results of _ ane _ and _ sne _ have been used for estimations of neutrino masses .
the values of neutrino masses obtained by _
ane _ are close to the ones , obtained by the _
sma msw _ solution .
the masses of @xmath0 , @xmath1 and @xmath2 are estimated to @xmath91 , @xmath92 and @xmath93 , respectively , and they support the normal hierarchy of neutrino masses . | based on the experimental data and estimations of the charged leptons and quarks masses , a close power law with exponent 3/4 has been found , connecting charged leptons masses and up quarks masses . a similar mass relation has been suggested for the masses of neutral leptons and down quarks .
the latter mass relation and the results of the solar and atmospheric neutrino experiments have been used for prediction of neutrino masses .
the obtained masses of @xmath0 , @xmath1 and @xmath2 are 0.0003 @xmath3 , 0.003 @xmath3 and 0.04 @xmath3 , respectively .
these values are compatible with the recent experimental data and support the normal hierarchy of neutrino masses .
pacs numbers : 14.60.lm ; 12.15ff ; 14.60.pq ; 12.10.kt key words : mass relation ; quark - lepton symmetry ; neutrino mass ; normal hierarchy = 1 |
quiver mutation is a central element in the recent theory of cluster algebras introduced by fomin and zelevinsky in @xcite .
it is an elementary operation on quivers which generates an equivalence relation .
the mutation class of a quiver @xmath2 is the class of all quivers which are mutation equivalent to @xmath2 .
the mutation class of quivers of type @xmath3 is the class containing all quivers mutation equivalent to a quiver whose underlying graph is the dynkin diagram of type @xmath3 , shown in figure [ fig : dynkin]@xmath4 .
the quivers in the mutation class were characterized by buan and vatne in @xcite . the corresponding task for type @xmath1 , shown in figure [ fig : dynkin]@xmath5 ,
was accomplished by vatne in @xcite .
furthermore , an explicit formula for the number of quivers in the mutation class of type @xmath3 was given by torkildsen in @xcite and of type @xmath1 by buan and torkildsen in @xcite .
in this article , we consider quivers of type @xmath6 .
that is , all quivers mutation equivalent to a quiver whose underlying graph is the extended dynkin diagram of type @xmath6 , i.e. , the @xmath7-cycle see figure [ fig : dynkin]@xmath8 .
if this cycle is oriented , then we get the mutation class of @xmath1 , see fomin et al . in @xcite and type iv in @xcite .
if the cycle is non - oriented , we get the mutation classes of @xmath6 , studied by the first named author in @xcite .
the purpose of this paper is to give an explicit formula for the number of quivers in the mutation classes of quivers of type @xmath6 .
since a cluster - tilted algebra @xmath9 of type @xmath6 is finite dimensional over an algebraically closed field @xmath10 , there exists a quiver @xmath2 which is in the mutation classes of @xmath6 ( see for instance assem et al . in @xcite ) and an admissible ideal i of the path algebra @xmath11 of @xmath2 such that @xmath12 .
therefore , we also obtain the number of non - isomorphic cluster - tilted algebras of type @xmath6 .
in fact , we prove a more refined counting theorem .
namely , one can classify these algebras up to derived equivalence , see @xcite .
each equivalence class is determined by four parameters , @xmath13 , @xmath14 , @xmath15 and @xmath16 ( up to interchanging @xmath13 , @xmath14 and @xmath15 , @xmath16 ) , where @xmath17 . given positive integers @xmath18 and @xmath19 with @xmath20 , the set of equivalence classes with @xmath21 and @xmath22 corresponds to one mutation class of quivers .
the number of cluster - tilted algebras in the derived equivalence classes with parameters @xmath23 and @xmath16 is given by @xmath24 if @xmath25 or @xmath26 , and @xmath27 if @xmath28 and @xmath29 .
( here @xmath30 is euler s totient function , i.e. , the number of @xmath31 coprime to @xmath32 . ) in particular , for @xmath33 and @xmath34 , we obtain the number @xmath35 of quivers mutation equivalent to a non - oriented @xmath7-cycle with @xmath18 arrows oriented in one direction and @xmath19 arrows oriented in the other direction : @xmath36 the paper is organized as follows . in section
[ sec : pres ] we collect some basic notions about quiver mutation .
furthermore , we present the classification of quivers of type @xmath6 according to the parameters @xmath18 and @xmath19 mentioned above , as given in @xcite . in section [ sec : comb - gramm ] we restate the classification as a combinatorial grammar . using generatingfunctionology we obtain the formulae for the assymmetric case where @xmath25 or @xmath26 . for the case @xmath28 and @xmath29 some additional combinatorial considerations , counting the number of quivers invariant under reflection , yield the result stated above . the formulae for @xmath35 are developed in parallel .
in fact , it is remarkable that the generating function including variables for the parameters @xmath14 and @xmath16 can be obtained by _ specialising _ the much simpler generating function having variables for the parameters @xmath18 and @xmath19 only .
moreover , extracting the coefficient of @xmath41 in a naive way from the equations obtained from combinatorial grammars results in a much uglier five - fold sum , instead of the three - fold sum stated in the main theorem .
finally , at the end of section [ sec : comb - gramm ] we prove the formula for the number of quivers in the mutation class of type @xmath1 , by exhibiting an appropriate bijection between these and a subclass of the objects counted in section [ sec : number - quivers ] .
* acknowledgements : * we would like to thank thorsten holm for invaluable comments on a preliminary version of this article , and ira gessel for providing the beautiful proof of lemma [ lem : gessel ] .
we also would like to thank christian krattenthaler who gave an elementary proof of the same lemma , of which at first we only had a computer assisted proof . 1 . a quiver in @xmath1 of type @xmath42 , where precisely one of the two distinguished arrows incident to the root is oriented towards it , or 2 . a quiver in @xmath1 of type @xmath43 , i.e. , a quiver having a unique oriented @xmath46-cycle
it remains to describe the bijection in the case where the main cycle of @xmath44 is a loop . in a first step ,
we delete the vertex of this loop and all arrows incident to it , to obtain a rooted quiver @xmath47 of type @xmath48 .
for the second and final step , we distinguish two cases : 1 .
the root of @xmath49 is incident to a single arrow @xmath50 . in this case
we obtain a quiver @xmath2 in @xmath1 of type @xmath42 by adding a second arrow , oriented in the same way as @xmath50 , to the other vertex @xmath50 is incident to
. 2 . on the other hand , consider the case that the root of @xmath49 is incident to an oriented @xmath51-cycle @xmath52 .
then , we glue a second @xmath51-cycle , oriented in the same way as @xmath52 , along the arrow of @xmath52 opposite to the root . in this way
we create a quiver in @xmath1 of type @xmath53 . * a quiver @xmath2 in @xmath1 of type @xmath42 has a uniquely determined root , and two distinguished arrows incident to it .
if they are oriented in opposite directions , then the main cycle in the preimage of the transformation is an oriented @xmath45-cycle . otherwise , the preimage is a loop .
* @xmath2 is of type @xmath53 , if and only if it has two oriented @xmath51-cycles sharing an arrow . * finally , @xmath2 is of type @xmath43 , if and only if it has a unique oriented @xmath46-cycle .
franois bergeron , gilbert labelle , and pierre leroux , _ combinatorial species and tree - like structures _
, encyclopedia of mathematics and its applications , vol .
67 , cambridge university press , cambridge , 1998 , translated from the 1994 french original by margaret readdy , with a foreword by gian - carlo rota .
aslak bakke buan and hermund andr torkildsen , _ the number of elements in the mutation class of a quiver of type @xmath1 _ , electronic journal of combinatorics * 16 * ( 2009 ) , no . 1 , research paper 49 , 23 pp . ( electronic ) , . | in this article we prove explicit formulae for the number of non - isomorphic cluster - tilted algebras of type @xmath0 in the derived equivalence classes .
in particular , we obtain the number of elements in the mutation classes of quivers of type @xmath0 . as a by - product , this provides an alternative proof for the number of quivers of dynkin type @xmath1 which was first determined by buan and torkildsen in @xcite . |
various schemes associated with quantum computing have been proposed for quantum information storage and transfers , such as superconducting coplanar waveguide resonators and dual waveguide devices @xcite .
a waveguide can be defined as a quantum wire in which the electron wave propagates in quantized modes without losing phase coherence at low temperature .
various proposals have been suggested to implement a semiconductor qubit @xcite . among these
, an idea is to use parallel quantum waveguides with a coupling - window placed between them @xcite .
tuning the window - coupling allows an electron wave interference between the waveguides and can switch the electron motion from one waveguide to the other @xcite .
the interference in the coupling - region specifies the possible types of qubit - operations : the electron wave can be transferred through either waveguide only ( pass gate - operation ) .
it can travel through both waveguides , split equally , implementing the so called square - root - of - not ( @xmath0 ) operation @xcite .
it can switch totally from one waveguide to the other ( not - operation ) @xcite . or it might switch from the first waveguide to the second one and then re - enter the first waveguide ( cnot - operation ) @xcite .
there are several parameters and phenomena that can modify the efficiency of the kind of qubit - operation that are mentioned above such as external magnetic field , the coulomb interaction , electrostatic gate voltage , and the geometry of the coupling window between the waveguides .
ferry s group proposed magnetically switching transport in an asymmetric double quantum waveguide , in which an external magnetic field is applied to switch electron motion from one quantum - waveguide to the other @xcite .
another suggestion for a switching - qubit is to vary the length of the coupling window ( cw ) @xmath1 . by varying the length of the cw
, a maximum inter - waveguide tunneling can be found that increases the efficiency of the device @xcite .
another approach for a switching - qubit is to define a saddle potential in the coupling window instead of a hard - walled well potential .
the saddle - potential makes a soft barrier between the two waveguides that washes out resonance peaks , but decreases the efficiency of the qubit @xcite . in the present work ,
we consider a double quantum waveguide ( dqw ) system with a cw between the waveguides in a static perpendicular magnetic field .
both ends of the dqw are connected to semi - infinite leads with an applied source - drain bias .
the coherent electron transport is investigated in the system by using a non - markovian master equation @xcite .
smooth gaussian potentials are used to define the barrier between the waveguides and the saddle - like cw potential .
in addition , two dots are embedded in the waveguides close to the cw to enhance inter - waveguide transport .
we will investigate coherent electron transport in the system by varying the cw length .
our results reveal the following conclusions : first , we predict a maximum conductance at resonant energy - levels . at a resonance energy , inter - waveguide scattering is enhanced and the charge density splits equally between the waveguides in the absence of the coulomb interaction as two - electron states are activated in the transport . in this case
, the dqw essentially works as a @xmath0-operation qubit . in the presence of the coulomb interaction ,
the charge current density does not split equally between the waveguides anymore at the resonance since two - electron states are blocked for the bias range used .
the strength of the blocking can be varied by the geometry or the material of the system .
the transient transport through two capacitively coupled parallel quantum dots has been investigated by moldoveanu et al . in a lattice model for a weaker coulomb interaction than is assumed here @xcite .
second , an external static magnetic field can increase inter - waveguide backward scattering leading to a decrease in the efficiency of the qubit .
these results make clear the difference to previous schemes where variable magnetic field was used to control the switching between asymmetric waveguides @xcite .
third , the influence of the side quantum dots ( qds ) can be to induce more energy - levels into resonance , and consequently increase the coupling between the waveguides and the inter - waveguide transport mechanism . in the presence of side qds , a not - operation qubit is realized .
the efficiency is slightly decreased in the presence of side qds because inter - waveguide backward scattering is enhanced . at high magnetic field
such that the effective magnetic length is comparable to the radius of the dots , variation of length of the cw does not influence the electron transport characteristics anymore because the electrons are localized in the side dots .
thus , the net charge current is extremely suppressed .
the paper is organized as follows : in sec .
[ sec : ii ] , we present the model describing the window - coupled double waveguide system based on the quantum master equation ( qme ) approach . section [ sec : iii ] presents our numerical results and discussion .
concluding remarks are addressed in sec .
[ sec : iv ] .
in this section , we present a method for calculating the transient ballistic transport in the coupled - window quantum waveguides without and with side qds .
we consider two laterally coupled waveguides connected to two semi - infinite wires from both ends .
the lower waveguide ( control ) and the upper waveguide ( target ) are coupled through a cw with length @xmath1 as is schematically shown in .
we consider a control - qd ( qd@xmath2 ) and a target - qd ( qd@xmath3 ) embedded , respectively , in the control- and the target - waveguide near the cw . the total system is exposed to an external magnetic field in the @xmath4-direction b@xmath5 .
the system is designed such that the electrons are injected from the left lead to only the control - waveguide ( red arrow ) .
our aim here is to study the charging of , and the output from each of the waveguides by varying the cw length @xmath1 ( blue dashed arrow ) .
in addition , the effects of the coulomb interaction , the external magnetic field , and the side dots on the conductance will be explored .
( blue dashed arrow ) .
a quantum dot qd@xmath2 ( qd@xmath3 ) is embedded in the control ( target ) waveguide .
an electron from left side enters the control waveguide ( red arrow ) .
asymmetric contacts are indicated by the light red rectangles marked l and r.,scaledwidth=35.0% ] the dqw is a two dimensional system with hard - wall confinement in the contact area to the external leads at @xmath6 with @xmath7 the length of the waveguides in the transport direction and parabolic confinement in the @xmath8-direction .
the single - electron hamiltonian of the dqw including side qds in an external magnetic field is @xmath9 where @xmath10 indicates the canonical momentum , @xmath11 is vector potential of the external magnetic field , and @xmath12 is the effective mass of an electron .
the confining potential of the dqw is @xmath13 , where @xmath14 stands for a hard - wall confining potential and @xmath15 refers to a parabolic confining potential with characteristic energy @xmath16 .
the gaussian potential barrier between the two waveguides is presented by @xmath17 and the side qds are defined by @xmath18 .
we diagonalize the hamiltonian of the central system presented in using a basis @xcite built with the eigenfunctions of the first two terms of ( [ h_s ] ) to find the single - electron energy spectrum @xmath19 , where @xmath20 is the composite quantum number of a single - electron state .
below , we shall demonstrate how the dqw is connected to the leads and show the time evolution of electron in the system .
we connect the dqw to two leads that act as electron reservoirs via a contact region .
the central system and the leads have the same confinement energy and the same strength of the perpendicular magnetic field .
the total hamiltonian of the system is then @xmath21 .
\nonumber \label{ht}\end{aligned}\ ] ] the first term describes the central system , where @xmath22 is the single - electron energy of state @xmath23 of the central system , and @xmath24 @xmath25 denotes the electron creation ( annihilation ) operator respectively .
the second term is the mutual coulomb interaction between electrons in the central system .
the third term of the hamiltonian denotes the semi - infinite @xmath26 lead , where @xmath27 refers to the left l or the right r lead with the dummy index @xmath28 representing the `` momentum '' of an electron in the lead and its subband index @xcite , @xmath29 is the single - electron energy spectrum in lead @xmath27 , and the electron creation and annihilation operator in lead @xmath27 are @xmath30 and @xmath31 , respectively . the last term is the time - dependent coupling hamiltonian that connects the central system , the dqw to the external leads via a coupling tensor @xmath32 , and @xmath33 dictates the time - dependence of the coupling .
the coupling hamiltonian describes the transfer of an electron between the single - electron state of the central system @xmath23 and the single - electron state of the leads @xmath34 @xmath35 @xmath36 and @xmath37 are the corresponding single - electron wave functions of the dqw system and the leads .
@xmath38 is a nonlocal coupling kernel @xmath39 } \nonumber \\ & \times \exp{\left ( -\delta_{n}^l(\mb{q } ) / \delta \right ) } , \label{cf}\end{aligned}\ ] ] where @xmath40 is the coupling strength , @xmath41 and @xmath42 are the coupling range parameters in the @xmath43- and @xmath8-direction , respectively , @xmath44 controls together with @xmath45 the energy affinity between states in the leads and the central system @xcite , and @xmath46 is a skewing parameter that shifts the weight of the coupling to the lower waveguide for an appropriate value stated below . in this subsection ,
we present how the time evolution of the open system is calculated in order to study time - dependent transport properties within the qme approach @xcite .
the time evolution of our system obeys the quantum liouville - von neumann equation @xmath47,\ ] ] where @xmath48 is the density operator of the total system . in order to focus our attention on the open central system we introduce a reduced density operator ( rdo ) with respect to it , by taking the trace over the fock space with respect to the leads @xmath49 with the initial condition @xmath50
= @xmath51 , where @xmath52 and @xmath53 are the density operators of the isolated left and the right leads , respectively .
the density operator of the @xmath54 lead can be defined as @xmath55 , where @xmath56 and @xmath57 is the boltzmann constant , and @xmath58 is the number of electrons in the leads .
the density operator of the central system at @xmath59 being @xmath60 @xcite allows us to obtain the equation of motion for the rdo in the following form @xcite @xmath61 where @xmath62/\hbar$ ] is the liouvillian with respect to the time - independent hamiltonian @xmath63 of the central system and @xmath64 is an integral kernel with its roots in the dissipative time - dependent coupling to the leads @xcite . in order to explore the switching processes between the waveguides
, we define the net charge current as @xmath65 where @xmath66 refers to the partial current from the left lead into the control - waveguide and @xmath67 indicates the partial current into the right lead from both waveguides @xcite . in the negative sign in front of @xmath68 is since a positive charge current is defined as the current into the central system , but the currents @xmath69 and @xmath70 are defined positive from left to right . to monitor the dynamic evolution of the electrons in the central system , we calculate the expectation value of charge current density operator @xmath71 where the charge current density operator is defined by @xmath72 \\ & + \frac{e^2}{m^ * } \mathbf{a}_{\rm ext}({\bf r } )
\psi^\mathrm{s*}_{n}({\bf r } ) \psi^\mathrm{s}_{n'}({\bf r } ) \bigg ) d_n^{\dagger } d_{n'}.\end{aligned}\ ] ] in the following , we shall demonstrate the influence of the length of the cw , external magnetic field , and the side qds on the coherent electron transport through the system and the charge current density in the central system in order to investigate inter - waveguide forward and backward scattering processes .
in this section , we shall discuss our numerical results of the ballistic transport properties through the double waveguide system made of a gaas semiconductor with the electron effective mass @xmath73 .
the length of the central system is @xmath74 nm which is assumed to be much smaller than the phase coherent length @xmath75 . at low temperature @xmath76
k , the phase coherent length of semiconductor based ( gaas / al@xmath77ga@xmath78as ) electron waveguide is @xmath79 nm @xcite .
the coherence in the electron transport in the double waveguide is an essential requirement for constructing a qubit . for the system placed in a static or vanishing external magnetic field ,
its length scales can be characterized by the effective magnetic length @xmath80 , where the cyclotron frequency @xmath81 and the transverse confinement energy is @xmath82 mev .
numerically the effective magnetic length can be expressed as @xmath83{1 + 2.982[b({\rm t})]^2 } } \ { \rm nm}.\end{aligned}\ ] ] for the case of a vanishing magnetic field
we assign to it a finite very small value in order to avoid numerical problems caused by exact degeneration of states with opposite spin .
we assume the central system is connected to the external leads acting as electron reservoirs with chemical potential in the left ( right ) lead @xmath84 mev ( @xmath85 mev ) , implying an applied potential difference or bias window @xmath86 mev .
furthermore , the temperature of the leads is fixed at @xmath87 k. the potential of the double quantum waveguides is shown in with the side qds located near the cw in order to influence the inter - waveguide transport .
@xmath88 t , @xmath89 nm , @xmath90 nm , and @xmath91 nm.,title="fig:",scaledwidth=48.0% ] + in the following , we shall explore the influence of the cw and the side qds on the charging of and the transport through the system .
we start by considering an ideal window - coupled dqw without side qds .
the injected electrons are assumed to come from the left lead only into the control - waveguide .
this asymmetry in the injection is accomplished with the choice of the skewing parameter @xmath92 defined in eq .
( [ cf ] ) and only used for the left lead .
in addition , we assume the two quantum waveguides to be of the same width .
they are separated by a gaussian potential barrier , in which the control- and target - waveguide interact with each other via a cw .
the dqw system is described by @xmath93 where @xmath94 is the barrier potential between the quantum waveguides defined as @xmath95 with @xmath96 mev and @xmath97 nm@xmath98 , as well as a cw potential @xmath99 with @xmath96 mev , and @xmath100 nm@xmath98 implying a barrier width @xmath101 nm for the first subband which can prevent electron tunneling between the waveguides through the barrier .
@xmath102 is a parameter that determines the cw length .
thus , the length of the cw can be defined as @xmath103 .
we begin to analyze the results by presenting which shows the net charge current @xmath104 as a function of the cw length @xmath1 at @xmath105 ps and @xmath106 t for noninteracting electrons ( solid blue ) and for coulomb interacting electrons ( dashed red ) .
the coulomb interaction reduces the overall current slightly .
the oscillations in the net charge current are indicative of a possible charge transport between the control - waveguide and the target - waveguide .
the oscillations give rise to a peak and a dip in the net charge current at @xmath107 and @xmath108 nm , respectively .
similar oscillation features have been found by zibold _
et al_. and gong _
et al_. in the absence of the coulomb interaction , and put in relation to energy - dependent inter - waveguide transmission in the quantum regime .
they pointed out that the inter - waveguide transmission can be enhanced when the energy levels of the coupled - waveguide system achieve a resonance conditions for a specific cw length @xcite . versus cw length @xmath1 without ( w / o ) ( blue solid ) and with ( w ) ( red dashed )
electron - electron ( e - e ) interaction for magnetic field @xmath106 t the chemical potentials are @xmath84 mev and @xmath85 mev implying @xmath109.,scaledwidth=23.0% ] to demonstrate the resonant energy levels in our system , we present which shows the many - electron ( me ) energy spectra for the dqw system as a function of the cw length at @xmath106 t including one - electron states ( 1es , red dots ) and two - electron states ( 2es , blue dots ) .
the me spectrum without the mutual coulomb interaction is displayed in fig . [
fig04](a ) , but the interacting spectrum in fig .
[ fig04](b ) .
the only difference between the energy spectra is that the coulomb interaction has raised most of the 2es well above the bias window in fig .
[ fig04](b ) .
the 2es do not contribute to the electron transport in the presence of the coulomb interaction because they are far from the chosen bias window , leading to a decrease in the net charge current . in the case of no coupling window
the two waveguides are only coupled through tunneling and the coulomb interaction @xcite leading to nearly degenerate states .
when the length of the cw @xmath1 is increased we thus find : first , the energy spectrum decreases monotonically @xcite .
second , the near degeneration of the energy - levels is lifted . as a result
, this leads to a crossover of energy levels ( resonance ) between the control- and target - waveguide at @xmath110 nm ( left green rectangular ) which corresponds to the current - peak in .
the resonance between the waveguides enhances the inter - waveguide electron transport and increases the net charge current .
the splitting of the energy levels increases at a higher window - coupling length such as @xmath111 nm ( right green rectangular ) indicating weaker resonances and more back - scattering inter - waveguide transport . without ( a ) , and with ( b ) the electron - electron coulomb interaction including one - electron states ( 1es , red dots ) and two - electron states ( 2es , blue dots ) at @xmath106 t. the one - electron states in the left green rectangle are close to be in resonance , but the one - electron states in the right green rectangle are not .
the chemical potentials are @xmath112 and @xmath113 ( black ) implying @xmath109.,title="fig:",scaledwidth=23.0% ] without ( a ) , and with ( b ) the electron - electron coulomb interaction including one - electron states ( 1es , red dots ) and two - electron states ( 2es , blue dots ) at @xmath106 t. the one - electron states in the left green rectangle are close to be in resonance , but the one - electron states in the right green rectangle are not .
the chemical potentials are @xmath112 and @xmath113 ( black ) implying @xmath109.,title="fig:",scaledwidth=23.0% ] to further explain the peak and the dip in the net charge current , we present the charge current density at @xmath106 t and @xmath114 ps in for the noninteracting system ( left panels ) and the interacting one ( right panels ) for the peak and the dip net charge current at @xmath110 nm ( a - b ) and @xmath108 nm ( c - d ) , respectively , shown in .
the dynamical evolution occurring in the cw implements different types of quantum logic gates in our system . in ( a )
the charge current density in the absence of the coulomb interaction reveals the following electron motion in the waveguides at @xmath110 nm : the charge is injected into the control input , then it exits equally from both control and target output . in the dqw
the charge exhibits partial inter - waveguide forward scattering which is similar to the condition of having equal electron transmission ratio in the double waveguides at a resonance @xcite .
the dqw here can be defined as a beam splitter or a @xmath0-operation quantum logic gate ( @xmath0-operation qubit ) @xcite .
figure [ fig05](c ) shows the charge current density in the current - dip at @xmath111 nm .
the charge current density switches from the control- to the target - waveguide with a small ratio of inter - waveguide backward scattering .
the charge switching property of the system here can be introduced as a not - operation quantum logic gate @xcite .
the net charge current is suppressed here due to the presence of an inter - waveguide backward scattering that forms the current - dip .
the cause of the inter - waveguide backward scattering is the geometry of the system , the symmetric waveguides , while in the asymmetric system the back scattering can be avoided @xcite . with the coulomb interaction
the charge current density is very similar for the dip near 110 nm ( comparing fig s [ fig05](c ) and ( d ) ) , but different for the peak at 40 nm , where very little scattering into the target waveguide is seen ( comparing fig . [
fig05](a ) and ( b ) ) . without the coulomb interaction
2es contribute up to 1/3 of the charge current , but the interaction reduces this with coulomb blocking to a negligible quantity . the forward inter - waveguide scattering in fig . [ fig05](a )
is facilitated by two - electron processes and states . in the present system the coulomb interaction blocks these processes to a large extent .
+ + the effects of a static external magnetic field are presented in ( a ) which shows the net charge current @xmath104 as a function of the cw length @xmath1 at @xmath105 ps in the presence of the coulomb interaction for different values of the magnetic field @xmath106 t ( blue solid ) , @xmath115 t ( green dashed ) , and @xmath116 t ( red dashed ) . for comparison
we display in ( b ) the left and the right currents @xmath117 and @xmath68 at @xmath106 t and the same point in time .
the comparison shows that at @xmath105 ps the system is still in the charging phase , but the overall shape of the currents is similar . versus cw length @xmath1 in the presence of the coulomb interaction for different values of the magnetic field
@xmath106 t ( blue solid ) , @xmath115 t ( green dashed ) , and @xmath116 t ( red dotted ) at time @xmath114 ps .
( b ) the current from the left lead @xmath117 ( pink dotted ) , the current into the right lead @xmath68 ( blue dashed ) , and the net charge current @xmath104 ( blue solid ) for @xmath106 t versus the cw length .
the chemical potentials are @xmath84 mev and @xmath85 mev implying @xmath109.,title="fig:",scaledwidth=23.0% ] versus cw length @xmath1 in the presence of the coulomb interaction for different values of the magnetic field @xmath106
t ( blue solid ) , @xmath115 t ( green dashed ) , and @xmath116 t ( red dotted ) at time @xmath114 ps .
( b ) the current from the left lead @xmath117 ( pink dotted ) , the current into the right lead @xmath68 ( blue dashed ) , and the net charge current @xmath104 ( blue solid ) for @xmath106 t versus the cw length .
the chemical potentials are @xmath84 mev and @xmath85 mev implying @xmath109.,title="fig:",scaledwidth=23.0% ] the role of the external magnetic field in our system is different from the role it plays in many models where researchers have applied a variable magnetic field to switch between different processes in the waveguides . the external magnetic field has been used to switch an electron current from one quantum - waveguide to the other one in a window coupled double waveguide @xcite . in our model , we observe switching characteristic in _ absence _ of a magnetic field @xmath106 t. in the presence of stronger magnetic field , the electron motion is affected by the lorentz force in which the electrons tend to a circular motion in both control- and target - waveguides increasing the electron dwell - time in the system .
ps in the presence of the coulomb interaction in the peak at @xmath110 nm ( a ) , and the dip at @xmath111 nm ( b ) net charge current of ( a ) in the case of @xmath118 t ( red dotted ) .
the effective magnetic length is @xmath119 nm.,title="fig:",scaledwidth=45.0% ] + ps in the presence of the coulomb interaction in the peak at @xmath110 nm ( a ) , and the dip at @xmath111 nm ( b ) net charge current of ( a ) in the case of @xmath118 t ( red dotted ) .
the effective magnetic length is @xmath119 nm.,title="fig:",scaledwidth=45.0% ] the inter - waveguide backward scattering is enhanced while a suppression in the inter - waveguide forward scattering is observed in the presence of strong magnetic field .
therefore , the net charge current decreases as shown in ( a ) in the case of @xmath120 t ( green dashed ) and @xmath116 t ( red dotted ) . to understand the reasons for the suppression of the net charge current in the presence of a higher magnetic field , we show in the charge current density at magnetic field @xmath118 t in the peak at @xmath110 nm ( a ) and the dip at @xmath121 nm ( b ) of the net charge current shown in ( a ) ( red dotted ) .
an obvious explanation is : the perpendicular magnetic field reduces the effective barrier height and increases the splitting of the energy levels at the crossover energy or resonance energy ( not shown ) .
therefore the charge current density does not equally split between the waveguides or localize in only control - waveguide , but rather an inter - waveguide backward scattering is observed due to the lorentz force , consequently the net charge current is suppressed and we observe edge states forming in each of the waveguides . in this case
the double - waveguide system does neither play the role of a @xmath0-operation nor a not - operation qubit in the presence of a higher magnetic field . in the following
, we shall study the influence of the side qds in the dqw for different values of the magnetic field .
in this section , we consider a side dot embedded in each waveguide to enhance the inter - waveguide transport as is shown in .
the dots to the sides of the cw may be expected to increase the dwell time in the coupling region and change the resonance condition between the two waveguides .
the window - coupling dqw potential with side qds is described by @xmath122 where @xmath123 is the side dot potential that is defined as @xmath124 with @xmath125 mev , @xmath126a@xmath127 , and @xmath128 nm@xmath98 implying the radius of each side - dot @xmath129 nm .
figure [ fig08](a ) shows the net charge current as a function of the cw length at @xmath114 ps in the presence of the coulomb interaction for different values of the magnetic field @xmath106 t ( blue solid ) , @xmath115 t ( green dashed ) , and @xmath116 t ( red dashed ) .
an oscillation in the net charge current is again established in a vanishing magnetic field @xmath130 t , while at higher magnetic field values @xmath115 and @xmath116 t some extra fluctuations in the current oscillation characteristic are observed .
the existence of fluctuations correlates with dynamic motion of the charge in the system in which electrons participate in an inter - waveguide backward or forward scattering at different cw length .
if the magnetic field is increased to @xmath131 t , such that the effective magnetic length is close to the dot radius @xmath132 the net charge current is suppressed to vanishing values ( not shown ) . in that case , varying of the cw length does not affect the electron transport in the system .
there are several reasons that lead to the almost vanishing net charge current at high magnetic field such as edge effects , inter - waveguide backward scattering and localized electrons in the dots as the effective magnetic length is close to the dot radius .
we present ( b ) to compare the left partial current ( pink dotted ) , the right partial current ( blue dashed ) to the net charge current ( blue solid ) at @xmath106 t. it can be seen that the partial currents have the same oscillation features and develop a peak and a current dip at the same cw lengths .
versus the window - coupling length @xmath1 in the presence of the coulomb interaction for different magnetic field @xmath106 t ( blue solid ) , @xmath115 t ( green dashed ) , @xmath116 t ( red dotted ) , @xmath133 t ( pink dotted ) at time @xmath114 ps .
( b ) the current from the left lead @xmath117 ( pink dotted ) , the current into the right lead @xmath68 ( blue dashed ) , and the net charge current @xmath104 ( blue solid ) for @xmath106 t versus the cw length .
the chemical potentials are @xmath84 mev and @xmath85 mev implying @xmath109.,title="fig:",scaledwidth=23.0% ] versus the window - coupling length @xmath1 in the presence of the coulomb interaction for different magnetic field @xmath106 t ( blue solid ) , @xmath115 t ( green dashed ) , @xmath116 t ( red dotted ) , @xmath133 t ( pink dotted ) at time @xmath114 ps .
( b ) the current from the left lead @xmath117 ( pink dotted ) , the current into the right lead @xmath68 ( blue dashed ) , and the net charge current @xmath104 ( blue solid ) for @xmath106 t versus the cw length .
the chemical potentials are @xmath84 mev and @xmath85 mev implying @xmath109.,title="fig:",scaledwidth=23.0% ] in order to understand the characteristics of net charge current , we refer to the energy spectrum and the charge current density .
figure [ fig09 ] shows the me - energy versus the window - coupling length at @xmath106 t including one - electron states ( 1es , red dots ) and two electron states ( 2es , blue dots ) .
it should be noted that more energy levels get into resonance at @xmath107 nm ( left green rectangular ) in the presence of the side - gate dots which indicates a stronger coupling between the control- and the target - waveguide . at larger cw length such as @xmath111 nm ( right green rectangular ) ,
the energy splitting increases resulting in different electron motion in the system .
including one - electron states ( 1es , red dots ) and two electron states ( 2es , blue dots ) at @xmath106 t. the one - electron states in the left green rectangular are close to be in resonance , but the one - electron states in the right green rectangular are not .
the chemical potentials are @xmath112 and @xmath113 ( black ) implying @xmath109.,title="fig:",scaledwidth=35.0% ] + the electron motion in the waveguides is completely changed in the presence of the side qds .
figure [ fig10 ] displays the charge current density at @xmath106 t and @xmath107 nm ( a ) and current - dip at @xmath134 nm ( b ) shown in ( a ) ( blue solid ) . in ( a ) the charge current density is displayed for @xmath107 nm . the charge in the control - waveguide is switched to the target - waveguide , thus a not - operation is realized . comparing to the system without side - gate dots in ( b ) , the inter - waveguide transport is enhanced due to a stronger coupling between the waveguides at the resonant - energy levels at @xmath107 nm . therefore , the side qds play an essential role to convert the qubit operation in the system .
in addition , the inter - waveguide backward scattering here reduces the efficiency of the qubit .
as we mention previously , the existence of inter - waveguide backward scattering might be due to symmetric waveguides while the backward - scattering in an asymmetric waveguides can be avoided or reduced @xcite .
figure [ fig10](b ) shows the charge current density at the current - dip @xmath134 nm . the incoming charge into the control - waveguide encounters the control - qd leading to electron back - scattering in the control - waveguide , thus the net charge current is suppressed forming a current - dip .
we can say that the side - dots play a scattering role in the waveguides in such way that the electrons partially get forward- and backward - scattered in and into both waveguides .
ps in the presence of the coulomb interaction in the peak @xmath135 nm ( a ) , and the dip @xmath134 nm ( b ) of the net charge current of ( a ) in the case of @xmath106 t. other parameters are @xmath136 nm , and @xmath129 nm.,title="fig:",scaledwidth=45.0% ] + ps in the presence of the coulomb interaction in the peak @xmath135 nm ( a ) , and the dip @xmath134 nm ( b ) of the net charge current of ( a ) in the case of @xmath106 t. other parameters are @xmath136 nm , and @xmath129 nm.,title="fig:",scaledwidth=45.0% ] by tuning the external magnetic field to a higher value , @xmath137 t , the electrons manifest different dynamical motion in the waveguides . in the charge current density is shown in this stronger magnetic field @xmath118 t at the cw length @xmath138 ( a ) and @xmath108 nm ( b ) .
the graphs show circular motion , edge states , in each waveguide .
in ( a ) is shown the current charge density at @xmath107 nm , the electrons in the control - waveguide do not totally switch to the target - waveguide , because of increased energy level splitting at the resonance - energy in the presence of higher magnetic field .
in addition , charge is partially localized in the control - qd as the effective magnetic length approaches the radius of the quantum dot @xmath139 .
ps in the presence of the coulomb interaction in the peak @xmath135 nm ( a ) , and the dip @xmath134 nm ( b ) net charge current of ( a ) in the case of @xmath118 t. other parameters are @xmath140 nm , and @xmath129 nm.,title="fig:",scaledwidth=45.0% ] ps in the presence of the coulomb interaction in the peak @xmath135 nm ( a ) , and the dip @xmath134 nm ( b ) net charge current of ( a ) in the case of @xmath118 t. other parameters are @xmath140 nm , and @xmath129 nm.,title="fig:",scaledwidth=45.0% ] in ( b ) , the electrons are mostly localized in the control - qd corresponding to the current - dip at @xmath141 nm .
this electron localization is expected because at a larger cw length the splitting of energy levels increases and leads to suppression the inter - waveguide scattering .
therefore , the electrons move along the control - waveguide and get localized in the control - qd at this higher magnetic field .
we have presented a model for a window coupled double quantum waveguide in an external perpendicular magnetic field .
the dqw is weakly connected to two leads with different chemical potentials in which a non - markovian master equation is utilized to investigate coherent switching in electron transport between the waveguides for the implementation of quantum logic gates .
a comparison between the charge current and the left and right currents in figures [ fig06 ] and [ fig08 ] at the time point in time @xmath142 ps shows that the system has not completely reached a steady state yet , but the sought after function of the dqw is already present in the late transition regime .
this is important .
the exact steady state takes a long time to reach due to a low rate of charging for 2 and 3 electron states .
valuable time can be gained by operation of a device in the transition regime . by tuning the length of the cw ,
we have demonstrated two important physical characteristics of the waveguide system : resonant energy - levels and oscillations in the net charge current .
the resonant energy levels indicate a strong coupling between the dqw and the oscillations denote that the resonance conditions are governed by the length of the cw . in the case of an ideal non - coulomb - interacting window - coupled dqw system ,
the charge current density splits equally between the waveguides at resonant energies , therefore , the net charge current reaches it s maximum value and the waveguide system works as a @xmath0-operation qubit . in the presence of strong magnetic field , the inter - waveguide backward scattering is enhanced , and the efficiency of the logic gate decreases .
this inter - waveguide forward scattering at resonant levels without a magnetic field is not seen without including two - electrons states in the model .
it is strongly reduced by the coulomb interaction that lifts the 2es out of the group of active states in the transport .
this aspect can though be controlled by the exact system size and design as the coulomb blocking is reduced in a larger system , or by using a higher bias window or different material .
our models only includes sequential tunneling of electrons from the leads to the system , but the addition of higher order tunneling would only increase the importance of the many - electron structure and the coulomb interaction on the operation of the system . two parallel waveguides or dots can be coupled in many distinct ways .
moldoveanu@xcite et al .
attain it capacitively , while zibold@xcite et al .
fine tune the system by tunneling coupling .
both groups stress the strong dependence of the device on its exact geometry and the structure and correlations of many - electron states that are active in the transport .
our investigation with a different coupling scheme for the waveguides confirms this sensitive dependence . in the presence of side qds ,
more energy - levels of the two waveguides are brought into resonance which can lead to stronger coupling between them .
effectively , the dots can increase the density of states around the resonant transport states .
the electrons from the control - waveguide switch to the target - waveguide indicating a not - operation qubit .
but on the other hand , the side - dots can also increase the inter - waveguide backward scattering and reduce the qubit efficiency . in a stronger external magnetic field ,
the electrons get localized in the dots as the effective magnetic length is comparable to the radius of the side dot . in this case , the geometry of the cw does not affect the electron transport anymore .
the external magnetic field here is considered static , but one more variable to influence the transport is the shape of the side qds in the waveguides . we show that the varying of the cw length between the waveguides can specify the type of the qubit logic gate to be implemented by the system .
this work was financially supported by the icelandic research and instruments funds , the research fund of the university of iceland , the nordic high performance computing facility in iceland , and the national science council in taiwan through contracts no.nsc100-2112-m-239-001-my3 and no . most103 - 2112-m-239 - 001-my3 . 50 ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevb.89.115417 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.89.117901 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physreva.72.032330 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase
10.1103/physreva.70.052330 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.84.5912 [ * * , ( ) ] @noop * * , ( ) http://stacks.iop.org/1367-2630/11/i=7/a=073019 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.81.155442 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.195325 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.085311 [ * * , ( ) ] @noop ( ) link:\doibase
10.1103/physreva.3.1723 [ * * , ( ) ] @noop * * ( ) @noop * * , ( ) @noop link:\doibase 10.1103/physrevb.76.195301 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physreva.69.042303 [ * * , ( ) ] | we investigate coherent electron - switching transport in a double quantum waveguide system in a perpendicular static or vanishing magnetic field .
the finite symmetric double waveguide is connected to two semi - infinite leads from both ends .
the double waveguide can be defined as two parallel finite quantum wires or waveguides coupled via a window to facilitate coherent electron inter - wire transport . by tuning the length of the coupling window ,
we observe oscillations in the net charge current and a maximum electron conductance for the energy levels of the two waveguides in resonance .
the importance of the mutual coulomb interaction between the electrons and the influence of two - electron states is clarified by comparing results with and without the interaction .
even though the coulomb interaction can lift two - electron states out of the group of active transport states the length of the coupling window can be tuned to locate two very distinct transport modes in the system in the late transient regime before the onset of a steady state .
a static external magnetic field and quantum - dots formed by side gates ( side quantum dots ) can be used to enhance the inter - waveguide transport which can serve to implement a quantum logic device .
the fact that the device can be operated in the transient regime can be used to enhance its speed . |
the high energy frontier will soon be explored by four detector experiments recording the results of collisions of the large hadron collider ( lhc)located at the european center for nuclear research ( cern ) in geneva , switzerland . designed to record ,
identify , and study the higgs boson and a wide array of potential new physics signatures as well as @xmath0 mesons and quark - gluon plasma , these experiments ultimately hope to observe an inconsistency between nature and the standard model ( sm ) of high energy physics .
however , since the typical production and detection probabilities for interesting processes are eleven orders of magnitude smaller than that of uninteresting sm backgrounds , these experiments must sift through forty million events per second while the lhc is colliding beams , recording roughly one in two hundred thousand events in order to accommodate bandwidth constraints .
the resultant 1 to 2 petabytes ( pb ) of monthly data must then be processed and analyzed offline into a form which is suitable for extraction of measurements .
the unprecedented scale of the required computing resources and the complexity of the computing challenges have made computing an important element of hep .
each lhc experiment has deployed its own `` computing model '' ( cm ) , which consists of several classes ( `` tiers '' ) of facilities which together comprise a grid of resources held bound by fast links and special middleware . in atlas experiment
, the tier-0 facility at cern will perform the first - pass processing of the data .
the ten national tier-1 facilities will reprocess these data with better calibrations within two months after data collection .
meanwhile , the roughly thirty tier-2 facilities placed at specific universities and labs will focus on simulations and data analysis .
in addition , considerable effort is directed towards development of : * monte carlo tools ( which link theory and experiment ) , * detector simulation frameworks , * algorithmic and statistical analysis tools , * data processing frameworks and algorithms , * grid middleware , * data production and management systems , and * underlying data persistency and database infrastructure .
ultimately the performance of these software components , many of which are used by multiple experiments , determines the scale of computing resources required by hep experiments . despite the impressive scale of the lhc computing grid and the sophistication of underlying software technologies ,
a dearth of computing resources will be one of the primary bottlenecks in extracting measurements from lhc data .
for example , figure [ fig : cm ] shows the percentage of atlas tier 2 resources required in 2010 for fast and full simulation production as function of the fraction of recorded data .
the atlas cm model expects that physics analysis activity will require roughly half of the resources at atlas tier 2s , leaving the other half for simulation .
but we see that the allocating the nominal 50% of tier 2 resources to simulation limits the volume of fully simulated data to roughly 20% of the data the atlas detector will collect .
this deficiency will fundamentally limit the significance of the comparisons between recorded data and theoretical predictions ( via detailed detector modeling ) which are necessary to make any statements about nature . therefore atlas
must rely on less accurate fast simulations to produce monte carlo statistics comparable to the recorded data . and
if data needs to be resimulated , as is often the case in the first years of an experiment , cpu resources must be borrowed from analysis activity , thereby stalling extraction of measurements .
in addition , the experiment cms do not typically provide physicists the necessary resources for the cpu - intensive activities which in the past decade have come to characterize analyses of hep data at the tevatron and the @xmath0 factories .
these activities include sophisticated fits , statistical analysis of large `` toy '' monte carlo models , matrix element calculations , and use of the latest discriminant techniques , such as boosted decision trees .
physicists must therefore rely on leveraged resources and emerging technologies to accomplish such tasks .
in tandem with these developing needs in high - energy physics , the role of computing in both business and everyday life has evolved significantly . in the early 1990s ,
the html protocol was developed in the context allowing the hundreds of collaborating physicists in hep experiments at cern to communicate with their global group of colleagues .
this development was crucial to the later transformation of the internet from an academic tool to a global medium . today
, computing is seen as a commodity , and a critical component of the world economy .
major companies like google and amazon manage huge data centers and sell cpu cycles , disk , and bandwidth by the minute and megabyte .
regular innovations in data processing , delivery , organization , and communication continuously drive significant business and social progress .
figure [ fig : googletrends ] compares the google search volume for `` grid computing '' with technologies such as solid - state drives ( ssds ) , general purpose graphics processing units ( gpgpu ) , virtualization , and cloud computing .
we see that while hep has been building large and expensive grid sites along with the necessary grid software , virtualization and cloud computing have developed a wider appeal .
fortunately , efforts to take advantage of these technologies have recently begun in hep .
virtualization is likely to help address problems in hep computing that can be traced back to the fact that hep experiment software generally require specific operating systems ( os ) and are typically difficult to install .
virtualization provides a means of providing single file ( ie a virtual hard drive image ) which can be pre - packaged with all necessary software , including the os , and can run on any modern `` host '' machine regardless of the host hardware and os .
the first benefit of virtualization is that average physicists can simply download such an file and then instantly be able to use their experiment s software on their personal computer without a great deal of expertise .
the cernvm @xcite effort have been particularily successful in this area .
similarly system administrators can use virtualization to simplify deployment of the complicated set of software packages required by grid sites .
this route is particularly attractive for tier 3 sites that do not necessary have a full - time system administrator .
but perhaps the most promissing application of virtualization is in the area of oppertunistic computing where idle desktop computers ( for example , in university offices , class - rooms , and labs ) , can be used to assist data simluation . the appeal of cloud computing is the promise of on - demand access to vast computing resources hosted by companies that can provide attractive pricing due to the economy of scale .
so , for example , a hep experiment can short - term lease huge cpu resources for data simulation .
it is noteworthy that cloud computing has an implicit reliance on virtualization for delivering the appropriate software environment to the cloud resource . at this point , it is not clear that such cloud computing is appropriate for hep .
current cloud computing offerings such as amazon s ec2 are targeted for web - hosting rather than large data processing tasks and therefore offer prohibitive pricing and performance for hep applications .
what s more , the large cloud computing providers such as amazon and microsoft favor their own proprietry software solutions that lock their clients to their cloud and have avoided open efforts such the `` open cloud manifesto '' by ibm .
in addition some in the industry view the clound computing vision as an unrealistic `` myth '' .
nonetheless , many within the hep community are exploring the potential of hep computing .
for example nimbus now provides a means of turning amazon ec2 resources into a self - configuring cluster for hep computing . a very promising technology , that is now becoming cost - effective , is ssd .
because these storage devices have no mechanical or moving parts , they provide impressive input / output ( i / o ) rates , in particular huge gains in random read access in comparison to traditional hard drives ( hd ) .
two high i / o hep applications that can benefit from ssds immediately come to mind .
first is ntuple data analysis , a task that is typically characterized by rapid iterations over upto terabytes of data .
the second is in computing sites where hundreds of processes running on multiple cores access data stored on a single storage device .
since ssds present the same interface as hds , their deployment in hep environments is rather easy .
the primarly limitation is then the additional cost of the ssds , which we may expect to lead to ssd / hd hybrid solutions where hds are used for long - term storage and data is prestaged to ssds for faster access .
we may also note that while hep has traditionally interpreted moore s law as predicting faster processors , the industry has shifted to more cores per processor .
thus we find that though many hep applications in principle lend themselves to parallelization , very few existing hep software were fundamentally designed with parallel processing on multiple cores in mind .
recently , the omnipresence of dual and quad - core processors has prompted some parallelization efforts within the hep community .
these efforts take advantage of the embarrassingly parallel nature of hep computations by simply running multiple instances ( or threads ) of the software .
the only challenge then is sharing memory across these parallel threads in applications that have a large memory footprint such as simulation and reconstruction . assuming that memory cost is nt a factor ,
the problem then becomes the processor to memory bandwidth which can be saturated when a large number of cores access large amount of memory .
this bandwidth limitation constrains performance gain when scaling to large number of cores per processor . as a result ,
most hep multi - core optimization efforts are concentrated on processing forking and thread safety . while central processing units ( cpus ) have been evolving towards more cores , graphics processing
units ( gpus ) that were originally targeted to the personal computer gaming enthusiasts have evolved to be capable of computing traditionally performed on the cpu .
these modern processors that are also now present in many desktops and computers are known as the general purpose graphical processor units ( gpgpus ) . despite that fact that for specific computing tasks , these gpgpus can marshall thousands of simplified processing units in parallel in order to reduce computing times by orders of magnitude , they have yet to capture the attention of hep as a whole .
a very simple study of read / write rates of root @xcite analyses illustrates the potential of ssds and gpgpus on the data analysis iteration rate . for this study , we consider two simple root applications , one which creates an ntuple ( ttree ) with random data ( simple types such as bools , ints , floats , and vectors of simple types ) , and another that reads histograms all quantities in these ntuples . we find
that simple read / write rates stablilize with about 20 variables of each type per event ( 3 kb / event ) and 600 events .
table [ fig : ssdhd ] summarizes the results of the study . with
root s data compression turned off , a single instance of each application achieve approximately 25 mb / s read or write rate on a hard drive which provide 70 mb / sec sequential read .
with compression turned on ( providing 30% file size reduction ) , this figure falls to 4 and 16 mb / s read and write , respectively , illustrating that ( de)compression is the main bottleneck in input / output bound root analyses .
ideally , the gpu can be used to eliminate this bottleneck . in order to observe the benefits of the fast random access of ssds
, we ran eight instances of these applications with the data stored on a single hd or ssd .
we also observe that only uncompressed data writing appears to be limited by disk access . and though ssds generally provide faster rates , the improvement is significantly less where the i / o is limited by ( de)compression . [ cols="<,^,^,^,^ " , ] [ fig : ssdhd ]
from the cell processor in the playstation 3 ( ps3 ) to in newer - generation graphic processing units ( gpus ) used in desktop and laptop computers , we find the building blocks for high - performance computing ( hpc ) systems already present in devices we daily use .
originally driven by the gaming industry , gpu architectures have recently been developed to also support general - purpose computing .
these gpus offer impressive power consumption / performance and price / performance ratios .
the omnipresence of these general purpose gpus have pushed industry leaders such as microsoft and apple into a race to develop strategies that take advantage their capabilities .
the raw computational horsepower of gpus is staggering .
a single modern gpus provides nearly one tflops ( @xmath1 floating - point operations per second ) @xcite , roughly 20 times more than a typical multi - core cpu .
what s more , the trend over the past decade exhibits an exponential growth in the ratio of the computational power of gpu to cpu .
while use of computer graphics hardware for general - purpose computation has been an area of active research for many years ( eg @xcite , @xcite @xcite ) , the wide deployment of gpus in the last several years has resulted in an increase in experimental research with graphics hardware .
some notable work include password cracking @xcite , artificial neural networks @xcite , solving partial differential equations ( pdes ) @xcite , line integral convolution and lagrangian - eulerian advection @xcite , and protein folding ( folding@home ) @xcite . in hep
, the developers of the fairroot demonstrated a two orders of magnitude acceleration of track fitting in high multiplicity environments using gpus @xcite .
currently , the primary players in the gpu arena are nvidia and amd ( through the purchase of ati ) .
nearly all of nvidia s gpus , from the low - end laptop gpu to the professional tesla line , can be programmed using their compute unified device architecture ( cuda ) , a parallel programming compiler and software environment designed for issuing and managing general purpose computations on the gpu through extensions to the standard c language .
there are also standard numerical libraries for fft ( fast fourier transform ) and blas ( basic linear algebra subroutines ) .
ati / amd has also developed a proprietary gpgpu hardware and software known as data parallel virtual machine ( dpvm ) .
however , their focus seems to now shifted to open computing language ( opencl ) , a gpu software standard originally developed by apple and then released to an open consortium which includes all the relevant companies as signatories @xcite .
the primary appeal of opencl is it s architecture independence which is achieved through run - time compilation of computing software kernels . the first publicly available implementation of opencl was recently released last week as part of apple s snow leopard operating system .
other , less popular , gpu architectures and software include clearspeed @xcite , brookgpu @xcite , amd stream computing , and sh @xcite . finally , a very promising upcoming product is intel s larabee , a x86-based many - core gpu which is rumored to be release later this year .
the processing model for gpu is very different than cpu .
whereas cpus are optimized for low latency , gpus are optimized for high throughput .
gpus are essentially stream processors , hardware that operates in parallel by running a single computation on many records in a stream at once .
the limitation is that each parallel computation may only process independent memory elements that can not share memory with others .
the result is that gpus are generally suitable for computations that exhibit specific characteristics
. they must be compute intensive with large number of arithmetic operations per i / o or global memory reference .
they must be amiable to data parallelism , where the same function is applied to all records of an input stream and a number of records can be processed simultaneously without waiting for results from previous records . and they must exhibit data locality , a specific type of temporal locality common in signal and media processing applications , where data is produced once , read once or twice later in the application , and never read again .
hep applications that are good candidates for gpgpu acceleration include monte carlo integration for matrix element methods , discriminant training or calculation in multivariate analysis , maximum - likelihood fitting , compression / decompression during data input / output , event generation , full or fast detector simulation , event reconstruction ( in particular for the high level trigger ) , and detector alignment .
the difficulty with employing gpgpus is that existing applications can not be simply rebuilt to run on gpgpus , but must rather treat the gpu as a co - processor .
the software must explicitly adopt a data - parallel processing model where the data is broken into chunks and independently processed by algorithms which are highly constrained in both their memory access and complexity .
a practical approach of gpgpu accelerating existing applications is to rewrite computational bottlenecks to prepare the data on the cpu , transfer it to the gpu memory , execute the computation on the gpu , and then transfer the results back .
even modest acceleration of detector simulation and reconstruction using gpgpu will have a significant impact on hep computing . for reconstruction , where thousands of tracks and hundreds of thousands calorimeter cells must be processed
, any gain directly translates to higher trigger output . for simulation , where thousands of particles must be propagated through detector and magnetic fields ,
gains translate into fewer computing resource requirements .
the practical time - scale for deployment of such gpgpu accelerated strategies is for lhc upgrade .
while strategies for gpgpu acceleration of tracking and calorimetry are rather straight - forward , the complexity of geant4 @xcite and the fact that it was nt written with parallelization in mind , make simulation a much more difficult problem . taking full advantage of gpgpus in simulation will likely only be possible in the next generation software ( perhaps geant5 )
one promising interim strategy is to employ multiple parallel geant4 threads running on the cpu ( see geant4 parallelization efforts of @xcite ) that offload specific calculations ( eg magnetic field extrapolation ) to a service that can batch perform the calculation using the gpu . | while in the early 90s high energy physics ( hep ) lead the computing industry by establishing the http protocol and the first web - servers , the long time - scale for planning and building modern hep experiments has resulted in a generally slow adoption of emerging computing technologies which rapidly become commonplace in business and other scientific fields .
i will overview some of the fundamental computing problems in hep computing and then present the current state and future potential of employing new computing technologies in addressing these problems . |
the detection of first - overtone carbon monoxide ( co ) emission near 2.3 in sn 1987a opened exciting new possibilities for the study of supernovae ( sne ) .
unfortunately , for nearly a decade , sn 1987a remained the sole supernova with detected molecular emission .
however , with the maturing of near - infrared ( nir ) spectrographs , observations of several sne in the late - time nebular phase have been made . as a result
, co emission has now been detected in other type ii supernovae : sn 1995ad @xcite , sn 1998s @xcite , sn 1998dl , and sn 1999em @xcite .
these observations suggest that molecule formation may be a common occurrence in type ii sne
. the study of molecular emission in sne can provide valuable information about the composition , explosion dynamics , and the late - time temperature evolution and energy balance in the ejecta .
for example , the detection of co formation can place constraints on the mixing in the sn ejecta . because co is quickly destroyed by the presence of ionized helium , the detection of co emission in a type ii or ib supernova implies either that the co is not microscopically mixed with helium or that the helium is not ionized @xcite . in the case of a type ic supernova
there is likely little or no helium left in the outer envelope at the time of core collapse . as a result ,
co emission in a type ic does not place strong constraints on the mixing between ejecta layers .
due to its very large number of collisionally excitable energy levels , co can be an important coolant of the sn ejecta .
indeed , at temperatures of a few thousand k , co emission may be the dominant cooling mechanism @xcite .
co cooling may , in turn , play a key role in dust formation in supernova ejecta .
sn 1987a and sn 1998s , two supernovae with detected co emission , also showed strong evidence of dust formation in the ejecta . in sn
1987a , thermal emission from dust was detected in the mid and far - infrared while , at the same time , line emission from the ejecta shifted to the blue ( @xcite , and references therein ) . similarly , late - time spectra of sn 1998s exhibited h and he lines with a multi - peak line profile , the red side of which faded dramatically with time , while the blue side remained nearly constant . in both cases ,
the changes in line emission were attributed to the formation of dust in the ejecta , which obscured the far side of the supernova causing the redshifted emission to fade . thermal emission from dust
was also seen in sn 1998s @xcite , but this was not due to dust formation in the ejecta . in this case , the emission was likely from pre - existing dust in an extended cloud which was heated by x - rays and uv light from a strong interaction between the supernova and dense circumstellar gas @xcite . analysis of near - infrared co emission can also provide information about the co - rich ejecta .
for example , the overall shape of the co emission profile can be used as a temperature diagnostic for the co emitting gas ( e.g. @xcite ) .
the fine structure in the near - infrared is formed by transitions between different vibrational levels , with higher level transitions emitting at longer wavelengths .
as the temperature decreases , the higher vibrational levels become de - populated and emission from the red end of the co profile decreases . for high s / n data , it is possible to determine the velocity of the co emitting gas from the shape of the co profile @xcite . at low velocities , the band structure of the co emission
is well defined and the blue edge of the emission profile is quite sharp .
as the expansion velocity increases the bands blend together , smearing out the pattern of peaks and troughs .
also , the short - wavelength end of the profile creeps farther toward the blue and the rise becomes progressively shallower .
high velocity co can have important implications for the progenitor star . to accelerate a significant amount of co out to high velocity in a type ii supernova
, the progenitor must be massive ( @xmath0 for @xmath1 km s@xmath2 ) and must have lost a large portion of its h and he rich mantle prior to core - collapse @xcite .
the large progenitor mass is required to build up a large carbon / oxygen layer which will lie far enough out in the ejecta to accelerate to high velocity . in a smaller progenitor , the c / o rich layer will be buried too deeply beneath the outer mantle . on the other hand , for stripped - envelope supernovae , where the pre - collapse mass loss becomes extreme , the requirement for a massive progenitor may be relaxed .
for instance , in a type ic supernova there is likely little or no mass left above the c / o rich layer and quite large c / o velocities might be expected from a fairly low mass progenitor .
however , if the c / o velocity becomes too high , then the density might drop too fast for a significant amount of co to form . here
we present the first detection of co in a type ic supernova , sn 2000ew . in 2 ,
we describe the observations and data reduction . in
3 we present the data and compare the co emission observed in sn 2000ew to that seen in the nir spectrum of sn 1987a .
we also discuss the conclusion of @xcite that both co emission and an unidentified feature seen in the spectrum of sn 1987a near 2.26 are ubiquitous features in the nir spectra of type ii sne .
sn 2000ew was discovered in ngc 3810 on 28 nov 2000 by t. puckett and a. langoussis with the puckett observatory 0.30-m automated supernova patrol telescope @xcite .
sn 2000ew was originally classified as a type ia supernova by @xcite , but was subsequently found to be of type ic @xcite . according to vsnet white light magnitude estimates
, it reached a maximum of @xmath3 mag around 4 dec 2000 .
the recession velocity of ngc 3810 , corrected for virgocentric infall , is 1053 km s@xmath2 ( taken from the lyon - meuden extragalactic database . )
using h@xmath4 km s@xmath2 mpc@xmath2 @xcite yields an approximate distance to sn 2000ew of 15 mpc and implies a maximum light brightness of @xmath5 mag .
this is somewhat faint for type ic supernovae which typically show peak magnitudes around @xmath6
@xmath7 @xcite .
however with the crudeness of our magnitude estimate it is difficult to determine if sn 2000ew is unusually sub - luminous or just on the faint end of the luminosity spectrum for type ic sne . on 5 mar 2001
, we obtained a 1.92.5 spectrum of sn 2000ew using the infrared camera and spectrograph ( ircs ; @xcite ) on the 8.2 m subaru telescope .
a 06 slit was used yielding a resolution of about 850 km s@xmath2 .
the spectrum was collected in eight 400 s exposures dithered between two positions along the slit in an abba pattern .
first order sky subtraction was obtained by subtracting ` b ' frames from ` a ' frames .
1-d spectra were then extracted using standard iraf tasks .
the spectra were wavelength calibrated using an argon lamp spectrum obtained at the end of the night , and the calibration was tested by measuring the wavelengths of night - sky lines in the data .
the data was corrected for telluric absorption using nearby av and gv stars selected from the gemini spectroscopic standard star catalogues .
the g - dwarf was divided by a solar spectrum @xcite to remove stellar features .
the result was used to correct for telluric absorption in the a dwarf spectrum .
the stellar features were fit from the corrected a - star spectrum and removed from the raw a - star spectrum . the raw a - star spectrum with the stellar features removed was then used to correct the target data for telluric absorption .
( see @xcite for a more detailed discussion of this technique . )
the instrumental response was removed by matching the observed spectra of the a - star telluric standards to model spectra from the stellar atmosphere calculations of @xcite .
the absolute flux level was set using nir photometry taken a week later on 12 mar 2001 using tifkam @xcite on the 2.4 m hiltner telescope at mdm observatory .
this night was not photometric and the sn brightness was measured relative to a field star 93 s and 26 e of the supernova .
boot - strapping the nir photometry of this star from calibrated images taken in jan 2001 resulted in @xmath8 for sn 2000ew on 12 mar 2001 .
this was then converted to flux units using the zero - points of @xcite , and the flux was set by matching the flux of the spectrum averaged over the combined filter and atmospheric bandpass of @xcite .
we estimate that the absolute fluxing of the spectrum is accurate to 20% .
figure [ sn00ew_fig1 ] shows the full 1.92.5 spectrum of sn 2000ew on 5 mar 2001 , some 97 days after discovery and about 90 days after maximum light .
the spectrum has been doppler shifted to remove the 993 km s@xmath2 heliocentric radial velocity of ngc 3810 @xcite .
strong emission from the first - overtone band of co is seen from 2.29 to the red end of the spectrum at 2.5 .
the spectrum also shows three broad emission features near 1.94 , 1.98 and 2.22 .
we identify the 1.94 and 1.98 features as emission from the 4p@xmath9p@xmath103d@xmath9d multiplet , and the 2.22 feature as 4s@xmath11s4p@xmath11p@xmath10 .
these features were identified in the spectrum of sn 1987a @xcite .
all three lines have a fwzi of about 4000 km s@xmath2 , peak around 500 km s@xmath2 and exhibit relatively boxy profiles that fade slightly from the blue to the red .
in addition to these broad emission lines , the spectrum also contains two narrow emission features at 2.04 and 2.06 which we identify as [ ] 2.0466 and 2.0587 , respectively .
both lines are unresolved and the narrowness of these features suggests that they are associated with circumstellar emission rather than supernova ejecta .
both lines have been identified in clumps of dense circumstellar gas in the cas a supernova remnant , along with two other lines in the spectral region covered here : 2.1661 ( br@xmath12 ) and [ ] 2.2244 @xcite .
the latter [ ] line would be lost in the broad feature , but the br@xmath12 line lies in a relatively empty region of the spectrum and no trace of this line can be seen in our spectrum of sn 2000ew .
this suggests that the circumstellar clumps detected here may be hydrogen poor , perhaps originating in the he mantle of the progenitor star and shed prior to core - collapse .
such an interpretation is consistent with the general picture that type ic progenitors shed a substantial portion of their helium rich mantle prior to core collapse ( @xcite ; @xcite , and references therein ) .
the lack of hydrogen emission also suggests that the narrow lines are not associated with a background region or other ism gas in the host galaxy .
lcccc event & sn type & epoch of last & epoch of first & reference + & & co non - detection & co detection + + sn 1987a & iip pec & 112 d & 192 d & 1 + sn 1995ad & iip & & 105 d & 2 + sn 1998s & iin & 44 d & 109 d & 3 + sn 1999dl & iip & & 152 d & 4 + sn 1999em & iip & 118 d & 178 d & 4,5 + sn 1999gi & iip & 74 d & 126 d & 5 + sn 2000ew & ic & 39 d & 97 d & 6,7 + + sn 1990w & ic & 90 d & & 8 + sn 1995v & iip & 84 d & & 9 + sn 2001b & ib & 60 d & & 7 + * references : * ( 1 ) meikle et al .
( 1989 ) ; ( 2 ) spyromilio & leibundgut ( 1996 ) ; ( 3 ) fassia et al .
( 2001 ) ; ( 4 ) spryomilio , leibundgut & glimozzi ( 2001 ) ; ( 5 ) gerardy & fesen 2002 , in prep . ;
( 6 ) gerardy et al .
( 2002 , this paper ) ; ( 7 ) gerardy et al .
2002 , in prep .
( 8) wheeler et al . (
1994 ) ; ( 9 ) fassia et al .
( 1998 ) ; @xcite have concluded that co formation is common in type ii supernovae . with the discovery of co in the type ic event
sn 2000ew , we have the first evidence that this conclusion might be extended to include core - collapse supernovae in general .
table [ sn00ew_tab1 ] lists all the core - collapse supernovae for which _
k_-band spectra have been published to date as well as a couple of objects for which papers are in preparation . for each object
, we list the epoch of the last non - detection of co ( if there were any ) , as well as the epoch that co was first detected .
the epochs quoted are relative to the supernova discovery date or the date of the earliest pre - discovery detection . of the 10 objects with _
k_-band spectra , seven have co detections .
furthermore , it may be that the three non - detections were simply not observed at a late enough epoch .
the results in table [ sn00ew_tab1 ] suggest that co tends to form around 100200 days , and none of the three objects without detections were observed past 100 days .
a quantitative analysis of co emission in sne requires a good understanding of the chemical and kinematic structure of the ejecta , as well as detailed modeling of the co formation , destruction , and excitation processes .
( e.g. @xcite . )
unfortunately , such an analysis for sn 2000ew requires more information than is available in our single spectrum .
furthermore , the s / n of the data does not place strong constraints on the fine - structure in the co bands
. we will therefore restrict ourselves to a very basic analysis of the co emission along with a comparison to the spectrum observed in sn 1987a .
the shape of the co profile provides a lower limit on the temperature of the emitting region .
lte modeling of co has shown that for temperatures below about 2000 k , the first two bandheads have comparable strengths , but higher order bandheads are down by at least 2030% @xcite .
this creates a noticeable break in the spectrum redward of the second bandhead near 2.32 .
since no break is evident in the spectrum of sn 2000ew , we conclude that the temperature is at least 2000 k in the co emitting region .
it is useful to compare the observed co emission in sn 2000ew with the much higher s / n co detection in sn 1987a .
figure [ sn00ew_fig2 ] shows the co emission from sn 1987a on day 192 @xcite overplotted on our spectrum of sn 2000ew .
the sn 1987a data have been shifted vertically and scaled to match the continuum and co brightness of sn 2000ew .
both spectra are plotted in the rest frame of the host galaxy .
longward of 2.28 , the sn 2000ew data matches the sn 1987a data fairly well .
the bandhead structure is not visible in sn 2000ew but this is likely due to the low s / n of the sn 2000ew detection rather than velocity smearing .
in fact , the blue edge of the co feature in sn 2000ew matches that in sn 1987a quite well , suggesting that the co has a similar velocity in both objects : @xmath13 km s@xmath2 ( e.g. , @xcite ) . while a 2000 km s@xmath2 co velocity is not unexpected for a type ii supernova like sn 1987a , it is quite low for some hydrodynamic models of type ic sne .
spherically symmetric hydrodynamic explosion models of the type ic sn 1994i predict a minimum velocity for the c / o rich layer of around 7000 km s@xmath2 @xcite .
for such a model to be consistent with an observed co velocity of 2000 km s@xmath2 would require a kinetic energy release well below @xmath14 erg . on the other hand , the low co velocity may be consistent with strongly non - spherical jet - driven explosion models . in the asymmetric explosion models used by @xcite to analyse the type ic hypernova sn 1998bw , ejecta velocities are large along the jet , but much more modest perpendicular to the jet axis . in their @xmath15 erg model
, the c / o rich gas has a minimum velocity around 3000 km s@xmath2 . with
a lower kinetic energy release such a model might be able to account for the low co velocities observed in sn 2000ew . as shown in figure [ sn00ew_fig2 ]
, sn 1987a exhibited an emission feature peaked near 2.265 which is absent in the sn 2000ew spectrum .
to date , this line has not been satisfactorily identified .
@xcite identify the feature as co@xmath16 but detailed models of co chemistry in sn 1987a have ruled out this identification @xcite .
furthermore , this feature did not evolve with the co bands in sn 1987a , remaining visible long after the co emission had faded @xcite .
@xcite ( @xcite , hereafter slg ) suggest that the sn 1987a feature is `` ubiquitous '' in type ii supernovae and must be accounted for when measuring the velocity of co emission .
however , the presence of this 2.26 feature in other supernovae is somewhat uncertain .
detections have been reported for sn 1995ad @xcite , sn 1998dl , and sn 1999em ( slg ) . however , in all three cases the purported detections are not clear features , but rather apparent excesses of emission above the continuum level which depend critically on where the continuum baseline is defined in these low s / n spectra . furthermore , this feature is not seen in the co spectrum of the type iip sn 1999gi ( gerardy & fesen 2002 , in prep . ) nor is it seen in our co spectrum of sn 2000ew , which otherwise shows many similarities to sn 1987a and other type ii sne .
as evidence of the presence of the 2.26 sn 1987a feature in other sne , slg reported that this feature was blended with the blue edge of the co feature in the 20 june 1998 spectrum of sn 1998s presented by @xcite ( hereafter gfhw ) .
slg referenced figure 6 of gfhw which is reproduced here as figure [ fig3 ] .
this figure shows the co detection of sn 1998s with that of sn 1987a overplotted .
the sn 1987a spectrum has been shifted vertically and scaled to approximately match the continuum level and co brightness of the sn 1998s spectrum , and the wavelengths of both spectra have been shifted to the rest frame of the supernovae . near 2.4
the sn 1998s spectrum is somewhat fainter than the sn 1987a .
gfhw express some concern that this may be due to residual telluric absorption in the data .
if real , the steeper profile is indicative of a somewhat lower temperature in sn 1998s compared to sn 1987a .
both spectra show excess emission to the blue of the co rise near 2.8 .
gfhw interpret the emission as the extended blue wing of a high velocity co profile , but admit that there may be an unidentified line contributing to this emission .
they noted that the emission can not be due to the 2.26 feature as that feature appears farther to the blue in sn 1987a .
slg reached the opposite conclusion , stating that the emission in sn 1998s was the same feature as that seen in sn 1987a , and attributed the difference in wavelength to the different recession velocities of the supernovae .
however , since both spectra in figure [ fig3 ] ( and figure 6 of gfhw ) have been doppler shifted to the supernova rest frame , we do not agree with their conclusion .
a sn 1998s spectrum obtained by @xcite casts further doubt on the presence of this 2.26 feature . in their spectrum
taken one day before that of gfhw they find little excess emission blueward of the sharp edge of the co bands and suggest that the differences between the two spectra are largely due to noise .
consequently , we suggest that the sn 1987a feature is not ubiquitous , and may in fact be relatively rare , with sn 1987a remaining the only unambiguous detection of this feature .
we have presented _ k_-band spectra of the type ic sn 2000ew which shows strong first - overtone emission of carbon monoxide .
this is the sixth core - collapse supernova for which a co detection has been published and it is the first detection in a non - type ii supernova .
the spectrum also shows narrow emission lines of [ ] and but is conspicuously lacking narrow ( br@xmath12 ) emission .
we interpret this emission as coming from dense clumps of hydrogen - poor circumstellar gas .
@xcite have concluded that co formation probably occurs in all type ii supernovae .
we present a table listing ten core - collapse sne for which _
k_-band spectra were obtained , seven of which show co emission .
co emission was typically first observed between 100 and 200 d , and the three sne without co detections were not observed past 100 d. the detection of co in the type ic sn 2000ew provides the first indication that co formation may also be common in stripped - envelope core - collapse sne ( types ib , ic , and `` iib '' ) .
the observed co emission profile in sn 2000ew is quite similar to that of sn 1987a , and indicates an expansion velocity around 2000 km s@xmath2 , which is surprisingly low for a type ic supernova .
the low observed co velocity may be another indication that strongly non - spherical explosion models are needed for type ic sne . to date
, there have been no observations that provide strong evidence for high velocity co emission .
an apparent blue wing in the co spectrum of sn 1998s presented by gfhw might be due to high velocity co. however a coeval spectrum taken by @xcite does nt show this blue extension to the co profile and they derive a much lower co velocity from their observations .
thus the co velocity in sn 1998s is uncertain . in all of the other co detections , the observed velocity is around 2000 km s@xmath2 or lower .
it could be that a low c / o velocity is a requirement for co formation .
co formation is a highly density sensitive process , and it is possible that in higher velocity gas the density decreases too fast to allow much co to form .
however , for most type ii supernovae , the c / o rich layers are buried too deep in the ejecta to expect co velocities much higher than 2000 km s@xmath2 .
looking for co emission in type ib / c supernovae will provide a better test as much higher c / o velocities are expected in these objects .
we find no evidence in sn 2000ew for the presence of a 2.26 feature seen in sn 1987a and , contrary to @xcite , we suspect it is either faint or absent in other type ii and type ib / c objects . sn 1987a remains the only clear detection of this feature , and its identification remains a mystery .
+ we would like to thank the subaru observatory staff for their excellent support , especially dr . hiroshi terada . c. l. g. and r. a. f.s research
is supported by nsf grant 98 - 76703 .
k. n. has been supported in part by the grant - in - aid forscientific research ( 07ce2002 , 14047206 , 14540223 ) of the ministry of education , culture , sports , and technology in japan .
research of jcw is supported by nsf grant 0098644 .
+ clocchiatti , a. , et al .
2000 , , 529 , 661 dennefeld , m. , & patris , j. 2000 , 7532 fassia , a. , meikle , w. p. s. , geballe , t. r. , walton , n. a. , pollacco , d. l. , rutten , r. g. m. , & tinney , c. 1998 , , 299 , 150 fassia , a. , et al .
2001 , , 325 , 907 filippenko , a. v. 1997 , , 35 , 309 filippenko , a. v. , chornock , r. , & modjaz , m. 2000 , 7547 freedman et al .
2001 , , 553 , 47 gearhart , r. a. , wheeler , j. c. , & swartz , d. a. 1999 , , 510 , 944 gerardy , c. l. , fesen , r. a. , hflich p. , & wheeler , j. c. 2000 , , 119 , 2968 ( gfhw ) gerardy , c. l. & fesen , r. a. 2001 , , 121 , 2781 gerardy , c. l. , fesen , r. a. , nomoto , k. , garnavich , p. m. , jha , s. , challis , p. m. , kirshner , r. p. , hflich , p. , & wheeler , j. c. , 2002 , , submitted gerardy , c. l. , et al .
2002 , in prep .
gerardy , c. l. , & fesen , r. a. 2002 , in prep .
hanson , m. m. , conti , p. s. , & reike , m. j. 1996 , , 107 , 281 hanson , m. m. , rieke , g. h. , & luhman , k. l. 1998 , , 116 , 1915 hawarden , t. g. , leggett , s. k. , letawski , m. b. , ballantyne , d. r. , casali , m. m. 2001 , , 325 , 563 haynes , m. p. , van zee , l. , hogg , d. e , roberts , m. s. , & maddalena , r. j. 1998 , , 115 , 62 hflich , p. , sharp .
c. , & zorec , j. 1989 , in : workshop on particle astrophysics , ed .
norman , e. b. ( singapore , world scientific ) , 205 iwamoto , k. , nomoto , k. , hflich , p. , yamaoka , h. , kumagi , s. , & shigeyama , t. 1994 , , 437 , 115 kobayashi , n. et al .
2000 , proc .
spie 4008 : optical and ir telescope instrumentation and detectors , eds .
m. iye & a. f. moorwood , 1056 kurucz , r. l. 1994 , cd - rom 18 , solar abundance model atmospheres for 0,1,2,4 and 8 km / s ( cambridge : smithsonian astrophys . obs . ) lepp , s. , dalgarno , a. & mccray , r. 1990 , , 358 , 262 liu , w. , dalgarno , a. , & lepp , s. 1992 , , 396 , 679 liu , w. , & dalgarno , a. 1995 , , 454 , 479 livingston , w. , & wallace , l. 1991 , nso tech . rep .
91 - 001 maeda , k. , nakamura , t. , nomoto , k. , mazzali , p. a. , patat , f. , & hachisu , i. 2002 , , 565 , 405 mccray , r. 1993 , , 31 , 175 meikle , w. p. s. , allen , d. a. , spyromilio , j. , & varani , g .- f .
1989 , , 238 , 193 meikle , w. p. s. , spyromilio , j. , allen , d. a. , varani , g .- f . &
cumming , r. j. 1993 , , 261 , 535 nomoto , k. , suzuki , t. , & iwamoto , k. 1995 , phys . rep .
256 , 173 pogge , r. w. , depoy , d. l. , atwood , b. , obrien , t. p. , byard , p. l. , martini , p. , stephens , a. , gatley , i. , merrill , k. m. , vrba , f. j. , & henden , a. a. 1998 , , 3354 , 414 puckett , t. , & langoussis , a. 2000 , 7530 schultz , a. , et al . 2001 , `` nicmos instrument handbook , '' version 4.1 ( baltimore : stsci ) .
sharp , c. & hflich , p. 1989 , highlights of astronomy 8 , ed .
d. mcnally , iau general assembly ( dordrecht : kluwer ) , 207 spyromilio , j. , meikle , w. p. s. , learner , r. c. m. , & allen , d. a. 1988 , , 334 , 327 spyromilio , j. , & leibundgut , b. 1996 , , 283 , l89 spyromilio , j. , leibundgut , b. , & gilmozzi , r. 2001 , , 376 , 188 ( slg ) wallace , l. , hinkle , k. , & livingston , w. 1993 , nso tech . rep .
93 - 001 wheeler , j. c. , harkness , r. p. , clocchiatti , a. , benetti , s. , brotherton , m. s. , depoy , d. l. , & elias , j. 1994 , , 436 , 135 | we present _ k_-band ( 1.9 2.5 ) spectra of the type ic sn 2000ew observed with ircs on the subaru telescope .
these data show the first detection of carbon monoxide ( co ) emission in a type ic supernova .
the detection of co in sn 2000ew provides further evidence that molecule formation may be a common occurrence in core - collapse supernova ejecta .
the spectrum also contains narrow emission lines of [ ] and probably from dense clumps of hydrogen - poor circumstellar gas surrounding sn 2000ew .
our spectrum of sn 2000ew shows no trace of an unidentified feature seen near 2.26 , just blueward of the co emission , in the spectrum of sn 1987a and we discuss proposed detections of this feature in other type ii supernovae . |
apart from their relevance to the description of ferrimagnetic materials , mixed - spin models are interesting from a purely theoretical point of view , being among the simplest models to exhibit tricritical behavior .
so they are especially convenient for studying the effects of inhomogeneities on the phase diagrams and the multicritical behavior of magnetic systems . from a few exact @xcite and several approximate @xcite calculations
, we now have a good picture of the phase diagrams of mixed spin-@xmath0spin-@xmath1 ising models in the presence of a crystal field .
our aim in this work is to use this model hamiltonian to investigate the effects of disorder on the location of the transition lines and the tricritical point .
the mixed - spin ising model is described as a two - sublattice system , with spin variables @xmath2 and @xmath3 , on the sites of sublattices @xmath4 and @xmath5 , respectively . restricting the interactions to nearest - neighbors ( belonging to different sublattices ) and single - ion terms , the most general spin hamiltonian in even spin space
is written as @xmath6 where the first sum is over nearest - neighbor pairs , the second sum is over the sites of sublattice @xmath5 , and the parameter @xmath7 is assumed to be positive ( ferromagnetic exchange ) . for @xmath8 ,
the crystal field favors the @xmath9 states ; the competition between anisotropy and ferromagnetic exchange terms leads to the appearance of a tricritical point .
it should be pointed out that one needs three parameters to describe the even space of the better known spin-@xmath1 blume - emery - griffiths ( beg ) model . in the present case of a mixed - spin ising system , from the point of view of the calculations ,
the reduction of the analysis to a two - parameter space is a particularly attractive feature .
there are exact calculations for the thermodynamic functions associated with the model hamiltonian given by eq .
( [ smp ] ) on a simple chain and on some three - coordinated two - dimensional structures . on a honeycomb lattice
, the problem can be mapped onto a spin-@xmath0 ising model on a triangular lattice , which does not display a tricritical point.@xcite this mixed - spin model can also be exactly solved on a bethe lattice @xcite ( the deep interior of cayley tree ) , leading to the same results of a recent cluster - variational calculation.@xcite the results on a bethe lattice with coordination @xmath10 indicate the absence of a tricritical point for @xmath11 , as confirmed by migdal
kadanoff renormalization - group calculations.@xcite in the infinite - coordination limit of the bethe lattice , one regains the well - known results for the tricritical point displayed by the curie
weiss ( mean - field ) version of the model .
an earlier approximate effective - field calculation @xcite predicted a tricritical point for @xmath12 , but this result has been recently challenged.@xcite in order to analyze the effects of disorder , we consider the hamiltonian @xmath13 where @xmath14 is a set of independent , identically distributed random variables associated with the binary probability distribution @xmath15 with this choice of disorder , and for @xmath16 , the ground state can be mapped onto a percolation problem in which the dilution affects the sites belonging to only one of the sublattices ( corresponding to spin @xmath17 ) .
this association is easy to see if we note that a uniform crystal field @xmath16 leads to @xmath9 for all @xmath18 , which breaks the connectivity between the spin-@xmath19 variables .
the presence of a randomly located distribution of @xmath20 crystal fields recovers that connectivity and , for sufficiently high values of @xmath21 , leads to the formation of a percolating cluster . in the rather artificial case of annealed disorder , on the honeycomb lattice
, there is also an exact solution @xcite for the thermodynamic properties of the mixed - spin model described by eqs .
( [ sphd ] ) and ( [ smprob ] ) .
it is interesting to remark that this solution in the annealed case reproduces the critical concentration of the percolation problem associated with the ground state of the model with quenched ( frozen ) disorder , which is equivalent to the usual percolation problem on the triangular lattice .
for the physically more relevant case of quenched disorder , there are approximate calculations using an effective - field theory with correlations,@xcite which point to the ( expected ) weakening of the tricritical behavior due to the presence of disorder . in the present work ,
we first analyze the temperature anisotropy ( @xmath22 ) phase diagram of the curie
weiss version ( mean - field limit ) of the spin hamiltonian given by eqs .
( [ sphd ] ) and ( [ smprob ] ) . depending on the concentration @xmath21
, there appear novel transition lines and multicritical points .
to include the effects of thermal fluctuations , we then resort to a standard self - consistent bethe peierls approximation ( which is analogous , in the case of the corresponding uniform model , to performing an exact bethe - lattice calculation ) . in section
ii , we present the curie weiss results . the bethe
peierls approximation is discussed in section iii . in the appendices , we report an exact solution in one - dimension as well as some low - temperature expansions to supplement the numerical results from the bethe peierls approximation .
the curie weiss version of the mixed - spin ising model is given by the hamiltonian @xmath23 where the sums are over all sites belonging to each one of the sublattices .
for a given disorder configuration @xmath24 , we calculate the partition function by performing a partial trace over the set of spin variables @xmath25 . in the thermodynamic limit
, we use the saddle - point method and average over disorder to obtain the free energy functional @xmath26\nonumber\\% & & -\frac{1}{2\beta}\int\wp(d_{b})\ln\left[% 1 + 2e^{-\beta{d_{b}}}\cosh(\beta{j\sigma})\right]dd_{b}.%\end{aligned}\ ] ] @xmath27 -\frac{1}{2\beta } \int\wp ( d_{b})\ln \left [ 1 + 2e^{-\beta d_{b}}\cosh ( \beta j\sigma ) \right ] dd_{b}.\ ] ] 2from the minimization of @xmath28 with respect to @xmath29 , we have the @xmath4-sublattice magnetization , @xmath30 , \ ] ] where the random variable @xmath31 satisfies the probability distribution in eq . ( [ smprob ] ) .
we can now calculate various expectation values .
for example , we have @xmath32 the critical line comes from the condition @xmath33 where @xmath34 and @xmath35 .
the thermodynamic stability of the critical line depends on the sign of the fourth derivative of @xmath36 at @xmath37 .
there is then the possibility of a tricritical point , given by the additional condition @xmath38 the stability of the tricritical point is determined by @xmath39 which means that the tricritical behavior is suppressed by disorder concentrations larger than approximately @xmath40 . in fig .
[ fig1 ] , we plot some @xmath41 phase diagrams for a set of typical values of the concentration @xmath21 . in the uniform case ( @xmath42 ) ,
there is just a tricritical point , @xmath43 . for @xmath44 ,
a tricritical point still exists ( see fig .
[ fig1 ] , for @xmath45 ) . however , at low temperatures and sufficiently large values of @xmath46 , there appears a low - density ( @xmath47 as @xmath48 ) ferromagnetic phase , which we call ferro - ii phase ; at fixed values of @xmath46 , an increase of temperature induces a second - order transition from the ferro - ii to the paramagnetic phase .
this transition is represented by a critical line that meets the first - order line at a critical end point , @xmath49 .
this critical end point separates the first - order line into two distinct regions : ( i ) at higher temperatures , there are transitions between the usual , high - density ( @xmath50 as @xmath48 ) ferromagnetic phase ( ferro - i ) and the paramagnetic phase ; ( ii ) at lower temperatures , the transitions are between the ferro - i and ferro - ii phases , the first - order boundary ending at a finite - temperature critical point , @xmath51 . for @xmath52 ,
the tricritical point is replaced by a critical end point and a simple critical point , separated by a first - order transition line between the ferromagnetic phases ( see inset in fig .
[ fig1 ] , for @xmath53 ) . for @xmath54 ,
the critical line is fully stable ( see fig .
[ fig1 ] , for @xmath55 ) .
however , for @xmath56 , there still exists a small finite - temperature region where there are ( first - order ) transitions between the ferromagnetic phases .
to give an estimate of the effects of thermal fluctuations , which are not accounted for by the curie
weiss calculations , we now resort to a standard self - consistent bethe peierls approximation .
as the model is defined on a bipartite lattice , we have to consider two distinct clusters of coordination @xmath10 ( see fig . [ fig2 ] ) . in one of them , which we call @xmath4 ,
the central site is occupied by a spin @xmath57 , connected to @xmath10 spins of the @xmath17 type . in the other cluster , called @xmath5 , there is a central @xmath17 spin surrounded by @xmath10 spin-@xmath0 variables . according to the standard prescription of the bethe
peierls approximation , we assume that the boundary spins of cluster @xmath58 are under the action of an effective magnetic field @xmath59 and an effective crystal field @xmath60 , while the boundary spins of cluster @xmath5 are in an effective magnetic field @xmath61 .
the crystal field acting on the central site of a cluster @xmath5 is a random variable @xmath31 .
we also consider external magnetic fields , @xmath62 and @xmath63 , acting on the central sites of clusters @xmath4 and @xmath5 , respectively . for the sake of simplicity , we assume that the same effective crystal field @xmath60 acts on all boundary spins of cluster @xmath4 ( and impose self - consistency between both thermal and disorder averages associated with the two clusters ) . in a more refined approach
, we might introduce different effective crystal fields to mimic the extended disorder of real materials .
however , we expect that , in the absence of external magnetic fields , the assumption of a single @xmath60 is reasonable , at least in the paramagnetic phase , where there is no long - range order ; in particular , we expect that the paramagnetic phase boundaries obtained by the two approaches are equivalent within the bethe
peierls approximation .
we will see that this is indeed supported by the calculations .
the partition functions associated with the two clusters are given by @xmath64 ^{q}+e^{-\gamma _ { a}}\left [ 1 + 2e^{-\tilde{\delta}}\cosh ( % \tilde{\gamma}_{b}-k)\right ] ^{q}\ ] ] and @xmath65 ^{q}+e^{-\delta _ { b}}\left\ { e^{\gamma _ { b}}\left [ 2\cosh ( \tilde{\gamma}_{a}+k)\right ] ^{q}+e^{-\gamma _ { b}}\left [ 2\cosh ( \tilde{\gamma}_{a}-k)\right ] ^{q}\right\ } , \ ] ] 2 where @xmath66 , @xmath34 and @xmath35 . the effective fields @xmath67 , @xmath68 and @xmath69 are determined by the consistency equations @xmath70 _ { \text{av}}=% \frac{\partial \ln z_{a}}{\partial \gamma _ { a}}=\frac{1}{q}\int \wp ( d_{b})% \frac{\partial \ln z_{b}}{\partial \tilde{\gamma}_{a}}dd_{b } , \label{smcem}\ ] ] @xmath71 _ { \text{av}}=\frac{1}{q}% \frac{\partial \ln z_{a}}{\partial \tilde{\gamma}_{b}}=\int \wp ( d_{b})\frac{% \partial \ln z_{b}}{\partial \gamma _ { b}}dd_{b } , \label{smcen}\ ] ] and @xmath72 _ { \text{av}}=-\frac{1}{q% }
\frac{\partial \ln z_{a}}{\partial \tilde{\delta}}=-\int \wp ( d_{b})\frac{% \partial \ln z_{b}}{\partial \delta _ { b}}dd_{b } , \label{smceq}\ ] ] where @xmath73 and @xmath74 _ { % \text{av}}$ ] indicate thermal and disorder averages , respectively .
we point out that the introduction of the effective crystal field @xmath60 is essential to achieve consistency between the equations for the two clusters . in order to analyze the critical behavior ,
it is convenient to choose the magnetization @xmath29 , the temperature @xmath75 , and the external fields @xmath63 and @xmath31 , as the independent thermodynamic variables .
thus , the external field @xmath62 is written as a function of those variables , and the second - order transitions in zero external field ( @xmath76 ) are given by @xmath77 where the derivative is taken for fixed values of the remaining independent variables , and @xmath78 to calculate the derivative in eq .
( [ smbcp ] ) , we take the ( implicit ) derivative of the consistency equations with respect to @xmath29 , imposing the condition @xmath37 and eliminating the derivatives involving @xmath79 , @xmath80 and the effective fields .
also , for @xmath37 , we have @xmath81 , since those variables are odd functions of @xmath29 for @xmath76 .
thus , the consistency equation for @xmath80 leads to @xmath82 and the final result is @xmath83 v_{0 } + ( q-1)^{2}v_{0}^{2}}{1+(q-2)v_{0}+(q-1)^{2}v_{0}^{2 } } , \label{derga}\ ] ] where @xmath84 . for @xmath85 , eq .
( [ derga ] ) reproduces the exact one - dimensional expression for the @xmath4-sublattice susceptibility ( see appendix [ app1 ] ) .
in fact , for @xmath85 , it is not difficult to check that we regain all the exact one - dimensional results .
it is easy to see that , in the uniform case , corresponding to @xmath86 , the critical line is given by @xmath87 \right\ } , \ ] ] in agreement with the results from the bethe - lattice @xcite and the cluster - variational @xcite calculations .
using the binary distribution in eq .
( [ smprob ] ) , we obtain @xmath88 therefore , the critical line is given by @xmath89 where @xmath90 in the @xmath48 ( @xmath91 ) limit , we have @xmath92 which has a real solution for @xmath93 if @xmath94 this last result should be anticipated for a bethe lattice , as we can see from the following arguments . consider a cayley tree where the sites belonging to every other shell , say those on odd - numbered shells , are occupied with probability @xmath21 , while the remaining sites are always occupied . if @xmath10 is the coordination of the tree , the average number of paths from the root ( shell @xmath95 ) to the first shell is given by @xmath96 , while we have @xmath97 paths from there to the second shell . according to this reasoning ,
the average number of paths from the root to the @xmath98th shell is given by @xmath99 . in order to have at least one path to the surface of the tree ( @xmath100 ) ,
it is required that @xmath101 , which is just the condition in eq .
( [ smpc ] ) .
this result , together with the reproduction of the exact one - dimensional solution , would suggest that the present treatment also gives exact results on the bethe lattice even in the presence of disorder .
however , as remarked in previous similar treatments,@xcite this works for the paramagnetic phase only , because only then it is correct to assume that all boundary sites are under the action of the same ( zero ) effective field .
the existence of a percolating cluster , which we do not take into account in this treatment , prevents this approximation from still giving correct results for the ordered phases .
we now consider eq .
( [ smclp ] ) , in the infinite coordination limit ( @xmath102 , @xmath103 , @xmath104 ) .
we then have @xmath105 which agrees with eq .
( [ smcwc ] ) for the curie
weiss version of the model .
the tricritical points are determined by eq .
( [ smbcp ] ) supplemented by the condition @xmath106 which is equivalent to @xmath107 where @xmath108 is given by @xmath109 the tricritical points are stable if @xmath110 to calculate this derivative , we again take implicit derivatives of the consistency equations ( up to fifth order ) with respect to @xmath29 , at @xmath111 , and eliminate all derivatives involving @xmath79 , @xmath80 and the effective fields . unlike the previous analysis
, we have not been able to obtain closed - form expressions for the stability condition of the tricritical point , but it is not difficult to perform a number of numerical calculations . for the uniform model
, we have @xmath112 .
therefore , eq . ( [ smbptri ] ) takes the form @xmath113 which is again identical to the result obtained from the bethe - lattice @xcite and the cluster - variational @xcite calculations . notice that this equation has real solutions only if @xmath114 .
thus , the bethe
peierls approximation does not predict a tricritical point for the square lattice ( @xmath115 ) . for the binary distribution in eq .
( [ smprob ] ) , we have @xmath116 .\ ] ] in the infinite - coordination limit we can write @xmath117 , \ ] ] which leads to the equation @xmath118 -2=0,\ ] ] at the tricritical point . indeed , one of the solutions of this equation corresponds to eq .
( [ smcwtri ] ) , for the curie
weiss version of the model , while the other solution represents a thermodynamically unstable situation . in table
[ tab1 ] , for various values of the coordination number @xmath10 , and using the binary distribution given by eq .
( [ smprob ] ) , we give the corresponding values of the concentration @xmath119 , at which the tricritical point becomes unstable , and the critical percolation concentration @xmath120 .
we see that , for @xmath121 , the tricritical behavior is suppressed for @xmath122 , while , for @xmath123 , that suppression occurs for @xmath124 .
as shown in table [ tab1 ] , @xmath119 increases with @xmath10 , which indicates that disorder is more effective for small coordination numbers . as the effects of binary disorder strongly depend on the coordination
, we now discuss the phase diagrams for the typical cases .
for @xmath125 and @xmath126 , there are no tricritical points .
the @xmath41 phase diagram displays just a fully stable critical line .
the main effect of disorder is to make the paramagnetic phase unstable at @xmath127 , regardless of the value of @xmath46 , for @xmath21 larger than the critical percolation concentration @xmath120 .
the phase diagrams in fig .
[ fig3 ] , for @xmath125 , are in qualitative agreement with the exact results for the honeycomb lattice ( which is also three - coordinated ) under annealed disorder.@xcite at @xmath127 , there is even quantitative agreement with the value of the critical crystal field at @xmath120 , given by @xmath128 , although of course this agreement does not extend to the value of @xmath120 itself .
our results for @xmath125 and @xmath115 are also in qualitative agreement with those obtained by a real - space renormalization - group approach for the two - dimensional blume - emery - griffiths model in a random crystal field .
@xcite for @xmath129 , the concentration @xmath119 above which the tricritical point becomes unstable is lower than @xmath120 . for @xmath130 , disorder depressses the tricritical temperature , and shortens the first - order transition line . for @xmath131 , the tricritical point
is replaced by a critical end point , @xmath49 , and a simple critical point , @xmath51 , as in the curie
weiss version of the model .
however , the paramagnetic phase is stable at @xmath127 if @xmath16 , and the first - order line reaches @xmath132 at @xmath127 . as @xmath21 increases , first the critical end point @xmath49 and then the simple critical point @xmath51 reach the @xmath127 axis , at values of @xmath21 which can be determined by a low - temperature expansion of the consistency equations ( see appendix [ app2 ] ) . in fig .
[ fig4 ] , we plot the @xmath41 phase diagram for @xmath133 and @xmath134 . to determine the first - order lines shown in that figure
, we numerically solve the consistency equations to obtain the conditions @xmath135 and @xmath136 which correspond to a maxwell construction . for @xmath123 , we have @xmath124 , so the behavior of the system is quite similar to the predictions of the curie
weiss version of the model .
we performed detailed calculations for the phase diagram of a random - anisotropy mixed - spin ising model both in the mean - field limit ( curie weiss version of the model ) , in which thermal fluctuations are neglected , and according to a standard self - consistent bethe
peierls approximation ( which turns out to be exact in one dimension ) . for a binary distribution of crystal fields
, we obtained closed - form expressions for the critical lines and the location of the tricritical points . depending on the concentration @xmath21 , the mean - field results for the @xmath41 phase diagrams predict novel first - order lines and multicritical points ( besides a ferromagnetic region , for all values of the crystal field , extending down to the lowest temperatures ) .
the bethe
peierls approximation shows that this additional ferromagnetic region is suppressed for concentrations below a certain percolation threshold .
also , the bethe
peierls results point out to the absence of a tricritical behavior for lattices with coordination @xmath137 .
all results reported in this paper are in agreement with general predictions for the effects of disorder on first - order transitions and multicritical points ( for a recent review , see a paper by cardy @xcite ) .
for an open chain with @xmath138 sites ( @xmath139 even ) , and in zero external field , the hamiltonian of the mixed - spin ising model can be written as @xmath140 given a disorder configuration @xmath141 , we perform a partial trace over the spin variables @xmath142 to write @xmath143\right\ } , \end{aligned}\ ] ] where @xmath35 and @xmath144 . introducing a prefactor @xmath145 , @xmath146\ ] ] and an effective interaction @xmath147 , such that @xmath148 the partition function can be written as the factorized form @xmath149 .
\label{smz1d}\end{aligned}\ ] ] from eq .
( [ smz1d ] ) , we obtain the thermal average @xmath150 which depends on the value of the crystal field on the @xmath18th site only .
since we are considering a nearest - neighbor one - dimensional model in zero field , the thermal averages @xmath151 and @xmath152 are zero . performing the disorder average , we obtain the expectation value @xmath153 for a given disorder configuration , the magnetic susceptibilities of the @xmath154 and @xmath79 sublattices are given by @xmath155 and @xmath156 the two - spin correlation functions , @xmath157 and @xmath158 can be calculated if we introduce the transformation @xmath159 after some algebraic manipulations , for @xmath160 , we have @xmath161 and @xmath162 from which we obtain the expectation values @xmath163 and @xmath164 which depend on the distance between sites @xmath18 and @xmath165 .
the expectation values of the susceptibilities are given by @xmath166 _
{ \text{av}}=\frac{1}{t}\left [ 1 + 2\sum_{r=1}^{\infty } g_{\sigma } \left ( r\right ) \right ] = \frac{1}{t}\frac{% 1+q\tanh ^{2}k}{1-q\tanh ^{2}k}\ ] ] and @xmath167 _ { \text{av}}=\frac{1}{t}\left [ q+2\sum_{r=1}^{\infty } g_{s}\left ( r\right ) \right ] = \frac{q}{t}\frac{1+q\tanh ^{2}k}{1-q\tanh ^{2}k},\ ] ] where @xmath80 is determined by eq .
( [ smq1d ] ) .
for the binary distribution , in the low - temperature limit ( @xmath168 ) , if we neglect terms of order @xmath169 and higher , the consistency equations ( [ smcem])-([smceq ] ) for cluster @xmath4 lead to the expressions @xmath170 @xmath171 and @xmath172 where @xmath173 for cluster @xmath5 , we have @xmath174 solving eqs .
( [ ltsa ] ) and ( [ ltqa ] ) for @xmath179 in terms of @xmath154 , @xmath79 and @xmath80 , and using eqs .
( [ ltma])-([ltqb ] ) , we can write eq .
( [ ltga ] ) in the form @xmath180 where @xmath181 is determined from the solution of eq .
( [ ltma ] ) .
notice that , according to eqs .
( [ ltstate ] ) and ( [ ltma ] ) , @xmath182 and @xmath181 depend on the temperature through the parameter @xmath183 only .
as @xmath48 , this parameter goes to zero ( if @xmath16 ) , or infinity ( if @xmath184 ) , except in the vicinity of the point @xmath185 with coordinates @xmath132 , and @xmath127 , where @xmath186 can assume any value .
since the equation of state ( [ ltstate ] ) becomes asymptotically exact as @xmath187 , it can be used to determine the values of @xmath21 at which the critical end point and the simple critical point reach @xmath185 , and thus disappear . to do that calculation
, we impose the conditions @xmath188 from which we obtain the values of @xmath189 , @xmath190 and @xmath191 at which the critical end point reaches @xmath185 , and the conditions @xmath192 which give the corresponding values @xmath193 , @xmath194 and @xmath195 for the simple critical point .
.values of the critical percolation concentration @xmath120 and the concentration @xmath119 at which the tricritical point becomes unstable , as functions of the coordination @xmath10 in the bethe
peierls approximation .
[ cols="^,^,^ " , ] | we investigate the phase diagram of a mixed spin-@xmath0spin-@xmath1 ising system in the presence of quenched disordered anisotropy .
we carry out a mean - field and a standard self - consistent bethe peierls calculation .
depending on the amount of disorder , there appear novel transition lines and multicritical points . also , we report some connections with a percolation problem and an exact result in one dimension . 2 |
the error box of the batse gamma - ray burst grb 970616 ( connaughton et al .
1997 ) was scanned by rxte four hours after the event ( marshall et al .
observations revealed a previously unknown x - ray source inside the error box whose position was consistent with the one provided later on by the interplanetary network ( ipn ) , yielding a combined rxte / ipn trapezoidal error box ( hurley et al .
four days later , the asca satellite observed the rxte / ipn error box and detected four x - ray sources .
one of them , a#1 according to the nomenclature of murakami et al .
( 1997 ) , was suggested as the x - ray counterpart to grb 970616 .
two candidates were initially proposed by galama et al .
( 1997 ) and udalski ( 1997 ) , but other observers ( pahre et al . 1997 , dey et al . 1997 , wheeler et al . 1997 , castro - tirado et al .
1997 ) failed to confirm these .
observations performed by rosat during june 23 - 25 revealed eleven faint x - ray sources , lying three of them within the rxte / ipn error box .
one of the three sources , r#2 , following greiner et al .
( 1997 ) , was consistent with the asca variable source a#1 , showing a flux five times lower than the one measured by asca .
the 3@xmath0 upper limit to the unabsorbed rosat flux of the source a#4 , implies a factor of seven lower than the previously measured by asca .
the fading observed during the asca observation could suggest that a#1 is the x - ray counterpart to the grb .
however , the presence of another decaying x - ray source inside the rxte / ipn error box keeps open the association with the grb .
a massive cluster of galaxies @xmath1 away from the rxte / ipn error box is reported by bentez et al .
fig 1 shows the locations of the cluster and of the candidates proposed by galama et al .
( 1997 ) and udalski ( 1997 ) .
observations in the k@xmath2-band were performed with the 3.5-m telescope ( + omega ) of the calar alto observatory on 1997 june 21.18 , 25.17 and 26.18 ut .
also j - band images were taken with the 2.2-m telescope ( + magic ) on june 25.14 .
the frames were centered on the candidate proposed by galama et al .
additional k@xmath2-band observations were performed on 1998 oct .
5.125 with the 3.5-m telescope .
optical observations were mainly taken at la silla with the danish 1.54-m telescope equipped with dfosc , whose field of view ( @xmath3 ) enabled us to cover the four asca sources ( see fig .
1 ) . table 1 displays the observing log .
.[970616table2 ] observing log of the optical observations performed with the 1.54-m danish telescope .
[ cols="<,^,^,^,^,^ " , ] additional r - band imaging was obtained on 1997 july 20.199 , 20.207 and 20.257 with the 0.8-m iac80 telescope at the observatorio del teide . a 1020 @xmath4
1024 pixels ccd provides a @xmath5 field of view . limiting magnitude
was r @xmath6 20 on the 360-s exposures .
optical and ir images of the rxte error box were obtained starting 106 hours after the event .
no variable optical counterpart was found within the a#1 and a#4 x - ray source error circles in the rxte / ipn annulus , being the variation @xmath7b @xmath8 0.3 , @xmath7v @xmath8 0.2 , @xmath7r @xmath8 0.2 and @xmath7i @xmath8 0.3 for optical objects with b @xmath8 23 , v @xmath8 23 , r @xmath8 24 and
i @xmath8 22 .
in fact , no objects brighter than r = 24.2 are seen within the error box of the rosat r#2 source detected at the asca position , consistently with the magnitudes reported by groot et al .
therefore , either there is a considerable intrinsic absorption ( as in the case of grb 970828 , yoshida et al .
1999 ) or the transient optical emission displayed a fast decline , as seen in grb 980326 ( groot et al .
1998 ) and grb 980519 ( hjorth et al .
1999 ) .
no ir variability has been found in a @xmath9 region centered at the position of the candidate proposed by galama et al .
this object was observed at j = 18.3 @xmath10 0.2 , and k@xmath2 = 17.4 @xmath10 0.3 and did not show any obvious ir / optical variation in the k@xmath2-band .
the color is consistent with this of a late - type star , as suggested by pahre et al .
( 1998 ) and it seems to be a non - variable star ( groot et al .
any fading or increase was @xmath8 0.5 mag for k@xmath11 17.0 . | we report on near - infrared and optical observations of the grb 970616 error box and of the x - ray sources discovered by asca and rosat in the region .
no optical transient was found either within the ipn band or in the x - ray error boxes , similarly to other bursts , and we suggest that either considerable intrinsic absorption was present ( like grb 970828 ) or that the optical transient displayed a very fast decline ( like grb 980326 and grb 980519 ) . |
many observed phenomena can be associated with wave - like motions , increasing interest in the study of wave dynamics .
linear perturbation theory considers an arbitrary disturbance as a superposition of independently evolving eigenmodes , thus simplifying the description of the process .
however , interactions between different harmonics as well as between different kind of waves leads to the appearance of substantially new phenomena . in the case of large - amplitude acoustic waves
, nonlinearity leads to the generation of higher harmonics which cause steepening of the wave front and consequently the formation of shock waves . also developments in plasma theory
raise interest in the study of interactions between different waves .
it is shown that nonlinear interaction leads to the generation of resonant triplets ( or multiplets ) in the plasma @xcite . the nonlinear interaction between magnetohydrodynamic ( mhd )
waves has been studied in various astrophysical situations @xcite . additionally , mhd wave coupling due to inhomogeneity of the medium @xcite or a background flow @xcite has also been developed .
recently , a new kind of interaction between sound and alfvn waves has been discussed by zaqarashvili [ 16 ] .
the physical basis of this interaction is the parametric influence ; sound waves cause a periodical variation in the medium s parameters , which affects the velocity of transversal alfvn waves and leads to a resonant energy transformation into certain harmonics . in a high @xmath0 plasma , it is shown that periodical variations of the medium s density , caused by the propagation of sound waves along an applied magnetic field , results in alfvn waves being governed by mathieu s equation ( here @xmath1 , where @xmath2 is the plasma pressure and @xmath3 is the magnetic field ) .
consequently , harmonics with half the frequency of sound waves grow exponentially in time .
the same phenomenon was developed in the case of standing sound waves @xcite .
the process of energy exchange between these different kinds of wave motion is called _ swing wave - wave interaction_. this terminology arises from an analogy with a swinging pendulum , as described below . in this paper
we further develop the theory for interactions between fast magnetosonic waves and alfvn waves .
for clarity of presentation we first recall the pendulum analogy and show that under certain conditions the energy of spring oscillations along the pendulum axis is transformed into the energy of transversal oscillations of the pendulum , and _
vice versa_. following a discussion of the general physics of swing interaction we go on to consider the example of coupling between fast magnetosonic waves propagating across an applied magnetic field and alfvn waves propagating along the field .
finally , we briefly describe the applications of the theory to various astrophysical situations .
it is useful to begin with a mechanical analogy of the wave dynamics in a medium ( see @xcite in the case of three - wave interaction ) . consider a mathematical pendulum with mass @xmath4 and equilibrium length @xmath5 ( see fig .
part of the pendulum length consists of a spring with stiffness constant @xmath6 .
there are two kinds of oscillation in this system : transversal oscillations due to gravity and spring oscillations along the pendulum axis due to the elasticity of the spring .
this is a _
swing pendulum_. in equilibrium , gravity is balanced by the stiffness force @xmath7 of the spring so that @xmath8 where @xmath9 is the gravitational acceleration and @xmath10 is the equilibrium length of the spring ( the natural length of the spring is supposed negligible ) .
the swing pendulum in equilibrium ( left ) and in oscillation ( right).,width=264 ] for displacement @xmath11 along the pendulum axis , the stiffness force becomes @xmath12 newton s second law applied along the pendulum axis , when the pendulum makes an angle @xmath13 with the vertical ( see fig.1 ) , gives the equation of motion ( the centrifugal force due to the transversal oscillation is neglected ) @xmath14 due to the oscillation of the spring along the axis , the pendulum length is a function of time and the equation of transversal motions of the pendulum under gravity is @xmath15 for clarity of presentation a term @xmath16 is here neglected ; it does not affect the physical nature of the phenomenon ( for general consideration , see @xcite ) .
so we have two different oscillations of the pendulum , which are coupled , and each oscillation influences the other . considering small amplitude oscillations , we find two coupled equations governing the dynamics of the pendulum : @xmath17 @xmath18 where @xmath19 and @xmath20 are the fundamental frequencies of the system
. from equations ( 1 ) and ( 2 ) we can see that longitudinal oscillations of the pendulum causes a periodical variation of the pendulum length . in certain conditions
this can lead to the well known parametric amplification of transversal oscillations .
when @xmath11 is a periodical function of time , then equation ( 2 ) becames mathieu s equation and it has a resonant solution when @xmath21 corresponding to @xmath22 . under these conditions , initial spring oscillations , @xmath11 , along
the pendulum axis can amplify small transversal perturbations , @xmath23 ( see equation ( 2 ) ) . on the other hand ,
transversal oscillations may be considered as an external periodic force ( see equation ( 1 ) ) which causes the damping and consequent amplification of longitudinal oscillations .
so , in the absence of dissipation , there is a subsequent energy exchange between different oscillations in the system .
but if some kind of external force supports the spring oscillations then they can amplify the transversal oscillations until nonlinear effects became significant .
the generalisation of the above analogy to waves in a medium leads to interesting phenomena .
spring oscillations do work against gravity and cause periodical variations of the parameter ( pendulum length @xmath5 ) of transversal oscillations . as a result of this work
, the energy of spring oscillations transforms into the energy of transversal oscillations .
so we may expect a similar process in a medium when one kind of waves cause a periodical variation of another wave parameters .
there are three main forces in the equation of motion for an ideal conductive fluid : the pressure gradient @xmath24 , gravity @xmath25 , and the lorentz force @xmath26 . here
@xmath2 and @xmath27 denote the plasma pressure and density , @xmath28 is the gravitational potential and @xmath29 is the current in a magnetic field @xmath30 .
each of these forces represents the restoring force against the fluid inertia and thus leads to the generation of different kinds of wave motion .
of these forces , only the lorentz force does not include the density ( in the pressure gradient the density arises from the equation of state ) .
this fact leads to the appearance of the density in the expression for the magnetic speed , the alfvn speed @xmath31 , which describes the propagation of magnetic waves and depends on the medium density @xmath32 . for a similar reason , the frequency of pendulum oscillations does not depend on the pendulum mass ( because the gravitational force depends on it ) , while the frequency of spring oscillation does depend on the mass ( because the stiffness force does not depend on it ) .
on the other hand , compressible waves cause density variations in the medium and therefore they may affect the propagation properties of magnetic waves .
this suggests a coupling between longitudinal , compressible waves ( leading to density perturbations ) and transversal magnetic waves propagating with a velocity which depends on the density .
the later can be associated with alfvn waves which are transversal and represent the purely electromagnetic properties of the medium .
the compressible waves cause a periodical variation of the density and so of the alfvn speed , and may lead to the effective energy transmission into certain harmonics of alfvn waves .
the swing coupling between sound and alfvn waves propagating along an applied magnetic field [ 16 - 17 ] is a good example of this phenomenon . on the other hand , magnetosonic waves propagating at an angle to the magnetic field
also cause periodical variations of the alfvn speed and may lead to similar phenomena .
it is worth noticing that , contrary to the alfvn waves , the magnetosonic waves can be easily excited in a medium by any force ( even of non electromagnetic origin ) .
therefore , the coupling between magnetosonic and alfvn waves allows the transmission of energy into purely transversal magnetic oscillations through compressible magnetosonic oscillations . to show the mathematical formalism of swing wave interaction we consider the case of magnetosonic wave propagation across the applied magnetic field . in this case
we have fast magnetosonic waves .
for simplicity we consider a rectangular geometry , which then can be generalised to cylindrical and spherical symmetries .
consider motions of a homogeneous medium , with zero viscosity and infinite conductivity , as described by the ideal mhd equations : @xmath33 @xmath34 + { { ( { { \bf b}{\cdot}{\nabla}}){\bf b}}\over { 4{\pi}}},\ ] ] @xmath35 where @xmath36 is the fluid velocity .
we consider adiabatic processes , so the pressure @xmath2 and density @xmath32 are connected by the relation @xmath37 where @xmath38 and @xmath39 are the unperturbed uniform pressure and density and @xmath40 is the ratio of specific heats .
we neglect gravity , though it may be of importance under some astrophysical conditions .
the unperturbed magnetic field @xmath41 is directed along the @xmath42 axis.the system is bounded in the @xmath11 direction ( @xmath43 is the size of the system ) .
standing fast magnetosonic waves are polarized in the @xmath11 direction , while alfvn waves are polarized along the @xmath44 axis and propagate along the @xmath42 axis.,width=264 ] linear analysis of equations ( 4)-(7 ) show the existence of three kinds of mhd waves : alfvn and magnetosonic ( fast and slow ) waves .
the difference between these waves is that the restoring force of alfvn waves is the tension of magnetic field lines , @xmath45 , acting alone , while the restoring force of magnetosonic waves is mainly the gradient of ordinary and magnetic pressures , @xmath46 $ ] .
the various waves can be distinguished by their different speeds and polarizations .
the linear evolution of the waves in a homogeneous medium is governed by the usual linear wave equations .
consider a uniform , unperturbed , magnetic field @xmath47 directed along the @xmath42-axis , and the case of magnetosonic wave propagation across the field in the @xmath11 direction ( see fig .
then there are only fast magnetosonic waves ( the slow wave is absent ) which in the linear approximation is described by the equations : @xmath48 @xmath49,\ ] ] @xmath50 where @xmath51 and @xmath52 are the perturbations of magnetic field and velocity , respectively , and @xmath53 is the sound speed . here and afterwards
@xmath32 denotes the perturbation of density ( in equations ( 4)-(5 ) @xmath32 was the total density ) . the wave equation for linear
fast magnetosonic waves then follows , @xmath54 where @xmath55 is the phase velocity of fast waves and @xmath56 is the alfvn speed
. the solution of the wave equation can be either propagating or standing patterns .
the boundedness of the medium leads to the formation of a discrete spectrum of harmonics which represent the normal modes ( eigenmodes ) of the system .
we consider the standing fast magnetosonic waves , which have a straightforward extention to cylindrical ( pulsating magnetic tube ) and spherical ( pulsating sphere with dipole - like magnetic field ) geometries .
the solutions for standing ( plane ) fast magnetosonic waves are : @xmath57 where @xmath58 ( @xmath59 ) is the eigenvalue for a system of size @xmath43 in the @xmath11 direction , @xmath60 is the corresponding eigenfrequency , and @xmath61 is the relative amplitude of the waves .
eigenvalues and eigenfrequencies are related by the dispersion relation @xmath62 .
it is seen from the expressions ( 12 ) that standing fast magnetosonic waves cause a local periodical variation in both the density and the magnetic field .
this variation is maximal near the nodes of the velocity and approaches to zero near the antinodes .
the amplitude of the variation is considered to be small @xmath63 , and so does not affect the fast magnetosonic wave itself .
consider now the influence of the density and the magnetic field variations ( 12 ) on alfvn waves , considered to be polarised in the @xmath64 plane .
then the velocity fields of fast magnetosonic and alfvn waves are decoupled .
the linear equations for alfvn waves are : @xmath65 @xmath66 where @xmath67 and @xmath68 are small perturbations of the magnetic field and the velocity .
these equations lead to the wave equation @xmath69 the influence of the fast magnetosonic waves can be expressed by modifying equations ( 13 ) and ( 14 ) , which now became @xmath70 @xmath71 here we have neglected the advective terms @xmath72 and @xmath73 for several reasons . at the initial stage , the perturbations @xmath67 and @xmath68 of alfvn waves propagating along the @xmath42 axis do not depend on the @xmath11 coordinate ; each magnetic surface across @xmath11 evolves independently .
the @xmath11 dependence arises due to the action of the fast magnetosonic waves , and so the neglected terms are second order in @xmath74 .
moreover we can consider the alfvn waves at the velocity node of standing fast magnetosonic waves , where these terms are zero . in principle
, the coordinate @xmath11 stands as a parameter in equations ( 16 ) and ( 17 ) of alfvn waves .
equations ( 16 ) and ( 17 ) lead to the hill type second order differential equation @xmath75 @xmath76 where @xmath77 denotes the time derivative of the perturbing field . introducing @xmath78 and neglecting terms of order
@xmath74 leads to the equation @xmath79 { { \partial^2 h_y}\over { \partial z^2}}=0.\ ] ] comparing equations ( 20 ) and ( 15 ) we can see that the influence of standing fast magnetosonic waves is expressed through a periodical variation of the alfvn speed . performing a fourier transform of @xmath80 with @xmath81 , equation ( 20 ) leads to mathieu s equation @xcite @xmath82{\hat h_y}=0,\ ] ] where @xmath83 with @xmath11 playing the role of a parameter .
equation ( 21 ) has main resonant solution if @xmath84 and it can be expressed as @xmath85,\ ] ] where @xmath86 .
the solution has a resonant character within the frequency interval @xmath87 equation ( 24 ) shows that the harmonics of alfvn waves with half the frequency of fast magnetosonic waves grow exponentially in time .
the growth rate of alfvn waves is maximal at the velocity nodes of fast magnetosonic waves and tends to zero at the antinodes ( see equations ( 12 ) and ( 22 ) ) .
the amplitude of the magnetic field component in alfvn waves depends on the @xmath11 coordinate , i.e. there is the periodical magnetic pressure gradient along this direction .
energy conservation implies that this gradient leads to the damping of initial fast magnetosonic waves , i.e. the energy transformed into alfvn waves is extracted from fast magnetosonic waves . to show this
, we consider the backreaction of amplified alfvn waves on the initial fast magnetosonic waves .
the dependence of @xmath67 on the @xmath11 coordinate leads to an additional term in the equation of motion ( 9 ) for fast waves , @xmath49 - { { \partial}\over { \partial x}}\left [ { { b_y^2}\over { 8{\pi}}}\right ] .\ ] ] therefore the wave equation ( 11 ) now becomes @xmath88.\ ] ] the additional term has the frequency of the initial fast magnetosonic waves @xmath60 ( within the order of @xmath74 ) and can be considered as the external , periodic force . at the initial stage
it can be neglected as of second order of smallness .
however , it becomes significant because of the exponential growth of amplitudes ( see equation ( 24 ) ) .
it oscillates out of phase with respect to the initial fast waves ( 12 ) , thus leading to their damping ( as expected from physical considerations ) .
note that equations ( 1 ) and ( 2 ) describing the pendulum oscillations are similar to equations ( 20 ) and ( 27 ) describing alfvn waves and fast magnetosonic waves ; the longitudinal oscillations of the pendulum correspond to the fast magnetosonic waves and the transversal oscillations correspond to the alfvn waves .
swing coupling between fast magnetosonic and alfvn waves may be generalised from rectangular geometry to other symmetries , though a detailed description is beyond the scope of this paper .
the suggested mechanism of energy transformation from fast magnetosonic into alfvn waves has important consequences .
it can be noted that the alfvn waves hardly undergo either excitation or damping processes , while magnetosonic waves can be easily excited by external , even nonelectromagnetic , forces .
then swing interaction leads to the intriguing but natural suggestion that the energy of the nonelectromagnetic force which supports the magnetosonic waves in the system can be transmitted into the energy of purely magnetic incompressible oscillations .
this result has many astrophysical applications .
we briefly describe several of them .
resonant interaction between mhd waves , due to the inhomogeneity of the medium , was proposed by ionson [ 8 ] .
it arises where the frequency of an incoming wave matches the local frequency of the medium .
then a resonant energy transformation may take place , known as resonant absorption .
similar phenomenon may arise also due to swing wave interaction . in this case fast magnetosonic waves can transform their energy into alfvn waves , even in a homogeneous medium . for given medium parameters ( magnetic field , density )
the energy of fast waves may be absorbed by harmonics with wavelengths satisfying the resonant condition ( 23 ) .
consequently , fast magnetosonic waves can transmit their energy into alfvn waves in any spatial distribution of density or magnetic field .
the process can be called _ swing absorption_. the particular point of _ swing absorption _ is that energy absorption occurs through the harmonics with half the frequency of incoming waves ( see equation ( 23 ) ) .
the process may be of importance in the earth s magnetosphere and in the solar atmosphere . swing wave interaction may play an important role in the excitation of torsional alfvn waves in solar coronal loops .
it may be suggested that any external action on the magnetic tube , anchored in the highly dynamical photosphere , causes a radial pulsation at the fundamental frequency , like a tuning fork ( see also @xcite ) . for a tube of radius @xmath89
the fundamental frequency of pulsation will be of order @xmath90 , where @xmath91 is the phase velocity of fast magnetosonic waves at the photospheric level .
if we consider the alfvn and sound speeds to be of order @xmath92 kms@xmath93 and the radius of order @xmath94 km , then the period of fundamental mode of pulsation will be a few tens of seconds .
radial pulsations of the tube may lead to the resonant ( exponential ) amplification of torsional alfvn waves with half the frequency of the pulsations .
these high frequency torsional alfvn waves can propagate upward and carry energy from the photosphere into the magnetically controlled corona or they may be damped in chromospheric regions leading to the heating of the chromospheric magnetic network .
swing wave interaction may be of importance in stellar interiors . a radial pulsation of a spherically symmetric star with dipole - like magnetic field may lead to the amplification of torsional oscillations .
there are a number of energy sources which can support pulsations : radiation , nuclear reactions , tidal forces in binary stars , convection , etc .
( e.g. , @xcite ) .
then the transformation of pulsational energy into torsional oscillations may lead to new sources for stellar magnetic activity .
the swing wave - wave interaction @xcite is developed here in the case of fast magnetosonic waves propagating across a magnetic field and alfvn waves propagating along the field . in the case of oblique propagation ,
slow magnetosonic waves also exist and they may transmit their energy into alfvn waves . in some cases
the coupling between slow magnetosonic and alfvn waves may be of importance .
also , the coupling in the case of different geometries ( cylindrical , spherical ) may be important in astrophysical situations .
the most important result of swing wave interaction is that it reveals a new energy channel for alfvn waves , permitting the transformation of energy of nonelectromagnetic origin into the energy of electromagnetic oscillations . | we suggest a mechanism of energy transformation from fast magnetosonic waves propagating across a magnetic field to alfvn waves propagating along the field . the mechanism is based on _ swing wave - wave interaction _ [ t.v .
zaqarashvili , astrophys .
j. lett .
* 552 * , 107 ( 2001 ) ] .
the standing fast magnetosonic waves cause a periodical variation in the alfvn speed , with the amplitude of an alfvn wave being governed by mathieu s equation . consequently , sub - harmonics of alfvn waves with a frequency half that of magnetosonic waves grow exponentially in time .
it is suggested that the energy of nonelectromagnetic forces , which are able to support the magnetosonic oscillations , may be transmitted into the energy of purely magnetic oscillations .
possible astrophysical applications of the mechanism are briefly discussed . |
the present baryon asymmetry in the universe is one of the most mysterious problems in cosmology and particle physics , see a review @xcite . within the standard model ( sm ) , the electroweak baryogenesis
was suggested to explain the inequality between baryon and anti - baryon number , and the recent developments are shifted into constructing the theory of reheating the universe @xcite .
the electroweak baryogenesis satisfies the well - known sakharov s three conditions required for successful baryogenesis @xcite , namely baryon number production , charge parity ( cp ) violation and the process taking place out - of - equilibrium ; however , the predicted cp violation is too small to explain the present observed baryon number . by satisfying the above three conditions , the affleck - dine ( ad ) baryogenesis @xcite , which was proposed in the theoretical framework beyond the sm , namely , the minimal super - symmetric standard model ( mssm ) , is a more successful scenario to tackle this puzzle , since it may solve problems of gravitino and moduli overproduction and give rise simultaneously to the ordinary matter and dark mater in the universe .
the mssm has many gauge - invariant flat directions along which r parity is preserved .
the flat directions are lifted by super - symmetry ( susy ) breaking effects arising from nonrenormalisable terms , which give a u(1 ) violation through a - terms . in the original scenario of the ad baryogenesis , one can parametrise one of the flat directions in terms of a complex scalar field known as an ad field ( or ad condensate which consists of a combination of squarks and/or sleptons fields )
. the ad field evolves to a large field expectation value during an inflationary epoch in the early universe .
after inflation , the orbit of the ad field can be kicked along the phase direction due to the a - terms which generate the u(1 ) charge ( baryon / lepton number ) , and then the a - terms become negligible , where the ad field rotate towards the global minimum of the scalar potential .
hence , the generated global u(1 ) charge is fixed and the orbit of the ad field rotates around the origin of the complex field - space , _ c. f. _ the anomaly mediated models @xcite . after the ad condensate decays into usual baryons and leptons
, ad baryogenesis becomes complete .
the trajectory of the ad field is identical to the planetary orbits in the well - known kepler - problem as we will show later , replacing the newtonian potential by an isotropic harmonic oscillator potential @xcite .
this coincidental classical - mechanics reduction was noted for the orbits of a probe brane in the branonium system @xcite . as general relativity predicted that planetary orbits precess by adding the relativistic correction to the newtonian potential , we will see the similar events occur for the orbits of ad fields , which are disturbed by quantum and nonrenormalisable effects . by including quantum corrections @xcite and/or thermal effects @xcite in the mass term of the standard ad scalar potentials ,
the ad condensate is classically unstable against spatial perturbations due to the presence of negative pressure @xcite , and fragments to bubble - like objects , eventually evolving into @xmath0-balls @xcite .
lee pointed out @xcite that @xmath0-balls may form due to bubble nucleation ( first order phase transition ) @xcite , even in the case that the condensate is classically stable against the linear spatial perturbations .
a @xmath0-ball is a nontopological soliton @xcite whose stability comes from the existence of a continuous global or local charge @xmath0 ( see a review @xcite and references therein ) .
tsumagari _ et .
@xcite showed previously the complete stability analysis of @xmath0-balls at zero - temperature in both polynomial potentials and mssm flat potentials .
@xcite investigated the stability of @xmath0-balls in a thermal bath .
the stability of the thermal susy @xmath0-balls is different from the one of the standard `` cold '' @xmath0-balls , since they suffer from evaporation @xcite , diffusion @xcite , dissociation @xcite , and decays into light / massless fermions @xcite .
therefore , most susy @xmath0-balls are generally not stable but long - lived , and it may thermalise the universe by decaying into baryons on their surface @xcite , which could solve the gravitino and moduli over - production problems without fine - tuning .
the susy @xmath0-balls in gravity - mediated ( grv - m ) models are quasi - stable decaying into the lightest susy particles ( lsp dark matter ) , and the fraction of the baryons from the @xmath0-balls may give the present baryon number , which can explain the similarity of the energy density between the observed baryons and dark matter @xcite .
the susy @xmath0-balls in gauge - mediated ( gau - m ) models , however , can be extremely long - lived so that those @xmath0-balls are a candidate of cold dark matter @xcite and may give the present observed baryo - to - photon ratio @xcite .
the dynamics and formation of @xmath0-balls have been investigated numerically . with different relative phases and initial velocities , the authors @xcite found a charge transfer from one @xmath0-ball to the other and interesting ring formation after the collision .
it has been found @xcite that similar ring - like solutions are responsible for the excited states from the ground state ( @xmath0-ball ) by introducing extra degrees of freedom : spatial spins @xcite and twists @xcite .
the formation of @xmath0-balls after inflation have been extensively investigated in both grv - m models @xcite and gau - m models @xcite , in which susy is broken by either gravity or gauge interactions .
as we will show , the @xmath0-ball formation involves a nonequilibrium dynamics which is related to reheating problem in cosmology . the reheating process after the inflation period involves nonlinear , out - of - equilibrium , and nonperturbative phenomena so that it is extremely hard to construct a theory for the whole process , see the 2 particle irreducible effective action as a remarkable approach @xcite . in the first stage of reheating ( _ preheating _ ) , it is currently well known that the fluctuations at low momenta are amplified , which leads to explosive particle production .
after preheating , the subsequent stages towards equilibrium are described by the wave kinetic theory of turbulence ; micha _ et .
_ @xcite recently estimated the reheating time and temperaute .
these turbulent regimes appear in a large variety of nonequilibrium process , and indeed , the evolution of @xmath0-ball formation experiences the active turbulence at which stage , many bubbles collide as observed in the next stage of tachyonic preheating @xcite . during this bubble - collision stage within the reheating scenario ,
it is believed that gravitational waves may be emitted from the stochastic motion of heavy objects @xcite .
the problem of gravitational wave emissions has been discussed only in the fragmentation stage of @xmath0-ball formation so far @xcite , but not in the collision stage as opposed to the preheating cases . in this paper , we show analytically and numerically that in grv - m and gau - m models the approximate trajectory of the ad fields is , respectively , precessing spiral or shrinking trefoil due to the quantum , nonrenormalisable , and the hubble expansion effects .
moreover , we explicitly present the exponential growth of the linear spatial perturbations in both models . by introducing @xmath1 ( and @xmath2)-dimensional lattice simulations
, we identify that the evolution in @xmath0-ball formation involves nonequilibrium dynamics , including turbulent stages . following the pioneering work on the turbulent thermalisation by micha _ et .
@xcite , we obtain scaling laws for the evolution of variances during the @xmath0-ball formation .
the paper is divided as follows .
we explore both analytically and numerically the dynamics of the ad field in sec .
[ sectorbit ] . in sec .
[ sectinst ] , we study the late evolution of the ad fields and the process of @xmath0-ball formation , introducing detailed numerical lattice results . finally , we conclude and discuss our results in sec .
[ concl ] .
two appendices are included .
we obtain the equations of motion and their perturbed equations for multiple scalar fields in appendix [ multi ] . in appendix
[ app2 ] , we find elliptic forms for the orbits of ad fields .
in this section , we investigate the equation for the orbit of an ad condensate .
this orbit coincides with the well - known orbit equation in the centre force problem in classical dynamics , i.e. planetary motions so that we call the ad condensate `` ad planet '' sometimes . for the bound orbits , the effective potential should satisfy the condition where the curvature at the minimum of the effective potentials should be positive . in the presence of hubble expansion ,
the effective potential depends on time ; thus , the full solution of the orbit equations can be obtained numerically except the case that the ad field is trapped by a quadratic potential . in appendix
[ app2 ] , we obtain the exact orbit in this exceptional case when hubble expansion is assumed to be small and adiabatic . the orbit of the ad planet , or more precisely an eccentricity of the elliptic motion in the complex field - space , is determined by the initial charge and energy density . in order to obtain analytic expressions of the orbit in more general potential cases in which we are more interested , we restrict ourself to work in minkowski spacetime and on the orbit which should be nearly circular . in appendix
[ app2 ] , we also obtain the perturbed orbit equation and necessary conditions for closed orbits where the orbits come back to their original positions after some rotations around the minimum of the effective potential . in bertrand
s theorem @xcite , there are only two potential forms allowed to be closed orbits : isotropic harmonic force and the inverse - squared force .
each of the central forces gives dynamical symmetries , namely fradkin tensor @xcite and runge - lenz vector @xcite , respectively .
these dynamical charges are obtained both classically by the algebra of poisson bracket @xcite and quantum - mechanically by the corresponding lie algebra in the abelian case @xcite as well as nonabelian case @xcite . by approximating phenomenologically motivated models that appear in the mssm and using the results in appendix [ app2 ] , we present , in this section , analytic motions of the nearly circular orbits and the pressure of the ad planets .
further , we check these analytic results with full numerical solutions .
let us consider a motion of ad fields in an expanding universe with scale factor @xmath3 and hubble parameter @xmath4 , where a over - dot denotes the time derivative .
we investigate the ad field after they start to rotate around the origin of the effective potentials and the value of the u(1 ) charge @xmath5 is fixed due to negligible contributions from a - terms . by decomposing the complex ( ad ) field
@xmath6 as @xmath7 , where @xmath8 and @xmath9 are real scalar fields , the equations of motion for @xmath10 and @xmath11 ( see in appendix [ multi ] ) are + 3h+&=&0 , + [ phshomo]+3h+&= & 0=0 , where the conserved comoving charge density is defined by @xmath12 , and the effective potentials are @xmath13 .
note that we will use @xmath14 shortly .
from , the energy density @xmath15 and pressure @xmath16 are given by _
e=^2+v_+ , p=^2 -v_- . with various values of the charge density @xmath5 , shows typical effective potentials @xmath17 in minkowski spacetime where we set @xmath18 .
the models shown in will be used later . given an initial charge and energy density ( or equivalently initial momenta and positions ) , the ad field oscillates around the value @xmath19 , which is defined by .|__cr=0 , where the orbit becomes circular when it starts from there , i.e. @xmath20 .
this orbit is bounded when the curvature is positive w^2.|__cr>0 .
for example , given a power - law potential such that @xmath21 where @xmath22 is a coupling constant and @xmath23 is real and a power of the homogeneous field @xmath8 , the condition given by implies that bound orbits exist for @xmath24 if @xmath25 and for @xmath26 if @xmath27 , where we used .
another example is the case that a scalar potential is logarithmic , i.e. @xmath28 where the coupling constant @xmath29 is positive . in this case , is automatically satisfied .
we investigate these two cases in more detail in appendix [ app2 ] .
let us rescale the field @xmath10 as @xmath30 where @xmath31 is the value of @xmath3 at an initial time .
it follows that the equations of motion in are --+^3=0 , = 0 , where we defined @xmath32 , and the terms involving @xmath33 and @xmath34 are negligible under the assumption of an adiabatic hubble expansion , i.e. @xmath35 . by introducing a new variable , @xmath36 , and using the second expression in , the first expression in becomes the well - known orbit equation in the centre force problem such that + = -^3j(,t ) .
notice that @xmath37 depends on time caused by the hubble expansion , whereas the time - dependence in @xmath38 vanishes when the potential @xmath39 is given by a quadratic mass term , @xmath40 , where @xmath41 is a mass of the ad field , @xmath6 .
let us introduce two models which appear in the mssm in which susy is broken due to either gravity or gauge interactions .
the former case , so - called , the gravity - mediated ( grv - m ) model , has a scalar potential v = m^2 ^ 2 + ^n , where @xmath42 is of order of the susy breaking scale , which could be the gravitino mass scale @xmath43 evaluated at the renormalisation scale @xmath44 @xcite .
also , @xmath45 is a coupling constant for the nonrenormalisable term , which is suppressed by a high energy scale , e.g. the planck scale @xmath46 gev . here , @xmath47 is a factor from the gaugino - loop correction , whose value is typically @xmath48 $ ] when the flat direction does not have a large top quark component @xcite ; thus , we concentrate on the case of @xmath49 from now on .
the power @xmath50 of the nonrenormalisable term depends on flat directions . as examles of the directions involving squarks ,
the @xmath51 direction has @xmath52 , whilst the @xmath53 direction is @xmath54 . for @xmath55 , the first two terms in can be approximated by @xmath56 , we then find that v()^l+^n n > l which we call model a , where we set @xmath57 and @xmath41 has a mass - dimension , @xmath58 , since @xmath59 for @xmath55 . for small values of @xmath8
, we confirm that the power @xmath23 is not approximately @xmath60 , so we will find a value of @xmath23 numerically in that case later . in another scenario in which susy is broken by gauge integration , so - called , gauge - mediated ( gau - m ) model , the scalar potential has the curvature with the electroweak mass at a low energy scale , whilst it grows logarithmically at the high energy scale ( which means that the potential is nearly flat as similar as the case of @xmath61 in ) .
therefore , vm^4 _
+ ^n , where @xmath62 is the messenger scale ( @xmath63 gev ) above which the potential grows logarithmically and @xmath64 is the same scale as @xmath62 .
we , thus , set @xmath65 for later convenience .
then , the scalar potential at high energy scale is approximately given by @xcite vm^4 _
what follows is that we assume the orbit of the ad condensate is determined by the high energy scale where @xmath66 , calling this case , , model b. using the results in appendix [ app2 ] , we obtain the following quantities , @xmath67 and @xmath68 by assuming that the dominant contribution in model a and b is , respectively , either power - law term or logarithmic term , each of which corresponds to the first term in , respectively . here
, we defined @xmath69 as a phase difference when the radial field @xmath8 goes from the minimum value through the maximum one and back to the same minimum point , see ; in addition , @xmath68 is given by a value of the equation of state averaged over a rotation of the orbit , see .
the sub - dominant terms ( nonrenormalisable terms ) perturb the orbits by introducing infinitesimally small quantities @xmath70 where the subscripts correspond to the names of models introduced above .
thus , the main contributions are either or . by recalling , we obtain the following relations for @xmath72 : w^2= , where we defined a positive parameter , @xmath73 , which is assumed to be infinitesimally small .
we also obtain @xmath74 , where @xmath75 is defined in . substituting @xmath75 into , we obtain @xmath69 and @xmath68 : & & , + [ modela - w]&=&. from ,
the orbits for @xmath76 are nearly closed , but it is perturbed by the nonrenormalisable term involved with @xmath77 .
it results in that the periapsis appears to precess where the precession rate can be obtained from .
the reader should notice that the denominator of the term involving @xmath77 in the second expression of has @xmath78 , which implies that it would be possible to have the non - negligible contribution from the term , even though @xmath79 .
from now on , we restrict ourself that this is not the case ; therefore , the dominant contributions appear as the leading orders in and , which correspond to and . from with @xmath80 ,
our results recover the result published in @xcite , @xmath81 . by introducing another infinitesimally small positive parameter , @xmath83
, we find w^2 & & , , + [ modelc - w ] & = & -1 .
since we are working in the high - energy regime , @xmath66 , the pressure of the ad condensate is likely to be negative , see . from the second expression for @xmath69 of ,
the orbits are not closed and it should look like trefoil , see . in an expanding universe ,
the above orbits for model a and b suffer from the hubble damping so that the orbits are naively expected to be precessing spiral or shrinking trefoil in the field space , respectively . in this subsection , we present numerical results to check the analytic results which we found in the previous subsection . to do so
, we use the full potentials , , instead of , and then solve numerically in minkowski spacetime as well as in an expanding universe .
we adopt the 4th order runge - kutta method with various sets of initial conditions , such as @xmath5 and @xmath84 .
since our analytical work holds as long as @xmath85 , we are concern with the two cases : a nearly circular orbit with @xmath86 and a more elliptic orbit with @xmath87 .
first of all , we parametrise by introducing dimensionless variables : @xmath88 in grv - m model and @xmath89 in gau - m model . since we know that @xmath90 gev , @xmath91 gev , @xmath46 gev
; hence , we can set @xmath92 in grv - m model , where we choose @xmath93 .
notice that these choices are same as the ones used in @xcite .
on the other hand , we know that @xmath94 gev ; hence , we can set @xmath95 in gau - m model , where we choose @xmath96 as used in the grv - m case .
notice that we can obtain the rescaled charge density @xmath97 and energy density @xmath98 , such that @xmath99 in grv - m model and @xmath100 in gau - m model .
therefore , our rescaled potentials in grv - m and gau - m models for a @xmath52 flat - direction are , respectively , v&=&^2 + b^2 _ * ^6 , + [ numgau ] v&=&+b^2 ^ 6 , where we omit over - rings for simplicity . the reader should notice that these variables that appear within the rest of this sub - section are dimensionless .
we can also obtain the ratio defined by an energy density divided by a mass multiplied by a charge density , where the mass corresponds to @xmath42 or @xmath62 in either grv - m or gau - m model , respectively . in order to obtain appropriate initial values of @xmath101 and @xmath102 satisfying the conditions @xmath103 and giving not too small charge densities
, we shall show that we need to choose only the initial values of @xmath102 in both grv - m and gau - m models .
first , by ignoring the nonrenormalisable term in for grv - m model , we can obtain @xmath104 from , where we set @xmath105 , which implies that we are setting that the initial phase is @xmath106 .
since @xmath107 has the maximum value at @xmath108 , we can set @xmath109 from , which imply that @xmath110 from the definition .
we notice that @xmath111 for @xmath112 ; hence , it breaks the condition , @xmath113 .
we can also see that @xmath114 for @xmath115 , so the charge density is suppressed exponentially .
therefore , we are concern with the following two cases : @xmath116 and @xmath117 , which give , respectively , @xmath118 and @xmath119 . similarly , in gau - m model , we choose that @xmath120 , @xmath121 and @xmath122 from the definition of @xmath123 . here
, we also set the initial phase is @xmath106 due to @xmath105 .
with this fact and the approximation , @xmath124 , we need to have @xmath112 . in addition
, we should have @xmath125 due to the condition , @xmath126 .
therefore , we choose @xmath127 and @xmath128 which gives , respectively , @xmath129 and @xmath130 . upon the above initial conditions , we initiate the numerical simulations with 8 different sets of the initial values in grv - m model and gau - m model summarised in table [ parameterset ] where we call each of the parameter - sets set-1 , set-2 , ... , and set-8. in , we also show , with the various charges which we introduced above , the effective potentials @xmath17 for the grv - m potential given by in the left panel and for the gau - m potential given by in the right panel .
our time - step , @xmath131 , in the numerical simulations is @xmath132 in the grv - m case and @xmath133 in the gau - m case . [ !
|c|c|c|c|c|c|c|c| set & model & @xmath102 & @xmath134 & @xmath5 & @xmath77 or @xmath123 & @xmath84 & @xmath135 + 1 & & & & & & 0.1 & 1.46 + 2 & & & & & & 0.3 & 1.51 + 3 & & & & & & 0.1 & 1.06 + 4 & & & & & & 0.3 & 1.09 + 5 & & & & & & 0.1 & @xmath136 + 6 & & & & & & 0.3 & @xmath137 + 7 & & & & & & 0.1 & @xmath138 + 8 & & & & & & 0.3 & @xmath139 + first , we present numerical results in minkowski spacetime in order to check our analytical results .
we then give anstze which are motivated by our analytic solutions in an expanding universe in the next sub - subsection .
[ [ section ] ] in , we show the numerical solutions using the grv - m potential with ( left ) and using the gau - m potential with ( right ) , and compare them with the corresponding analytic solutions which are given by . using the initial values whose parameter sets can be seen in table [ parameterset ] , we plot the numeric and analytic solutions in . in the top - left panel , the numerical plots ( red - plus dots for set-1 and blue - cross dots for set-2 )
have the same amplitudes as the analytical ones ( gree - dashed line for set-1 and purple - dotted - dashed line for set-2 ) , we , however , can see the significant differences for the frequencies of each oscillation . we notice that these discrepancies come from the artifact of our choice as @xmath140 in , since the choice is not appropriate for @xmath141 , recalling @xmath142 in set-1 and set-2 .
shortly , we will obtain numerically this power @xmath23 , and the obtained semi - analytic solutions match with the numerical ones .
with set-3 and set-4 , we can see that @xmath134 is not so small as opposed to the previous cases , i.e. @xmath143 ; thus , in the left - bottom panel of we can see a nice agreement between numerical plots ( red - plus dots for set-3 and blue - cross dots for set-4 ) and analytic plots ( skyblue - dotted - dashed line for set-3 and black - dotted line for set-4 ) .
similarly , we show the numerical and analytic plots for the gau - m potential in the right - panels of using the parameter - sets : for set-5 and set-6 in the right - top panel and for set-7 and set-8 in the right - bottom panel . by changing the values of the third eccentricity @xmath84 ( see table [ parameterset ] ) , the numerical plots deviate slightly from our analytic lines in the right - top and right - bottom panels of as we can expect ; in particular , we can see that our analytic values of both the frequencies and amplitudes of @xmath144 are larger than the numerical ones , and this difference can be significantly reduced when the orbits of the ad planets is nearly circular with @xmath86 . as we have seen in the left - top panel of , our analytic value , @xmath145 , in are not good enough to reproduce the numerical solutions since @xmath146 .
therefore , we set a trial function , @xmath147 , where a numerical value @xmath148 is found by using the fit command in the numerical software called gnuplot. we find that @xmath149 is the best value of @xmath148 , where we fitted this trial function @xmath150 onto the numerical full potential in for the range of @xmath151 $ ] , recalling @xmath142 in set-1 and set-2 . using this value of @xmath148 as the value of @xmath23 instead of @xmath145
, we plot the semi - analytic evolution for @xmath144 in ( green - dashed line for set-1 and purple - dotted - dashed line for set-2 ) against the corresponding numerical plots ( red - plus dots for set-1 and blue - cross dots for set-2 ) .
now , our semi - analytic solutions match with the numerical solutions . + [ [ section-1 ] ] using
, we show both numerical values @xmath152 and ( semi-)analytical values @xmath153 of the averaged equation of state in . for all cases , the ad condensate has a negative pressure and one can say that the numerical values are of the same order as analytic values . [ !
|c||c|c|c|c||c|c|c|c| @xmath68 & & + & set-1 & set-2 & set-3 & set-4 & set-5 & set-6 & set-7 & set-8 + @xmath152 & & @xmath154 & @xmath155 & @xmath156 & @xmath157 & + @xmath153 & & & & + [ [ section-2 ] ] in table [ parameterphi ] , we show the numerical and ( semi-)analytic values of @xmath69 in both grv - m model and gau - m model , which are analytically obtained in sec .
[ model - abc ] .
our analytical values agree well with the numerical values .
these values suggest that the orbits in grv - m model and gau - m model are nearly either elliptical or trefoil , respectively . [ !
|c||c|c|c|c||c|c|c|c| @xmath69 & & + & set-1 & set-2 & set-3 & set-4 & set-5 & set-6 & set-7 & set-8 + @xmath158 & 1.591 & 1.590 & 1.605 & 1.604 & 2.210 & 2.206 & 2.221 & 2.217 + @xmath159 & & & & + we carry out our numerical simulation in an expanding universe when the inflaton field , which is trapped by a quadratic potential , starts to coherently oscillate around the vacuum during the reheating era .
then the evolution of hubble expansion , @xmath160 , and scale factor , @xmath3 , follow as an ordinary nonrelativistic ( zero - pressure ) matter , see . for @xmath161 , we find @xmath162 and @xmath163 , where @xmath31 is given by the value of @xmath3 at @xmath164 and we set @xmath165 .
we also set the initial time as @xmath166 for grv - m model and @xmath167 for gau - m model .
notice that with this choice of @xmath168 our simulation starts from the same physical time because we rescaled the time by either @xmath90 gev or @xmath169 gev , respectively .
we again solve the equation of motion , , numerically using the 4th order runge - kutta method and compare them with the following anstze . in order to see the significant effects from the hubble expansion , we use set-3 in grv - m model and set-7 in gau - m model as the initial parameters . in an expanding spacetime
, one can guess that our analytical results in minkowski spacetime should be changed .
in particular , the amplitude of @xmath10 may decrease due to the hubble damping as we saw in the quadratic case in appendix [ sec2 ] , and similarly the frequency @xmath170 in should be changed .
hence , the orbit of the ad planet is precessing spiral or shrinking trefoil in either grv - m or gau - m model as one can see @xcite .
let us give an ansatz for @xmath144 , ^2(t)=^_1 ^ 2 . here , we use the minkowskian values of @xmath171 and @xmath172 , and will obtain the possible values of @xmath173 in both models . from , by ignoring the nonrenormalisable term and recalling @xmath163 , we can find the following proportion relations : @xmath174 and @xmath175 in model a , where we used @xmath140 . in model b ,
we obtain @xmath176 and @xmath177 . therefore , we set @xmath178 in model a , and @xmath179 in model b. we believe that our anstze are valid as long as the nonrenormalisable term does not play a role and the frequency of the coherent rotation , @xmath180 , is rapid compared to hubble expansion rate , @xmath181 .
the latter restriction implies that the rotation time scale is much shorter than the time scale of the hubble expansion , i.e. @xmath182 @xcite . [ [ section-3 ] ] in , we plot the evolution of @xmath144 with the numerical data ( red - plus dots ) for grv - m model ( left ) and for gau - m model ( right ) and with the analytic data ( green - dotted lines ) using our anstze .
the readers should compare the minkowskian cases of set-3 ( left bottom panel ) and set-7 ( right bottom panel ) in with the corresponding expanding background cases .
for both cases , the amplitudes of @xmath144 decrease in time as we expected , and our analytic plots excellently agree with the corresponding numerical results . in the left panel of
, the difference between the analytic line and the numeric plots arises for the late time .
we believe that this comes from the artifact of the approximation on @xmath140 in grv - m model , , since the values of @xmath144 decrease to the region where the above approximation does not hold , i.e. for @xmath141 as we saw in the left - top panel of .
[ [ section-4 ] ] in , we plot the numerical values of the equation of state , which is given by @xmath183 , where @xmath184 and @xmath15 in are the pressure and energy density of the ad condensate .
the averaged pressure over the rotations seems to be negative in grv - m model , see the left panel ; whereas , the pressure in gau - m model is always negative , see the right panel .
the frequencies of the rotation for @xmath185 in both cases are , respectively , similar as the corresponding frequencies of @xmath144 , see ; however , the phases are different from the phases of @xmath144 by @xmath186 . in summary
, we have analytically obtained the nearly circular orbits for both grv - m model and gau - m model in approximated by .
we then checked that the ( semi-)analytic results in and and our anstze in agree well with the corresponding numerical results obtained by numerically solving . in the rest of this paper
, we investigate the late evolution for the ad condensates once the spatial perturbations generated by quantum fluctuations or thermal noise from the early oscillation @xcite become non - negligible due to the negative pressure presented in and .
in this section , we analyse the late evolution of the ad condensates in both grv - m and gau - m models , in which we find that the spatial perturbations are amplified exponentially due to the presence of the negative pressure , and the presence of negative pressure supports the existence of nontopological solitons , i.e. @xmath0-balls . as a process of reheating the universe , the dynamics of the @xmath0-ball formation is nonequilibrium , nonperturbative , and nonlinear process , and it includes three distinct stages : _ pre - thermalisation _ ( linear perturbation ) , _ driven turbulence _ ( bubble collisions ) , and _ thermalisation _ towards thermal equilibrium . as opposed to the reheating process , we report that the driven turbulence stage lasts longer and the subsequent thermalisation process is different , which is caused by the presence of nontopological soliton solutions . during the turbulent stages
, we find scaling laws for the variances of fields and for the spectra of the charge density .
in addition , we adopt numerical lattice simulations to solve classical equations of motion , where our numerical code is developed from latfield @xcite , and we present detailed nonlinear and nonequilibrium dynamics ( some videos are available @xcite ) .
the late evolution , after the ad condensate forms , depends on the properties of models . in the standard ad baryogenesis scenario @xcite ,
the condensate govern by the quadratic potential , , decays into thermal plasma which may give our present baryons / leptons in the universe . by including quantum and/or thermal corrections in the mass term as in
, the subsequent evolution may be different from the standard ad scenario since the ad condensate has a negative pressure .
the negative pressure , which causes the attraction force among particles in the condensate , amplifies exponentially the linear spatial fluctuations .
we see this exponential growth for the linear perturbations in nearly circular orbit cases with the growth rate @xmath187 , and obtain the most amplified wave - number @xmath188 , which give a rough estimate on the nonlinear time @xmath189 and the radii of bubbles created just after the system enters into a nonlinear regime . as long as the perturbations are much smaller than the background field values
, we call this initial linear perturbation stage _pre - thermalisation_. let us consider the linear spatial instability for an ad condensate in minkowski spacetime .
first , we perturb the ad field @xmath6 with the linear fluctuations , @xmath190 and @xmath191
. equations of motion for @xmath190 and @xmath191 are given by , --2=0 , + [ a9 ] + -^2+=0 .
let us rescale @xmath190 and @xmath191 in the following form ~_0 e^s(t)+i,~_0 e^s(t)+i .
notice that both of the exponents @xmath192 should be the same in terms of a function of the wave number @xmath193 , because we are concerning with linear perturbations . substituting into ,
we obtain ( cc + ^2+^2 -^2+v^ & -2 + 2 & ^2 + + ^2 + ) ( cc + + ) 0 , where @xmath194 and we ignore the terms @xmath195 , assuming that the linear evolution is adiabatic , i.e. @xmath196 ( wkb approximation ) . notice that this assumption is violated only at the beginning of this linear evolution as we will see in the numerical subsection , sec .
[ qb_num ] . the nontrivial solution for @xmath197 can be obtained by taking the determinant of the matrix in , namely @xmath198 notice that the terms involving @xmath107 vanish if the orbit of the ad field is exactly circular . by looking for the most amplified mode @xmath199 , which is defined by @xmath200 from
, it implies that k^2_m=->0 , where the inequality comes from the reality condition for @xmath188 . by concerning with this mode in and by solving @xmath201 in , the solution for @xmath202 is _
m= , in which we are interested in the growing mode , i.e. @xmath203 . substituting into ,
we may obtain the most amplified mode . although it is rather hard to analytically solve , we know that the only one instability band exists for exactly circular orbits where @xmath204 ; 0<^2 < ^2-v^ ( ) , where @xmath205 and @xmath108 are time - independent due to the circular orbits . in addition , we can estimate a possible nonlinear time @xmath189 when the spatial averaged variance , var@xmath206 , becomes comparably large to the corresponding homogeneous mode @xmath8 . here , we defined var@xmath207 , and a hat and a bar denote an original field and a spatial average of the field , respectively .
notice that the nonlinear time in @xcite is defined by the time when the linear fluctuation @xmath208 for the most amplified mode becomes comparable large to the homogeneous - mode ; however , our definition is of more advantages as we see in the numerical subsection , sec .
[ qb_num ] .
the nonlinear time with our definition can be given by var()&~ & ^2_0 ~^2_0 ~^2_0 , + [ nlt]t_nl & ~ & t _ * + .
here , we approximated that @xmath209 and that the orbits over @xmath210 rotations with the period @xmath211 , , can be expressed by the integral form as shown in .
as we assumed , the spatial averaged variance of this field is not fully developed over all modes except @xmath212 until @xmath213 , where @xmath214 is a typical time scale when the variance starts to grow with the growth rate @xmath215 .
our main interest in this pre - thermalisation stage is the evolution of the number of particles in terms of modes , so that we consider @xmath5 as the particle number here . for a free field theory , both of the positive and charged particle occupation number
develop equally .
the present case , however , gives different consequences due to the presences of nonlinear interactions and the initial inequality of a charge density ( baryon asymmetry ) . without loss of generality
, we can focus on the case where the positive charge is initially present . since the charge density is given by @xmath216
, we can approximately obtain the evolution in the linear regime using , _
hence , the charge density evolves due to the linear fluctuation of the phase field .
let @xmath217 be the amplitude of fourier - transformed positive and negative charge density , @xmath218 , which are defined through the following decomposition , @xmath219 .
notice that the fourier transformed functions , @xmath220 , are related to , but are potentially different from the corresponding quantum mechanical expressions , @xmath221 and @xmath222 . here ,
@xmath223 are occupation numbers for positive and negative charged particles in a free field theory , and @xmath224 and @xmath225 are the annihilation / creation operators for both of the particles , respectively .
since we are interested in the growing mode for the positive charge density @xmath226 in which is initially zero except the zero - momentum mode , it implies that using n^+_k(t ) & & k^2|_0|^t_t_0 d ^2()e^ , + [ n+k ] & ~ & k^2|_0| ^2_cr e^(t - t_0 ) , where @xmath168 is a numerical value and we assumed @xmath227 , going from the first line to the second one .
therefore , the evolution of the positive charged particle number for a mode @xmath228 is proportional to @xmath229 .
our results , , are generalisations of the known results @xcite , in which the orbit of the affleck - dine condensate is exactly circular .
we also obtained the nonlinear time @xmath189 in and the exponential growth of the particle number in . using the results obtained in the previous subsection
, we can compute the most amplified mode @xmath230 and the growing mode @xmath231 averaged over one rotation of the nearly circular orbits for the models introduced in section [ model - abc ] , i.e. model a and model b. we shall confirm that these values are the same as the cases when the orbits are exactly circular , which implies that the instability band , , could exist even for the present nearly circular orbit cases . [ [ section-5 ] ] substituting the expressions , @xmath232 and @xmath233 ( _ c.f . _ and ) , into , we obtain the averaged growing factor for model a where @xmath234 : & & , + [ ext13 ] & & , where we substituted into to obtain @xmath230 and these results are consistent with the case for the exactly circular orbit . in order to satisfy @xmath235 , we should have @xmath236 , and implies that the condensate is unstable against spatial fluctuations when the pressure is negative with @xmath237 , see . we can check the results @xcite that @xmath238 and @xmath239 by setting @xmath140 in and by ignoring the nonrenormalisable term as did in @xcite , i.e. @xmath240 and @xmath241 .
these are of the same order as their results , recalling that @xmath242 due to @xmath55 .
[ [ section-6 ] ] similarly , we can also obtain the averaged growing factor for model b from , whose leading orders reproduce the results @xcite in which case that the ad orbit is assumed to be exactly circular and ignored the nonrenormalisable term . before we finish this subsection , let us remark the classical and absolutely stability of ad condensates .
lee found @xcite that the dispersion relations for the waves of linear fluctuations from when the orbits of the ad field are bounded .
for the longwave - length limits , there exist one massive and one massless modes
. the massless mode can be interpreted as the sound wave whose sound speed should be real for the classical stability reason , and the squared value of the sound speed is related to the value of @xmath68 in . therefore , this stability condition for the sound waves corresponds to the sign of the pressure in the ad condensates . in other words ,
the ad condensate has a negative pressure when the sound speed is imaginary ; equivalently , it is classically unstable against spatial fluctuations .
the zero - pressure ad condensate whose energy density is minimised with respect to any degrees of freedom is equivalent to the @xmath0-matter phase as coleman discussed in @xcite , where the absolutely stable @xmath0-matter can be excited by classically stable sound waves . even when the perturbations are fully developed to support the nonlinear solutions ,
the system is still far from equilibrium . indeed
, the system enters into more stochastic stages , turbulence regimes , where the strength of the turbulent behaviour depends on the `` reynolds '' number @xcite .
as a theory of reheating universe , a general nonequilibrium system goes through two different turbulence stages , going from driven turbulence to free turbulence stages .
a major energy transfer from the zero mode takes place during driven turbulence .
garcia - bellido _ et .
_ @xcite observed that bubbles form and collide during this stage in tachyonic preheating , and they suggested that the bubble collisions can be an active source of gravitational waves @xcite . in the usual reheating scenarios ,
this stage terminates when the energy left out in the zero - mode becomes smaller than the energy stored in other modes ( created particles ) .
since the energy exchange between zero - mode and other modes becomes negligible , the particle distribution is self - similar in time ( free turbulence ) and evolves towards thermal equilibrium . in the free turbulence stage ,
the quantum effects change the late evolution significantly , and the created particles are distributed followed by bose - einstein statistics rather than by a classical manner .
as long as an active and stable energy source exists in momentum space , we expect that the driven turbulence stage lasts for a long time . in the case of @xmath0-ball formation
, we expect that this active energy source corresponds to the excited states of @xmath0-balls ; hence , we can expect that the driven turbulence stage lasts longer compared to the linear perturbation regime as opposed to the usual reheating universe scenarios .
note that during this thermalisation stage the transition from the classical to quantum regime becomes important @xcite ; in the rest of this paper we concentrate on the case where the system is govern by classical evolution all the time . in turbulent stages , the scaling law can be found @xcite : ( ) t^p , where the power @xmath16 depends on the parameters of models , e.g. the relativistic values of @xmath16 are @xmath243 in the driven turbulence regime and @xmath244 in the free turbulence regime .
here , @xmath42 is the number with which particles interact . for the free turbulence regime ,
the particle number distribution follows a scaling law from the time @xmath245 when the free turbulence turns on , namely n_k(t)=t^-n_k_*(t = t_free ) , where @xmath246 . in this sub - subsection , we show that the condition of the negative pressure is the same as the existence condition of @xmath0-balls , known in @xcite .
this does not always mean that the spatially unstable condensate evolves towards to @xmath0-balls ; with given initial conditions , the condensate may evolve into other thermo - dynamically favour states in which the free energy is minimised .
the ansatz of @xmath0-balls claims that @xmath247 , which corresponds to the `` chemical potential '' @xmath248 , is constant , and that the radial field @xmath249 should be time - independent and depend on the radius @xmath250 of the @xmath0-ball , i.e. @xmath251 .
hence , the existence condition of @xmath0-balls at zero - temperature is ^2
this condition implies that the potential should grow less quickly than a quadratic term ; thus , it is equivalent to the fact that the ad condensate has a negative pressure for @xmath252 in , see . notice that this condition only tells that @xmath0-balls may appear after a unstable ad condensate fragments .
the evolution to the thermal equilibrium state is rather hard to compute analytically , and it is related to stability problems of the @xmath0-balls @xcite . therefore , we conduct numerical lattice simulations that give the entire processes of nonlinear as well as nonequilibrium evolution . in this subsection ,
we present detailed numerical results involved with lattice simulations for both grv - m and gau - m models with the parameter sets , set-3 and set-7 shown in ; we then check our analytical results obtained in the previous sections . in order to solve the second - order partial differential equations , @xmath253 , with the potentials introduced in
, we use the following appropriate parameters : @xmath254 in grv - m model and @xmath255 in gau - m model , which minimise the numerical errors . here
, @xmath256 is the fundamental lattice space and @xmath131 is the time step .
note that the variables in this subsection are normalized by appropriate energy scales as in sec .
[ numad ] .
we then conduct @xmath1 ( and @xmath2)-dimensional lattice simulations with @xmath257 ( and @xmath258 ) lattice units , imposing a periodic boundary condition .
our initial conditions are , @xmath259 and @xmath260 , where the initial fluctuations , @xmath261 and @xmath262 , are of a gaussian noise , which are responsible for `` quantum '' fluctuations .
their fluctuations , @xmath261 and @xmath262 , are of order @xmath263 in grv - m case and of order @xmath264 in gau - m case . in order to visualise these detailed evolution
, we use a 3d software , vapor @xcite , and some videos of our numerical results are available in @xcite .
[ [ section-7 ] ] in the top two panels of , we plot the amplitude of @xmath226 , where we took the average of @xmath265 over the axes of @xmath193 .
we show the amplitudes of @xmath226 for grv - m model in the left panel and for gau - m model in the right panel with two different time steps . in the panels ,
we indicate the analytical values of the most amplified modes @xmath188 obtained from with black - dashed vertical lines . in grv - m model ,
the amplitude with @xmath266 ( green - dashed line ) is a little noisy to see the first peak @xmath267 in terms of @xmath228 .
our analytical estimate , @xmath268 , is located at a more infrared region than the point @xmath269 , and the periodic structure can be seen in the higher - momentum space . in gau - m model , on the other hand , we can confirm that our analytical value , @xmath270 , agrees with the numerical value , @xmath271 , in the green - dashed line ; however , the analytical value appears in a slightly more infrared region .
we also observe the periodic structure in the higher - momentum modes as it was reported in @xcite . in the middle panels ( grv - m model on left and gau - m model on right )
, we compare both the zero - mode , @xmath272 ( red - solid lines ) , and the homogeneous field , @xmath273 ( green - plus dots ) , shown in the bottom panels of .
the middle panels in both cases show that the zero - mode does not decay quickly , and it oscillates around @xmath274 .
we can also check that our numerical parameters are appropriate , minimising numerical errors . in the bottom panels of
, we plot the evolution of @xmath275 for the modes both @xmath188 ( red - solid lines ) and @xmath267 ( green - dashed lines ) . in the left bottom panel
, we can see the exponential growth of the amplitude in grv - m model for both modes , and step - like particle production exists at the beginning of the evolution as broad resonant preheating @xcite ( c.f . ) , and it begins to create the particles exponentially afterwards .
the particles are produced quickly when the zero - mode @xmath144 increases in time at the beginning , see the middle panels .
this is the different feature of the evolution compared to the case of resonant preheating , where particle production for the broad resonance occurs nonadiabatically when the zero mode ( inflaton field ) crosses the zero axis . in the right bottom panel
, we can see more clearly the step - like particle creations for both modes , and then this step - like evolution smoothes out , which leads to the exponential particle production as in the grv - m case .
we believe that the adiabatic condition , @xmath276 , is `` softly '' violated only in this initial stage since we can not see the clear exponential growth at the beginning of this evolution . in the next paragraph
, we discuss the late linear evolution when this nonadiabatic evolution ceases , and show that our analytical results agree with numerical ones more nicely .
+ + [ [ section-8 ] ] in , we show the evolution of the various physical quantities in the late stage of linear perturbations : @xmath277 and var@xmath206 . the top panels plot the amplitude of @xmath278 with various times in both grv - m model ( left ) and gau - m model ( right ) . notice that we plot it against the logarithmic scale of @xmath228 as opposed to the linear scale seen in the top panels of .
for all time steps shown there , our analytical values of @xmath188 agree well with the first peak mode @xmath267 , at which the amplitudes are most amplified .
notice that the zero - momentum mode does not decay in both cases .
after the first peak of the amplitude is well developed , the second peak appears in the spectra , and later the third peak can be barely observed . roughly speaking ,
the @xmath279 peaks appear around the values which are @xmath188 multiplied by @xmath50 .
these higher peaks are suppressed by rescattering processes in which a particle from the first peak transfers some of its momentum to a particle from the zero - momentum modes ( ad condensates ) @xcite .
later , all modes of the particle spectra , @xmath278 , develop quickly , but the first peak is still visible .
the middle panels illustrate the evolution of a zero - mode field @xmath272 and the variance of the field var@xmath206 up to the nonlinear time @xmath280 . as we saw in the top panels , the zero mode does not decay even after the nonlinearity comes in , whilst the variance of the field develops exponentially from @xmath281 in grv - m model ( left ) and from @xmath282 in gau - m model ( right ) .
this delay of the exponential growth comes from the fact that the other modes do not evolve initially except the mode @xmath188 ; thus , we can set these times as @xmath214 defined in .
we fit a function , @xmath283 , against the exponential evolution for the variations , where @xmath284 is a numerical value , and we obtained @xmath285 in grv - m model and @xmath286 in gau - m model , which match satisfactorily with the analytical ones in , where we computed as @xmath287 in grv - m model and @xmath288 in gau - m model . from the middle panels , the nonlinear time is approximately both @xmath289 in grv - m model and @xmath290 in gau - m model , and these values agree well with the analytical estimates in , where the analytical values are @xmath291 in grv - m model and @xmath292 . in the bottom panels ,
we plot the evolution of the amplitude @xmath278 for the first peak mode ( red - plus dots ) , second peak mode ( green - cross dots ) and the analytical most amplified modes ( purple squared - cross dots ) .
the numerical values of the exponents for the most amplified modes @xmath188 in blue long - dotted lines , ( @xmath293 in grv - m model and @xmath294 in gau - m model ) match with the analytical ones in .
the second peaks @xmath295 in black short - dotted lines start to grow at @xmath296 in grv - m model and at @xmath297 in gau - m model , and we can set these values as @xmath168 defined in .
the initial behaviour of the amplitude of second peak of @xmath278 seems to be quasi - periodic , which implies that @xmath298 for the mode , @xmath295 , is pure imaginary , see ( c.f .
the bottom panels of fig . 5 in @xcite ) .
surprisingly , the growth rates are about twice larger than the values of both @xmath231 and @xmath284 for var(@xmath8 ) and @xmath267 .
note that the initial evolution is not adiabatic , so that the growth rates are not strictly exponential as we have seen in the bottom panels of .
for example , the growth of the first peaks , @xmath188 ( or @xmath267 ) , in gau - m model is not exponential initially , but it becomes exponential as the growth of the second peak mode @xmath295 .
+ + [ [ section-9 ] ] in , we show the snapshots of the positive charge density @xmath299 for grv - m model ( left panels ) and gau - m model ( right panels ) around @xmath300 , where timestep in the panels denotes the actual time divided by @xmath301 in grv - m model and the actual time divided by @xmath302 in gau - m model .
the colour bars illustrate the values of the positive charge density .
we can see long - wavelength objects ( sometimes called filaments ) in both cases , and the charge in some regions is compactified into spheres , see bottom panels .
these filaments and bubbles correspond to the nonlinear solutions , which may be nontopological strings @xcite and the excited states of @xmath0-balls , respectively .
the radii of these bubbles are of the same order as the wave - length which corresponds to the most amplified modes , @xmath188 .
as we will see in the next subsection , these bubbles grow by colliding and merging each other .
note that this bubble creation is nothing to do with bubble nucleation in first - order phase transition as opposed to the case in @xcite , in which case the ad condensate is classically stable against special perturbation , but not quantum mechanically . for grv - m model ( left panels ) and gau - m model ( right panels ) around @xmath300 , where timestep in the panels denotes the actual time divided by @xmath301 in grv - m model and the actual time divided by @xmath302 in gau - m model , and the colour bars illustrate the values of the positive charge density .
after the nonlinearity is fully developed , many bubbles form , which are pinched out of `` highly '' charged filaments.,title="fig : " ] for grv - m model ( left panels ) and gau - m model ( right panels ) around @xmath300 , where timestep in the panels denotes the actual time divided by @xmath301 in grv - m model and the actual time divided by @xmath302 in gau - m model , and the colour bars illustrate the values of the positive charge density .
after the nonlinearity is fully developed , many bubbles form , which are pinched out of `` highly '' charged filaments.,title="fig : " ] + for grv - m model ( left panels ) and gau - m model ( right panels ) around @xmath300 , where timestep in the panels denotes the actual time divided by @xmath301 in grv - m model and the actual time divided by @xmath302 in gau - m model , and the colour bars illustrate the values of the positive charge density .
after the nonlinearity is fully developed , many bubbles form , which are pinched out of `` highly '' charged filaments.,title="fig : " ] for grv - m model ( left panels ) and gau - m model ( right panels ) around @xmath300 , where timestep in the panels denotes the actual time divided by @xmath301 in grv - m model and the actual time divided by @xmath302 in gau - m model , and the colour bars illustrate the values of the positive charge density .
after the nonlinearity is fully developed , many bubbles form , which are pinched out of `` highly '' charged filaments.,title="fig : " ] [ [ section-10 ] ] in , we show the snapshots of the positive charge density for grv - m model in different time steps up to @xmath303 , where timestep in the figure denotes the actual simulation time divided by @xmath302 and the colour bars illustrate the values of the positive charge density .
after the system goes into a nonlinear regime , we can see a few lumps in the first few panels of the snapshots , and those lumps merge into larger lumpy objects .
finally , we can see a large cluster which consists of a complicated inner structure , see the last snapshot . recall that we are using the periodic boundary condition . and @xmath304 ) , where timestep in the figure denotes the actual simulation time divided by @xmath302 and the colour bars illustrate the values of the positive charge density .
a few created lumps collide and merge into a large cluster in the end.,title="fig : " ] and @xmath304 ) , where timestep in the figure denotes the actual simulation time divided by @xmath302 and the colour bars illustrate the values of the positive charge density .
a few created lumps collide and merge into a large cluster in the end.,title="fig : " ] and @xmath304 ) , where timestep in the figure denotes the actual simulation time divided by @xmath302 and the colour bars illustrate the values of the positive charge density .
a few created lumps collide and merge into a large cluster in the end.,title="fig : " ] + and @xmath304 ) , where timestep in the figure denotes the actual simulation time divided by @xmath302 and the colour bars illustrate the values of the positive charge density .
a few created lumps collide and merge into a large cluster in the end.,title="fig : " ] and @xmath304 ) , where timestep in the figure denotes the actual simulation time divided by @xmath302 and the colour bars illustrate the values of the positive charge density .
a few created lumps collide and merge into a large cluster in the end.,title="fig : " ] and @xmath304 ) , where timestep in the figure denotes the actual simulation time divided by @xmath302 and the colour bars illustrate the values of the positive charge density .
a few created lumps collide and merge into a large cluster in the end.,title="fig : " ] shows the detailed evolution of the positive charge density for gau - m model in different time steps up to @xmath305 , where timestep in the figure denotes the actual simulation time divided by @xmath306 and the colour bars illustrate the values of the positive charge density
. a large number of small bubbles can be observed , and nearby bubbles collide and merge into larger bubbles . in the final panel , there are smaller number of bubbles left ( compare to the first panel ) .
we believe that this time arrow is followed by the fact that the total energy of large bubbles is smaller than the total energy of smaller bubbles , c.f .
fission stability of @xmath0-balls in @xcite .
these large bubbles are able to carry a large amount of charge inside of them as we saw in the left panel of fig . 9 in @xcite in the `` thin - wall '' @xmath0-ball limit .
the differences of the evolution between grv - m and gau - m models come from a number of facts , e.g. different initial conditions , stability conditions and momentum fluxes due to asymptotic profiles at a large distance from the cores , see @xcite . and @xmath307 ) , where timestep in the figure denotes the actual simulation time divided by @xmath306 and the colour bars illustrate the values of the positive charge density .
there are smaller number of bubbles left in the end.,title="fig : " ] and @xmath307 ) , where timestep in the figure denotes the actual simulation time divided by @xmath306 and the colour bars illustrate the values of the positive charge density .
there are smaller number of bubbles left in the end.,title="fig : " ] and @xmath307 ) , where timestep in the figure denotes the actual simulation time divided by @xmath306 and the colour bars illustrate the values of the positive charge density .
there are smaller number of bubbles left in the end.,title="fig : " ] + and @xmath307 ) , where timestep in the figure denotes the actual simulation time divided by @xmath306 and the colour bars illustrate the values of the positive charge density .
there are smaller number of bubbles left in the end.,title="fig : " ] and @xmath307 ) , where timestep in the figure denotes the actual simulation time divided by @xmath306 and the colour bars illustrate the values of the positive charge density .
there are smaller number of bubbles left in the end.,title="fig : " ] and @xmath307 ) , where timestep in the figure denotes the actual simulation time divided by @xmath306 and the colour bars illustrate the values of the positive charge density .
there are smaller number of bubbles left in the end.,title="fig : " ] [ [ section-11 ] ] we show the snapshots of the negative charge density for grv - m model ( left panel ) at @xmath303 and gau - m model ( right panel ) at @xmath308 in , where the colour bars illustrate the values of the negative charge density .
these times correspond to the same times as in the final snapshots of .
the values of charge density in both models are much smaller than the values of positive charge density in .
this implies that we are observing the plots of thermal plasma rather than the charged ( nonlinear ) lumps .
their distributions are quite different each other .
the negative charge density for grv - m model is surrounded by the large positive charged cluster seen in the last panel of , and it is distributed all over the lattice ; whereas , for gau - m model the distributions of the negative charged plasma are highly concentrated only around the surface of the lumps ( compare the last panel of ) . and gau - m model ( right panel ) at @xmath309 , where the colour bars illustrate the values of the negative charge density . the negative charge density for grv - m model is surrounded by the large positive charged cluster ; however , the distribution spreads out over the lattice space , whereas the negative charge density for gau - m model is concentrated around the positive charged lumps ( compare them to the last panels of ) .,title="fig : " ] and gau - m model ( right panel ) at @xmath309 , where the colour bars illustrate the values of the negative charge density .
the negative charge density for grv - m model is surrounded by the large positive charged cluster ; however , the distribution spreads out over the lattice space , whereas the negative charge density for gau - m model is concentrated around the positive charged lumps ( compare them to the last panels of ) .,title="fig : " ] [ [ section-12 ] ] the top panels of show the evolution of zero - mode ( red - solid lines ) and the variations for @xmath8 ( dotted - dashed purple lines ) , whose latter evolution are fitted by a function , @xmath310 , ( black dashed lines ) , where @xmath311 is a numerical value as the power of . for both models ( grv - m model on the left panel and gau - m model on the right panel ) ,
the asymptotic evolution after the linear perturbation regime is overlapped by the function , where @xmath312 for grv - m model and @xmath313 for gau - m model .
our analytic values can be matched by setting @xmath314 with @xmath315 in grv - m model and @xmath316 with @xmath317 in gau - m model , see .
hence , we could identify this regime as driven ( stationary ) turbulence , and the main dynamics in each model is caused by either a five - particle interaction or three particle interaction , respectively .
note that our nonrenormalisation term has a @xmath318 term in both models . in the middle and bottom panels of , we plot , respectively , the amplitudes of @xmath278 and @xmath319 in different times for grv - m model ( left panels ) and gau - m model ( right panels ) . for @xmath320 of grv - m model ,
the amplitudes of the high momentum modes grow in time , whilst the lower momentum modes do not decay completely and stay for a long time .
we fit a function , @xmath321 , ( yellow dotted lines ) where @xmath322 is a numerical value onto the spectra at @xmath323 for the region where the function is fitted as shown in black dashed lines .
we find that @xmath324 for the @xmath278 case and @xmath325 for the @xmath319 case . in the right middle and bottom panels , we plot the amplitudes of @xmath320 for gau - m model in various times .
the amplitudes of the high momentum modes decrease as opposed to the grv - m case , and the slopes of the spectra for @xmath320 at @xmath326 in yellow - dotted lines are steeper than the grv - m case , where we fit the numerical spectra by the following values shown in black dashed lines : @xmath327 for the @xmath278 case and @xmath328 for the @xmath319 case .
+ + in order to significantly reduce the simulation time , we carry out @xmath2-dimensional lattice simulations with the same initial conditions used in @xmath1-dimensional cases , where our lattice units are reduced from @xmath257 to @xmath258 . in the top panels of ( grv - m model in the left panels and gau - m model in the right panels ) , we illustrate the evolution of zero - mode and the variances of @xmath8 , and in the bottom panels we plot the energy density ( at @xmath329 in the left - bottom panel and at @xmath330 in the right - bottom panel ) instead of the charge density to compare with the @xmath0-ball profiles at zero - temperature , which we obtained in fig . 3 and fig . 7 in @xcite .
the colour bars in the bottom panels of illustrate the values of energy density .
note that we are using the same parameters for grv - m model as the ones used in @xcite , whilst the potential for gau - m model used there is a generalised version of our present potential , so the profiles in gau - m model should look similar only qualitatively , but not quantitatively . from the top panels , we can also see , in particular grv - m model , the scaling exponent evolution during the driven turbulence stage after the pre - thermalisation ends as confirmed in the top panels of .
the subsequent evolution , however , is different each other and also unique apart from a characteristic free turbulence stage .
these features of thermalisation process are caused by stable nonlinear solutions , namely `` @xmath0-balls '' ; in grv - m model ( left panels ) , the variance does not evolve that much after the driven turbulence stage ends and we can see thin walled like charged lumps in the end , see the left - bottom panel . in gau - m model
( right panels ) the variance has a step - like evolution , at which stage we confirmed that two ( or sometimes more ) charged lumps collide and merge into a larger lump .
the collision rate is very low since the motions of these `` heavy '' bubbles are nonrelativistic , but we expect that there will be only one single @xmath0-ball left ultimately as similar as the grv - m case .
generally , we observe that almost all of the total energy is trapped into these lumps , where we also confirm that the total charge is absorbed into these lumps , as reported in @xcite . as the `` thin - wall '' @xmath0-balls in gau - m model
do not have an extremely thin - wall thickness @xcite , the profiles seen in the right bottom panel do not have such a thin - shell thickness .
note that the `` thick - wall '' @xmath0-balls in gau - m model may suffer from classical instability and fission against spatial perturbations around the @xmath0-ball solutions , and decay into smaller @xmath0-balls as opposed to the case of `` thick - wall '' @xmath0-balls in grv - m model .
the reader should also notice that the potential for gau - m model in the present case is different from eqs .
( 40 ) and ( 41 ) in @xcite , which may change the classical stability of the @xmath0-balls in the `` thick - wall '' limit .
furthermore , the stability of @xmath0-balls is related to their own charge @xmath0 so that the initial ratio , @xmath331 , can also cause the different evolution .
therefore , we believe that the evolution is very sensitive to the parameters of models used and the initial conditions .
it is worth mentioning , in the left - bottom panel , that the value of charge density within the charged cluster is slightly larger than the value of the thin - wall @xmath0-balls in the zero - temperature case ( compare to right bottom panel of fig . 3 in @xcite ) .
we believe that this is because this charged cluster appears in the thermal background , in which the thermal effects change their profiles .
+ dimensions : the top panels show the evolution of zero - mode ( red - solid lines ) and the variations for @xmath8 ( dotted - dashed purple lines ) . in the bottom panels ,
we plot the energy density ( at @xmath329 in the left - bottom panel and at @xmath330 in the right - bottom panel ) instead of the charge densty to compare the @xmath0-ball profiles seen in fig . 3 and fig . 7 of @xcite where the colour bars illustrate the values of energy density .
we can see that almost all of the charge is trapped into bubbles which may be `` thin - wall '' @xmath0-balls , recall that we are imposing a periodic boundary condition.,title="fig : " ] dimensions : the top panels show the evolution of zero - mode ( red - solid lines ) and the variations for @xmath8 ( dotted - dashed purple lines ) . in the bottom panels , we plot the energy density ( at @xmath329 in the left - bottom panel and at @xmath330 in the right - bottom panel ) instead of the charge densty to compare the @xmath0-ball profiles seen in fig . 3 and fig . 7 of @xcite where the colour bars illustrate the values of energy density
we can see that almost all of the charge is trapped into bubbles which may be `` thin - wall '' @xmath0-balls , recall that we are imposing a periodic boundary condition.,title="fig : " ] let us recap our findings in this section .
we have shown in both grv - m and gau - m models that the ad condensate that has a negative pressure is generally unstable against linear fluctuations , and the fluctuations evolve exponentially .
the condition for the presence of the negative pressure corresponds to the existence condition of @xmath0-balls , and under our initial conditions shown in , we observed that almost all of the total charge is trapped into a single ( and a few ) spherical lump(s ) ( `` thermal @xmath0-balls '' ) in the end of our numerical simulations . in the intermidiate regions between the initial exponential amplification stage and thermalisation stage in the presence of the nonlinear solutions
, we identified that the driven turbulence is active ; we then found the scaling exponent evolution for the variance of @xmath8 , and we saw that this stage lasts much longer than the case of tachyonic reheating .
in this paper , we discussed both analytically and numerically two main issues : affleck - dine ( ad ) dynamics and their subsequent nonequilibrium dynamics in the presence of nonlinear solutions .
we showed that the ad dynamics has the same features as orbital motions of planets , replacing the gravitation force by an isotropic harmonic oscillator force .
as the relativistic correction on the newtonian potential gives a precession for the planetary orbit motion , the orbits of ad fields are disturbed by the nonrenormalisable and quantum correction terms .
note that the essential origin of these corrections is physically different . in the presence of a negative pressure of the ad condensate
, we have shown that the condensate is classically unstable , and the evolution of the system is similar to the dynamics of reheating the universe , i.e. _ pre - thermalisation _ , _ bubble collisions _ and _ thermalisation_. we found that the thermalisation process occurs in the presence of charged lumps , which merge into a single ( or a few ) `` thin - walled @xmath0-ball(s ) '' absorbing most of the homogeneous charge distributed initially in a lattice . in sec .
[ sectorbit ] , we introduced two phenomenological models which are motivated by the mssm , i.e. gravity - mediated ( grv - m ) model and gauge - mediated ( gau - m ) model .
we obtained the frequencies of the rotation for the nearly circular orbits , and showed that the condensate can have a negative pressure in both cases , see sec .
[ model - abc ] .
furthermore , we checked numerically our analytic results with the various cases in both a non - expanding and expanding universe .
our analytic expressions have a number of advantages . in the existing literature on preheating for complex scalar fields @xcite , the motion of the complex scalar field
is assumed to be of an elliptical form , but their ansatz does not hold ( compare our expressions in and and their ansatz ) . in the multi - flat direction cases , our analytic expressions of the ad field
give the exact mathieu equation when the interaction term between the ad field @xmath6 and another field @xmath332 , which parametrises another flat direction , is given by @xmath333 , where @xmath334 is a coupling constant between them .
the previous literature @xcite suggested that the resonant susy preheating for nearly circular orbits is not effective since the characteristic dimensionless quantity @xmath335 is much less than unity , recalling that broad resonant preheating ( nonadiabatic evolution ) occurs for @xmath336 .
this statement also holds for our case when the orbit of the ad field is nearly circular since @xmath337 where @xmath84 is the third eccentricity of the orbits , recalling that nearly circular orbits correspond to the case of @xmath338 .
we obtained the successful anstze , , for nearly circular orbits in an expanding universe ( see also the top panels in ) , but our analytical expressions could be improved by the action variable technique as a real scalar field case @xcite .
these issues on understanding the analytic orbit forms are related to the dynamics of spinning scalar fields , which can be responsible for the early- and late- time exponential expansions of the universe ( spinflation @xcite and spintessence @xcite ) since the ad condensate can possess a negative pressure , which can satisfy the condition of slow - roll inflation , @xmath339 .
it has been discussed in @xcite about an oscillating field responsible for dark energy ( see a recent review @xcite ) , and it gives a constraint on the power of a power - law potential in order to obtain the attractor solutions @xcite . as in real scalar fields , a complex scalar field has been investigated , see for example @xcite .
following our analytical work , one can investigate the further analysis on dark energy for a complex scalar field and their late evolution in order to place constraints on parameters of the models , avoiding @xmath0-ball formation . in sec .
[ sectinst ] , we explored the late evolution of ad fields in minkowski spacetime in both grv - m and gau - m models . as a usual nonequilibrium dynamics , we proposed that the dynamics of the @xmath0-ball formation goes through three distinct regimes : _ pre - thermalisation _ , _
bubble collision _ ( driven turbulence ) and _ thermalisation_. we showed analytically that the ad condensate is unstable against spatial perturbations when the condensate has a negative pressure , and the perturbations grow exponentially .
the presence of the negative pressure satisfies the existence of @xmath0-balls as well as the fact that the sound wave of the perturbation has an imaginary value of the sound speed .
assuming the adiabatic linear evolution , we have analytically presented that the perturbations for the most amplified mode @xmath212 in grow with the exponent @xmath187 in , which we obtained by taking the average over one rotation of the orbits of the ad field . in the previous literature @xcite , these values were obtained by ignoring the nonrenormalisable term and by assuming that the orbit is circular . by including the nonrenormalisable term and considering more general elliptic orbits , we recovered their results as the leading order of our solutions in sec .
we also showed that the nonlinear time is delayed compared to the time which the authors @xcite obtained , since the other modes are not well developed when the most amplified mode starts to grow exponentially . with our @xmath1-dimensional numerical lattice simulations that are run for much longer time with much larger simulation sizes than the past lattice simulations in @xcite ,
our analytic results are well checked .
in addition , we found that the adiabatic condition is violated at the beginning stage of the linear perturbations as seen in broad resonant preheating . in the driven turbulence stage , we observed that many bubbles form and collide / merge into larger bubbles in both grv - m and gau - m models .
note that these bubbles are nothing to do with bubbles due to first order phase transition . by concerning the variance of the radial field @xmath8 ,
we have seen that the evolution follows a scaling exponent law as a signature of the driven turbulence @xcite . as opposed to the case of tachyonic preheating , this driven turbulence stage , in our case , lasts for longer time , which may be caused by the presence of classical nonlinear solutions , i.e. `` @xmath0-balls '' .
we saw in @xmath2-dimensional numerical results that a thermalisation stage actually exists where the evolution for the variance of a field has a different scaling law from the one which appears in the driven ( first ) turbulence stage .
we believe that quantum effects should be non - negligible in this late turbulence stage , and the classical thermalisation process , in our case , should be different from the corresponding quantum - mechanical thermalisation . since the thermalisation process is generally extremely long , a lattice simulation in an expanding background encounters serious problems in the ultra - violet limits ; thus , we ignored hubble expansion in this paper . by considering the quantum - mechanical effects as well as hubble expansion , it is worth to investigate the cosmological consequences as our future work . in the context of a ( p)reheating scenario , it has been suggested @xcite that the collision of bubbles during the driven turbulence stage can be an effective source of gravitational waves , which can be detected by ligo @xcite and lisa @xcite in the near future .
we noticed that this analysis should be applicable to the same driven turbulence stage of the @xmath0-ball formation , which was initially proposed in @xcite .
the problem of gravitational waves emitted in the fragmentation stage has been discussed @xcite , while the analysis of the gravitational wave emissions in the driven turbulence stage still remains .
this work was partially done when the author visited the cosmology group , university of helsinki .
this visit was financially supported by the bests scholarship , university of nottingham .
our numerical code is developed from the code , latfield @xcite .
the numerical simulations were carried out by uk national cosmology supercomputer , cosmos , funded by stfc , hefce and silicon graphics , and carried out by nottingham hpc facility .
the author would like to thank e. copeland for continuous encouragement and positive comments on the final draft of this manuscript .
the author is also grateful to n. bevis , d. buck , k. enqvist , m. kawasaki , k. maeda , p. saffin and o. seto for valuable discussions , a. mazumdar for leaving useful comments , and a. josan for carefully reading this manuscript .
in this appendix , we obtain euler - lagrange equations for multiple scalar fields @xmath340 with a symmetric field space metric @xmath341 , following the notations @xcite .
our aim is to obtain equations of motion for the background homogeneous ( zero - mode ) fields @xmath342 and the perturbed fields @xmath343 in a fixed unperturbed background ( friedmann - robertson - walker ) metric , @xmath344 where @xmath3 is scale factor of the universe and @xmath4 is hubble parameter . here , an over - dot denotes the time - derivative .
as the simplest nontrivial example of the multiple scalar fields , we find equations of motion for a complex scalar field @xmath345 where @xmath346 and @xmath347 are real scalar fields and the system possesses a u(1 ) symmetry .
let us start off with the action s = d^4x , where @xmath348 and @xmath349 is a potential for @xmath350 .
because of the action principle , we obtain the euler - lagrange equation for @xmath351 _
= g^g_ab , c_^a _
^b + v_,c , and the energy momentum tensor t_=g_ab_^a _ ^b + g_. here , we defined @xmath352 , and so on .
the energy density and pressure can be given by @xmath353 @xcite _
e&=&-g^ g_ab _ ^a _ ^b + v ( ) , + [ press - multi ] p&= & -g^ g_ab _ ^a _ ^b - v ( ) . by pertubing the fields as @xmath354 where @xmath355 , the homogeneous part gives , from , ^a+3h ^a + g^abv_,b=0 , where the covariant derivative , @xmath356 , can be defined by the `` christoffel symbols '' @xmath357 @xmath358
; thus , @xmath359 . on the other hand ,
we can obtain the equations of motion for the pertubed fields @xmath360 from ^a+3h^a-^2^a-^a_bcd ^b^c^d+(v^;a)_;d^d = g^abg_bc , dg^cev_,e^d , where we used @xmath361 , defined the `` riemann tensors '' as @xmath362 , and denoted the covariant derivative as a usual notion @xmath363 .
notice that we used @xmath364 .
when the system has a @xmath365 symmetry as @xmath366 and a flat field metric is @xmath367 , we can obtain @xmath368 .
then , with a potential @xmath369 gives + 3h-^2+&=&0 , + [ thetaeom ] + 3h+&=&0 .
here , the third term in corresponds to the `` centrifugal force '' due to the spin in a field space , and the third term in corresponds to the `` colliori force '' .
in addition , the energy density and pressure can be given by from _
e=+v , p =- v . furthermore , gives & & + 3h--2=0 , + [ pertth]&&+-^2+=0 .
we use to concern with the orbits of ad condensates in sec .
[ sectorbit ] , and use to investigate the linear spatial instability of the condensates in sec .
[ sectinst ] .
in this appendix , we obtain an exact orbit form in a quadratic potential case when hubble expansion is assumed to be small and adiabatic .
the orbit of a ad field , or more precisely an eccentricity of the elliptic motion in the complex field - space , is determined by the initial charge and energy density . in order to obtain analytic expressions of the orbit in more general potential cases in which we are more interested , we restrict ourself to work in minkowski spacetime and on the orbit which should be nearly circular .
we then obtain the perturbed orbit equation and necessary conditions for closed orbits , where the orbits come back to their original positions after some rotations around the minimum of the effective potential . by including the effects of hubble expansion
, we shall introduce anstze in sec .
[ numad ] , which are inspired by our solutions obtained in minkowski spacetime , and our numerical results support the anstze , assuming that the rotation frequency @xmath170 is always much greater than hubble expansion @xmath370 @xcite .
the exact orbit expressions of an ad field in an expanding universe can be obtained with a quadratic potential , v=^2=^3 ^ 2 , where @xmath41 is a mass of the field @xmath6 and we rescaled the field @xmath8 , @xmath30 . from now on ,
we solve the orbit equations , , for @xmath371 at first , and then solve them for @xmath372 , replacing the time - dependence by a phase variable @xmath9 .
we then show that the orbits are of usual elliptic forms with a third eccentricity @xmath84 for @xmath371 and @xmath372 .
in this subsection we obtain an expression for the orbit @xmath371 by solving . substituting into and ignoring the terms involving @xmath33 and @xmath34
, we obtain -+m^2=0 = 0 , where @xmath373 , which is approximately conserved .
since @xmath374 , leads to a harmonic oscillator form , ( ^2)=-4m^2 whose solution is ^2(t)&=&+ a , + [ tisig]&=&. here , @xmath375 is some constant value and we set @xmath168 as a time when the ad field starts to rotate .
we also defined a third eccentricity @xmath376 , where the apocentral and pericentral points are , respectively , given by @xmath377 and @xmath378 .
notice that the circular orbit case corresponds to @xmath379 , which implies that @xmath380 , and also note that the eccentricity is real and has a value between 0 and 1 in the present quadratic potential .
we can obtain a period @xmath211 of this orbit , = . substituting into @xmath381
, we obtain @xmath382 . from the above expressions for @xmath84 and @xmath383
, we can obtain @xmath384 . using this and
, it ends up with ( t)==. for the circular orbits @xmath379 , @xmath205 is time - independent as we can expect , and the ratio , @xmath385 , is unity . while for the radial orbits @xmath386 , @xmath387 as we can also expect , and we can find @xmath388 .
what follows is that we express @xmath371 as a function of @xmath9 by using the second expression in .
we then obtain ( -_0)= , where @xmath390 is an integration constant and we used the following integral formula , @xmath391 with some real values @xmath392 and @xmath393 . without loss of generality , we can choose @xmath394 , which imply that the orbit at @xmath395 gives , respectively , @xmath396 , recalling . by comparing to , we obtain ^2()&= & , + [ uel1]^2()&=&=+ , + [ uel ] & = & . hence , we can see that @xmath397 when @xmath398 and @xmath399 when @xmath400 .
finally , we obtained the expressions for the orbits as usual elliptic forms in .
for the circular orbits , @xmath379 , we can obtain @xmath401 from as we can expect . without hubble expansion
, we can investigate a nearly circular bounded orbit of an ad field in general potentials which satisfy . because of the above reasons , we concentrate on non - expanding background in this subsection , and obtain a time - dependent expression for the nearly circular orbits as .
we then find the expression that depends on the phase @xmath9 as .
moreover we obtain conditions for closed orbits , in which the perturbations are expanded up to 1st order ( for the complete proof of the condition up to 4th order , see bertrand s theorem @xcite ) . in minkowski spacetime , we can find an expression for the orbit @xmath10 in a general potential @xmath369 as in .
notice that the tilde variables are the same as un - tilde ones in the present non - expanding background .
recall the equation of motion in minkowski spacetime , + = 0 .
suppose that the orbit that is nearly circular as @xmath402 , where @xmath403 , recalling @xmath19 is defined by .
substituting this expression of @xmath8 into and keeping @xmath404 terms up to 1st order , we obtain a harmonic oscillator form + w^2 = 0 , where the reader should recall the condition , , for the bound orbits , and @xmath170 is constant since we are working in minkowski spacetime .
thus , the solution of is ( t)=_cr b(wt ) , where @xmath375 is a small positive dimensionless constant , i.e. @xmath405 due to @xmath403 , and we set the differentiation constant as @xmath406 to ensure that @xmath407 .
we can find that @xmath408 , @xmath409 , and @xmath410 , which give @xmath411 , @xmath412 , and @xmath413 , where we used the definition of the third eccentricity .
we can check that the condition , @xmath414 , is consistent with the fact that the orbit is nearly circular .
since @xmath415 and @xmath15 is constant , we can equate @xmath375 with @xmath15 and @xmath5 using : b==_cr 1 , where a prime denotes the differentiation with respect to @xmath8 .
finally , we obtain ^2(t ) = ^2_cr , where @xmath170 is given by ( compare with ) .
now , we can define the period @xmath211 = , which reproduces the case for @xmath416 in .
using , we can also find .
using , let us compute the pressure of this ad condensate whose orbit is described by . by expanding @xmath417 around @xmath108 and using , we obtain @xmath418 where we assumed that the higher order terms in @xmath14 are negligible . therefore , p & & -v(_cr)+ , + [ pressure ] & & -v(_cr)+ . here
we defined an averaged value over an one rotation in the orbit , , namely @xmath419 where @xmath420 is some quasi - periodic quantity and @xmath211 is determined by .
the result , , can be easily understood by the fact that the averaged pressure corresponds to the value at @xmath108 since the orbit oscillate around @xmath19 and @xmath421 , c.f . a real scalar field case @xcite .
similarly , we can obtain the averaged energy density v(_cr)+ + , where we can see the contribution from the term involving @xmath422 .
hence , the averaged equation of state is given by = . in order to obtain a @xmath9-dependent expression of the orbit as ,
let us switch the variable @xmath8 to @xmath424 . in minkowski spacetime , where we can again drop the tilde variables there ,
the orbit equation is + u =- j(u ) .
let @xmath425 , which is independent of @xmath9 , be the value of a circular orbit ( i.e. @xmath426 ) .
we then consider an orbit @xmath427 that deviates slightly from @xmath425 with a fluctuation @xmath428 , i.e. @xmath429 , where @xmath430 .
since @xmath431 , implies that @xmath432 . by expanding @xmath433 around @xmath434 ,
we obtain @xmath435 , where we are evaluating the differentiations at @xmath425 .
hence , we can obtain the perturbed orbit equation for @xmath428 + ^2=0 , where @xmath436 which should be positive for bounded orbits .
note that this condition , @xmath437 , is equivalent to the previous condition , , since ^2=w^2=.|__cr , where we used the fact @xmath438 at @xmath108 from .
the solution of is = u_0c(+_0 ) , where @xmath439 and @xmath390 are constants , and @xmath440 due to the fact that @xmath430 .
we can then show @xmath441 by equating the value of @xmath439 with @xmath5 and @xmath15 . substituting @xmath442 into @xmath15 and expanding @xmath443 around @xmath434 up to second order
, we can find c==b , where we used @xmath444 from .
the relation , @xmath445 , is obtained by changing the variable @xmath442 back to @xmath8 ( compare with ) .
let us choose @xmath446 in , then we obtain u&=&u_0 , + [ uel3 ] u^2 & & u^2_0 . notice that @xmath447 which is consistent with the condition for nearly circular orbits , @xmath448 as we have seen in appendix [ sigorbitmin ] and .
we can also find that @xmath449 for @xmath450 and @xmath451 for @xmath452 . to show that the orbit @xmath427 in has a similar form as ,
let us compute the following relations : @xmath453 and @xmath454 .
hence , @xmath455 and @xmath456 , which imply that is of similar orbit form as . as we computed
going from to , where for this case we deduce from , we finally obtain u^2 + . in the next subsection
, we obtain the conditions for closed orbits using @xcite .
let us define an angle @xmath69 , which is the phase difference as the orbit goes from @xmath457 to @xmath458 , = , where we used .
for closed orbits , the angle must have the value that is @xmath186 multiplied by a rational number , i.e. @xmath459 where @xmath460 ; therefore , @xmath75 should be the rational number . in order to obtain the @xmath8-independent value for @xmath69 , potentials can be of the forms , @xmath461 , @xmath462 , and etc . here , @xmath41 and @xmath64 are constant real values , and we should have @xmath24 for bound orbits , whereas we may have @xmath26 for bound orbits when @xmath463 , recalling .
the constant terms in the potentials add an extra energy for the orbits , and it does not play a significant role , so that we consider the potentials without the constant terms . the former power - law potential case , @xmath464 ,
gives = , which implies that the closed orbits exist for @xmath465 .
using and , we obtain , , w^2= , which implies that the bound orbits of the ad condensate has a negative pressure for @xmath252 . in the computation of @xmath15 , ,
we safely ignored the @xmath422 term . the bound orbits for @xmath466 are closed . for the quadratic potential case @xmath161 , the averaged pressure is zero
, in which the ad condensate corresponds to an example of nonrelativistic cold dark matter @xcite .
in addition , using we can find . on the other hand ,
the latter logarithmic potential case , @xmath467 , leads to = ~ , which corresponds to the former power - law case with @xmath61 .
similarly , using and , we obtain m^4 _ , m^4 _ , w^2= , which implies that the ad condensate has a negative pressure for @xmath468 . in the computation of @xmath15 , ,
we safely ignored the @xmath422 term again . using , we also find . in
, we can not clearly see the correspondence with the case for @xmath61 , but we can find @xmath469 for @xmath470 and @xmath471 for @xmath472 as the case with @xmath61 .
let us comment on the pressure when the ad orbit is exact radial , which corresponds to the zero - charge density case as for real fields @xcite . in this case
, the field @xmath10 coherently oscillates around the vacuum if the potential follows a power - law , i.e. @xmath473 for @xmath474 , and @xmath68 is the same equation as , but it gives a negative pressure for @xmath475 .
note that the lower bound of @xmath23 ensures to be a coherent oscillation for the radially oscillating ad fields and real scalar fields . in summary
, we have obtained analytically the explicit expressions , , for the orbit of ad fields in a quadratic potential under an expanding universe , and approximately obtained the elliptic orbit expressions , , for nearly circular orbits in minkowski spacetime in potentials which satisfy the condition for bound orbits .
m. dine and a. kusenko , rev .
* 76 * ( 2004 ) 1 [ arxiv : hep - ph/0303065 ] ; a. riotto and m. trodden , ann . rev .
nucl . part .
* 49 * ( 1999 ) 35 [ arxiv : hep - ph/9901362 ] .
l. kofman , a. d. linde and a. a. starobinsky , phys .
d * 56 * ( 1997 ) 3258 [ arxiv : hep - ph/9704452 ] ; j. h. traschen and r. h. brandenberger , phys .
d * 42 * ( 1990 ) 2491 ; y. shtanov , j. h. traschen and r. h. brandenberger , phys .
d * 51 * ( 1995 ) 5438 [ arxiv : hep - ph/9407247 ] .
m. axenides , s. komineas , l. perivolaropoulos and m. floratos , phys .
d * 61 * ( 2000 ) 085006 [ arxiv : hep - ph/9910388 ] ; r. battye and p. sutcliffe , nucl .
b * 590 * ( 2000 ) 329 [ arxiv : hep - th/0003252 ] ; p. bowcock , d. foster and p. sutcliffe , j. phys .
a * 42 * ( 2009 ) 085403 [ arxiv:0809.3895 [ hep - th ] ] ; c. elphick , phys . rev .
d * 55 * ( 1997 ) 7749 . m. s. volkov and e. wohnert , phys .
d * 66 * ( 2002 ) 085003 [ arxiv : hep - th/0205157 ] ; l. campanelli and m. ruggieri , arxiv:0904.4802 [ hep - th ] . h. arodz , j. karkowski and z. swierczynski , arxiv:0907.2801 [ hep - th ] . s. kasuya and m. kawasaki , phys .
d * 62 * ( 2000 ) 023512 [ arxiv : hep - ph/0002285 ] ; k. enqvist , a. jokinen , t. multamaki and i. vilja , phys .
d * 63 * ( 2001 ) 083501 [ arxiv : hep - ph/0011134 ] ; t. multamaki and i. vilja , phys .
b * 535 * ( 2002 ) 170 [ arxiv : hep - ph/0203195 ] .
s. kasuya and m. kawasaki , phys .
d * 61 * ( 2000 ) 041301 [ arxiv : hep - ph/9909509 ] .
s. kasuya and m. kawasaki , phys .
d * 64 * ( 2001 ) 123515 [ arxiv : hep - ph/0106119 ] .
j. berges , aip conf .
proc . * 739 * ( 2005 ) 3 [ arxiv : hep - ph/0409233 ] ; a. arrizabalaga , j. smit and a. tranberg , jhep * 0410 * ( 2004 ) 017 [ arxiv : hep - ph/0409177 ] ; a. arrizabalaga , j. smit and a. tranberg , phys .
d * 72 * ( 2005 ) 025014 [ arxiv : hep - ph/0503287 ] .
g. n. felder , j. garcia - bellido , p. b. greene , l. kofman , a. d. linde and i. tkachev , phys .
* 87 * ( 2001 ) 011601 [ arxiv : hep - ph/0012142 ] ; j. garcia - bellido , m. garcia perez and a. gonzalez - arroyo , phys . rev .
d * 67 * ( 2003 ) 103501 [ arxiv : hep - ph/0208228 ] .
g. n. felder and l. kofman , phys .
d * 75 * ( 2007 ) 043518 [ arxiv : hep - ph/0606256 ] ; j. f. dufaux , a. bergman , g. n. felder , l. kofman and j. p. uzan , phys .
d * 76 * ( 2007 ) 123517 [ arxiv:0707.0875 [ astro - ph ] ] .
a. kusenko and a. mazumdar , phys
. rev .
* 101 * ( 2008 ) 211301 [ arxiv:0807.4554 [ astro - ph ] ] ; a. kusenko , a. mazumdar and t. multamaki , arxiv:0902.2197 [ astro-ph.co ] ; j. f. dufaux , arxiv:0902.2574 [ astro-ph.co ] .
laplace p s de 1799 _ trait de mcanique celeste _ ; runge c 1919 _ vektoranalysis _
70 ( leipzig : s. hirzel , leipzig ) ; w pauli 1926 z. phyisik * 36 * 336 - 63 ; lenz w 1926 z. phyisik * 24 * 197 - 207 ; heintz w h 1974 am . j. phys . * 42*.107882 see also goldstein h 1975 am . j. phys .
* 43 * 737 - 8 ; _
ibid_. 1976 * 44 * 1123 - 4 .
z. chacko , h. murayama and m. perelstein , phys .
d * 68 * ( 2003 ) 063515 [ arxiv : hep - ph/0211369 ] ; r. allahverdi and a. mazumdar , jcap * 0708 * ( 2007 ) 023 [ arxiv : hep - ph/0608296 ] ; r. allahverdi and a. mazumdar , phys . rev .
d * 78 * ( 2008 ) 043511 [ arxiv:0802.4430 [ hep - ph ] ] .
s. kasuya , phys .
b * 515 * ( 2001 ) 121 [ arxiv : astro - ph/0105408 ] ; m. alimohammadi and h. mohseni sadjadi , phys .
d * 73 * ( 2006 ) 083527 [ arxiv : hep - th/0602268 ] ; h. wei , r. g. cai and d. f. zeng , class . quant .
* 22 * ( 2005 ) 3189 [ arxiv : hep - th/0501160 ] . | we present both analytically and numerically the consistent analysis from the affleck - dine ( ad ) dynamics to the subsequent semiclassical evolution in both gravity - mediated and gauge - mediated models .
we obtain analytically the elliptic motions in the ad dynamics as the analogy of the well - known kepler - problem , and by solving the equations of motion in a lattice , we find that the semiclassical evolution goes through three distinct stages as a nonequilibrium process of reheating the universe : _ pre - thermalisation _ , _ bubble collisions _ and _ thermalisation_. we report that the second stage of our case lasts rather long compared to the second stage of the reheating case , and the thermalisation process is unique due to the presence of `` thermal @xmath0-balls '' . |
ever since the experimental discovery@xcite of superconductivity by jerome
_ et al . , _ in 1980 in a bechgaard@xcite salt ( tmttf)@xmath1pf@xmath2 ( with a t@xmath3 of 1.2 k under a pressure of 6.5 kbar ) , search for new higher t@xmath3 molecular superconductors has been a vigorous field of research@xcite .
a large number of molecular superconductors@xcite with varying degrees of dimensionality have been discovered since 1980 , and a continuing multidisciplinary search for new higher t@xmath3 molecular superconductors is currently underway .
theoretical studies of a possibility of superconductivity in molecular conductors and conducting polymers have had a positive influence in the development of this field .
indeed , littles@xcite prediction of high superconducting transition temperatures within an exciton mediated mechanism of superconductivity in a hypothetical conducting polymer chain ( with polarizable molecules periodically attached to the spine ) played a boosting role for this field@xcite . for a discussion on the future prospects for superconductivity in conducting polymers
see heeger @xcite .
one of the strategies in the search for higher t@xmath3 molecular superconductors continues to be to search for molecules with large electron-_intra_molecular vibration coupling ( emvc ) and to crystallize them with the hope that the solid will be metallic or can be made metallic by charge transfer from another molecule ( in charge - transfer salts ) and/or by application of pressure .
if the material is metallic and if emvc is the pairing glue , large emvc implies a large t@xmath3 within the frame - work of bcs theory@xcite . in a molecular conductor , the phonon spectrum forms two distinct groups : high - frequency _ intra_molecular phonons and _ inter_molecular phonons which have relatively low frequencies ( @xmath4 @xmath5 in the currently known organic superconductors ) . where calculations are available@xcite , the emvc s are larger than the electron-_inter_molecular vibration coupling , though there are exceptions to this trend@xcite .
but , when emvc s become large , there is a possibility of polaron formation@xcite , a possibility not included in the migdal - eliashberg@xcite extension of the bcs theory .
if emvc is large enough to lead to small polaron formation , a possibility for superconductivity to arise is by the polaron pairing through the exchange of low frequency _
inter_molecular phonons .
recently we investigated@xcite this possibility in a simple model molecular conductor . in that work we considered molecular orbitals , overlaps of which on nearest neighbor ( nn ) molecules
produces the conduction band , to have _ no orbital degeneracy_. now , depending on the symmetry of the molecule , the molecular orbitals which participate in the band formation can be degenerate or non - degenerate .
the purpose of the present paper is to investigate the possibility of superconductivity in a three dimensional molecular conductor in which the molecular orbitals participating in the conduction band formation are doubly degenerate and emvc between electrons and a single _ intra_molecular vibration mode is large enough to lead to the formation of jahn - teller ( jt ) small polarons .
this is the @xmath0 jt model and is the simplest jt system .
molecules with tetragonal symmetry are examples of such a system . the rest of the paper is organized as follows . in sec .
ii , we introduce our model for the molecular conductor , perform a lang - firsov transformation on it to obtain the jt polaron hamiltonian . in sec .
iii , a bcs - mean - field ( bcs - mf ) theory is presented which is applicable in the weak coupling regime of the effective attractive interaction between the jt small polarons . in sec .
iv , we analyze the strong coupling bose regime .
an insulating state obtained for the case of two electrons per site is discussed in sec .
v. the conclusions are given in sec .
we employ the following model to study the @xmath0 jt polaron formation and their effective interactions : @xmath6 in the above @xmath7 , @xmath8 and @xmath9 are the molecular site indices , @xmath10 ( = 1 , 2 ) is the orbital index of the two degenerate molecular orbitals , @xmath11 ( @xmath12 ) the hopping energy between similar orbitals on nn molecules , @xmath13 the molecular vibration frequency , @xmath14 the electron - phonon ( _ el - ph _ ) interaction energy , @xmath15 the mass , @xmath16 the coulomb interaction strength , and @xmath17 the number of sites .
furthermore , @xmath18 and @xmath19 are the destruction operators of electrons and phonons respectively , and @xmath20 .
a chemical potential will be introduced later when we have to fix the electron density . in the appendix , using an argument given in ref.@xcite we show that , for @xmath21 , the @xmath22 jt model reduces to @xmath0 jt model . a detailed numerical study of polaron and bipolaron formation in @xmath22 jt model in one dimension and for the case of one and two electrons was recently published by shawish _ et.al._@xcite .
our calculation proceeds along the following lines .
first we apply a multi - band lang - firsov ( lf ) transformation@xcite which produces the jt small polarons , reduces the _
intra_orbital coulomb repulsion , and increases the _ inter_orbital coulomb repulsion .
we will argue that the _ intra_orbital polaron - polaron interaction can be attractive for realistic range of frequencies and _ el - ph _ coupling realizable in molecular conductors .
this interaction then is the source of superconductivity in our model . in the next step ,
we project out _ inter_orbital double occupancies employing a gutzwiller approximation@xcite method in the weak and strong coupling regimes of the attractive interaction to obtain effective hamiltonians in these regimes .
these effective jt hamiltonians are used to study superconductivity in the weak - coupling bcs - mf and strong - coupling ( bose ) regimes .
the superconducting transition temperatures obtained are then analyzed as a function of phonon frequency , _ el - ph _ interaction strength , coulomb repulsion strength , original band - width , and doping .
we will also see that , for the case of two polarons per site , we obtain a novel insulating state with a polaron singlet occupying the lattice sites . for this paired polaron mott insulator
, we show that one obtains a novel orbital density wave state ( odw ) for large and finite _ inter_orbital repulsion .
we also point out a mapping of this odw sate to the ising model . from our analysis of the superconducting transition temperature in the bcs - mf regime and the bose regime
, it will be shown that superconducting t@xmath3 up to 100 k is achievable for this mechanism .
now we go to the details of the calculations .
as stated , we apply a multi - band lf transformation to the @xmath7 .
the transformation @xmath23 transforms @xmath7 into the polaron representation .
lf is a unitary transformation with @xmath24 , where @xmath25 ( @xmath26 ) is @xmath27 using this @xmath28 , we have @xmath29 @xmath30 and @xmath31\,.\ ] ] using the above operators , the transformed hamiltonian is @xmath32 where @xmath33 @xmath34 @xmath35 and @xmath36 in the above equations @xmath37 and @xmath38 the lf transformation produces three effects .
the first is the polaronic band - width reduction , of course .
the second and third effects are the enhancement of the _ inter_orbital coulomb repulsion and the reduction of the _ intra_orbital coulomb repulsion , respectively .
these second and third effects have different origins : while the second has its origin in the jt nature of the system , the third is a small polaron effect . on phonon averaging@xcite of @xmath39 as usual , neglecting a constant coming from @xmath40 , one obtains @xmath41 where @xmath42 in which @xmath43\,,\ ] ] and @xmath44 .
now , @xmath45 is the hamiltonian for a collection of jt small polarons and these polarons interact among themselves though in the _ intra _ and _ inter _ orbital interactions @xmath46 and @xmath47 , respectively . at this stage
, we demand that @xmath46 be an attractive interaction .
this is very much realizable for the values of @xmath16 , @xmath14 , and @xmath13 in a range possible in molecular conductors .
some idea about the range of @xmath16 , @xmath14 , and @xmath13 would be useful to get a better feel . the highest _
intra_molecular vibration frequency is unlikely to exceed the frequency ( @xmath48 @xmath5 ) of the lightest molecule ( h@xmath1 , the hydrogen molecule ) . among the calculated values@xcite of @xmath14 ,
the highest seems to be around @xmath49 mev ( obtained for a mode of frequency @xmath50 @xmath5 in benzene ) .
practically all the existing molecular superconductors are built from the molecules tmtsf , bedt - ttf , dmit , dmet , bets , or c@xmath51 .
this is arguably a small subset of already synthesizable molecules .
then , it is quite possible that higher values of @xmath14 are realized in many molecules .
as for @xmath16 , large and highly polarizable molecules can be expected to have small @xmath16 s . in the existing molecular superconductors
@xcite , @xmath16 seems to be in a range around @xmath52 ev .
we note that the @xmath16 is _ not _ for an isolated molecule , but for a molecule sitting on a lattice site in the solid so that @xmath16 includes the effects of polarization of molecules surrounding a given molecule and consequently is reduced from the @xmath16 value for the isolated molecule .
now , if we tentatively fix an upper limit for @xmath14 around @xmath53 mev and keep the upper limit of @xmath13 as @xmath48 @xmath5 , it is not unreasonable to expect @xmath46 to be negative in a range of @xmath16 which obtains in molecular conductors
. given the estimates of _ el - ph _ coupling to individual modes , this situation is very unlikely to be realized in the existing organic superconductors .
but , that does not preclude its realizability in another important class of materials : the molecular conductors yet to be discovered . in our study , we assume that @xmath46 is negative .
this attractive @xmath46 is then the origin of superconductivity in our theory . in the next two sections , we study superconductivity in the weak and strong coupling regimes of @xmath54 .
the weak coupling regime of @xmath54 is the range of @xmath54 in which the superconducting t@xmath3 and the pair formation temperature are the same . in this section ,
we project out _ inter_orbital double occupancies , using the gutzwiller approximation method , to obtain an effective hamiltonian in the weak coupling regime of @xmath54 and then study the superconductivity using a bcs - mf theory .
the gutzwiller projection renormalizes @xmath55 and @xmath54 , and decouples the orbitals .
first , we recall that the bare _ el - el _ repulsions [ in eq .
( 1 ) ] are strongly renormalized [ eqs . ( 11 ) and ( 12 ) ] : while the _ inter_orbital repulsions have increased , the _ intra_orbital repulsion is suppressed to an extent that it has become attractive ( see the previous paragraph ) .
the effect of @xmath47 is to strongly reduce the probability of polarons occupying different orbitals on a given site . on
the other hand , @xmath54 ( now attractive ) increases the probability of two polarons ( with opposite spins ) occupying the same orbital on a given site .
both these aspects have to be considered while doing the gutzwiller projection .
we take , for simplicity , the @xmath47 to be infinity and completely project out the _ inter_orbital double occupancies .
qualitatively one can see that projecting the _
inter_orbital double occupancies reduces the polaron band - width ( through a density dependent function ) since it is energetically unfavorable for a polaron to hop from an orbital ( say 1 ) on a site to a target site where the orbital 2 is occupied .
also , strong reduction of _ inter_orbital double occupancies increases the _
intra_orbital double occupancies and consequently modifies the _ intra_orbital attraction .
we do the _
inter_orbital double occupancy projection in an approximate way .
first consider the part @xmath56 of @xmath57 [ eq .
we write the gutzwiller wave function @xmath58 as : @xmath59{|\psi_{\circ}\rangle } \,,\ ] ] where @xmath60 for calculating the expectation value of @xmath61 and @xmath62 in @xmath58 , we use a simple intuitive method proposed by okabe@xcite . since the maximum number of polarons ( @xmath63 )
a site can accommodate is four , there are sixteen possibilities for occupancy for finite @xmath47 .
as mentioned earlier , we take @xmath47 infinity limit for simplicity [ @xmath64 in eq .
( 16 ) ] , and consequently the sixteen possibilities reduces to seven . in the paramagnetic case ( @xmath65 , the probability for a site to be vacant is
@xmath66 $ ] and the probability for a site to be singly occupied is @xmath67 . here
@xmath68 is the _
intra_orbital double occupancy .
then , considering the hopping process given in fig .
1 , the gutzwiller band narrowing factor is @xmath69 ^ 2}{\frac{n}{4}(1-\frac{n}{4})}\,,\ ] ] where the numerator and the denominator corresponds to hopping in the presence and absence of _
inter_orbital repulsion , respectively .
the ground state energy for @xmath56 is @xmath70 where @xmath71 $ ] is the average kinetic energy per polaron , and is independent of the indices @xmath10 and @xmath72 .
minimization of @xmath73 with respect to @xmath68 leads to the equation to determine @xmath68 for a given @xmath74 : @xmath75 for @xmath76 , the above equation implies that @xmath77 , which means that one of the orbitals on all the sites are occupied by a singlet . for a general value of @xmath74
, one has to solve eq .
( 20 ) to obtain the variation of @xmath68 as a function of @xmath74 .
the optimized values of the band narrowing factor @xmath78 and the _ intra_orbital double occupancy ( @xmath79 ) as a function of @xmath74 are shown in fig . 2 and fig .
3 , respectively . from the numerical results shown in the figures
, we have @xmath80 and @xmath81 furthermore , it is clear that as @xmath82 , the polaron band - widths are narrowed and a mott insulator is obtained for @xmath76 .
this mott insulator has a singlet in one of the orbitals on each lattice site .
the properties and excitations of this insulator will be discussed separately in sect . v. next consider the term @xmath83 in eq .
( 13 ) . to get the total ground state energy , we add to @xmath73 the contribution from @xmath83 obtained using @xmath58 calculated for optimized @xmath68 ( @xmath79 ) .
this contribution per site is @xmath84 so , the total minimized energy per site is @xmath85 using eqs .
( 21 ) and ( 22 ) , the minimized energy per site is @xmath86 using the results of the gutzwiller projection of inter - orbital double occupancies , we write an effective hamiltonian for the jt polarons as : @xmath87 the effects of gutzwiller projection are the density dependent band - width renormalization and a renormalization of the on - site _
intra_orbital attractive interaction from @xmath54 to @xmath88 . from eq .
( 25 ) , it follows that @xmath89 .
we notice that this enhancement is similar to the enhancement of antiferromagnetic exchange interaction ( @xmath90 ) obtained on gutzwiller projection in a @xmath91 model studied in ref@xcite . for the @xmath91 model
, they found that the bare @xmath90 steadily increases with increasing electron density to reach @xmath92 in the mott insulator at the half - filling of their non - degenerate band .
notice also , for our model , that orbitals are decoupled on gutzwiller projection .
now , @xmath93 is clearly an effective attractive hubbard model .
we note that , unlike the usual phonon mediated attractive interaction which is attractive only in the debye shell around the fermi surface , all the polarons experience the attraction @xmath88 or @xmath54 . in the next section ,
we study superconductivity in the weak coupling regime of @xmath88 using a bcs type mean - field ( bcs - mf ) analysis to obtain superconducting energy gap and @xmath94 .
to move forward , we have to fix the weak and strong coupling regimes of the single band attractive hubbard model ( @xmath95 .
we can approximately fix these ranges using the existing studies@xcite on three dimensional non - degenerate attractive hubbard model .
a comparison of results for superconducting @xmath94 obtained using three methods ( t - matrix approximation , determinant quantum monte carlo , and dynamical mean field theory ) and the bcs - mf is given in fig .
2 of ref.@xcite notice that these results are for quarter filling of the band .
the regime of weak coupling , for a given @xmath74 , is the range of @xmath96 within which the pair formation temperature and the superconducting transition temperature are the same . using the calculation of pauli susceptibility ( which would be suppressed above the superconducting @xmath94 in the case pairs form above @xmath94 )
, the authors of@xcite argued that @xmath97 is the bcs - mf regime for @xmath98 .
considering together all these results , we tentatively fixed the bcs - mf regime of the 3d attractive hubbard model to be @xmath99 .
as for the strong coupling regime , all the electrons are in paired states below a pair formation temperature of @xmath100 .
we fixed the bose regime to be for @xmath101 .
the intermediate coupling regime then is obviously @xmath102 . in the weak coupling regime of @xmath54 , the ratio equivalent to @xmath96 in our case
is , from eq .
( 26 ) , @xmath103 . in the strong coupling regime of @xmath54 studied in the next section , the gutwiller projection has to be done in a different way than that given in this section , and that gives @xmath104 [ where @xmath105 as the ratio equivalent to @xmath96 .
we can now proceed to the analysis of superconductivity in the bcs - mf regime of the effective attractive interaction @xmath88 .
we have from eq .
( 26 ) , @xmath106 where @xmath107 and @xmath108\,,\ ] ] for a simple cubic lattice .
we have introduced a chemical potential ( @xmath109 ) to fix the number density .
given the form of @xmath110 , only isotropic s - wave pairing is possible .
a bcs - mf theory of @xmath111 gives : @xmath112 where the superconducting gap parameter is @xmath113 the superconducting gap and the particle density is determined from @xmath114 and @xmath115 where @xmath116 here @xmath117 is the polaron density of states ( dos ) corresponding to the polaron band @xmath118 .
assuming a square one - polaron single - spin dos ( _ i.e. _ , @xmath119 , the solutions of the gap and chemical potential equations give : @xmath120 and @xmath121 in the above equations , @xmath122 , @xmath123 , @xmath124 , @xmath125 , and @xmath74 is the number of polarons per site . for the values of @xmath126 and @xmath13 we use in numerical calculations ,
the temperature dependence of the polaron band - width is practically nil for the the temperature range ( @xmath127 k ) of interest to us .
the equation for @xmath94 is valid for @xmath128\gg 1 $ ] .
see also the note in ref@xcite .
now we go to the estimates of @xmath129 and @xmath130 considering realistic values of @xmath131 , and @xmath16 applicable to molecular conductors . while making the estimates , one has to satisfy several constraints .
these are : ( i ) jt small polaron formation condition ( @xmath132 ) ; ( ii ) phonon frequency should be below the frequency ( @xmath48 @xmath5 ) of the lightest molecule ( h@xmath1 ) ; ( iii ) a realistic upper limit for _ el - ph _ interaction strength was earlier fixed at @xmath53 mev ; and ( iv ) range of validity of bcs - mf theory requires that @xmath133 .
these conditions enforce severe constraints on the parameters @xmath131 , and @xmath16 for which the system can undergo superconducting instability and if that happens on the achievable @xmath94 s .
the variation of @xmath134 ( at @xmath135 ) with @xmath136 is shown in figs .
4 - 6 , for different values of @xmath13 , @xmath137 , and @xmath126 , respectively .
since the trends are similar for @xmath134 and t@xmath3 , we will discuss the t@xmath3 curves shown in figs .
. the lower limit of @xmath136 for the curves in figs .
4 - 9 is determined by the condition on @xmath138 , and the upper cut - off as a function of @xmath136 enforces the small polaron condition .
the variation of @xmath94 with @xmath136 for various values of @xmath13 is shown in fig
. 7 . for fixed
@xmath16 , @xmath139 and @xmath126 , bcs - mf superconductivity is confined to the @xmath13 range : @xmath140 where @xmath141 and @xmath142 . in this allowed range of @xmath13 , moderate values of @xmath94
are possible .
increasing @xmath13 is found to increase the @xmath94 s achievable and expands the density range in which superconductivity is possible .
the variation of @xmath94 with @xmath136 for different values of @xmath139 is shown in fig .
8 , and it shows that in the range shown changes in @xmath139 do not have much effect . on the other hand , the @xmath94 vs. @xmath136 curves shown in fig .
9 show that small changes in @xmath126 severely narrow the range of @xmath136 for which superconductivity is possible .
it is clear that the various conditions mentioned earlier severely limit the range of @xmath136 for which superconductivity is possible and the values of @xmath94 s achievable .
we would also like to note that below the lower cut - off of @xmath136 for each curve in the figs . 4 - 9
, the @xmath138 moves out of the bcs - mf regime into the intermediate to strong coupling regime and whether superconductivity is possible in that range of @xmath136 can not be addressed using the weak coupling formulas used in this section .
the values of @xmath16 , @xmath126 , @xmath139 , and @xmath13 used in making the @xmath94 estimates are definitely not unrealistic considering the variety of molecules synthesizable and crystallizable .
though band - filling control has not yet been achieved in molecular conductors , such a possibility can not be ruled out in the future . even if band - filling control is not possible , our results show that jt polarons can undergo superconducting instability with moderate @xmath94 s in a realistic range of @xmath16 , @xmath126 , @xmath139 , and @xmath13 realizable in molecular conductors .
this completes the analysis of superconductivity in the bcs - mf regime of our model . in the next section ,
we consider the strong coupling regime ( the bose regime ) of @xmath54 .
in this regime , all the jt polarons are in paired states below a temperature of @xmath143 . to obtain the bose condensation temperature of these pre - formed pairs , we start from the hamiltonian @xmath57 [ see eq .
since the largest terms in @xmath57 are @xmath62 and @xmath83 ( in the strong coupling regime ) , we treat ( @xmath144 ) as the part which determines the ground state energy and take @xmath61 as a perturbation .
now , in the ground state of ( @xmath144 ) , a site has either a singlet occupying one of the orbitals or is empty .
inter_orbital pairs are projected out because of the strong _ inter_orbital repulsion ( @xmath47 ) .
when we switch on the hopping term ( @xmath61 ) , the pairs become mobile through a second order ( in @xmath61 ) process . for the ground state of ( @xmath144 )
we have @xmath145 where @xmath146 is the ground state energy
. the second order pair hopping term then is , @xmath147 in eq .
( 38 ) , the @xmath148 sum is over states other than the degenerate ground states . since @xmath149 and noting that @xmath150 , the second order term becomes @xmath151 the @xmath152 is non - zero for ( a ) @xmath153 or ( b ) @xmath154 .
then , @xmath155 the hopping processes corresponding to the first and second terms in @xmath152 are shown in fig .
@xmath152 can be calculated considering the probability amplitudes for the processes shown in the figure and we obtained it to be the ground state energy of the hamiltonian @xmath156 where @xmath157 is the gutzwiller projection factor and @xmath158 is the co - ordination number of a site in the lattice .
the bose condensation temperature ( @xmath159 ) in the strong coupling regime of a three dimensional non - degenerate attractive hubbard model was obtained in ref.@xcite . comparing our results with theirs , one can immediately write down the @xmath159 for our case , and it is @xmath160 the above formula is relevant for @xmath161 . here
it must be pointed out that the finite temperature analysis in this section was done in real space and hence one need not worry about the orthogonality of the gutzwiller projected excited state wave functions obtained by varying the occupation numbers @xmath162 , as was done in ref.@xcite .
note that for @xmath76 , we have a mott insulator and @xmath159 is zero . while making @xmath159 estimates
, we have to again satisfy the constraints ( i)-(iii ) given in the previous section .
the small polaron formation condition in the strong coupling regime is @xmath163 where @xmath139 is the original half - band - width equal to @xmath164 for a simple cubic lattice we have considered .
the variation of @xmath159 as a function of @xmath136 for various values of @xmath13 is shown in fig .
11 , and it is seen that decreasing @xmath13 increases the @xmath159 s and contracts the range of @xmath136 in which bose condensation is possible . the upper cut - off in figs .
11 - 13 is due to the violation of the small polaron condition . in fig .
12 , we have displayed @xmath159 vs. @xmath136 for different values of @xmath139 . decreasing @xmath139 decreases @xmath159 while expanding the @xmath136 range for bose condensation .
finally , @xmath159 vs. @xmath136 for various values of @xmath126 is shown in fig . 13 .
increase of @xmath126 is seen to sharply reduce @xmath159 .
overall , it is clear from the figures that @xmath159 s up to @xmath165 k can be possible in the strong coupling regime .
furthermore , we note that slightly higher values of @xmath159 can be obtained in the intermediate coupling regime as compared to the strong coupling regime threshold ( @xmath166 ) ( see fig . 2 in ref.@xcite ) .
we have noted earlier that the ground state of the hamiltonian @xmath57 is a mott insulator for the case of two electrons per site .
this mott insulating state obtains independent of the value of @xmath54 ( @xmath167 ) ( see also the note in ref.@xcite ) .
the insulating state is obtained because of the strong _ inter_orbital repulsion @xmath47 between the jt polarons .
the repulsion @xmath47 narrows the polaron band as the density increases and eventually drives it zero at @xmath76 .
now , due to the presence of finite _ intra_orbital attraction between the polarons , the insulating state has a polaron singlet occupying one of the orbitals on site .
since the singlet can occupy either of the two orbitals available on a site , the insulting ground state is highly degenerate when @xmath47 is infinity .
when @xmath47 is large and finite , a pair on a site can gain energy by virtual hopping to a nearest neighbor vacant orbital .
this gain in energy , in second order perturbation theory , is @xmath168 per polaron .
hence for finite and large @xmath47 the degeneracy of the insulating state reduces to two and the singlets order in alternate orbitals on nn sites and thus we have an orbital density wave state .
we also note that one can map the paired polaron mott insulator to an ising model since at each site the pair could be in either of the orbitals . assigning pseudo - spins @xmath169 to occupation of orbital 1 and @xmath170 to that of orbital 2
, the ising model is @xmath171 with @xmath172 .
the ground state corresponds to the orbitally antiferromagnetic state mentioned earlier .
when the electron density deviates from @xmath76 , the insulator undergoes an insulator to metal transition and the superconducting instability of this metallic state was analyzed in the previous sections .
in this paper we presented a theoretical study of superconductivity in a three dimensional molecular conductor in which the molecular orbitals participating in the conduction band formation are doubly degenerate and emvc for electrons in these orbitals interacting with non - degenerate _
intra_molecular vibration mode is large enough to lead to the formation of jahn - teller ( jt ) small polarons .
we argued that the effective polaron - polaron interaction can be attractive for material parameters realizable in molecular conductors , and this interaction is the source of superconductivity in our model .
on analyzing superconducting instability in the weak and strong coupling regimes of this attractive interaction , we found that superconducting transition temperatures up to @xmath165 k are achievable in molecular conductors within this mechanism .
we also find , for two particles per molecular site , that the ground state is a novel mott insulating state in which a polaron singlet occupies one of the doubly degenerate orbitals on each site . in the infinite _
inter_orbital repulsion case , this mott insulator is highly degenerate since the singlet has the freedom to occupy either of the orbitals on a site . on the other band , in the case of large but finite _ inter_orbital repulsion case
, we found that the degeneracy of the ground state is reduced to two with the singlets orbitally antiferromagnetically ordered .
when the number of particles per site deviates from two , the mott insulator undergoes an insulator to metal transition .
one of the authors ( s. y. ) would like to thank sanjoy datta for useful discussions . here
we show that , for strong coupling ( _ i.e_. , @xmath173 ) , @xmath22 jt model is equivalent to the @xmath0 model in eq .
( 1 ) . for @xmath22
, @xmath174 now we do a rotation transformation on @xmath7 . for this purpose , let @xmath175 where @xmath176 and @xmath177 using the above , @xmath7 is transformed to : @xmath178 as argued in ref.@xcite , for @xmath21 , @xmath179 and @xmath180 are unimportant , and we are left with an effective single mode equation . then using @xmath181 ,
one gets the eq .
( 1 ) given in the text . here
it should be pointed out that the single normal mode approximation for @xmath22 works only for the low lying excitations of the mode . for the transition temperatures obtained and the frequencies considered @xmath182 , and hence phonon averaging after the lang - firsov transformation yields for the single mode approximated @xmath22 system the same result given by eq .
( 15 ) in the text .
d. jerome , a. mazaud , m. ribault , and k. bechgaard , j. phys .
( france ) lett . *
41 * , l95(1980 ) .
k. bechgaard , c. s. jacobsen , k. mortensen , h. j. pedersen , and n. thorup , solid state . commun . * 33 * , 1119(1980 ) .
t. ishiguro , k. yamaji , and g. saito , _ organic superconductors _ ( springer , berlin , 1998 ) .
w. a. little , phys . rev . * 134 * , a1416(1964 ) .
d. jerome , science * 252 * , 1509(1991 ) .
a. j. heeger , rev .
mod . phys . * 73 * , 681(2001 ) .
j. bardeen , l. n. cooper , and j. r. schrieffer , phys .
rev . * 108 * , 1175(1957 ) .
a. j. berlinsky , j. f. carolan , and l. weiler , solid state commun .
* 15 * , 795(1974 ) ; n. o. lipari , m. j. rice , c. b. duke , r. bozio , a. girlando , and c. pecile , intl
. j. quan .
qm symp . * 11 * , 583(1977 ) ; k. yamaji , solid state . commun . * 61 * , 413(1987 ) ; j. c. r. faulhaber , d. k. k. ko , and p. r. briddon , syn . met .
* 60 * , 227(1993 ) ; v. m. yartsev and a. graja , intl .
b * 12 * , 1643(1998 ) ; a. devos and m. lannoo , phys . rev .
b * 58 * , 8236(1998 ) ; t. kato and t. yamabe , j. chem . phys . * 115 * , 8592(2001 ) ; t. kato , k. yozhizawa , and k. hirao , j. chem . phys . * 116 * , 3420(2002 ) ; j. sinoa , j. schliemann , a. s. nez , and a. h. macdonald , phys . rev . lett . * 87 * , 226802(2001 ) ; c. m. varma , j. zannen , and k. raghavachari , science * 254 * , 989(1991 ) ; v. p. anthropov , o. gunnarson , and a. i. liechtenstein , phys .
b * 48 * , 7651(1993 ) ; o. gunnarson , rev .
phys . * 69 * , 575(1997 ) ; m. cote , j. s. grossmann , m. l. cohen , and s. g. louie , phys .
* 81 * , 697(1998 ) .
r. ramakumar , y. tanaka , and k. yamaji , phys .
b * 56 * , 795(1997 ) ; r. ramakumar and k. yamaji , eur . phys . j. b * 24 * , 71(2001 ) .
j. appel , in _ solid state physics _ ,
h. ehrenreich , f. seitz , and d. turnbull ( academic press , new york , 1967 ) , p193 ; a. s. alexandrov and n. f. mott , _ polarons and bipolarons _
( world scientific , singapore , 1995 ) .
a. b. migdal , zh .
34 * , 1438(1958)[sov .
jetp * 34 * , 996(1958 ) ] . g. m. eliashberg , zh . eksp
. teor . fiz . *
38 * , 966(1960 ) [ sov .
jetp * 11 * , 696(1960 ) ] ; * 39 * , 1437(1960)[*12 * , 1000(1961 ) ] .
t. holstein , ann .
( n. y. ) * 8 * , 325(1959 ) ; * 8 * , 343(1959 ) .
r. ramakumar and s. yarlagadda , phys .
rev . b*67 * , 214502(2003 ) .
m. d. sturge , in _ solid state physics _ ,
h. ehrenreich , f. seitz , and d. turnbull ( academic press , new york , 1967 ) , p 91 .
s. el shawish , j. bonca , li - chung ku , and s. a. trugman , phys .
b * 67 * , 014301(2003 ) .
i. g. lang and yu .
a. firsov , zh .
. teor . fiz . * 43 * , 1843(1962 ) [ sov .
jetp * 16 * , 1301(1962 ) ] .
m. c. gutzwiller , phys .
* 10 * , 159(1963 ) ; phys . rev . * 134 * , a923(1964 ) .
s. yarlagadda , phys .
b * 62 * , 14828(2000 ) . t. kato and t. yamabe in ref@xcite .
t. okabe , j. phys .
soc . of japan
* 65 * , 1056(1996 ) .
f. c. zhang , c. gros , t. m. rice , and h. shiba , supercond .
* 1 * , 36(1988 ) .
r. micnas , j. ranninger , and s. robaszkiewicz , rev .
62 * , 113(1990 ) .
m. keller , w. metzner , and u. schollwck , phys .
86 * , 4612(2001 ) .
a. sewer , x. zotos , and h. beck , phys .
b * 66 * , 140504(2002 ) .
the reader may be concerned about the validity of calculating @xmath94 from a theory on an effective hamiltonian obtained using gutzwiller approximation .
since bcs - mf t@xmath3 is much smaller than the effective fermi temperature , the corrections brought about by the modification required in the occupation number ( @xmath162 ) to incorporate the finite temperature effects in gutzwiller projection is expected to be small .
see : t. m. rice , k. ueda , h. r. ott , and h. rudigier , phys .
b * 31 * , 594(1985 ) ; k. seiler , c. gros , t. m. rice , k. ueda , and d. vollhardt , j. low . temp . phys . *
64 * , 195(1986 ) .
s. robaszkiewicz , r. micnas , and k. a. chao , phys .
b * 23 * , 1447(1981 ) . for @xmath76 ,
a mott insulating state is also possible for @xmath183 . | we present a theoretical study of a possibility of superconductivity in a three dimensional molecular conductor in which the interaction between electrons in doubly degenerate molecular orbitals and an _ intra_molecular vibration mode is large enough to lead to the formation of @xmath0 jahn - teller small polarons . we argue that the effective polaron - polaron interaction can be attractive for material parameters realizable in molecular conductors . this interaction is the source of superconductivity in our model . on analyzing superconducting instability in the weak and strong coupling regimes of this attractive interaction , we find that superconducting transition temperatures up to 100 k are achievable in molecular conductors within this mechanism .
we also find , for two particles per molecular site , a novel mott insulating state in which a polaron singlet occupies one of the doubly degenerate orbitals on each site .
relevance of this study in the search for new molecular superconductors is pointed out . 2 |
chiral symmetry is believed to govern interactions among hadrons at low energies where the relevant degrees of freedom are not the quark and gluon fields of the qcd lagrangian , but composite hadrons . in order to make contact with experiment one must resort to methods such as chiral perturbation theory ( chpt ) which incorporates the symmetries and symmetry breaking patterns of underlying qcd and is written in terms of the active degrees of freedom .
a systematic loop expansion can be carried out which inherently involves a characteristic scale @xmath9 gev at which the chiral series is expected to break down .
the limitation to very low - energy processes is even enhanced in the vicinity of resonances .
the appearance of resonances in certain channels constitutes a major problem to the loopwise expansion of chpt since their contribution can not be reproduced at any given order of the chiral series .
this can be prevented by including the resonance exchanges explicitly with the couplings fixed from electromagnetic and hadronic data . in that case
, however , the theory loses its predictive power and does not provide a stringent test of chiral symmetry .
recently , considerable effort has been undertaken to combine the effective chiral lagrangian approach with non - perturbative methods , both in the meson - baryon sector @xcite and in the purely mesonic sector @xcite .
the combination with non - perturbative schemes have made it possible to go to energies beyond @xmath10 and to generate resonances dynamically .
two prominent examples in the baryonic sector are the @xmath11 and the @xmath12 .
the first one is an @xmath1-wave resonance just below the @xmath13 threshold and dominates the interaction of the @xmath14 system
. the properties of the @xmath11 which appears at the correct position with the right width were reproduced remarkably well in @xcite .
a similar analysis was performed in @xcite where in addition to the @xmath12 the @xmath15 was also obtained via the inclusion of the @xmath16 channel .
the @xmath12 is of particular interest since it decays very strongly into the @xmath17 channel and provides insight into the @xmath17 interaction .
applying the same formalism as in @xcite , the authors in @xcite could generate the @xmath12 , which they identify as a quasi - bound @xmath18-@xmath19 state .
information on the @xmath12 was experimentally extracted from precise eta photoproduction data off protons close to threshold at mami ( mainz ) @xcite and together with an analogous electroproduction experiment performed at elsa ( bonn ) @xcite the data covered the whole range of the @xmath12 resonance . in general , photoproduction of mesons
is a tool to study baryonic resonances and the investigation of transitions between these states provides a crucial test for hadron models .
the dominance of the @xmath20 in the photoproduction of pions , e.g. , has allowed to extract information on its electromagnetic transition amplitudes .
because of their hadronic decay modes nucleon resonances have large overlapping widths , which makes it difficult to study individual states , but selection rules in certain decay channels can reduce the number of possible resonances .
the isoscalars @xmath6 and @xmath2 are such examples since , due to isospin conservation , only the isospin-@xmath21 excited states decay into the @xmath17 and @xmath22 channels .
electroproduction experiments are even more sensitive to the structure of the nucleon due to the longitudinal coupling of the virtual photon to the nucleon spin and might in addition yield some insight into the possible onset of perturbative qcd .
perturbative qcd should apply at sufficiently high photon virtuality @xmath23 , see e.g. @xcite , however there is no consensus about how high the momentum transfer must be .
it has been found experimentally , that in the case of electroproduction of the @xmath20 resonance at momentum transfers up to @xmath24 gev@xmath25 perturbative qcd is not applicable , @xcite , whereas a possible onset of scaling in the reaction @xmath26 at @xmath27 gev@xmath25 is reported in @xcite .
clas at jlab has also electroproduced @xmath6 mesons for invariant momentum transfers @xmath28 between 0.375 and 1.375 gev@xmath25 @xcite .
it was found that the @xmath5 photocoupling @xmath29 decreases much slower with @xmath28 than , e.g. , the nucleon dipole form factor , which is unusual and difficult to explain theoretically .
furthermore , there is still some controversy about the nature of the @xmath30 and @xmath31 couplings . for @xmath6 photoproduction in the @xmath32 resonance region ,
e.g. , both an effective lagrangian approach @xcite and coupled channel models @xcite have been employed . in these approaches the coupling of the @xmath6 to the nucleons
is described by both a pseudovector and a pseudoscalar term and the coupling constant and the coupling structure of the born terms is unknown . in @xcite
it has been shown that differential cross sections are rather sensitive to the assumptions about this vertex , but within the framework of chiral perturbation theory this coupling is fixed at lowest order by making use of the chiral @xmath33 symmetry of the lagrangian , whereas explicitly chiral symmetry breaking terms appear at higher orders .
the @xmath33 symmetric limit provides therefore a convenient starting point which overcomes the problem of fixing the @xmath30 vertex .
the @xmath34 chiral meson - baryon lagrangian has been used in a coupled channel model @xcite and by adjusting a few parameters a large amount of low - energy data was described . all the above mentioned investigations
have in common that they treat the @xmath6 meson as a pure @xmath34 octet state @xmath35 and mixing of @xmath35 with the corresponding singlet state @xmath36 which yields the physical states @xmath6 and @xmath2 is neglected .
the @xmath2 is interesting by itself .
the qcd lagrangian with massless quarks exhibits an @xmath33 chiral symmetry which is broken down spontaneously to @xmath37 , giving rise to a goldstone boson octet of pseudoscalar mesons which become massless in the chiral limit of zero quark masses . on the other hand ,
the axial @xmath3 symmetry of the qcd lagrangian is broken by the anomaly .
the corresponding pseudoscalar singlet would otherwise have a mass comparable to the pion mass @xcite .
such a particle is missing in the spectrum and the lightest candidate would be the @xmath2 with a mass of 958 mev which is considerably heavier than the octet states . in conventional chiral perturbation theory the @xmath2
is not included explicitly , although it does show up in the form of a contribution to an lec .
however , it is also possible to include the @xmath2 explicitly in the chiral lagrangian framework as has been done , e.g. , in @xcite for the purely mesonic sector which was then extended to include baryons in @xcite . within this framework the @xmath2 is combined with the goldstone bosons @xmath38 into a nonet and the @xmath30 and @xmath31 couplings are constrained by chiral symmetry .
the @xmath2 photoproduction has been investigated theoretically in @xcite . in the effective lagrangian approach of @xcite
a pseudoscalar coupling of the @xmath2 to the nucleons was chosen and it was concluded that the @xmath22 decay channel is dominated by the not so well established @xmath39 resonance , whereas in the quark model used in @xcite the off - shell effects of the @xmath32 were prominent .
in contrast , the experimental data from elsa @xcite suggested the coherent excitation of two resonances @xmath40 and @xmath41 . in @xcite @xmath6 and
@xmath2 photoproduction has been studied in the coupled channel formalism working along the lines of @xcite . however , the treatment of the @xmath2 is incomplete and only the leading terms in the meson - baryon potentials are taken into account .
their results fail to reproduce the @xmath6 photoproduction data and are unable to describe the appearance of a @xmath5 resonance for @xmath2 photoproduction as reported at elsa . in order to predict the cross sections for electroproduction , which will become available from experiments at jefferson lab in the near future @xcite ,
a much more thorough investigation is needed . in the present work
we will focus on @xmath0 photo- and electroproduction off nucleons in a framework which allows a unifying description of these processes with chiral symmetry and unitarity being the main ingredients .
this is achieved by combining the effective lagrangian with a non - perturbative scheme based on coupled channels and the bethe - salpeter equation .
the approach contains only a few parameters , both coupling constants in the lagrangian , so - called low - energy constants ( lecs ) , and finite range parameters which appear in the evaluation of the loop integrals . with this small set of parameters
it will be a highly non - trivial task to reproduce the data on meson photo- and electroproduction and meson - baryon scattering experiments which are described simultaneously within this multi - channel analysis .
hence this investigation will provide a test whether processes up to energies of @xmath42 gev are still constrained by chiral symmetry and whether the @xmath2 meson can be included in the effective lagrangian with baryons as proposed in @xcite .
we will perform a global fit to a wide range of meson - baryon scattering and photoproduction data . to this end , we restrict ourselves to @xmath1-waves and therefore the comparison with data should only be valid in the near threshold region . at higher energies @xmath43-waves
start dominating , as can be seen , e.g. , in the work by caro ramon et al .
one of the purposes of this work is to shed some light on the @xmath1-wave resonance @xmath40 which can be studied in the @xmath1-wave approximation .
our results must be compared to the cross section reported in @xcite .
we will continue giving predictions for @xmath6 and @xmath2 electroproduction processes and compare our results with the @xmath6 electroproduction data from clas @xcite .
as we will see , the present framework also provides an at least qualitative explanation of the slow fall - off with increasing @xmath28 of the @xmath32 photocoupling .
effects of @xmath6-@xmath2 mixing and the importance of the @xmath2 contributions in this coupled channel formalism even for processes where it appears only as a virtual state are discussed in detail .
this may eventually lead to a better understanding of gluonic effects and the significance of the axial @xmath3 anomaly in low - energy hadron physics . in the next two sections ,
we introduce the effective lagrangian and the coupled channel formalism which is then generalized to electroproduction processes .
the results of our analysis are presented and compared to existing @xmath0 photoproduction and @xmath6 electroproduction data in sec . 5 , where also predictions for @xmath2 electroproduction are made .
we summarize our findings in sec .
in this section , we will shortly review a systematic way of including the @xmath2 in the chiral effective lagrangian . for details the reader
is referred to @xcite in the purely mesonic sector or @xcite the presence of baryons .
the @xmath45 chiral effective lagrangian of the pseudoscalar meson nonet @xmath46 coupled to the ground state baryon octet @xmath47 can be decomposed as @xmath48 with the mesonic piece up to second chiral order @xcite @xmath49 and a part @xmath50 which describes the meson - baryon interactions @xcite . at lowest order it reads @xmath51 \rangle - { \stackrel{\circ}{m}}\langle \bar{b}b \rangle + i u_1 \frac{\eta_0 ^ 2}{f_\pi^2 } \big ( \langle [ d^{\mu},\bar{b } ] \gamma_{\mu } b \rangle - \langle \bar{b } \gamma_{\mu } [ d^{\mu},b ] \rangle \big ) { \nonumber}\\ & & - \frac{1}{2 }
d \langle \bar{b } \gamma_{\mu } \gamma_5 \{u^{\mu},b\ } \rangle - \frac{1}{2 } f \langle \bar{b } \gamma_{\mu } \gamma_5 [ u^{\mu},b ] \rangle - \frac{1}{2 } d_s \langle \bar{b } \gamma_{\mu } \gamma_5 b \rangle \langle u^{\mu } \rangle , \end{aligned}\ ] ] where only the terms that are necessary for the present calculation are kept and @xmath52 denotes the trace in flavor space . the pseudoscalar meson nonet is summarized in @xmath53 with @xmath54 where @xmath55 mev is the pion decay constant and @xmath56 contains the goldstone bosons @xmath57 .
the covariant derivative of the meson fields includes the coupling to an external photon field @xmath58\ ] ] with @xmath59 being the quark charge matrix and @xmath60 the photon field .
explicit chiral symmetry breaking is induced via the quark mass matrix @xmath61 which enters in the combinations @xmath62 with @xmath63 the order parameter of the spontaneous symmetry violation .
the second and third term of eq .
( [ mes ] ) appear already in conventional chpt whereas the first and fourth one are due to the axial @xmath3 anomaly .
the first one is the mass term of the singlet field @xmath36 which remains in the chiral limit of vanishing quark masses .
the coefficient @xmath64 is a parameter not fixed by chiral symmetry and in the large @xmath65 limit it is proportional to the topological susceptibility of gluodynamics .
the fourth term yields @xmath35-@xmath36 mixing , which can be described in terms of a single mixing angle , see eq .
( [ app : mix]).-@xmath36 since terms from the fourth order lagrangian yield off - diagonal elements for the derivative pieces of the lagrangian so that the mixing in the @xmath6-@xmath2 system can not be parameterized in terms of a single mixing angle if large @xmath65 counting rules are not imposed .
the presentation of counterterms of the fourth order mesonic lagrangian is beyond the scope of this work and for our purposes it will be sufficient to assume a mixing scheme in terms of one mixing parameter . ]
the ground state baryon octet @xmath66 is summarized in a @xmath67 matrix @xmath68 , @xmath69 is the common baryon octet mass in the chiral limit and @xmath70 are the axial vector couplings of the baryons to the mesons .
the values of @xmath71 and @xmath72 are extracted phenomenologically from the semileptonic hyperon decays and a fit to data delivers @xmath73 , @xmath74 @xcite .
the parameters @xmath75 and @xmath76 do not enter in conventional chpt and , since their values have not yet been determined , we will extract them from our fit . finally , the covariant derivative of the baryon fields is given by @xmath77 = \partial_\mu b + [ \gamma_\mu , b]\ ] ] with the chiral connection @xmath78 + i e { \cal a}_\mu q .\ ] ] note that there is no pseudoscalar coupling of @xmath36 to the baryons of the form @xmath79 . such a term is in principle possible but can be absorbed by the @xmath76-term in eq .
( [ bar ] ) by means of the equation of motion for the baryons . at
next - to - leading order the terms relevant for @xmath1-wave meson - baryon scattering are @xmath80 \rangle + b_0 \langle \bar{b}b \rangle \langle \chi_+ \rangle { \nonumber}\\ & & + i \frac{c_d}{f_\pi } \eta_0 \langle \bar{b } \{\chi_-,b\ } \rangle + i \frac{c_f}{f_\pi } \eta_0 \langle \bar{b } [ \chi_-,b ] \rangle + i \frac{c_0}{f_\pi }
\eta_0 \langle \bar{b } b\rangle \langle \chi_+ \rangle { \nonumber}\\ & & + d_1 \langle \bar{b } \{u_{\mu},[u^{\mu},b]\ } \rangle + d_2 \langle \bar{b } [ u_{\mu},[u^{\mu},b ] ] \rangle + d_3 \langle \bar{b } u_{\mu } \rangle \langle u^{\mu } b \rangle + d_4 \langle \bar{b } b \rangle \langle u^{\mu } u_{\mu } \rangle { \nonumber}\\ & & + d_5 \langle \bar{b } \ { u_{\mu } , b \ } \rangle \langle u^{\mu } \rangle + d_6 \langle \bar{b } [ u_{\mu } ,
b ] \rangle \langle u^{\mu } \rangle + d_7 \langle \bar{b } b \rangle \langle u^{\mu } \rangle \langle u_{\mu } \rangle .\end{aligned}\ ] ] we made use of the cayley - hamilton identity , in order to eliminate @xmath81 .
the terms @xmath82 and @xmath83 are already present in the @xmath34 case , whereas @xmath84 and @xmath85 are new terms of the extended theory .
the lecs @xmath86 and @xmath87 are responsible for the splitting of the baryon octet masses at leading order in symmetry breaking .
working in the @xmath88 limit of equal @xmath89 and @xmath90 quark masses , @xmath91 , one obtains @xmath92 since the three baryon mass differences are represented in terms of two parameters , there is a corresponding sum rule the gell - mann okubo mass relation for the baryon octet @xcite : @xmath93 which experimentally has only a 3% deviation .
a least - squares fit to the mass differences ( [ massdiff ] ) yields @xmath94 gev@xmath95 and @xmath96 gev@xmath95 .
the @xmath97 term , on the other hand , can not be determined from the masses alone .
one needs further information which is provided by the pion - nucleon @xmath98-term and reads at leading order @xmath99 employing the empirical value of @xcite , @xmath100 mev , one obtains @xmath101 gev@xmath95 .
recently , this value has been questioned @xcite and the authors of this work arrive at a value @xmath102 mev which would translate into a value of @xmath103 gev@xmath95 .
in particular the latter value yields a large strangeness content of the proton , however both values may change if loop effects are included @xcite . due to these uncertainties in the value of @xmath97 , any fitted value which lies in the range @xmath104 gev@xmath105 gev@xmath106 is still acceptable . in the @xmath107 formalism three more explicitly symmetry breaking terms enter , @xmath108 and @xmath109 , which have unknown values .
these are suppressed by one order of 1/@xmath65 with respect to their counterparts @xmath110 and @xmath97 , and following 1/@xmath65 arguments one can assume them to be significantly smaller in magnitude , e.g. @xmath111 .
we will vary these three parameters only within small ranges around zero and compare the results with existing data .
the situation is less clear for the derivative terms @xmath112 . in @xcite
the parameters @xmath113 to @xmath114 were determined in a coupled channel analysis while being subject to two constraints from @xmath115 and @xmath116 scattering lengths .
the authors obtain the values @xmath117 , @xmath118 , @xmath119 and @xmath120 in units of gev@xmath95 . however , they employ the @xmath34 chiral lagrangian which does not include the @xmath2 explicitly and work in the heavy baryon formulation treating the baryons as heavy sources while loop diagrams are regularized by an explicit cutoff .
we expect the values of @xmath121 to change in our approach since we derive the potentials from the relativistic @xmath107 lagrangian and the loops are evaluated in dimensional regularization .
an estimate for the parameters @xmath122 can alternatively be obtained by assuming resonance saturation . within this model
the decuplet @xmath123 of spin-@xmath124 baryon fields which includes the @xmath125 is expected to determine approximately the values of the @xmath122 parameters .
the interaction of the decuplet states with the octet baryons and the pseudoscalar mesons reads at lowest order @xmath126 where the coupling constant @xmath127 1.21.8 can be determined from the decays @xmath128 .
the decuplet contributions to the lecs @xmath122 yield at leading order in the averaged decuplet mass @xmath129 via resonance saturation the values @xmath130 , @xmath131 , @xmath132 , @xmath133 , @xmath134 , @xmath135 and @xmath136 in units of gev@xmath95 for @xmath137 and @xmath138 gev . while @xmath139 are in qualitative agreement with the fitted values from @xcite , the values for @xmath140 differ considerably .
contributions from other resonances have been neglected here since we are only interested in a rough order of magnitude estimate for these parameters .
we will therefore vary @xmath141 , @xmath142 , within small ranges around the values most unknown couplings are constrained within certain ranges and they will be fixed from a fit to many scattering data . after setting up the lagrangian which we will be using in this work , we can now proceed by explaining our coupled channel approach .
the lagrangian of the preceding section could in principle be used to calculate the one - loop diagrams of @xmath2 electroproduction , but the perturbative expansion of the chiral series will be useless for this case .
first , the @xmath2 with a mass of 958 mev close to the scale of chiral symmetry breaking @xmath143 gev introduces a new massive scale in the effective theory which spoils the strict chiral counting scheme unless one imposes large @xmath65 counting rules @xcite , @xmath144 , it has been shown that the usage of infrared regularization yields a well - behaved chiral series since loops with an @xmath2 are suppressed @xcite ] .
large @xmath65 counting rules in turn imply that the @xmath2 mass must be treated as a small quantity . in our opinion , this assumption is phenomenologically not justified and we consider the @xmath2 to be a massive state .
second , @xmath2 electroproduction has a threshold close to @xmath145 gev far beyond the scale @xmath143 gev at which the perturbative chiral expansion is expected to break down .
one must therefore resort to non - perturbative schemes . for the investigation of hadronic resonances the combination of the effective chiral lagrangian with coupled channel approaches
have been proven to be useful .
e.g. , in @xcite the @xmath12 nucleon resonance emerged as a quasi - bound state of @xmath18 and @xmath19 . in the present investigation
, we employ a relativistic chiral unitary approach based on coupled channels . by imposing constraints from unitarity
we perform the resummation of the amplitudes obtained from the tree level potentials and the loop integrals .
let us describe first the coupled channel approach for meson - baryon scattering , which will then be generalized to electroproduction of mesons . to this end , one expands @xmath146 and @xmath147 in eqs .
( [ mes ] ) , ( [ gama ] ) in terms of the mesons @xmath148 @xmath149 + \ldots\ ] ] in our model the relativistic tree level amplitude @xmath150 for the meson - baryon scattering process @xmath151 is obtained by the diagrams shown in fig .
( [ figmes ] ) . in the present work ,
we restrict ourselves to the @xmath1-wave partial wave amplitude @xmath152 which is given by @xmath153 where we have averaged over the spin @xmath98 of the baryons and @xmath1 is the invariant energy squared .
it is most convenient to work in the isospin basis and to characterize the meson - baryon states in the electroproduction process @xmath154 by their total isospin @xmath155 or @xmath156 .
the analysis reduces then to five channels with total isospin @xmath157 which are @xmath158 , @xmath159 , @xmath160 , @xmath161 , @xmath162 ( labeled with indices 1 , 2 , 3 , 4 , 5 , respectively ) and two channels with @xmath156 , @xmath163 , @xmath164 ( labeled by indices 6 and 7 ) .
the potentials read @xmath165 where @xmath166 and @xmath167 label the channels involved and @xmath168 and @xmath169 , @xmath170 are the energy and mass of the incoming ( outgoing ) baryon . note that we have replaced the common baryon octet mass @xmath69 by the physical masses @xmath170 which is consistent at the order the potentials @xmath152 are calculated and use of the physical masses is also mandatory in order to reproduce the correct threshold positions of the different channels involved .
the coefficients @xmath171 are given in app .
[ app : pot ] . for each partial wave @xmath172 unitarity imposes a restriction on the ( inverse ) @xmath123-matrix above the pertinent thresholds @xmath173 with @xmath174 being the three - momentum in the center - of - mass frame of the channel under consideration .
hence the imaginary part of @xmath175 is similar to the imaginary piece of the fundamental scalar loop integral @xmath176 above threshold @xmath177 [ l^2 - m_\phi^2 + i \epsilon ] } \ ] ] with @xmath178 and @xmath179 being the physical masses of the baryon and the meson , respectively . for the finite part , @xmath180 of @xmath176 , one obtains , e.g. , in dimensional regularization @xmath181 + ( m_\phi^2 - m_b^2 ) \ln\left(\frac{m_\phi^2 } { m_b^2}\right ) { \nonumber}\\ & & - 8 \sqrt{q^2}|\mbox{\boldmath$q$}_{cm}| \ ; \mbox{artanh } \left(\frac{2 \sqrt{q^2}|\mbox{\boldmath$q$}_{\scriptstyle{cm}}| } { ( m_\phi + m_b ) ^2 - q^2 } \right ) \bigg\}\end{aligned}\ ] ] where @xmath182 is the regularization scale .
we allow this scale to vary for the different loops , in order to simulate higher order contributions . alternatively , in @xcite the real piece
has been adjusted by introducing a scale dependent constant for each channel in analogy to a subtraction constant of a dispersion relation for @xmath175 . in a more general way
, one could model the real parts by taking any analytic function in @xmath1 and the baryon and meson masses .
this option has been successfully applied for the case of @xmath88 chpt in @xcite , but a straightforward generalization to the @xmath107 case would introduce a number of unknown new parameters and thus reduce the predictive power of the present approach .
the inverse of the amplitude @xmath175 can be decomposed into real and imaginary parts @xmath183 where @xmath184 and re@xmath185 $ ] give the real part and im@xmath185 $ ] gives the imaginary part required by unitarity , eq .
( [ unit ] ) . inverting ( [ invers ] ) yields @xmath186^{-1 } \ ; \tau\ ] ] which is understood to be a matrix equation .
the matrix @xmath180 is diagonal and includes the expressions for the loop integrals in each channel .
expanding expression ( [ tau ] ) @xmath187 and from matching to our tree level amplitude it follows @xmath188 our final expression for the @xmath123 matrix reads then @xmath189^{-1 } \ ;
v\ ] ] which amounts to a summation of a bubble chain in the @xmath1-channel .
this is equivalent to a bethe - salpeter equation with @xmath152 as potential .
we are now in a position to extend the above developed formalism to electroproduction processes . following @xcite
we assume that the @xmath1-wave electroproduction process can be described by a similar bethe - salpeter equation as for the strong interactions .
our starting point are the contact , born and meson - pole diagrams for meson electroproduction with vertices of the lagrangian as presented in sec . [
sec : lagran ] , see fig . 2 .
- and @xmath89-channel and meson pole diagram for meson electroproduction .
solid and dashed lines denote the baryons and pseudoscalar mesons , respectively .
the photon is represented by a wavy line.,title="fig:",width=102 ] - and @xmath89-channel and meson pole diagram for meson electroproduction .
solid and dashed lines denote the baryons and pseudoscalar mesons , respectively .
the photon is represented by a wavy line.,title="fig:",width=124 ] - and @xmath89-channel and meson pole diagram for meson electroproduction .
solid and dashed lines denote the baryons and pseudoscalar mesons , respectively .
the photon is represented by a wavy line.,title="fig:",width=143 ] - and @xmath89-channel and meson pole diagram for meson electroproduction .
solid and dashed lines denote the baryons and pseudoscalar mesons , respectively . the photon is represented by a wavy line.,title="fig:",width=124 ] from these diagrams
we extract the transverse and longitudinal @xmath1-wave multipoles at leading order which are identified with the electroproduction potentials @xmath190 and @xmath191 , respectively . for
the proton the expressions @xmath190 in the isospin basis , which can be obtained from the formulas appearing for example in @xcite , read @xmath192 where @xmath193 is the mixing angle ( see the appendix ) .
the functions @xmath194 and @xmath195 for the electroproduction process @xmath196 in the channel @xmath166 are given by @xmath197 @xmath198 { \nonumber}\\ & & + \frac{1}{4 |\mbox{\boldmath$k$}| |\mbox{\boldmath$q$}| } \big((m_a^2-m_b^2- 2 e_a e_\phi + m_\phi^2)^2 + 4 |\mbox{\boldmath$k$}|^2 ( ( \sqrt{s } -m_a)(e_b + m_b ) - & & \qquad \qquad + \frac{2 |\mbox{\boldmath$k$}|^2 } { n_a^2 } ( \sqrt{s } + m_a ) ( m_a^2-m_b^2- 2 e_a e_\phi + m_\phi^2)\big ) { \nonumber}\\ & & \qquad \qquad \qquad \qquad \times \ln \frac{m_b^2 -m_a^2 -m_\phi^2 + 2 e_a e_\phi + 2|\mbox{\boldmath$k$}| |\mbox{\boldmath$q$}| } { m_b^2 -m_a^2 -m_\phi^2 + 2 e_a e_\phi - 2|\mbox{\boldmath$k$}| |\mbox{\boldmath$q$}| } \ ; \bigg)\end{aligned}\ ] ] with @xmath199 the three - momenta of the photon and the meson in the center - of - mass frame .
the energies of the baryons and meson are given by @xmath169 and @xmath200 , respectively , whereas @xmath170 and @xmath179 denote their masses .
the function @xmath194 describes charged meson electroproduction processes on protons , whereas @xmath195 is obtained when neutral mesons are produced . in the case of the neutron one obtains @xmath201 the expressions for the longitudinal @xmath1-wave potentials @xmath191 have a similar structure , but are much more lengthy and will therefore not be presented for brevity .
once a meson and a baryon have been electroproduced , they may rescatter into all possible meson - baryon pairs , as shown in figs . 3 and 4 .
we sum this infinite interaction chain in the bethe - salpeter approach to obtain the final results for the electric dipole amplitude @xmath202 and the longitudinal @xmath1-wave @xmath203 .
this procedure ensures that the resonances that appear in meson electroproduction are the same ones that appear in meson - baryon scattering .
@xmath204{fig3a.eps } } = \parbox{2cm}{\centering\includegraphics[scale=0.8]{fig3b.eps } } + \parbox{3.6cm}{\centering\includegraphics[scale=0.8]{fig3c.eps } } + \parbox{4.6cm}{\centering\includegraphics[scale=0.8]{fig3d.eps } } + \,\,\ldots\ ] ] @xmath205 the electric dipole moment @xmath202 is given by @xmath206^{-1 } \ ; b_{0 + } \ ; , \ ] ] while the longitudinal @xmath1-wave @xmath203 reads @xmath207^{-1 } \ ; c_{0 + } .\ ] ] the total cross section for the electroproduction of mesons on the nucleon is @xmath208 with @xmath209 where @xmath210 is the virtual photon polarization . within this formalism
the baryons and mesons are treated as pointlike particles .
however , with increasing momentum transfers the composite structure of hadrons will become important , affecting the @xmath28 behavior of the initial electroproduction potentials @xmath190 and @xmath191 . in order to analyze these effects ,
we will compare in the next section the results for pointlike particles with those for composite hadrons by inserting monopole form factors in @xmath190 and @xmath191 .
( 300,435 ) ( 0,310)(100,120 ) ( 200,310)(100,120 ) ( 0,165)(100,120 ) ( 200,165)(100,120 ) ( 0,20)(100,120 ) ( 200,20)(100,120 ) ( 100,395)@xmath211 ( 300,395)@xmath212 ( 100,250)@xmath213 ( 300,250)@xmath214 ( 6,105)@xmath215 ( 300,105)@xmath216
in this section we will present the results of our calculation .
we start by performing a global fit to available data for meson - proton and photon - proton reactions which constrains the parameters in the approach .
this allows us to give predictions for further processes such as the cross sections for @xmath217 and @xmath6 and @xmath2 electroproduction .
we conclude this section by studying the effects of @xmath6-@xmath2 mixing and the contributions of the @xmath218 meson to those reactions where it is not produced as a final particle . we have performed a global fit to a large amount of data , consisting of meson - proton and photon - proton reactions for values of @xmath219 between 1.5 and 2.0 gev .
our cross sections include only @xmath1-wave contributions , which are dominant in this energy range for most processes . from comparison with the work of @xcite
we expect @xmath43-waves to exceed the @xmath1-wave contribution for most processes at high energies , further away from threshold , but for a few channels they constitute the main contribution in the whole energy domain .
this explains why our results are below the experimental data for some of the discussed processes .
( 300,265 ) ( 0,165)(100,120 ) ( 200,165)(100,120 ) ( 0,20)(100,120 ) ( 200,20)(100,120 ) ( 0,255)@xmath211 ( 200,255)@xmath212 ( 0,105)@xmath213 ( 200,105)@xmath214 the fit yields the values @xmath220 and @xmath221 for the unknown parameters of the leading order lagrangian , eq .
( [ bar ] ) , @xmath222 gev@xmath95 , @xmath223 gev@xmath95 and @xmath224 gev@xmath95 for the mass counterterms , and @xmath225 , @xmath226 , @xmath227 , @xmath228 , @xmath229 , @xmath230 and @xmath231 ( in units of gev@xmath95 ) for the derivative couplings in the next - to - leading order lagrangian . for
the regularization scales our fitted values are @xmath232 gev , @xmath233 gev , @xmath234 gev , @xmath235 gev and @xmath236 gev .
the parameters @xmath108 and @xmath109 are set to zero , since according to large @xmath65 estimates they are expected to be suppressed and small variations of their values around zero have not shown any substantial impact on the cross sections .
the pion - proton cross sections are summarized in fig .
[ fig : pip ] : fig .
[ fig : pip].a shows the cross section for the reaction @xmath237 . at @xmath238 energies below 1 gev
the cross section is dominated by the @xmath1-wave resonance @xmath12 , which is nicely reproduced in our calculation . in fig .
[ fig : pip].b the results for the reaction @xmath239 are given .
again the cross section is dominated by @xmath1-waves , but in this case the main contribution stems from the @xmath240 which is accompanied by a cusp effect due to the opening of the @xmath19 threshold . in figs .
[ fig : pip].c and [ fig : pip].d the @xmath19 production reactions are depicted .
they receive some ( small ) contribution from the isospin @xmath241 component of the amplitude and are dominated in the low - energy region by @xmath1-wave contributions , but do not signal the appearance of a resonance . in fig .
[ fig : pip].e we show the cross section for the reaction @xmath242 .
this reaction receives only contributions from the isospin @xmath241 amplitude , and it turns out that the @xmath1-wave contribution is negligible as already observed in @xcite . the last pion - proton cross section is shown in fig .
[ fig : pip].f , where the @xmath2 is produced in @xmath243 collisions .
albeit we show this plot with the rest of the pion - proton cross sections , it is important to note that the last reaction is _ not _ included in our fit .
once the involved parameters are constrained by the other reactions , our model provides a prediction for this process which can be checked by future experiments .
let us now turn to the photoproduction cross sections which are shown in fig .
[ fig : gamp ] .
we present in fig .
[ fig : gamp].a the @xmath6 photoproduction data measured at mami @xcite and elsa @xcite and our fitted result confirms the dominance of the @xmath12 , which is responsible for almost the entire cross section in the low - energy region .
results for the reaction @xmath244 are given in fig .
[ fig : gamp].b , where at low energies the cross section is dominated by the @xmath240 , while @xmath43-waves become important for @xmath245 energies above 1.1 gev .
one also observes a cusp effect due to the @xmath19 threshold . for the next two reactions , @xmath246 and @xmath247 photoproduction , the cross sections are @xmath43-wave
dominated , and our results are able to account for small @xmath1-wave cross sections , describing the data only in the very near threshold region .
finally , we show in fig .
[ fig : fig7 ] our results for the reaction @xmath248 , which has been measured at elsa @xcite , where the coherent contribution of two resonances , the @xmath249 and @xmath41 , was observed .
our formalism is capable of reproducing the appearance of an @xmath1-wave resonance around 1900 gev in contradistinction to the work reported in @xcite which remains far below the measured decay rates .
so far , we have discussed with the exception of @xmath217 in fig . [
fig : pip].f only the channels that were used in order to constrain the undetermined chiral parameters in our approach . using the same values for the parameters we can now predict the cross sections for further processes which will provide a true test for the applicability of the model .
the electroproduction of @xmath6 and @xmath2 mesons on a nucleon , e.g. , is an ideal testing ground for the approach , being much more sensitive to its details .
detailed information about the electromagnetic structure of the nucleon is supplied by electroproduction experiments , which yield important constraints for hadron models .
after having fixed the parameters in our approach , we can now compare our predictions for @xmath6 and @xmath2 electroproduction with available experimental data .
we first show our predictions for pointlike hadrons .
the @xmath6 electroproduction on the proton has been measured in detail at clas at jlab @xcite and the data is shown together with our results in fig .
[ fig : elp].a .
( 330,125 ) ( 0,20)(100,120 ) ( 220,20)(100,120 ) ( 115,115)@xmath211 ( 332,115)@xmath212 the invariant momentum transfer @xmath28 of the presented data ranges from 0.375 to 0.875 gev@xmath25 and the applicability of our approach to such high momentum transfers may be regarded as questionable .
nevertheless , we should be able to capture qualitative features of the @xmath28 evolution of the cross section , in particular its slow fall - off which is unusual and in sharp contrast , e.g. , to the fall given by a nucleon dipole form factor .
our results , although showing less decrease with @xmath28 than that of a simple nucleon dipole form factor , produce a faster reduction at low @xmath28 than the experimental data , and then flatten at higher momentum transfers .
we employed in all plots a virtual photon polarization of @xmath250 which is roughly the average of the polarizations of the clas experiment and small variations around this value yield changes that are almost negligible .
our predictions for electroproduction of the @xmath2 on the proton are depicted in fig .
[ fig : elp].b for the same range of momentum transfers and data for this process will soon become available from experiments at jefferson lab @xcite . in this case , the features of the @xmath28 evolution of the cross section are even more striking , since they exhibit a fast increase instead of the usual decrease . as already mentioned in sec .
[ sec : ext ] , the composite structure of baryons and mesons will become increasingly important with rising invariant momentum transfers @xmath28 .
it is therefore natural to include form factors in the initial electroproduction potentials @xmath190 and @xmath191 , in order to account for the electromagnetic structure of hadrons .
one may be inclined to multiply the potentials by an overall form factor which is the same for all hadrons that interact with the photon ( and thus the same for all channels ) .
obviously , this would reduce all electroproduction cross sections , e.g. , it would be easy to produce a decreasing @xmath28 evolution of the @xmath2 electroproduction cross section , in accordance with other electroproduction processes .
however , this procedure will also yield smaller cross sections for @xmath6 electroproduction leading to stronger disagreement with the data . in this case
, the inclusion of form factors which accounts for the electromagnetic structure of the baryons and mesons seems to worsen the situation .
as we will see now , there is indeed a way of including form factors for the particles , while at the same time improving the situation _ both _ for @xmath6 _ and _ @xmath2 electroproduction .
based on the observation that form factors provide a more realistic description of the electromagnetic response of hadrons , we will present a plausible qualitative explanation for the unusually hard transition form factor of the @xmath12 .
this hard form factor has been difficult to understand theoretically , and only recent work within a constituent quark model using a hypercentral potential led to better agreement with experiment @xcite . at the same time
, it has been claimed that the hard form factor is in contrast to models that interpret the @xmath12 as a quasi - bound @xmath18-@xmath19 state @xcite .
here , we will show that this is not necessarily the case .
( 330,125 ) ( 0,20)(100,120 ) ( 220,20)(100,120 ) ( 115,115)@xmath211 ( 332,115)@xmath212 in order to model the electromagnetic structure of @xmath190 and @xmath191 , we include monopole form factors @xmath251 with a mass parameter @xmath252 depending on the outgoing channel @xmath166 . for @xmath6 and @xmath2 electroproduction the channels @xmath253 and @xmath254 dominate the amplitude .
we choose @xmath255 gev for the first channel , so that the form factor is almost identical with the dirac form factor of the proton , and @xmath256 gev for the latter .
the remaining three channels play only a minor role and we set @xmath257 gev in their case . the results for @xmath6 and @xmath2 electroproduction after including these form factors are given in fig .
[ fig : elpform ] .
surprisingly , the cross sections for @xmath6 electroproduction are increased , while they are reduced in the case of the @xmath2 .
this is due to the fact that we employed different form factors for the participating channels and that some of these channels may compensate each other in the final state interactions . to be more precise , for @xmath6 electroproduction
the @xmath253 and @xmath254 channels tend to counterbalance each other , for @xmath2 electroproduction , on the other hand , they add up .
in contrast to common belief , the inclusion of form factors can increase the cross section , e.g. in @xmath6 electroproduction , yielding a @xmath28 evolution closer to experiment and indicating a hard transition form factor .
the claim that the hard form factor is counterintuitive to an interpretation of this state as a bound hadronic system @xcite is not justified , since within the model the photon couples directly to one of the ground state octet baryons or a meson and only after this initial reaction the produced meson forms a bound state with the baryon .
we do not expect our results for electroproduction to reproduce precisely the experimental data .
nevertheless , the inclusion of simple form factors for the electroproduction potentials is able to explain qualitatively the slow decrease of the @xmath12 photocoupling .
the same form factors also flatten the increase in the @xmath28 evolution in @xmath2 electroproduction but keeping the results still above the cross section for photoproduction .
again , we can not regard it as very likely that the results for the @xmath2 will be in exact agreement with future experimental data , in particular due to the uncertainties involved with the implemented form factors .
the qualitative behavior of the @xmath28 evolution , however , indicates a hard transition form factor also for @xmath2 electroproduction . within our model
it would be very unlikely to accomodate both a hard form factor for @xmath6 electroproduction and faster decreasing cross sections for electroproduction of the @xmath2 .
upcoming experiments will clarify this issue and provide a further test of the model .
next , we consider @xmath6 and @xmath2 electroproduction on the neutron , see fig .
[ fig : elnform ] .
( 330,125 ) ( 0,20)(100,120 ) ( 220,20)(100,120 ) ( 115,115)@xmath211 ( 332,115)@xmath212 we have employed the same monopole form factors as in the case of the proton .
photoproduction of the @xmath6 has a cross section of about 6 @xmath258 at its peak , being less than half of the cross section on the proton .
this ratio was measured to be approximately 2/3 @xcite which is clearly above our prediction .
this problem was already recognized in @xcite and is due to the fact that the photon can not couple directly via the charge to a neutral baryon or meson and thus some of the electroproduction amplitudes @xmath190 and @xmath191 vanish , e.g. , the @xmath18 channel to the order we are working . to cure this problem ,
the leading corrections for the coupling of the photon to a neutral baryon via its anomalous magnetic moment have been taken into account in @xcite leading to better agreement with experiment . as these corrections originate from the higher order lagrangian , which is beyond the scope of the present investigation , we refrain from including these terms , but keep in mind that sizable corrections may occur in electroproduction on the neutron .
the @xmath28 evolution of the cross section exhibits a sharp decrease both for @xmath6 and @xmath2 , a behavior which is remarkably different from the proton case and provides another prediction to be checked .
this is due to the inclusion of form factors , otherwise the cross sections for @xmath6 and @xmath2 electroproduction on the neutron , not shown here for brevity , would grow with @xmath28 . in this section
, we wish to investigate the effects of @xmath6-@xmath2 mixing and the importance of the @xmath259 virtual state in our coupled channel formalism .
this is done in a two - step procedure : first , @xmath6-@xmath2 mixing is turned off , and secondly , we eliminate the @xmath259 channel from the model . in both cases we do not repeat the fit which would actually compensate most of the changes .
we are particularly interested in the reaction @xmath239 which exhibits the most prominent @xmath22 cusp of all the channels .
omission of both @xmath6-@xmath2 mixing and the @xmath259 channel do not lead to substantial differences in the remaining reactions ( where the @xmath2 is not produced ) . in fig .
[ fig : effe ] we have chosen to present in addition to @xmath239 the photoproduction process @xmath260 , in order to give a measure for the changes in the other channels .
( 330,125 ) ( 0,20)(100,120 ) ( 220,20)(110,120 ) ( -5,115)@xmath211 ( 225,115)@xmath212 for @xmath239 variation in @xmath6-@xmath2 mixing has almost no impact ( dash - dotted line ) , like in most other channels in which the @xmath6 is not produced as a final particle .
eliminating the @xmath259 channel makes the @xmath22 cusp disappear and lowers the cross section , bringing it to better agreement with the data ( dashed line ) .
it suggests that the region around the @xmath22 cusp is overemphasized within our model .
this feature may change after the inclusion of @xmath43-waves , since then a new overall fit to the different reaction channels will lower the @xmath1-wave contribution reducing the absolute importance of the cusp . for the photoproduction of the @xmath6 on the proton @xmath6-@xmath2 mixing
plays a slightly more prominent role ( dash - dotted line ) , as the @xmath6 is produced in the final state .
if the @xmath259 channel is turned off , the changes are again quite moderate ( dashed line ) .
overall we can conclude , that the results for the production of the goldstone bosons are not modified substantially after omitting the @xmath2 which is in accordance with intuitive expectation , since the @xmath259 channel is much higher in mass than the other channels .
we can therefore confirm that it was justified in previous coupled channel analyses to neglect @xmath6-@xmath2 mixing and treat the @xmath6 as a pure octet state , see e.g. @xcite .
in this work we presented a coupled channel approach to meson - baryon scattering and electroproduction processes including the @xmath2 meson based on chiral symmetry and unitarity . since we restrict ourselves to the threshold region , only @xmath1-wave contributions are taken into account which are dominant at the pertinent energies for most of the discussed reactions .
the most general chiral effective lagrangian for the strong interactions which includes the @xmath2 explicitly is presented up to next - to - leading order and the effective @xmath1-wave potentials are derived which are then iterated in a bethe - salpeter equation . following the work of @xcite we extended the formalism to include electroproduction processes where the photon can couple to the hadrons via their charge . with only a few chiral parameters these
are , on the one hand , the unknown coupling constants of the lagrangian and , on the other hand , the regularization scales of the loop integrals within the bethe - salpeter equation we performed a fit to a large amount of data , consisting of meson - baryon scattering and photoproduction of mesons on the proton .
the @xmath1-wave approximation yields good agreement with the low - energy data in many channels , whereas in the remaining channels @xmath43-wave contributions are expected to be dominant .
we clearly generate the appearance of the @xmath12 in the pion- and photo - induced production of the @xmath6 , while the @xmath240 is produced in the @xmath18 channels .
we are also able to produce a resonance - like shape in the photoproduction of the @xmath2 which has been assigned to the @xmath249 in a recent experiment at elsa .
the overall agreement of our results with data indicates that chiral dynamics and unitarity govern processes up to center - of - mass energies of @xmath2612 gev and that the @xmath2 can be included systematically in a chiral effective lagrangian with baryons .
having constrained the parameters of the approach , we can give predictions for further processes such as the pion - induced production of the @xmath2 in @xmath217 or @xmath6 and @xmath2 electroproduction , which is more sensitive to the structures of the nucleon due to the longitudinal coupling of the virtual photon to the nucleon spin and provides a better test for our model when compared with data .
the only available electroproduction data exists so far for the @xmath6 with invariant momentum transfers @xmath28 starting at 0.375 gev@xmath25 .
it exhibits an unusually slow @xmath28 evolution which is in general difficult to understand .
although our results are larger than that of a simple nucleon dipole form factor , for example , they exhibit a faster decrease with @xmath28 for small values of @xmath28 which then flattens at higher momentum transfers . in the case of @xmath2 electroproduction ,
the features of the @xmath28 evolution of the cross section are even more striking , since they exhibit a fast increase instead of the usual decrease . by choosing different form factors for the initial electroproduction potentials which take into account the electromagnetic structure of the hadrons instead of treating them as pointlike particles the cross sections for @xmath6 electroproduction are _
increased _ , but lowered for the @xmath2 .
this brings the @xmath28 evolution of @xmath6 electroproduction closer to experiment and signals a hard transition form factor within the model .
we argue that according to this approach a hard form factor is expected also for @xmath2 electroproduction which provides a further test of the model . in order to make more precise statements
, one must include not only @xmath43-waves , but also higher order corrections for the coupling of the photon to a baryon via its anomalous magnetic moment .
the omission of such contact interactions explains why we fail in reproducing the photoproduction cross section of the @xmath6 on the neutron .
finally , we have discussed the impact of @xmath6-@xmath2 mixing and the importance of the @xmath259 channel within the coupled channel formalism .
omission of the mixing and the @xmath259 channel do not lead to substantial changes in those reactions in which the @xmath2 is not produced as a final particle .
this is in accordance with intuitive expectation , since the @xmath259 channel is much higher in mass than the other channels .
we are grateful to n. kaiser who shared his insight in the coupled channel formalism with us .
useful discussions with t. hemmert and w. weise are gratefully acknowledged .
we would also like to thank j. mueller for providing us with the data for @xmath6 electroproduction from clas at jlab .
this work has been supported in part by the deutsche forschungsgemeinschaft . in this appendix , we list the coefficients @xmath171 from eq .
( [ ampl ] ) in the isospin basis .
there are five channels with total isospin @xmath157 which are @xmath158 , @xmath159 , @xmath160 , @xmath161 , @xmath162 ( labelled with indices 1 , 2 , 3 , 4 , 5 , respectively ) and two channels with @xmath156 , @xmath163 , @xmath164 ( labeled by indices 6 and 7 ) .
the physical states can be expressed in terms of the isospin states , @xmath262 where we have used the convention in which for the following states there is a sign difference between the physical and isospin basis : @xmath263 , @xmath264 , @xmath265 , @xmath266 , @xmath267 , @xmath268 , @xmath269 , @xmath270 .
the octet field @xmath35 and its singlet counterpart @xmath36 as they appear in the lagrangian are related to the physical mass eigenstates @xmath6 and @xmath2 in the one - mixing - angle scheme as follows @xmath271 where we have used @xmath272 .
this is the value which one obtains from the gell - mann okubo mass relation for the pseudoscalar mesons at lowest order .
as discussed in the main text , small variations in this value lead only to moderate changes which can even be partially compensated by repeating the fit .
the implementation of the more general two - mixing - angle scheme , see e.g. @xcite , does not change any of our conclusions and will therefore not be considered here .
the coefficients @xmath273 read @xmath274 \cos \vartheta - \big [ \sqrt{2 } ( b_d + b_f ) - \sqrt{3 } ( c_d + c_f ) \big ] \sin \vartheta \big ) m_\pi^2 { \nonumber}\\ & & - \big ( 2 \big [ d_1 + 3d_2\big ] \cos \vartheta -\sqrt{2 } \big [ 2 d_1 + 3d_5 + 3 d_6\big ] \sin \vartheta \big ) p_1 ^ 2{\nonumber}\\ c_1 ^ 3 & = & - \frac{3}{8 } [ m_n + m_\lambda - 2 \sqrt{s } ] - \frac{1}{4 } ( d+f ) ( d+3f ) \frac{(\sqrt{s } - m_n)(\sqrt{s } - m_\lambda)}{\sqrt{s } + m_n } { \nonumber}\\ & & - \frac{1}{2 } ( b_d + 3 b_f ) ( m_k^2 + m_\pi^2 ) + 3(d_1 + d_2 ) p_1 ^ 3 { \nonumber}\\ c_1 ^ 4 & = & - \frac{1}{8 } [ m_n + m_\sigma - 2 \sqrt{s } ] - \frac{3}{4 } ( d^2-f^2 ) \frac{(\sqrt{s } - m_n)(\sqrt{s } - m_\sigma)}{\sqrt{s } + m_n } { \nonumber}\\ & & + \frac{1}{2 } ( b_d - b_f ) ( m_k^2 + m_\pi^2 ) + ( d_1 - 7 d_2 + 2d_3 ) p_1 ^ 4 { \nonumber}\\ c_1 ^ 5 & = & \frac{1}{4 } ( d+f ) \big ( \sqrt{2 } ( 2d + 3d_s ) \cos \vartheta - ( d-3f ) \sin
\vartheta \big ) \frac{(\sqrt{s } -m_n)^2}{\sqrt{s } + m_n } { \nonumber}\\ & & + 2 \big ( [ b_d + b_f ] \sin \vartheta + \big [ \sqrt{2 } ( b_d + b_f ) - \sqrt{3 } ( c_d + c_f ) \big ]
\cos \vartheta \big ) m_\pi^2 { \nonumber}\\ & & - \big ( \sqrt{2 } \big [ 2 d_1 + 3d_5 + 3 d_6\big ] \cos \vartheta + 2 \big [ d_1 + 3d_2\big ] \sin \vartheta \big ) p_1 ^ 5{\nonumber}\\ c_2 ^ 2 & = & -\frac{1}{12 } \big ( \sqrt{2 } ( 2d+3d_s ) \sin \vartheta + ( d-3f ) \cos \vartheta \big)^2 \frac{(\sqrt{s } -m_n)^2}{\sqrt{s } + m_n } - 4 m_n u_1 \sin^2 \vartheta { \nonumber}\\ & & + \frac{8}{3 } \big ( - \big [ b_0 + b_d -b_f \big ] \big ( \sqrt{2 } \cos \vartheta + \sin \vartheta \big)^2 + \sqrt{6 } \big [ c_0 + c_d - c_f \big ] \sin \vartheta \big ( \sqrt{2 } \cos \vartheta + \sin \vartheta \big ) \big ) m_k^2 { \nonumber}\\ & & + \frac{2}{3 } \big ( \big [ 2 b_0 + 3b_d -5b_f \big ] \cos^2 \vartheta - 2 \big [ \sqrt{2 } ( -4b_0 -3b_d + b_f ) + \sqrt{3 } ( 4c_0 + 3c_d - c_f ) \big ] \cos \vartheta \sin \vartheta { \nonumber}\\ & & - \frac{4}{3 } \big [ b_0 + 2 b_f - \sqrt{6 } ( c_0 + 2 c_f ) \big ] \sin^2 \vartheta \big ) m_\pi^2 { \nonumber}\\ & & + 2 \big ( 2 \big [ d_4 + 2d_5 + 3 d_7\big ] \sin^2 \vartheta - \sqrt{2 } \big [ 2 d_1 -d_5 + 3 d_6\big ] \sin \vartheta
\cos \vartheta + \big [ -d_1 + 3d_2 + 2 d_4\big ] \cos^2 \vartheta \big ) p_2 ^ 2{\nonumber}\ ] ] @xmath275\cos \vartheta { \nonumber}\\ & & - \frac{1}{12 } ( d+3f ) \big ( \sqrt{2 } ( 2d + 3d_s ) \sin \vartheta + ( d-3f ) \cos \vartheta \big ) \frac{(\sqrt{s } - m_n)(\sqrt{s } - m_\lambda)}{\sqrt{s } + m_n } { \nonumber}\\ & & + \frac{1}{6 } [ b_d + 3 b_f ] ( 3 m_\pi^2 - 5 m_k^2 ) \cos \vartheta - \frac{2}{3 } \big [ \sqrt{2 } ( b_d + 3 b_f ) - \sqrt{3 } ( c_d + 3 c_f ) \big ] m_k^2 \sin \vartheta { \nonumber}\\ & & + \big ( \sqrt{2 } \big [ 2 d_1 + d_5 + 3 d_6\big ] \sin \vartheta + \big [ d_1 - 3d_2 + 2 d_3\big ] \cos \vartheta \big ) p_2 ^ 3{\nonumber}\\ c_2 ^ 4 & = & \frac{3}{8 } [ 2\sqrt{s } -m_n - m_\sigma ] \cos \vartheta { \nonumber}\\ & & - \frac{1}{4 } ( d - f ) \big ( \sqrt{2 } ( 2d + 3d_s ) \sin \vartheta + ( d-3f ) \cos \vartheta \big ) \frac{(\sqrt{s } - m_n)(\sqrt{s } - m_\sigma)}{\sqrt{s } + m_n } { \nonumber}\\ & & + \frac{1}{2 } [ b_d - b_f ] ( 3 m_\pi^2 - 5 m_k^2 ) \cos \vartheta - 2 \big [ \sqrt{2 } ( b_d - b_f ) - \sqrt{3 } ( c_d - c_f ) \big ]
m_k^2 \sin \vartheta { \nonumber}\\ & & - \big ( \sqrt{2 } \big [ 2 d_1 -3d_5 + 3 d_6\big ] \sin \vartheta + \big [ d_1 - 3d_2\big ] \cos \vartheta \big ) p_2 ^ 4{\nonumber}\\ c_2 ^ 5 & = & \frac{1}{12 } \big ( \sqrt{2 } ( 2d+3d_s ) \sin \vartheta + ( d-3f ) \cos \vartheta \big ) \big ( \sqrt{2 } ( 2d+3d_s ) \cos \vartheta - ( d-3f ) \sin \vartheta
\big ) { \nonumber}\\ & & \times \frac{(\sqrt{s } -m_n)^2}{\sqrt{s } + m_n } + 4 m_n u_1 \cos \vartheta \sin \vartheta { \nonumber}\\ & & + \frac{1}{3 } \bigg [ \big(8\sqrt{2}(b_0 + b_d - b_f ) -8 \sqrt{3 } ( c_0 + c_d - c_f ) \big ) m_k^2{\nonumber}\\ & & -\big(2 \sqrt{2}(4b_0 + 3b_d - b_f ) -2 \sqrt{3 } ( 4c_0 + 3c_d - c_f ) \big ) m_\pi^2 { \nonumber}\\ & & + 3 \sqrt{2 } ( 2 d_1 -d_5 + 3 d_6 ) p_2 ^ 5\bigg]\cos ( 2\vartheta ) { \nonumber}\\ & & -\frac{1}{3 } \bigg [ 4\big(b_0 + b_d - b_f + \sqrt{6}(c_0 + c_d - c_f)\big ) m_k^2 { \nonumber}\\ & & + \big(-4 b_0 - 3 b_d + b_f + 2 \sqrt{6 } ( c_0 + 2 c_f)\big ) m_\pi^2 { \nonumber}\\ & & + 3(d_1 -3 d_2 + 4 d_5 + 6 d_7)p_2 ^ 5 \bigg ] \sin ( 2 \vartheta)\end{aligned}\ ] ] @xmath276 \sin \vartheta{\nonumber}\\ & & + \frac{1}{12 } ( d+3f ) \big ( \sqrt{2 } ( 2d + 3d_s ) \cos \vartheta - ( d-3f ) \sin \vartheta \big ) \frac{(\sqrt{s } - m_n)(\sqrt{s } - m_\lambda)}{\sqrt{s } + m_n } { \nonumber}\\ & & + \frac{1}{6 } [ b_d + 3 b_f ] ( 3 m_\pi^2 - 5 m_k^2 ) \sin \vartheta + \frac{2}{3 } \big [ \sqrt{2 } ( b_d + 3 b_f ) - \sqrt{3 } ( c_d + 3 c_f ) \big ] m_k^2 \cos \vartheta { \nonumber}\\ & & - \big ( \sqrt{2 } \big [ 2 d_1 + d_5 + 3 d_6\big ] \cos \vartheta - \big [ d_1 - 3d_2 + 2 d_3\big ] \sin \vartheta \big ) p_3 ^ 5{\nonumber}\\ c_4 ^ 4 & = & \sqrt{s } - m_\sigma - \frac{3}{4 } ( d - f)^2 \frac{(\sqrt{s } - m_\sigma)^2}{\sqrt{s } + m_n } { \nonumber}\\ & & - 2 ( 2 b_0 + b_d - 2 b_f ) m_k^2 + 2 ( - 2d_1 + d_2 + 2 d_4 ) p_4 ^ 4 { \nonumber}\\ { \nonumber}\\ c_4 ^ 5 & = & \frac{3}{8}[2\sqrt{s } -m_n - m_\sigma ] \sin \vartheta{\nonumber}\\ & & + \frac{1}{4 } ( d - f ) \big ( \sqrt{2 } ( 2d + 3d_s ) \cos \vartheta - ( d-3f ) \sin
\vartheta \big ) \frac{(\sqrt{s } - m_n)(\sqrt{s } - m_\sigma)}{\sqrt{s } + m_n } { \nonumber}\\ & & + \frac{1}{2 } [ b_d - b_f ] ( 3 m_\pi^2 - 5 m_k^2 ) \sin \vartheta + 2 \big [ \sqrt{2 } ( b_d - b_f ) - \sqrt{3 } ( c_d - c_f ) \big ] m_k^2 \cos \vartheta { \nonumber}\\ & & + \big ( \sqrt{2 } \big [ 2 d_1 -3d_5 + 3 d_6\big ] \cos \vartheta - \big [ d_1 - 3d_2\big ] \sin \vartheta \big ) p_4 ^ 5{\nonumber}\\ c_5 ^ 5 & = & -\frac{1}{12 } \big ( \sqrt{2 } ( 2d+3d_s ) \cos \vartheta - ( d-3f ) \sin \vartheta \big)^2 \frac{(\sqrt{s } -m_n)^2}{\sqrt{s } + m_n } - 4 m_n u_1 \cos^2 \vartheta { \nonumber}\\ & & + \frac{8}{3 } \big ( - \big [ b_0 + b_d -b_f \big ] \big ( \sqrt{2 } \sin \vartheta - \cos \vartheta \big)^2 - \sqrt{6 } \big [ c_0 + c_d - c_f \big ] \cos \vartheta \big ( \sqrt{2 } \sin \vartheta - \cos \vartheta \big ) \big ) m_k^2 { \nonumber}\\ & & + \frac{2}{3 } \big ( \big [ 2 b_0 + 3b_d -5b_f \big ] \sin^2 \vartheta + 2 \big [ \sqrt{2 } ( -4b_0 -3b_d + b_f ) + \sqrt{3 } ( 4c_0 + 3c_d - c_f ) \big ] \sin \vartheta \cos \vartheta { \nonumber}\\ & & - 2 \big [ b_0 + 2 b_f - \sqrt{6 } ( c_0 + 2 c_f ) \big ] \cos^2
\vartheta \big ) m_\pi^2 { \nonumber}\\ & & + 2 \big ( 2 \big [ d_4 + 2d_5 + 3 d_7\big ] \cos^2 \vartheta + \sqrt{2 } \big [ 2 d_1 -d_5 + 3 d_6\big ] \cos \vartheta \sin \vartheta + \big [ -d_1 + 3d_2 + 2 d_4\big ] \sin^2 \vartheta \big ) p_5 ^ 5{\nonumber}\ ] ] @xmath277 - 2 ( 2 b_0 + b_d + b_f ) m_\pi^2 + 2 ( d_1 + d_2 + 2d_4 ) p_6 ^ 6{\nonumber}\\ c_6 ^ 7 & = & \frac{1}{4 } [ m_n + m_\sigma - 2 \sqrt{s } ] - ( b_d - b_f ) ( m_k^2 + m_\pi^2 ) + 2 ( -d_1 + d_2 + d_3 ) p_6 ^ 7{\nonumber}\\ c_7 ^ 7 & = & - \frac{1}{2 } [ \sqrt{s } - m_\sigma ] - 2 ( 2 b_0 + b_d + b_f ) m_k^2 + 2 ( d_1 + d_2 + 2d_4 ) p_7 ^ 7{\nonumber}\ ] ] for an isospin state labeled @xmath166 ( @xmath167 ) which consists of a baryon @xmath278 ( @xmath279 ) and a meson @xmath280 ( @xmath281 ) we have used the abbreviation @xmath282 and @xmath283 ( @xmath284 ) is the three - momentum of the incoming ( outgoing ) baryon in the center - of - mass frame . | the photo- and electroproduction of the @xmath0 mesons on nucleons are investigated within a relativistic chiral unitary approach based on coupled channels .
the @xmath1-wave potentials for electroproduction and meson - baryon scattering are derived from a chiral effective lagrangian which includes the @xmath2 as an explicit degree of freedom and incorporates important features of the underlying qcd lagrangian such as the axial @xmath3 anomaly .
the effective potentials are iterated in a bethe - salpeter equation and cross sections for @xmath4 photo- and electroproduction from nucleons are obtained .
the results for the @xmath2 photoproduction cross section on protons reproduce the appearance of an @xmath5 resonance around 1.9 gev observed at elsa .
the inclusion of electromagnetic form factors increases the predicted @xmath6 electroproduction cross sections on the proton , providing a qualitative explanation for the hard form factor of the photocoupling amplitude observed at clas . * @xmath7 , @xmath8 photo- and + electroproduction off nucleons * b. borasoy , e. marco , and s. wetzel http://www.ph.tum.de/[physik department ] + http://www.tum.de/[technische universitt mnchen ] + d-85747 garching , germany + [ cols= " < , < " , ] |
morphology - dependent resonances ( mdrs ) , which is defined as the resonances found in cylindrical , spherical and ellipsoidal shaped optical cavities , are a much studied topic in the field of optics due to their practical importance in many applications @xcite as well as their heuristic value .
more than a decade ago , johnson is the first to approach mdr in a spherical cavity @xcite as an analogy of quantum mechanical resonances , which is familiar in nuclear , atomic , and molecular scattering theory . by the same approach ,
nckel have studied mdr analytically in a cylindrical cavity @xcite as an unperturbed system in order to analyze a slightly deformed one .
recently , study of the mdr is resurfaced as a hot issue in connection with the application of deformed microcavity lasers because of its high potentiality of application for photonics and optoelectronics .
_ have reported the existence of a different set of resonances , which can be classified as shape ( _ outer _ ) resonances ( _ ors _ ) with high leakage along with feshbach ( _ inner _ ) resonances ( _ irs _ ) in a two - dimensional dielectric circular microdisk for transverse magnetic ( tm ) and transverse electric ( te ) polarizations @xcite .
the similar topic was treated in a numerical method for quantum billiard @xcite .
the evolutions of these two classes of resonances to the so - called _ small opening limit _ where the refractive index @xmath1 of the disk goes to infinity have been explicitly studied @xcite and the applicability of effective potential analogy for the ors has been checked @xcite .
one intriguing fact of the mdr dynamics from last two studies is that , for te polarization with magnetic field @xmath2 perpendicular to the cavity plane , both irs and ors are found in the same region of complex wavenumber plane where the regime of @xmath1 is practically relevant . in two - dimensional chaotic quantum billiards , it is well known that the level dynamics resembles dynamics of one - dimensional particles with repulsive interaction as a system parameter varies @xcite .
it is called avoided level crossing . similarly , avoided resonance crossing ( arc ) phenomenon in open hamiltonian systems
is generally categorized into two classes @xcite . in the case of a strong interaction , shortly _ strong arc _ , the imaginary parts of complex - valued energies are crossing and the real parts are avoided .
the identities of the two modes are exchanged via the arc and their wave patterns are strongly superposed during the arc . for a weak interaction , shortly _ weak arc _ , the real parts are crossing and the imaginary parts are avoided while their identities are retained . in two - dimensional chaotic dielectric cavities , there have been undertaken many theoretical @xcite and experimental @xcite researches related with the arc .
due to the open property , the resonance dynamics in a dielectric cavity is much complicated and shows diverse aspects compared to that of the closed billiard .
recently , it has been represented that , when a deformation parameter is varied in the rectangular or elliptic cavity , of which the corresponding billiard is integrable , arc can occur by the effect of openness @xcite .
another recent report have claimed that , even on a slight deformation from the disk shape , a symmetry breaking of escaping rays occurs , which gives rise to dramatic changes in emission patterns depending on the characteristics of deformations @xcite , and emission directions and resonance patterns are deeply influenced by the refractive index of the cavity , i.e. , openness of the system , as well as the underlying classical manifold structures @xcite .
these imply that analyzing the emission directions and mode patterns of slightly deformed cavities requires understanding of the mdr dynamics for openness . in this paper
, we uncover the mechanism behind the mdr dynamics in a circular disk for te polarization as the openness parameter @xmath3 is varied and observe the impacts of pure openness on the dynamics .
the perfect rotational symmetry provides the analysis possible .
the generic two - dimensional dielectric cavities have mainly two types of changeable system parameters , e.g. , degree of geometrical deformation and openness . in such cavities
, irs can generally interact with each other .
their resonance dynamics contains a complicated mixture of the aspect by deformation and the effect of openness . on the other hand
, the circular disk does not have any deformation parameter and the only relevant parameter is the openness .
moreover , its rotational symmetry makes that the angular component of the wavefunction is trivial . as a result
, it renders the system effectively one - dimensional and then leads more clear and basic study for the pure openness without the aspect by deformation . , @xmath4 , and @xmath5 , respectively .
[ fig : arcs_disk],width=321 ] the helmholtz equation of dielectric cavities is given by @xmath6 , where @xmath7 and @xmath8 are the wavefunction and the wavenumber , respectively .
in the disk with the radius @xmath9 , the complex - valued wavenumbers @xmath10 for resonances are obtained from the boundary matching conditions for tm and te polarizations under the purely outgoing wave condition @xcite . in the opening regime
@xmath11 , the irs , @xmath12 , are labelled by angular momentum quantum number @xmath13 and modal ordering number @xmath14 which corresponds to the number of radial intensity peaks inside the cavity .
similarly , the ors can be also represented by @xmath15 , where @xmath14 is assigned in accordance with the increasing order of @xmath16 $ ] . for a given @xmath13 , the maximum numbers of ors in tm and te polarizations are @xmath17 $ ] and @xmath18 $ ] , respectively ( the notation @xmath19 $ ] is integer value ) .
notice that , in te case , some ors have relatively low leakage and we have especially called them _ low - leaky outer resonances _ ( low - leaky ors ) @xcite .
since there is only one low - leaky or having the largest @xmath16 $ ] among the ors for each @xmath20 , low - leaky ors can be denoted by @xmath21)$ ] .
additionally , the notations @xmath22 and @xmath23 are used for ir and or in the small opening limit , respectively , where they can be unambiguously identified @xcite .
the subscript zero means @xmath24 .
when @xmath25)$ ] is evolved from the small opening limit toward the large opening regime , both weak arc and strong arc are unexpectedly observed in the dielectric disk .
figure [ fig : arcs_disk ] shows the two types of arcs between ir and low - leaky or with @xmath26 . at @xmath27 , @xmath5 interacts with @xmath4 via weak arc .
when the openness increases , @xmath5 interacts with @xmath28 via strong arc at @xmath29 .
the radial intensity patterns at five points on each line in fig . [ fig : arcs_disk ] are shown in fig .
[ fig : radial_wf ] . due to the superposition near the point of the weak arc at @xmath27 , the leakage of @xmath30 ( fig .
[ fig : radial_wf](g ) ) decreases while that of @xmath31 ( fig .
[ fig : radial_wf](b ) ) increases .
however their identities are retained after the arc as shown in figs .
[ fig : radial_wf](c ) and ( h ) .
likewise at the weak arc , @xmath30 and @xmath32 also undergo the change of leakage and their spatial patterns are superposed near the point of strong arc as shown in figs .
[ fig : radial_wf](h ) and ( m ) .
especially , just at the point of the strong arc , two resonances are strongly superposed and then their patterns are almost identical as shown in figs .
[ fig : radial_wf](i ) and ( n ) . after the strong arc
, the identities of @xmath32 and @xmath30 are exchanged as shown in figs .
[ fig : radial_wf](j ) and ( o ) .
consequently , @xmath5 becomes @xmath32 and @xmath28 becomes @xmath30 . on a further increase of @xmath33 over @xmath34 , @xmath28 , which has become @xmath30 , interacts with @xmath35 through another strong arc .
eventually @xmath22 can experience strong arcs up to two times over the whole regime of @xmath33 .
@xmath22 changes to @xmath21)$ ] at the first strong arc and @xmath22 to @xmath36 at the following second strong arc : @xmath37)\stackrel{_{\text{strong arc}}}{\longrightarrow}\text{i}(m , l-1).\ ] ] , @xmath5 , and @xmath28 , respectively .
five pictures in each row are at @xmath38 , @xmath39 , @xmath40 , @xmath41 , and @xmath42.[fig : radial_wf],width=321 ] on the other hand , @xmath4 and @xmath28 do not interact each other , even if they have the same symmetry and their imaginary values of @xmath10 cross at @xmath43 .
it has been known that two - dimensional optical dielectric cavities are typically non - integrable systems due to their open property and the openness breaks the integrability of the systems .
consequently , the avoided crossings between the irs occur in the optical cavities , even if their corresponding billiards are integrable .
it was explicitly shown in elliptic and rectangular cavity @xcite .
but , unlike the cavities which correspond to integrable billiards such as rectangular and elliptic cavity , the avoided crossings between irs intriguingly do not occur in the case of circular disk cavity even though the system is open .
this fact implicitly indicates that the irs remain with the properties of the normal modes in the corresponding billiard which is integrable in this case .
before we discuss the numerical results , we will consider the analytical approach through a toy model .
the arcs can be displayed with effective hamiltonian composed of a non - hermitian @xmath44 matrix and a coupling term as follows:@xmath45 where @xmath46 and @xmath47 are the complex energy @xmath48 of ir and or , respectively .
the off - diagonal element in the coupling term , @xmath49 ( @xmath50 ) , describes the coupling effect from or ( ir ) to ir ( or ) and it is assumed that the coupling effects are proportional to @xmath33 due to the fact that the coupling must becomes zero in the small opening limit .
equation leads the eigenvalues @xmath51 to simply illustrate the arcs , we substitute @xmath46 and @xmath47 with the intuitive relations of the mdr dynamics .
firstly , to obtain the basic mdr dynamics for te case except for the effects of mode interaction and brewster angle by openness , we borrow the dynamics for tm polarization , in which the interactions between the two types of resonances are infinitesimal in the experimentally feasible opening regime due to the absence of low - leaky ors . from this , @xmath16 $ ] and @xmath52 $ ] of an ir become almost linear to @xmath33 and to @xmath53 , respectively @xcite . here
, @xmath53 given by @xmath54 is the lower bound of @xmath52 $ ] of irs in tm case .
also an or in tm case maintains almost an constant complex valued @xmath10 for the change of @xmath33 @xcite .
secondly , we consider that the destinations of @xmath55 of irs and @xmath10 of ors in the small opening limit for te case , correspond to the zeros of bessel function @xcite , @xmath56 , and the zeros of hankel function derivatives @xcite , @xmath57 , respectively .
then , @xmath46 and @xmath47 for te case can be simply expressed as follows : @xmath58 even though we borrowed the mdr dynamics for tm case , the dynamics of @xmath59 and @xmath60 was well matched to the numerically obtained resonance tracing behaviors @xcite for te case in the opening regime where the interaction is very small .
figures [ fig : arcs_heff](a ) and [ fig : arcs_heff](b ) show weak arc of @xmath4 and strong arc of @xmath28 with @xmath5 , respectively , when the coupling constant @xmath61 .
the value of coupling constant was properly chosen as compared with the numerical result fig .
[ fig : arcs_disk ] .
the degree of mode coupling depends on the relative quantity between @xmath62 and coupling strength @xmath63 located in the second term of right - hand side of eq . .
according to the relative quantity , the crossing point of @xmath33 are determined and the transition of the type of arc can be occur .
it is a well - known phenomenon @xcite . in fig .
[ fig : arcs_heff ] obtained from the toy model , ( a ) is about @xmath64 and ( b ) is about @xmath65 . under
the our coupling constant matched to the numerical result ( fig .
[ fig : arcs_disk ] ) , the point of transition from weak arc to strong arc for @xmath26 is shown between these two irs . , title="fig:",width=158 ] , title="fig:",width=158 ]
for te polarization .
inverted triangles are @xmath21)$ ] and the others are @xmath12 decided by considering arc .
circles are @xmath12 changed from @xmath25)$ ] . dashed , dotted , and dot - dashed lines in ( a ) are @xmath66=m\omega$ ] , @xmath66=m$ ] , and @xmath67 , respectively .
the inset in ( a ) is for @xmath68 .
blue dotted line in ( b ) is @xmath53 .
[ fig : respos_n1.5_arc],width=321 ] for te polarization .
inverted triangles are @xmath21)$ ] and the others are @xmath12 decided by considering arc .
circles are @xmath12 changed from @xmath25)$ ] . dashed , dotted , and dot - dashed lines in ( a ) are @xmath66=m\omega$ ] , @xmath66=m$ ] , and @xmath67 , respectively .
the inset in ( a ) is for @xmath68 .
blue dotted line in ( b ) is @xmath53 .
[ fig : respos_n1.5_arc],width=321 ] the dynamics from the numerical result and the toy model is explained in terms that the coupling between cavity itself and environment becomes strong as openness increases , and then the irs are exquisitely affected by low - leaky ors in large opening regime .
hence , a care should be taken to the identification of mdrs according to the history of arcs . on a resonance position plot in the large opening regime
, various irs can coexist such as once and twice swapping resonances after strong arcs , those superposed near weak arcs and strong arcs , and the uncoupled ones .
figure [ fig : respos_n1.5_arc ] shows the resonance positions in consideration of the arcs .
we included only the irs and the low - leaky ors for the sake of convenience .
the dotted and the dashed lines in fig .
[ fig : respos_n1.5_arc](a ) are @xmath66=m$ ] and @xmath66=m\omega$ ] , respectively . in the effective potential analogy , @xmath66=m$ ]
corresponds to the top of the effective potential barrier , that is , the critical incident angle for irs and @xmath66=m\omega$ ] corresponds to the bottom of the barrier , that is , the incident angle @xmath69 .
effective potential analogy is successfully applied to schematic description of the properties for irs . for this reason
, it is often used on the research about the resonances in dielectric disk .
we , however , note that the effective potential analogy has some insufficient parts yet on applying to ors@xcite . for a given @xmath13 in fig .
[ fig : respos_n1.5_arc](a ) , a circle is @xmath12 changed from @xmath25)$ ] by the first strong arc and an inverted triangle is @xmath21)$ ] changed from @xmath22 by the final strong arc .
the triangles between the circle and the inverted triangle are @xmath36 changed from @xmath22 by two consecutive strong arcs .
if we are able to get anyhow a correct position of @xmath21)$ ] , it is natural that @xmath22 , whose @xmath16$]s are positioned between @xmath25)$ ] and @xmath21)$ ] , are @xmath36 @xcite , because @xmath21)$ ] must be excluded from the ordering of @xmath14 for @xmath12 . here , we newly use a line number for ors defined as @xmath70-l$ ] , @xmath71 and , then , the low - leaky ors are on the line of @xmath72 .
the individual line of @xmath73 except zero is generally parallel to the line of @xmath66=m$ ] .
however , the line of @xmath72 has a different slope .
it surprisingly corresponds to the brewster angle in large opening regime . in the viewpoint of semiclassics , the incident angle for an ir
can be estimated by @xmath74 $ ] and the brewster angle is represented as @xmath75 .
consequently , one can obtain @xmath66 $ ] relevant to the brewster angle , that is , @xmath76 in fig .
[ fig : respos_n1.5_arc](a ) and inset , @xmath16$]s of low - leaky ors are well - aligned on the @xmath67 line , whether they have experienced strong arc or not .
also , by strong arc in the opening regime where irs and low - leaky ors are coexistent in the same region of @xmath8 plane , @xmath52$]s of low - leaky ors settle down in the bottom region of the ir group as shown in fig [ fig : respos_n1.5_arc](b ) .
figure [ fig : res_dynamics](a ) is the three - dimensional representation of the dynamics of @xmath77 and @xmath5 in @xmath16 $ ] , @xmath52 $ ] and @xmath33 space . for more detailed depiction and notation of reference marks , we projectively plot the dynamics of @xmath16 $ ] and @xmath52 $ ] in fig . [
fig : res_dynamics](b ) and fig .
[ fig : res_dynamics](c ) , respectively .
the dynamics of @xmath16 $ ] for irs and ors in tm polarization exhibits almost linear and constant behaviors , respectively @xcite . on the other hand , in te case , the low - leaky ors breaks the monotonic behavior of irs as shown in fig .
[ fig : res_dynamics](b ) through the arcs . as @xmath33 increases
, @xmath5 interacts with @xmath78 , where @xmath79 , sequentially through a series of weak arcs before it interacts with @xmath28 through the first strong arc at @xmath29 , i.e. , there is a transition from weak arc to strong arc . in the large opening regime ,
the point of strong arc satisfies @xmath67 for given @xmath13 and @xmath33 , since @xmath16$]s of irs increase while @xmath16$]s of low - leaky ors tend to maintain near @xmath67 .
after the first strong arc , @xmath16 $ ] of @xmath28 stays near @xmath67 because the resonance turns to @xmath30 by the exchange of their identities .
the strong arcs between @xmath30 and @xmath78 , where @xmath80 , sequentially occur at @xmath16\simeq k_{b}$ ] in the region of @xmath81 until @xmath33 becomes unity , where all resonances merge into the continuum . by openness .
solid lines and a dashed line are @xmath77 and @xmath5 , respectively .
dotted line in ( b ) and ( c ) is @xmath67 and @xmath53 , respectively .
[ fig : res_dynamics],width=321 ] by openness .
solid lines and a dashed line are @xmath77 and @xmath5 , respectively .
dotted line in ( b ) and ( c ) is @xmath67 and @xmath53 , respectively .
[ fig : res_dynamics],width=321 ] the dynamics of @xmath52 $ ] shows more interesting behavior .
@xmath52 $ ] for a resonance is directly connected to the lifetime ( @xmath82 $ ] ) and @xmath0-factor ( @xmath83/2\text{im}[k]$ ] ) . on the space of
@xmath52 $ ] vs @xmath33 as shown in fig .
[ fig : res_dynamics](c ) , ir and low - leaky or approach each other near the point of arc . since the low - leaky or has lower @xmath0 than the irs for the same @xmath13 , @xmath84\right|$ ] of @xmath85 increases by the interaction with @xmath30 .
moreover , the increased @xmath84\right|$ ] after arc is always below the bound line @xmath53 due to the aftermath of the arc .
hence , at a fixed value of @xmath33 , the irs positioned below the @xmath53 line are the resonances influenced by the arcs .
this is the reason why the irs for te polarization have generally lower @xmath0 than the corresponding modes for tm case @xcite and their imaginary part falls below the @xmath53 line in the ordinary experimental regime of @xmath33 as shown in fig [ fig : respos_n1.5_arc](b ) .
also , the dynamics gives a clear explanation for the breaking of the decreasing order of @xmath0-factor in accordance with the increasing order of @xmath14 near the brewster angle .
in a local region of complex @xmath8 space under experimentally feasible circumstances , the grouping of resonances by the order of @xmath0 makes it possible for us to predict the characteristics of the resonances .
if the effects of arc are practically zero , the ir with higher @xmath14 has generally low @xmath0 according to the effective potential analogy since a resonance with higher energy is less trapped in the effective potential @xcite . in te case , however , the order of @xmath0 of irs is reversed from a certain value of @xmath14 due to arcs with ors as shown in fig .
[ fig : res_dynamics](c ) .
for instance , at @xmath68 , the order of @xmath0 corresponds to the order of @xmath14 in tm case as @xmath86 . on the other hand , in te case ,
the order is reversed as @xmath87 .
as @xmath33 increases , this phenomenon is extended to the resonances with lower @xmath14 due to the increase of the coupling strength .
such phenomena , only occur for te polarization case , indicate that the existence of brewster angle strongly activate the interactions between irs and ors as @xmath33 increases . in the dielectric disk for te polarization ,
the irs are sensitively affected by brewster angle for the internal cavity boundary .
so , the irs whose @xmath16 $ ] is near @xmath67 have larger leakage ( @xmath88\right|$ ] ) than the others @xcite .
i.e. , the more @xmath16 $ ] of irs approaches to @xmath67 , the more the leakage of the ir increases .
an ir whose leakage is increased near the brewster angle by the change of openness , as @xmath33 increases , approaches to a low - leaky or in complex @xmath8 space . then , the coupling with the low - leaky or becomes relatively strong and the interaction between them is more activated .
such arc for each ir occurs sequentially at other points of @xmath33 .
moreover , since the coupling strength becomes strong as @xmath33 increases , the transition to strong arc from weak arc occurs based on an ir with specific value of @xmath14 as shown in fig .
[ fig : res_dynamics ] .
summarizing , we have demonstrated that the mdr dynamics by openness in a dielectric circular microdisk for te polarization shows arcs over two types of resonances due to the coupling with environment .
the low - leaky outer resonances interact pronouncedly with the inner resonances through the opening channel near the brewster angle . as a result , in the opening regime relevant to experiments , the resonance classification between inner and outer resonances is more complicated and the change of leakage for inner resonances occurs .
the reverse order of @xmath0-factors of inner resonances which is discrepant from the effective potential analogy can be understood in terms of the arcs between inner and outer resonances .
this study will be useful as a basis to analyze intricate resonance dynamics in deformed dielectric microcavities .
we would like to thank i. kim , j .- w .
ryu , s .- y .
lee , and t. harayama for helpful discussions .
this work was supported by acceleration research ( center for quantum chaos applications ) of mest / kosef . | we have studied the parametric - evolution of morphology - dependent resonances according to the change of openness in a two - dimensional dielectric microdisk for te polarization . for the first time
, we report that the dynamics exhibits avoided resonance crossings between inner and outer resonances even though the corresponding billiard is integrable . due to these recondite avoidances ,
inner and outer resonances can be exchanged and @xmath0-factor of inner resonances is strongly affected .
we analyze the diverse phenomena arisen from the dynamics including the avoided crossings . |
the investigation of the factorization properties of scattering amplitudes at their singularities @xcite lead to important results in quantum field theory and theoretical particle physics , such as the development of new methods for phenomenological computations .
in particular , integrand reduction methods , developed for one - loop diagrams @xcite and recently extended to higher loops @xcite , use the knowledge of the analytic and algebraic structure of loop integrands in order to rewrite scattering amplitudes as linear combinations of master integrals . at one loop ,
integrand - reduction methods allow to express any integrand in dimensional regularization as a sum of contributions with at most five propagators in the loop , regardless of the number of external legs of the amplitude .
the numerators of these contributions are _ polynomial residues _ which have a universal parametric form that does not depend on the process .
this parametric form can be written as a sum of monomials in the components of the loop momentum , multiplied by unknown process - dependent coefficients .
after integration , the amplitude becomes a linear combination of known integrals .
the coefficients of this linear combination can be identified with a subset of the ones which parametrize the residues .
therefore , the problem of the computation of any one - loop amplitude can be reduced to the one of performing a polynomial fit of the coefficients of the residues .
the fit of the unknown coefficients can be efficiently performed by evaluating the numerator of the integrand on _ multiple cuts _
, i.e. on values of the loop momentum such that a subset of the loop denominators vanish .
the multiple - cut conditions can be viewed as _ projectors _ which isolate the residue sitting on the cut denominators .
a residue can be evaluated by putting on - shell the corresponding loop propagators and subtracting from the integrand the non - vanishing contributions coming from higher - point residues .
this leads to a top - down algorithm which allows to compute any one - loop amplitude with any number of external legs . within semi - numerical computations ,
the algorithm is usually implemented by sampling the integrand on several solutions of the multiple cuts and subtracting at each step of the reduction all the non - vanishing contributions coming from higher - point residues .
this yields a system of equations for the coefficients of each residue .
the method is suited for automation and it has been implemented in several codes , some of which are public ( e.g.cuttools @xcite and samurai @xcite ) .
its usage within several automated frameworks @xcite has been particularly successful and produced highly non - trivial phenomenological results . in this paper
we present a new public c++ library called ninja , which implements an alternative integrand - reduction algorithm first proposed in ref .
this is based on the systematic application of the laurent series expansion to an integrand on the multiple cuts . after performing a suitable laurent expansion on a multiple cut , in the asymptotic limit both the integrand and the subtraction terms
exhibit the same polynomial behavior as the residue .
this allows one to directly identify the coefficients of the residues ( and thus the ones of the master integrals ) with the ones of the laurent expansion of the integrand , corrected by subtractions terms which can be computed once and for all as functions of a subset of the higher - point coefficients .
this leads to a diagonal system of equations for each residue and to a significant reduction of the number of subtraction terms which affect the computation of lower - point contributions .
ninja takes as input the numerator cast into four different forms .
the first is a procedure which evaluates it as a function of the loop momentum .
the others instead compute the leading terms of properly defined laurent expansions of the numerator .
since the integrand of a one - loop amplitude is a rational function of the loop momentum , the laurent expansions for an integrand can be obtained via a partial fraction decomposition .
ninja implements it semi - numerically via a simplified polynomial division algorithm between the expansions of the numerator and the ones of the denominators .
the coefficients of the laurent expansion are then corrected by the subtraction terms and multiplied by the master integrals .
these are computed by interfacing ninja with an external library which can be specified by the user .
interfaces for oneloop @xcite and looptools @xcite are already provided with the distribution .
the simplified subtractions and the diagonal systems of equations make the algorithm implemented in ninja significantly simpler and lighter than the traditional one .
the library has been interfaced with the one - loop package gosam @xcite and has already been used to compute nlo corrections to higgs boson production in association with a top - quark pair and a jet @xcite and several six- , seven- and eight - point amplitudes involving both massive and massless particles as external states or propagating in the loop @xcite .
these applications showed that ninja has better performance and numerical stability than implementations of traditional integrand reduction algorithms .
in particular , ref .
@xcite provides a detailed analysis of the performance and accuracy of this library . with this paper ,
we make ninja publicly available as a standalone library which can be interfaced to other packages and frameworks for one - loop computations . in order to simplify the generation of the numerator expansions and the corresponding source code needed by ninja as input , we distribute with the library a small python package called ninjanumgen which uses form-4 @xcite in order to compute the expansions and produce an optimized source code which can be used by the library . in this paper , besides the description of the implementation of the algorithm and the usage of the library
, we also propose a method for the extension of the integrand reduction via laurent expansion for the computation of integrals whose rank is one unit higher than the number of loop denominators .
this method is not present elsewhere in the literature and has been implemented in ninja , allowing thus to use the library for computations in non - renormalizable and effective theories as well .
the paper is organized as follows . in section [ sec : laurent ] we give a review of the laurent expansion method for the integrand reduction of one - loop amplitudes . in section [ sec : ninja ] we discuss its semi - numerical implementation in the library ninja .
the usage of the library is explained , with the help of simple examples , in section [ sec : usage ] . in section
[ sec : examples ] we give a description of the examples which are distributed with the library , giving a better view of its usage and capabilities . appendix [ sec : ninjanumgen ] gives more details on the usage of the package ninjanumgen . in appendix
[ sec : hr ] we present the extension of the algorithm to higher - rank integrals .
appendix [ sec : intlib ] gives more information on the interface between ninja and the libraries of master integrals .
since this paper is rather technical , the reader which is mostly interested in the usage of the library might want to read sections [ sec : usage ] and [ sec : examples ] first , referring to the previous sections or the appendices at a later time if needed .
in this section we review the _ integrand reduction via laurent - expansion method _ @xcite for the computation of one - loop integrals .
the method is based on the systematic application of the laurent series expansion on the universal integrand decomposition of one - loop amplitudes , which allows to reduce any amplitude as a linear combination of known master integrals .
a generic contribution @xmath0 to an @xmath1-point one - loop amplitude in dimensional regularization has the form @xmath2 in the previous equation , the integrand @xmath3 is a _ rational function _ of the components of the @xmath4-dimensional loop momentum @xmath5 , with @xmath6 .
the numerator @xmath7 is a process - dependent polynomial in @xmath5 , while the denominators @xmath8 are quadratic polynomials in @xmath5 and correspond to feynman loop propagators , @xmath9 the function @xmath10 appearing in eq .
is a conventional normalization factor given by @xcite @xmath11 as a function of the renormalization scale @xmath12 and the dimension @xmath13 .
the @xmath4-dimensional loop momentum @xmath5 can be split into a four - dimensional part @xmath14 and a @xmath15-dimensional part @xmath16 as @xmath17 the numerator @xmath18 will therefore be a polynomial in the four components of @xmath14 and the extra - dimensional variable @xmath19 .
every one - loop integrand in dimensional regularization can be decomposed as sum of integrands having five or less loop denominators @xcite @xmath20 where the second sum on the r.h.s
. runs over all the subsets of the denominator indexes @xmath21 containing @xmath22 elements .
the _ residues _
@xmath23 appearing in eq .
are irreducible polynomials , i.e.polynomials which do not contain terms proportional to the corresponding loop denominators @xmath24 .
these residues have a universal , process - independent parametric form in terms of unknown , process - dependent coefficients . for any set of denominators @xmath24 with @xmath25
, one can build a @xmath26-dimensional basis of massless momenta @xmath27 @xcite .
the first two elements of the basis are linear combinations of two external momenta @xmath28 , @xmath29 of the sub - diagram identified by the considered set of loop denominators . more explicitly , we define @xmath30 with @xmath31 where the momenta @xmath32 were defined in eq . .
if the sub - diagram has less than two independent external momenta , the remaining ones are substituted by arbitrary reference vectors in the definition of @xmath33 and @xmath34 .
the momenta @xmath35 and @xmath36 are instead chosen to be orthogonal to the first two and can be defined using the spinor notation as @xmath37}{2 } , \qquad e_4^\mu = \frac{\langle e_2\ , \gamma^\mu\ , e_1]}{2}.\ ] ] they satisfy the property @xmath38 . for subsets of @xmath39 denominators , we also define the vectors @xmath40 and @xmath41 @xmath42 with @xmath43 .
we observe that the vector @xmath41 is orthogonal to all the external legs of the sub - diagram identified by the four denominators . by expanding the four dimensional part @xmath14 of the loop momentum @xmath44 in the basis @xmath45
, the numerator and the denominators can be written as polynomials in the coordinates @xmath46 , @xmath47 with @xmath48 the coordinates @xmath49 can be written as scalar products @xmath50 where @xmath51 .
for @xmath39 we also consider the alternative expansion of the loop momentum @xmath52 with @xmath53 the universal parametric form of the residues @xmath23 in a renormalizable theory is @xcite @xmath54 where we understand that the unknown coefficients @xmath55 depend on the indexes of the residue ( e.g. @xmath56 ) , while the scalar products @xmath49 and @xmath57 depend on both the indexes of the residue and the loop momentum @xmath14 .
the parametrization in eq . can be extended to effective and non - renormalizable theories where the rank of the numerator can be larger than the number of loop propagators @xcite .
more details on the higher - rank case are given in appendix [ sec : hr ] .
most of the terms appearing in eq .
are _ spurious _ , i.e. they vanish after integration and do not contribute to the final result .
the amplitude @xmath0 can thus be expressed as a linear combination of master integrals , corresponding to the non - spurious terms of the integrand decomposition , namely @xmath58 \bigg\ } \pagebreak[1 ] \nn & + \sum_{\{i_1 , i_2 , i_3\}}\bigg\ { c_{0}^ { ( i_1 i_2 i_3 ) } i_{i_1 i_2 i_3 } + c_{7}^ { ( i_1 i_2 i_3 ) } i_{i_1 i_2 i_3}[\mu^2 ] \bigg\ } \pagebreak[1 ] \nn & + \sum_{\{i_1 , i_2\}}\bigg\ { c_{0}^ { ( i_1 i_2 ) } i_{i_1 i_2 } + c_{1}^ { ( i_1 i_2 ) } i_{i_1 i_2}[(q+p_{i_1})\cdot e_2 ] \nn & \qquad + c_{2}^ { ( i_1 i_2 ) } i_{i_1 i_2}[((q+p_{i_1})\cdot e_2)^2 ] + c_{9}^ { ( i_1 i_2 ) } i_{i_1 i_2}[\mu^2 ] \bigg\ } \pagebreak[1 ] \nn & + \sum_{i_1 } c_{0}^ { ( i_1 ) } i_{i_1 } , \label{eq : integraldecomposition}\end{aligned}\ ] ] where @xmath59 \equiv h(\mu_r^2,d)\ , \int d^d \bar q { \alpha \over d_{i_1 } \cdots d_{i_k } } , \qquad i_{i_1 \cdots i_k } \equiv i_{i_1 \cdots i_k}[1].\ ] ] the coefficients of this linear combination can be identified with a subset of the coefficients of the parametric residues in eq . .
since all the master integrals appearing in eq . are known , the problem of the computation of an arbitrary one - loop amplitude can be reduced to the problem of the determination of the non - spurious coefficients appearing in the parametrization of the residues @xmath23 .
the coefficients appearing in the integrand decomposition can be computed by evaluating the integrand on _ multiple cuts _ ,
i.e. on values of the loop momentum @xmath5 such that a subset of loop denominators vanish @xcite .
more in detail , the coefficients of a @xmath22-point residue @xmath23 can be determined by evaluating the integrand on the corresponding @xmath22-ple cut @xmath60 .
for these values of the loop momentum , the only non - vanishing contributions of the integrand decomposition are the ones coming from the residue in consideration and from all the higher - point residues which have @xmath61 as a subset of their loop denominators . within the original integrand reduction algorithm @xcite ,
the coefficients are computed by sampling the numerator of the integrand on a finite subset of the on - shell solutions , subtracting all the non - vanishing contributions coming from higher - point residues , and finally solving the resulting linear system of equations .
this is therefore a top - down approach , where higher - point residues are computed first , starting from @xmath62 , and systematically subtracted from the integrand for the evaluation of lower - point contributions .
these are referred to as subtractions at the integrand level .
the integrand reduction via laurent expansion method , presented in ref .
@xcite , improves this reduction strategy by elaborating on techniques proposed in @xcite .
whenever the analytic dependence of the integrand on the loop momentum is known , this approach allows to compute the coefficients of a residue @xmath23 by performing a laurent expansion of the integrand with respect to one of the components of the loop momentum which are not fixed by the on - shell conditions of the corresponding multiple cut @xmath63 . in the asymptotic limit defined by this laurent expansion ,
both the integrand and the higher - point subtractions exhibit the same polynomial behavior as the residue
. therefore one can directly identify the unknown coefficients with the ones of the laurent expansion of the integrand , corrected by the contributions coming from higher - point residues .
hence , by choosing a suitable laurent expansion , one obtains a diagonal system of equations for the coefficients of the residues , while the subtractions of higher - point contributions can be implemented as _
corrections at the coefficient level _ which replace the subtractions at the integrand level of the original algorithm .
since the polynomial structure of the residues is universal and does not depend on the process , the parametric form of the coefficient - level corrections can be computed once and for all , in terms of a subset of the higher - point coefficients .
more in detail , the corrections at the coefficient level are known functions of a subset of the coefficients of 3- and 2-point residues . in particular , no subtraction term coming from 4- and 5-point contributions is ever needed .
this allows to skip the computation of the ( spurious ) 5-point contributions entirely , and to completely disentangle the determination of 4-point residues from the one of lower point contributions . in the following ,
we address more in detail the computation of 5- , 4- , 3- , 2- , and 1-point residues , also commonly known as _ pentagons _ , _ boxes _ , _ triangles _ , _ bubbles _ and _ tadpoles _ respectively . for simplicity , we focus on renormalizable theories , where ( up to a suitable choice of gauge ) the maximum allowed rank of the integrand is equal to the number of loop denominators and the most general parametrization of the residues is the one given in eq . .
ninja can also be used for the computation of integrals whose rank exceeds the number of denominators by one .
the extension of the method to the higher - rank case is discussed in appendix [ sec : hr ] .
[ [ point - residues ] ] 5-point residues + + + + + + + + + + + + + + + + as mentioned above , pentagon contributions are spurious . within the original integrand reduction algorithm ,
their computation is needed because they appear in the subtractions at the integrand level required for the evaluation of lower - point contributions .
a 5-point residue only has one coefficient , which can easily be computed by evaluating the numerator of the integrand on the corresponding 5-ple cut . within the laurent expansion approach ,
the subtraction terms coming from five - point residues always vanish in the asymptotic limits we consider , therefore their computation can be skipped .
for this reason , in the library ninja the computation of pentagons is disabled by default , even though it can be enabled for debugging purposes , as explained in section [ sec : runtime ] .
[ [ point - residues-1 ] ] 4-point residues + + + + + + + + + + + + + + + + the coefficient @xmath64 of a box contribution @xmath65 can be determined via four - dimensional quadruple cuts @xcite .
a quadruple cut @xmath66 in four dimensions ( i.e. @xmath67 , @xmath68 ) has two solutions , @xmath69 and @xmath70 .
the coefficient @xmath64 can be expressed in terms of these solutions as @xmath71 given the simplicity of eq .
, this is the only coefficient which ninja computes in the same way as the traditional algorithm .
the coefficient @xmath72 can instead be determined by evaluating the integrand on @xmath4-dimensional quadruple cuts in the asymptotic limit of large @xmath19 @xcite .
a @xmath4-dimensional quadruple cut has an infinite number of solutions which can be parametrized by the ( @xmath73)-dimensional variable @xmath19 .
these solutions become simpler in the considered limit , namely @xmath74 where the vector @xmath75 and the constant @xmath76 are fixed by the cut conditions .
the coefficient @xmath72 is non - vanishing only if the rank of the numerator is greater or equal to the number of loop denominators . in a renormalizable theory
, it can be found in the @xmath77 asymptotic limit as the leading term of the laurent expansion of the integrand @xmath78 the other coefficients of the boxes are spurious and , since they neither contribute to the final result nor to the subtraction terms , their computation can be skipped .
[ [ point - residues-2 ] ] 3-point residues + + + + + + + + + + + + + + + + the coefficients of the residues of a generic triangle contribution @xmath79 can be determined by evaluating the integrand on the solutions of the corresponding @xmath4-dimensional triple cut @xcite .
these can be parametrized by the variable @xmath19 and the free parameter @xmath80 , @xmath81 where the vector @xmath75 and the constant @xmath76 are fixed by the cut conditions @xmath82 .
the momentum @xmath75 is a linear combination of @xmath33 and @xmath34 and is therefore orthogonal to @xmath35 and @xmath36 . on these solutions ,
the non - vanishing contributions to the integrand decomposition are the ones of the residue @xmath79 , as well as the ones of the boxes and pentagons which share the three cut denominators . however , after performing a laurent expansion for large @xmath80 and dropping the terms which vanish in this limit , the pentagon contributions vanish , while the box contributions are constant in @xmath80 but they also vanish when taking the average between the parametrizations @xmath69 and @xmath70 of eq . .
more explicitly , @xmath83 moreover , the expansion of the integrand is given by @xmath84 and it has the same polynomial behavior as the expansion of the residue @xmath79 , @xmath85 by comparison of eq . , and one can directly identify the ten triangle coefficients as the corresponding terms of the expansion of the integrand , @xmath86 hence , with the laurent expansion method , the determination of the 3-point residues does not require any subtraction of higher - point terms .
[ [ point - residues-3 ] ] 2-point residues + + + + + + + + + + + + + + + + the coefficients of a generic 2-point residue @xmath87 can be evaluated on the on - shell solutions of the corresponding double cut @xmath88 , which can be parametrized as @xmath89 in terms of the three free parameters @xmath90 , @xmath80 and @xmath19 , while the constants @xmath91 and @xmath92 are fixed by the on - shell conditions . after evaluating the integrand on these solutions and performing a laurent expansion for @xmath93 , the only non - vanishing subtraction terms come from the triangles , @xmath94 even though the integrand and the subtraction terms are rational functions , in the asymptotic limit they both have the same polynomial behavior as the residue , namely @xmath95 \\ \frac{\delta_{i_1 i_2 j}(q_+,\mu^2)}{d_j } = { } & c_{s_3,0}^{(j)}+ c_{s_3,9}^{(j)}\ , { \mu^2}+ c_{s_3,1}^{(j)}\ , { x } + c_{s_3,2}^{(j)}\ , { x^2 } - \big ( c_{s_3,5}^{(j ) } + c_{s_3,8}^{(j ) } { x}\big ) { t } + c_{s_3,6}^{(j)}\ , { t^2}+\o(1/{t } ) \nn \frac{\delta_{i_1 i_2 j}(q_-,\mu^2)}{d_j } = { } & c_{s_3,0}^{(j)}+ c_{s_3,9}^{(j)}\ , { \mu^2}+ c_{s_3,1}^{(j)}\ , { x } + c_{s_3,2}^{(j)}\ , { x^2 } - \big ( c_{s_3,3}^{(j ) } + c_{s_3,7}^{(j ) } { x}\big ) { t } + c_{s_3,4}^{(j)}\ , { t^2}+\o(1/{t } ) \label{eq : bubblessubexp } \pagebreak[1 ] \\ \delta_{i_1
i_2}(q_+,\mu^2 ) = { } & c_0+c_9\ , { \mu^2}+c_1\ , ( e_1\cdot e_2)\ , { x } + c_2 \ , ( e_1\cdot e_2)^2\ , { x^2}\nn & + \big(c_5 + c_8 \ , ( e_1\cdot e_2)\ , { x}\big)\ , ( e_3\cdot e_4)\ , { t } + c_6\ , ( e_3\cdot e_4)^2\ , { t^2}+\o(1/{t } ) \nn \delta_{i_1 i_2}(q_-,\mu^2 ) = { } & c_0+c_9\ , { \mu^2}+c_1\ , ( e_1\cdot e_2)\ , { x } + c_2\ , ( e_1\cdot e_2)^2\ , { x^2 } \nn & + \big(c_3 + c_7 \ , ( e_1\cdot e_2)\ , { x}\big)\ , ( e_3\cdot e_4)\ , { t } + c_4\ , ( e_3\cdot e_4)^2\ , { t^2}+\o(1/{t } ) .
\label{eq : bubbledeltaexp}\end{aligned}\ ] ] the coefficients @xmath96 of the expansion of the subtractions terms in eq.s are known parametric functions of the triangle coefficients .
hence , the subtraction of the triangle contributions can be implemented by applying coefficient - level corrections to the terms appearing in the expansion of the integrand .
more explicitly , by inserting eq.s , and in eq .
one gets @xmath97 [ [ point - residues-4 ] ] 1-point residues + + + + + + + + + + + + + + + + the only non - spurious coefficient @xmath64 of a tadpole residue @xmath98 can be computed by evaluating the integrand on solutions of the single cut @xmath99 . for this purpose
, one can consider 4-dimensional solutions of the form @xmath100 parametrized by the free variable @xmath80 . in the asymptotic limit @xmath101 ,
only bubble and triangle subtraction terms are non - vanishing , @xmath102 similarly to the case of the 2-point residues , in this limit the integrand and the subtraction terms exhibit the same polynomial behavior as the residue , i.e.@xmath103 putting everything together , the coefficient of the tadpole integral can be identified with the corresponding one in the expansion of the integrand , corrected by coefficient - level subtractions from bubbles and triangles @xmath104 the subtraction terms @xmath105 and @xmath106 , coming from 2-point and 3-point contributions respectively , are known parametric functions of the coefficients of the corresponding higher - point residues .
the c++ library ninja provides a semi - numerical implementation of the laurent expansion method described in section [ sec : laurent ] .
the laurent series expansion is typically an analytic operation , but since a one - loop integrand is a rational function of the loop variables , its expansion can be obtained via a partial fraction decomposition between the numerator and the denominators .
this is implemented in ninja via a simplified polynomial - division algorithm , which takes as input the coefficients of a parametric expansion of the numerator @xmath18 and computes the leading terms of the quotient of the polynomial division with respect to the uncut denominators . in this section
we describe the input needed for the reduction performed by ninja and we give further details about the implementation of the reduction .
all the types , classes and functions provided by the ninja library are defined inside the ` ninja ` namespace . in particular , the types ` real ` and ` complex ` are aliases for ` double ` and ` std::complex < double > ` , unless the library was compiled in quadruple precision .
classes for real and complex momenta are defined as ` realmomentum ` and ` complexmomentum ` respectively .
they are wrappers of four - dimensional arrays of the corresponding floating - point types , which overload arithmetic and subscript operators .
more in detail , an instance ` p ` of one of these classes represents a momentum @xmath107 according to the representation @xmath108},\texttt{p[1]},\texttt{p[2]},\texttt{p[3]}\ } = \{e_p , x_p , y_p , z_p\},\ ] ] i.e. with the energy in the zeroth component , followed by the spatial components .
the inputs needed from the reduction algorithm implemented in ninja are the momenta @xmath32 and the masses @xmath109 of the loop denominators defined in eq . , besides the numerator @xmath110 of the integrand .
the latter must be cast in four different forms , one of which is optional .
the c++ implementation requires the numerator to be an instance of a class inherited from the abstract class ` ninja::numerator ` .
the latter defined as .... class numerator { public : virtual complex evaluate(const ninja::complexmomentum & $ \(q\)$ , const ninja::complex & $ \(\mu^{2 } \)$ , int cut , const ninja::partitionint partition [ ] ) = 0 ; virtual void muexpansion(const ninja::complexmomentum $ \(v\)$ [ ] , const ninja::partitionint partition [ ] , ninja::complex c [ ] ) { } virtual void t3expansion(const ninja::complexmomentum & $ \(v_{0}\)$ , const ninja::complexmomentum & $ \(v_{3}\)$ , const ninja::complexmomentum & $ \(v_{4}\)$ , const ninja::complex & $ \(\beta\)$ ,
int mindeg , int cut , const ninja::partitionint partition [ ] , ninja::complex c [ ] ) = 0 ; virtual void t2expansion(const ninja::complexmomentum & $ \(v_{1}\)$ , const ninja::complexmomentum & $ \(v_{2}\)$ , const ninja::complexmomentum & $ \(v_{3}\)$ , const ninja::complexmomentum & $ \(v_{4}\)$ , const ninja::complex $ \(\beta\)$ [ ] , int mindeg , int cut , const ninja::partitionint partition [ ] , ninja::complex c [ ] ) = 0 ; virtual ~numerator ( ) { } } ; .... the input parameters ` cut ` and ` partition ` are common to more methods and give information about the multiple cut where ninja is currently evaluating the numerator .
although this information is not always necessary , there might be occasions where it could be useful for an efficient evaluation of the numerator .
the integer ` cut ` is equal to @xmath22 if the numerator is being evaluated on a @xmath22-ple cut , with @xmath111 .
this parameter is not given in the method ` muexpansion ` because the latter is always evaluated on quadruple cuts ( i.e. @xmath112 ) .
the parameter ` partition ` points to an array of integers ( namely of integer type ` ninja::partitionint ` ) , with length equal to ` cut ` , containing the indexes of the cut numerators .
if the user asks to perform a global test ( see section [ sec : runtime ] ) , the numerator will also be evaluated outside the solutions of the multiple cuts , in which case the parameter ` cut ` will be set to zero . as an example , if the method ` t3expansion ` is evaluated on the 3-ple cut @xmath113 for the determination of the 3-point residue @xmath114 , we will have @xmath115 , @xmath116}=0 $ ] , @xmath117}=2 $ ] , and @xmath118}=5 $ ] .
the returned expansion only needs to be valid for the cut specified ( e.g. @xmath113 in the previous example ) .
here is a detailed description of each method of the class ` ninja::numerator ` , for a generic numerator of an @xmath1-point integrand of rank @xmath119 .
if the analytic expression of the integrand is available , all these methods can be easily generated with the help of the simple python package ninjanumgen , which is distributed with the library and whose usage is described in section [ sec : integrand ] and in appendix [ sec : ninjanumgen ] . the details which follow
are only needed to those who prefer to provide an alternative implementation of the required methods without ninjanumgen .
it must return the value of the numerator @xmath110 evaluated at the ( complex ) values of @xmath14 and @xmath19 given as input .
this is used for the computation of the laurent expansion in @xmath19 required to obtain the coefficient @xmath72 of the boxes . in the renormalizable case
, this method should compute the leading term of a parametric expansion in @xmath80 of the integrand defined by @xmath120 where @xmath41 is given by @xmath40`[0 ] ` , i.e. by the zeroth entry of the array of momenta @xmath40 .
for renormalizable theories this array will therefore only contain at most one element .
the method should write the leading coefficient of the expansion in the zeroth entry of the array pointed by the parameter ` c ` , i.e. .... c[0 ] = $ \(\n\)$[$\(t^{r}\)$ ] ; .... the generalization of this method to the higher - rank case is described in appendix [ sec : hr ] .
the implementation of this method is only required when @xmath121 .
it is also not needed if the user chooses to disable the @xmath19-expansion method for the boxes , but in that case more evaluations of the numerator will be needed and the computation of the pentagons will not be skipped . this method is used for the computation of the coefficients of the residues of both the triangles and the tadpoles .
it must compute the coefficients of the terms @xmath122 for @xmath123 , given by substituting into the numerator the parametric expansion of the loop momentum defined by @xmath124 as a function of the momenta @xmath125 and the constant @xmath76 which are passed as parameters .
the maximum value of the parameter ` mindeg ` is @xmath126 .
since in a renormalizable theory @xmath127 , and by definition of rank we have @xmath128 , in this case at most 6 terms can be non - vanishing in the specified range of @xmath129 .
the method should write these terms in the entries of the array pointed by ` c ` , ordered by decreasing powers of @xmath80 .
terms with the same power of @xmath80 should be ordered by increasing powers of @xmath19 .
a pseudo - implementation will therefore look like .... int idx = 0 ; for ( int j=$\(r\)$ ; j>=$\(r\)$-mindeg ; --j ) for ( int k=0 ; 2*k<=$\(r\)$-j ; + + k ) c[idx++ ] = $ \(\n\)$[$\(t^{j}\mu^{2k}\)$ ] ; .... in the current version of ninja , this method is called during the computation of the coefficients of the bubbles .
it must compute the coefficients of the terms @xmath130 for @xmath131 , given by the expansion @xmath132 as a function of the momenta @xmath125 and the constants @xmath133}$ ] , which are passed as parameters to the method .
the maximum value of ` mindeg ` is @xmath134 . in a renormalizable theory
, this implies that one can have at most 7 non - vanishing terms in this range of @xmath129 .
it is worth observing that the expansion in eq .
can be obtained from the previous one in eq . with the substitutions @xmath135 the terms of the expansion
must be stored in the entries of the array pointed by ` c ` , ordered by decreasing powers of @xmath80 .
terms with the same power of @xmath80 should be ordered from the lowest to the highest with respect to the lexicographical order in the variables @xmath136 .
a pseudo - implementation will have the form .... int idx = 0 ; for ( int j=$\(r\)$ ; j>=$\(r\)$-mindeg ; --j ) for ( int l=0 ; l<=$\(r\)$-j ; + + l ) for ( int k=0 ; 2*k<=$\(r\)$-j - l ; + + k ) c[idx++ ] = $ \(\n\)$[$\(t^{j}x^{l}\mu^{2k}\)$ ] ; .... for every phase - space point , ninja at run - time computes the parametric solutions of the multiple cuts corresponding to each residue .
the laurent expansion of the integrand on these solutions is performed via a simplified polynomial division between the expansion of the numerator and the set of the uncut denominators .
the coefficients of this expansion are corrected by the coefficient - level subtractions appearing in eq . and .
the non - spurious coefficients are finally multiplied by the corresponding master integrals in order to obtain the integrated result as in eq . .
the coefficients of the expansions of the numerator are written on a contiguous array by the numerator methods described in section [ sec : input ] .
the laurent expansion is obtained via a simplified polynomial division .
the latter is performed in - place on the same array , keeping only the elements which are needed for the final result . a possible implementation for an univariate expansion , with a numerator @xmath137}\ ,
t^r + \texttt{num[1]}\ , t^{r-1 } + \ldots + \texttt{num[nterms-1]}\ , t^{r-\texttt{nterms}+1 } + \o(t^{r-\texttt{nterms}})\ ] ] and denominator @xmath138}\ , t + \texttt{d[1]}\ , + \texttt{d[2]}\ , \frac{1}{t},\ ] ] would have the form .... void division(complex num [ ] , int nterms , complex den[3 ] ) { for ( int i=0 ; i < nterms ; + + i ) { num[i ] /= den[0 ] ; if ( i+1<nterms ) { num[i+1 ] -= den[1]*num[i ] ; if ( i+2<nterms ) num[i+2 ] -= den[2]*num[i ] ; } } } .... one can check that this routine correctly replaces the first ` nterms ` elements of the array ` num ` with the first ` nterms ` leading elements of the laurent expansion of @xmath139 . the actual implementation in ninja , having to deal with multivariate expansions , is significantly more involved than the ` division ` procedure presented here .
nevertheless , it qualitatively follows the same algorithm .
the coefficients obtained by the division are then corrected by the coefficient - level subtractions and thus identified with the corresponding coefficients of the residues , as explained in section [ sec : laurent ] .
once the reduction is complete , the non - spurious coefficients are multiplied by the corresponding master integrals .
ninja calls the routines implementing the master integrals through a generic interface which , as in the case of the numerator , is defined in the c++ code via an abstract class ( called ` integrallibrary ` ) .
this allows one to use any integral library which can be specified at run - time .
more details on the implementation of this interface are given in appendix [ sec : intlib ] .
the current version of ninja already implements it for two integral libraries . the first built - in interface , `
ninja::avhoneloop ` , is a c++ wrapper of the routines of the oneloop library @xcite .
this wrapper caches every computed integral allowing constant - time lookup of their values from their arguments .
the caching of the integrals can significantly speed up the computation , especially for complex processes .
every instance of the class ` avhoneloop ` has an independent cache of master integrals ( hence , one may safely use it in multi - threaded applications by using one instance of the class per thread ) .
the second implementation of the interface , ` ninja::looptools ` , uses instead the looptools library @xcite , which already has an internal cache of computed integrals .
in this section we explain how to use the library for the computation of a generic one - loop integral .
ninja can be interfaced to any one - loop generator capable of providing the input it needs , and in particular to packages which can reconstruct the analytic dependence of the numerators on the loop momentum .
an interface for the one - loop package gosam @xcite is already built in the library , and has been used for highly non - trivial phenomenological computations @xcite .
an interface with the package formcalc @xcite is currently under development .
the author is open to give his assistance in interfacing other packages as well . in this paper
we focus on the usage of ninja as a standalone library .
we will show how to generate the numerator methods needed as input , starting from an analytic expression of the numerator , with the help of the python package ninjanumgen which is distributed with the library .
we will then explain how to perform the reduction , and how to set the main available options .
ninja can be obtained at the url ` http://ninja.hepforge.org ` .
the library is distributed with its source code using the gnu build system ( also known as autotools ) .
it can be compiled and installed with the shell commands ....
./configure make make install ....
this will typically install the library and the header files in sub - directories of ` /usr / local ` .
the ` prefix ` option can be used in order to specify a different installation path . in this case
, you might need to update your ` ld_library_path ` ( or ` dyld_library_path ` on mac os ) environment variable accordingly . in order to use ninja for
the production of phenomenological results , one must interface it with a library of master integrals .
as already mentioned , interfaces to the oneloop and looptools libraries are provided ( see appendix [ sec : intlib ] for interfacing a different library ) .
they can be enabled by passing to the ` configure ` script the options ` with - avholo[=flags ] ` and ` with - looptools[=flags ] ` .
for instance , the following commands [ source , bash ] ---- ./configure
--prefix=$\$$home / ninja \
--with - avholo='-l / path / to / avh_olo / lib -lavh_olo ' \
fcinclude =- i / path / to / avh_olo / include make install ---- will install all the files in sub - directories of ` $ home / ninja ` and build the interface with the oneloop library , which must be already installed and linkable with the flags specified with the ` with - avholo ` option .
we also specified the ` fcinclude ` variable with the flags which are needed to find fortran-90 modules when they are not installed in a default directory .
given the importance of numerical precision in the calculation of scattering amplitudes , there is also the option to compile the library in quadruple precision ( ` with - quadruple ` ) , which uses the gcc libquadmath library . by using the ` ninja::real ` and ` ninja::complex ` floating point types , the same source code can be compiled both in double and quadruple precision , depending on how ninja has been configured .
a full list of optional arguments for the ` configure ` script can be obtained with the command ` ./configure
help ` . while most of them are common to every package distributed with the gnu build system , some are instead specific to the ninja library and they are described in table [ tab : configure ] . in
most of the cases , only the options for interfacing the integral libraries should be needed .
.options and environment flags for the ` configure ` script . only the options which modify the default behavior of ninja are listed .
[ cols="<,<",options="header " , ] in this example we consider six incoming photons @xcite .
this is a non - trivial case where the ` setcutstop ` method of an ` amplitude ` class can make the computation more efficient when lower point cuts do not contribute to the total result .
a generic six - photon diagram has an integrand of the form @xmath140 where we assumed the fermion running in the loop to be massless .
the momenta @xmath141 are defined by @xmath142 one can work out the algebra , define the corresponding spinor products and vectors , and generate the input for ninja in the same way as for the four - photons case .
one can also check that the terms proportional to @xmath19 in the final expression for the integral vanish upon integration .
therefore , we can perform the simplification @xmath143 , or equivalently @xmath144 , in the numerator .
moreover , one can exploit the knowledge that only the cut - constructible contributions of boxes and triangles contribute to the total result , hence we can ask ninja to stop the reduction at triple cuts with ....
amp.setcutstop(3 ) ; .... and remove the rational part from the result with .... amp.onlycutconstructible ( ) ; .... which will make the computation more efficient ( in the example implemented here , the run - time is reduced by about 33% ) . in the file ` 6photons.cc ` we call the method ` evaluate ` on all the independent permutations of the external legs , generated at run - time with the function ` std::next_permutation ` of the c++ standard library .
the results have been compared with the ones in ref.s @xcite as well as with a similar computation performed with samurai for several helicity choices . with this example
, we discuss a possible strategy for the generation of the input needed by ninja which can be suited for more complex computations where an efficient evaluation of the numerator methods at run - time can be important .
we consider the one - loop integral defined by the diagram depicted in figure [ fig : ttbarh ] , contributing to the 5-point helicity amplitude @xmath146 .
[ sec : ttbarh ] .
this picture has been generated using gosam.,scaledwidth=45.0% ] the analytic expression for the integrand of this example , which can be worked out from the feynman rules of the standard model , has been generated with the help of the package gosam and can be found in the form file ` ttbarh.frm ` .
this already contains some abbreviations which are independent from the loop momentum @xmath44 of the diagram . at run - time , these @xmath5-independent abbreviations are computed only once per phase space point , making thus the evaluation of the numerator and its expansions more efficient .
this analytic expression is processed by ninjanumgen which produces the numerator expansions .
we also add to the numerator class ` ttbarhdiagram ` an ` init ` method which uses the spinor library described in section [ sec:4photons ] in order to compute the relevant spinor products and polarization vectors , as well as the abbreviations which do not depend on the loop momentum .
these are stored as private data members of the class . for simplicity ,
our result neglects the coupling constants and an overall color factor .
even though we considered a single diagram and a specific helicity choice , this example illustrates a general strategy for the generation of an analytic numerator expression which is suited for the numerical evaluations performed by integrand - reduction algorithms like the one implemented in the library ninja .
the full amplitude for this process has been computed in ref.s @xcite , while an additional jet has recently been added to the final state in ref.s @xcite where ninja has been used for the reduction of the corresponding integrands generated by gosam . in this example
we show how ninja can be used in order to compute integrals whose rank is higher than the number of loop denominators .
this simple test is similar to the example presented in ref .
[ sec : usage ] , hence we will describe each step as in the previous case . we define a 5-point amplitude of rank 6 , with kinematics @xmath147 and integrand @xmath148 in terms of the reference vectors @xmath149 ( @xmath150 ) and the momenta @xmath141 running into the loop @xmath151 we follow the same steps outlined in section [ sec : usage ] . with ninjanumgen
we generate the methods for ninja . after writing the integrand in the form file ` mynumhr.frm ` we call the script with the command .... ninjanumgen mynumhr.frm --nlegs 5 --rank 6 -o mynumhr.cc .... which creates the file ` mynumhr.cc ` and a template for the header ` mynumhr.hh ` .
once again , we define the vectors @xmath149 as public members of the numerator class ` diagram ` , by inserting .... public : ninja::complexmomentum v0,v1,v2,v3,v4,v5 ; .... in the class definition .
a possible test program can be almost identical to the one we showed in section [ sec : dummyex4 ] , with obvious changes in the definition of the rank , the number of external legs and the reference vectors @xmath149 .
this is implemented in the file ` simple_higher_rank_test.cc ` . in order to run this example ,
the user must compile the library with the ` enable - higher_rank ` option , otherwise a call to the ` evaluate ` method of an ` amplitude ` object will cause a run - time error .
as one can see , when ninjanumgen is used for the generation of the expansions , the higher - rank case is handled automatically without any intervention by the user .
besides , the internal higher - rank routines of ninja will be automatically called whenever the rank @xmath119 is equal to @xmath152 ( where @xmath1 is the number of loop propagators ) , while in the public programming interface there is no difference with respect to the normal - rank case . in the last examples , we wish to illustrate the possibility of using ninja in a multi - threaded application .
these examples are implemented using posix threads , which are a standard in unix - like operating systems , but adapting them to different programming interfaces for threads ( such as openmp ) should be straightforward . in order to implement a thread - safe application , one should avoid race conditions which might occur if different threads try to write on the same variables .
in particular , one should avoid accessing global variables for writing from different threads .
the only global variables used directly by ninja are those controlling the global options described in section [ sec : globaloptions ] .
as explained in that section , these options are only meant to change the general behavior of the library for debugging purposes ( e.g. for checking that the provided numerator methods are correct ) , while in general the default options should not be changed during a phase - space integration , especially when performance is important . hence ,
on the side of the ninja library , there should be no issue and one can safely call the ` evaluate ` method from different ` amplitude ` objects in different threads . during a call of the ` evaluate ` method on an ` amplitude ` object ,
issues might however arise from global variables used by the chosen library of master integrals or the numerator methods .
as for the numerator methods , all the examples distributed with ninja define a thread - safe numerator class ( more specifically , one can safely call numerator methods from different instances of the class in different threads ) .
this is simply done by using data members of the class instead of global variables , making thus different instances of the same class completely independent .
if the procedures implemented by libraries of master integrals are thread - safe , one can therefore use ninja in multi - threaded applications . as an example
, one can use the class ` avhoneloop ` which , as explained in section [ sec : mis ] , wraps routines of the oneloop library and adds a cache of computed integrals .
the cache is a non - static data member of the class .
one can therefore create one instance of this class per thread and assign it accordingly to the ` amplitude ` objects to be evaluated in the same thread .
as an example , with ....
avh_olo.init(1 ) ; avhoneloop my_lib[n_threads ] ; amplitude < realmasses > amp[n_threads ] ; for ( int i=0 ; i < n_threads ; + + i ) amp[i].setintegrallibrary(my_lib[i ] ) ; .... we create ` n_threads ` amplitude objects whose ` evaluate ` method can be safely called in a separate thread ( in the first line , we called the ` init ` method on the global instance ` avh_olo ` defined in the library , in order to allow oneloop to perform its global initialization ) . in this way
, different threads will also have an independent cache of master integrals .
this strategy allows to build a multi - threaded application which uses ninja for the reduction of one - loop integrals .
recent versions of looptools ( namely looptools-2.10 or later ) can also be used in threaded applications , since they have a mutex regulating writing access to the internal cache of integrals . in
the following we discuss the possibility to build a multi - threaded application with ninja and any other ( not necessarily thread - safe ) library of master integrals .
indeed , even though ninja has obviously no control over possible issues arising from routines of external libraries , we offer an easy way to work around any potential problem . in this case
, there is no general way to ensure that calling routines of the same integral library from different threads will not cause conflicts .
however , one can avoid these conflicts by scheduling the calls of the external procedures in such a way that they are never evaluated at the same time from two or more threads .
if the computation of the master integrals takes only a small fraction of the total run time ( which is usually the case when a cache of integrals is present ) , the effects of this on the performance will in general be reasonably small . within ninja , implementing a scheduled access on the routines used by a library of master integrals
is straightforward . as explained more in detail in appendix
[ sec : intlib ] , the generic interface used by ninja in order to call master integral procedures , has two methods called ` init ` and ` exit ` which are evaluated exactly once in each call of the ` evaluate ` method , immediately before the computation of the first master integral and after the computation of the last master integral respectively .
therefore we can use mutexes ( such as the ones present the posix standard for threads ) in order to _ lock _ the calls to the master integrals in the ` init ` method and _ unlock _ them in the ` exit ` method .
this makes sure that , between the calls of the ` init ` and ` exit ` methods , no other thread will use the master integral routines , hence avoiding any possible conflict . in order to make a library of master integrals thread - safe
, we use the template class ` threadsafeintegrallibrary ` , which is included in the distribution .
this automatically wraps an existing class derived from ` integrallibrary ` and adds to it a mutex that schedules the calls to the master integrals as explained above . as an example , defining a thread - safe version of a generic library ` baselibrary ` can be simply achieved with .... # include < ninja / thread_safe_integral_library.hh > using namespace ninja ; threadsafeintegrallibrary < baselibrary > my_lib ; .... which defines a new interface ` my_lib ` that can be made the default by calling .... setdefaultintegrallibrary(my_lib ) ; .... before any thread is created ( alternatively , we could call the ` setintegrallibrary ` method on each ` amplitude ` object , either outside or inside the threads ) . in the files ` thread_4photons.cc ` and ` thread_6photons.cc ` we repeat the examples of the four- and six - photons amplitudes , but this time we compute several phase - space points in parallel on different threads . as mentioned before
, we do not need to implement other numerator classes , since the ones described in sections [ sec:4photons ] and [ sec : sixphotons ] can be safely used in multi - threaded applications . in the source files ,
we implement both the approaches described in this section
. the preprocessor will select the former if the oneloop interface has been enabled and the latter otherwise .
the multi - threaded examples can be compiled with .... make thread - examples .... if at least one between the oneloop and looptools libraries was enabled during configuration and your system supports posix threads .
a complete discussion on the implementation of multi - threaded applications for doing phenomenology at one - loop is beyond the purposes of this paper .
moreover , a detailed assessment of possible advantages of this approach would generally depend on the generator of the integrands and the phase space integration . in these examples we showed that the methods implementing the reduction via laurent expansion in ninja can be safely used in multi - threaded programs .
we presented the public c++ library ninja which implements the integrand reduction via laurent expansion method for the computation of one - loop amplitudes in quantum field theories .
the main procedures of the library take as input the numerator of the integrand and some parametric expansions of the same , which can be generated with the help of the simple python package ninjanumgen included with the distribution .
the expansions of the integrand on the multiple cuts are computed semi - numerically at run - time , via a simplified polynomial - division algorithm .
some of the coefficients of the laurent expansions are thus identified with the ones which multiply the master integrals .
the algorithm is light and proved to have good performance and numerical stability , hence it is suited for applications to complex one - loop processes , characterized by either several external legs or several mass scales .
we described the usage of the library , in particular the generation of the input , the calls of the procedures for the reduction , and the interface to libraries of master integrals .
this information can be used in order to interface the library with existing one - loop packages .
we thus expect that ninja will be useful for future computations in high - energy physics , especially for those involving more complex processes .
the author thanks all the other members of the gosam collaboration for the common development of a one - loop package which could be interfaced with ninja , and especially pierpaolo mastrolia , edoardo mirabella and giovanni ossola for innumerable discussions and exchanges .
the author also thanks thomas hahn for his support with looptools and comments on the draft .
this work was supported by the alexander von humboldt foundation , in the framework of the sofja kovalevskaja award project `` advanced mathematical methods for particle physics '' , endowed by the german federal ministry of education and research .
the reduction procedures implemented in ninja take as input a class derived from the abstract class ` ninja::numerator ` .
this must implement the required expansions in the corresponding methods .
if the analytic expression of the numerator can be provided by the user , the source code for the methods can be automatically generated with the help of the simple python package ninjanumgen , which is distributed with the library and can be installed as explained in section [ sec : installation ] . the package can be used both as a script or as a module within python . in section [ sec : integrand ] we already gave a simple example of its usage as a script .
as explained there , the user must create a file containing a form expression of the numerator of the integrand .
the package uses form-4 in order to generate the expansions which are needed and produce a c++ source file with the definitions of the corresponding methods . if not already present , an header file with a sketch of the definition of the class will also be created
. the user can complete it by adding data members and methods which are specific of this class .
form allows one to define symbols between square brackets ( e.g.`[symbol_name ] ` ) , containing characters which otherwise would not be permitted in a declaration .
ninjanumgen also allows the usage of such symbols in the expression of the numerator , and it will remove the brackets ( which would produce illegal c++ code ) when writing the final source files .
this gives the user a wider range of possibilities , for instance using symbols which correspond to variable names containing underscores or data members of structures ( e.g. with ` [ structure_instance.data_member ] ` ) .
we first give a few more details about the usage of the package as a script .
it is invoked with the command .... ninjanumgen --nlegs nlegs $ { \it optional - arguments}$ file .... where ` file ` is the name of the file which contains the numerator expression and ` nlegs ` is the number of external legs of the loop , which is equal to the number of loop denominators .
a description of all the allowed arguments can be obtained with the command .... ninjanumgen --help .... and the most important ones are : ` rank rank ` , ` -r rank ` : : rank of the numerator , by default it will be assumed to be equal to the number of external legs of the loop ` diagname diagname ` , ` -d diagname ` : : name of the numerator expression in the form file , by default it will be assumed to be ` diagram ` ` cdiagname cdiagname ` : : name of the numerator class in the generated c++ files , by default it will be the same as the form expression ` formexec formexec ` : : the form executable , the default is ` form ` ` qvar qvar ` : : name of the loop momentum variable @xmath14 defined in eq
. , the default is ` q ` ` mu2var mu2var ` : : name of the loop variable @xmath19 defined in eq
. , the default is ` mu2 ` ` output output ` , ` -o output ` : : name of the output source file , the default is ` ninjanumgen.cc ` ` header header ` : : c++ header file containing the definition of the numerator class : if the file does not exists , one will be created . by default
it will have the same name as the output but with ` .hh ` extension .
as mentioned , one can also use the package as a python module ( ` ninjanumgen ` ) .
this contains a class ` diagramexpansion ` which can be used for the generation of the source code which implements the numerator methods .
the input parameters of the constructor of this class roughly correspond to the arguments which can be used in the script .
a detailed description can be obtained , after installation , by invoking python in interactive mode ( usually done with the command ` python ` ) and typing .... import ninjanumgen help(ninjanumgen.diagramexpansion ) .... the method ` writesource ` generates the source files . as a simple example , the source for the integrand we defined in section [ sec : integrand ] could have been generated within python with the commands .... # import the module import ninjanumgen # define the mandatory arguments for the constructor n_legs = 4 input_file = ' mynum.frm ' output_file = ' mynum.cc ' # define an instance of the class diagramexpansion mynum = ninjanumgen.diagramexpansion(input_file , output_file , n_legs , rank=4 ) # generate the source mynum.writesource ( ) .... we suggest to look at the python files in the ` examples ` directory for other basic examples .
as pointed out in ref .
@xcite , the laurent expansion method can be generalized to non - renormalizable and effective theories with higher - rank numerators . in a renormalizable theory , with a proper choice of gauge the rank @xmath119 can not be greater than the number @xmath1 of loop propagators .
ninja , if configured with the ` enable - higher_rank ` option , can also be used for the computation of integrals with @xmath153 . here
we describe the generalization of the method to the higher - rank case , underlining the points where it differs from the renormalizable case . in eq .
, we gave the most general parametrization of the residues @xmath23 in a renormalizable theory .
in the higher - rank case with @xmath153 , such parametrization is generalized as follows @xcite @xmath154 the generalized integral decomposition is thus @xmath155 \pagebreak[1 ] \nn & + \sum_{\{i_1 , i_2\}}\bigg\ { c_{10}^ { ( i_1 i_2 ) } i_{i_1 i_2}[\mu^2\ , ( q+p_{i_1})\cdot e_2 ) ] + c_{13}^ { ( i_1 i_2 ) } i_{i_1 i_2}[((q+p_{i_1})\cdot e_2)^3 ] \bigg\ } \pagebreak[1 ] \nn & + \sum_{i_1 } \bigg\ { c_{14}\ , i_{i_1}[\mu^2 ] + c_{15}^ { ( i_1)}\ , i_{i_1}[((q+p_{i_1})\cdot e_3)((q+p_{i_1})\cdot e_4 ) ] \bigg\ } \label{eq : hrintegraldecomposition}\end{aligned}\ ] ] this higher - rank decomposition has been used for the computation of nlo corrections to higgs - boson production in association with two @xcite and three @xcite jets .
other libraries which implement the reduction of higher - rank integrals are xsamurai @xcite , which extends the more traditional integrand reduction algorithm of samurai , and golem95 @xcite . while the extension of the laurent expansion method for the computation of higher - rank 3-point and 2-point residues is straightforward , for 4-point and 1-point residues some further observations are in order . here
we propose a generalization of the laurent expansion method which allows to efficiently compute the non - spurious coefficients of 4- and 1-point residues without spoiling the nice features of the algorithm , such as the simplified subtractions of higher - point contributions and the diagonal systems of equations .
this generalization is not present elsewhere in the literature and has been implemented in the ninja library .
[ [ point - residues-5 ] ] 4-point residues + + + + + + + + + + + + + + + + the coefficient @xmath64 can be computed exactly as in the renormalizable case . for the coefficient @xmath156
, one needs instead to keep also the next - to - leading term in the @xmath19 expansion described before , so that the @xmath4-dimensional solutions of a quadruple cut , given in eq .
, in the asymptotic limit become @xmath157 where it is worth noticing that @xmath75 can be obtained as the average of the two solutions of the corresponding four - dimensional quadruple cut . in this limit
, the expansion of the integrand reads @xmath158 hence the leading term is now the spurious coefficient @xmath159 , but @xmath72 can still be obtained as the next - to - leading term .
this can be implemented semi - numerically , by keeping the two leading terms of the expansion of the numerator and performing a polynomial division with respect to the two leading terms in the expansion of the uncut denominators which have the form @xmath160 given the very limited number of terms involved , the division can be implemented very efficiently in a small number of operations .
more in detail , if ` num ` and ` den ` are arrays of length two containing the leading and next - to - leading terms in the expansion of the numerator and a denominator respectively , we can perform the division in place with the commands ....
num[0 ] /= den[0 ] ; num[1 ] -= den[1]*num[0 ] ; num[1 ] /= den[0 ] ; .... which will have the effect of replacing the entries of ` num ` with the ones of the expansion of @xmath139 .
we also observe that the computation and the subtraction of pentagons is not needed in the higher - rank case either .
[ [ point - residues-6 ] ] 1-point residues + + + + + + + + + + + + + + + + on higher - rank 1-point residues @xmath98 we consider @xmath4-dimensional solutions of the corresponding single cut of the form @xmath161 in terms of the free variables @xmath80 and @xmath19 . by taking the @xmath93 limit of the integrand and the subtraction terms evaluated on these solutions , we obtain an asymptotic polynomial expansion of the form @xmath162 one can check that the non - spurious coefficients of the tadpole are given in terms of the ones of the expansions above by @xmath163 in the higher - rank case , the ` muexpansion ` method of the numerator needs to compute both the leading and the next - to - leading term of the expansion in @xmath19 .
the package ninjanumgen , takes care of this automatically when the specified rank is higher than the number of external legs of the loop .
the information in the next paragraph is only needed for a custom implementation of the method without ninjanumgen .
the ` muexpansion ` method in the higher - rank case should compute the two leading terms of the expansion in @xmath80 of the numerator , defined by @xmath164 with @xmath165}}$ ] , where @xmath40 is the array of momenta passed as input parameter .
the leading and next - to - leading terms of the expansion should be written in the first two entry of the array pointed by the parameter ` c ` , i.e. .... c[0 ] = $ \(\n\)$[$\(t^{r}\)$ ] ; c[1 ] = $ \(\n\)$[$\(t^{r-1}\)$ ] ; ....
all the other methods should have instead the same definition described in section [ sec : input ] . as one can see from eq .
, in the higher - rank case five new types of integral appear in the final decomposition .
they are a 2-point integral of rank 3 , a 1-point integral of rank 2 , and three more integrals containing @xmath19 at the numerator which contribute to the rational part of the amplitude .
ninja contains an implementation of all these higher - rank integrals in terms of lower - rank ones .
this means that , should the user choose to interface a custom integral library ( see appendix [ sec : intlib ] ) , these higher - rank integrals would not be needed , although specifying an alternative implementation would still be possible .
all the integrals of eq . which contribute to the rational part of the amplitude
have already been computed in ref .
@xcite . with our choice of the normalization factor given in eq . , they read @xmath166 = { } & \frac{1}{6}\left ( \frac{s_{i_2 i_1 } + s_{i_3 i_2 } + s_{i_1 i_3}}{4 } - m_{i_1}^2 - m_{i_2}^2 - m_{i_3}^2 \right ) + \o(\epsilon)\\ i_{i_1 i_2}[\mu^2\ , ( ( q+p_{i_1 } ) \cdot v ) ] = { } & \frac{((p_{i_2}-p_{i_1 } ) \cdot v)}{12}\big ( s_{i_2 i_1 } - 2\ , m_{i_1}^2 - 4\ , m_{i_2}^2 \big ) + \o(\epsilon ) \\ i_{i_1}[\mu^2 ] = { } & \frac{m_{i_1}^4}{2 } + \o(\epsilon)\end{aligned}\ ] ] where @xmath167 were defined in eq .
and @xmath40 is an arbitrary vector .
the tadpole of rank 2 appearing in eq . can also be written as a function of the scalar tadpole integral @xmath168 as follows @xmath169 = { } m_{i_1}^2\ , \frac{(e_3\cdot e_4)}{4}\left ( i_{i_1 } + \frac{m_{i_1}^2}{2 } \right ) + \o(\epsilon).\ ] ] since the vector @xmath170 in the bubble integral of rank 3 appearing in eq .
is massless , the corresponding integral is simply proportional to the form factor @xmath171 , @xmath172 = ( ( p_{i_2}-p_{i_1})\cdot e_2)^3\ , b_{111}(s_{i_2 i_1},m_{i_1}^2 , m_{i_2}^2).\ ] ] the form factor can be computed using the formulas of ref .
@xcite , as a function of form factors of scalar integrals @xmath173 . in the special case with
@xmath174 we use eq .
( a.6.2 ) and ( a.6.3 ) of that reference . for the general case @xmath175
we implement instead the following formula @xmath176-m_{i_2}^2 \ , i_{i_2}-i_{i_2}[\mu^2 ] \nn & - 4 \ , i_{i_1 i_2}[\mu^2\ , ( ( q+p_{i_1})\cdot ( p_{i_2}-p_{i_1 } ) ) ]
\nn & - 4 \ , m_{i_1}^2 \ ,
i_{i_1 i_2}[(q+p_{i_1})\cdot ( p_{i_2}-p_{i_1})]\big ) \nn & + 4\ , ( m_{i_2}^2-m_{i_1}^2-s_{i_2 i_1 } ) \ , i_{i_1 i_2}[((q+p_{i_1})\cdot ( p_{i_2}-p_{i_1}))^2 ] \bigg\}.\end{aligned}\ ] ]
ninja already implements interfaces for the oneloop and the looptools integral libraries .
these libraries have been used in a large number of computations and provide very reliable results , hence they should suffice for most purposes . however , ninja has been designed considering the possibility of using any other library of master integrals .
the master integrals are computed by calling virtual methods of the abstract class ` ninja::integrallibrary ` , which is defined in the header file ` ninja / integral_library.hh ` .
therefore , any library of master integrals can be interfaced by implementing a class derived from ` integrallibrary ` .
each method of the library corresponds to a different master integral appearing in eq . , which should be implemented for both real and complex internal masses ( and optionally for the massless case ) .
an implementation of higher - rank integrals can also be provided but it is not needed , since ninja has a default implementation of them in terms of lower rank integrals .
there are two further methods , namely ` init ` and ` exit ` .
the former is called inside the method ` amplitude::evaluate ` just before the computation of the first needed master integral , while the latter is called after the last master integral has been computed .
the method ` init ` takes as input the square of the renormalization scale to be used in the subsequent calls of the methods implementing the master integrals .
it can also be used in order to perform any other initialization the library might need before computing the integrals .
the ` exit ` method instead , does not need to be implemented and by default it will not perform any action . in section [ sec : threads ] we gave an example of a case where a non - trivial implementation of the ` exit ` method could be useful .
- real masses virtual void getboxintegralrm(complex rslt[3 ] , real s21 , real s32 , real s43 , real s14 , real s31 , real s42 , real m1sq , real m2sq , real m3sq , real m4sq ) = 0 ; // - complex masses virtual void getboxintegralcm(complex rslt[3 ] , real s21 , real s32 , real s43 , real s14 , real s31 , real s42 , const complex & m1sq , const complex & m2sq , const complex & m3sq , const complex & m4sq ) = 0 ; .... and they must write the @xmath177 term of the result in the @xmath178-th entry of the array ` rslt ` , for @xmath179 .
the arguments are the invariants @xmath167 and the squared masses @xmath180 .
similar methods need to be provided for 3-point , 2-point and 1-point master integrals , as described in detail in the comments inside the header file ` ninja / integral_library.hh ` .
examples of implementation of this interface for the libraries oneloop and looptools can be found in the source code .
more in detail , we define the instances ` ninja::avh_olo ` and ` ninja::loop_tools ` of the classes ` ninja::avhoneloop ` and ` ninja::looptools ` respectively , which implement the methods described above as wrappers of the corresponding routines in each integral library .
the oneloop interface also implements a cache of master integrals on top of these routines .
the cache is implemented similarly to a hash table , which allows constant - time look - up of each computed integral from its arguments .
hence , the methods of the ` avhoneloop ` class will call the routines of the oneloop library only if a master integral is not found in the cache . the cache can be cleared with the class method ` avhoneloop::clearintegralcache ` . during a phase - space integration ,
we suggest calling this method once per phase space point , especially for more complex processes .
this method does not completely free the allocated memory , but keeps the buckets of the hash table available in order to store the integrals more efficiently in subsequent calls of the respective methods .
if the user wishes to completely free the allocated memory , the method ` avhoneloop::freeintegralcache ` can be used , although in general ` clearintegralcache ` should be preferred .
as already mentioned , every instance of ` avhoneloop ` has a cache of master integrals as data member
. this can be useful for building multi - threaded applications , as discussed in the examples of section [ sec : threads ] .
since looptools already has an internal cache of master integrals , the implementation of its interface is much simpler and only consists in wrapper of its routines .
we implemented a ` clearintegralcache ` method in the ` looptools ` class as well , which in this case simply calls the routine which clears the cache of integrals in looptools . | we present the public c++ library ninja , which implements the integrand reduction via laurent expansion method for the computation of one - loop integrals .
the algorithm is suited for applications to complex one - loop processes . |
objects with sizes and masses between those of globular clusters and compact ellipticals ( @xmath13100 pc ; @xmath14 to @xmath15 ) were first discovered in spectroscopic surveys of galaxy clusters ( hilker 1999 ; drinkwater 2000 ) .
they were quickly dubbed ultra - compact dwarf " galaxies ( ucds ) , even though their galaxian nature was unclear .
large populations of ucds have been discovered in fornax , virgo , and other galaxy clusters , as well as in group and field environments see reviews in chilingarian ( 2011 ) , norris & kannappan ( 2011 ) , and brodie ( 2011 ) .
ucd formation scenarios have coalesced around two poles : star cluster or galaxy . in the former scenario , ucds form the massive end of the normal sequence of globular clusters ( mieske 2012 ) .
further , if some star clusters form in gravitationally - bound complexes , these can merge to make objects that are larger and more massive than single clusters ( brns 2011 ) .
alternatively , ucds could be galaxies that formed in individual dark matter halos either in situ " , as unusual , extremely compact galaxies or as the products of tidal stripping of more massive progenitor galaxies ( e.g. , drinkwater 2003 )
. a reasonable synthesis of these scenarios may be that the least - massive ucds " , with @xmath16 , are largely star clusters , while the most massive objects ( @xmath17 ) are galaxies , or the tidally stripped remnants thereof . at intermediate masses
both star clusters and galaxies may co - exist ( e.g. , norris & kannappan 2011 ; brodie 2011 ) . there is more at stake than the natural desire to understand these novel stellar systems .
if a significant fraction of ucds contain dark matter , then they form a populous class that must be included in counts of subhalos for comparisons to cosmological theory .
further , if some ucds are formed by tidal stripping , their chemical and structural properties help trace galaxy transformation . here
we report the discovery of an extraordinary ucd around the virgo elliptical ngc 4649 ( m60 ) .
it has a half - light radius of 24 pc but a stellar mass of @xmath18 , giving it the highest surface density of any galaxy in the local universe .
we also present evidence that this ucd may contain a central supermassive black hole .
we discovered m60-ucd1 in the _ hubble space telescope_/advanced camera for surveys imaging of strader ( 2012 ) .
we have a single orbit of imaging split between @xmath19 and @xmath20 ( hereafter @xmath21 and @xmath22 ) .
m60-ucd1 is located at ( r.a . ,
= ( 190.8999 , 11.5347 ) in decimal j2000 coordinates .
this is at a projected distance of only @xmath23 kpc from the center of m60 ( figure 1 ; assuming a distance of 16.5 mpc ; blakeslee 2009 ) .
no mention is made of m60-ucd1 in previous virgo surveys , including the acs virgo cluster survey ( ct 2004 ) .
it is present in the sdss dr7 photometric catalog ( abazajian 2009 ) as j124335.96 + 113204.6 , and was classified by simard ( 2011 ) as a background galaxy .
a spectrum of m60-ucd1 was obtained on the night of 17 january 2012 with keck / deimos ( faber 2003 ) , utilizing the 1200 l / mm grating centered at 7800 and a 1 slit ( resolution @xmath24 ) .
we obtained three 30-min exposures in 0.8 seeing . using the spec2d pipeline ( cooper 2012 ) ,
the spectra were extracted , calibrated , and combined in the standard manner to produce a final one - dimensional spectrum . to improve stellar population constraints ,
further spectroscopy was undertaken with mmt / hectospec ( fabricant 2005 ) on 16 may 2012 , using the 270 l / mm grating with wavelength coverage from 3700 to 9100 and 5 resolution .
three 20-min exposures were taken in 0.9 seeing .
these hectospec data were pipeline - reduced in a standard manner as described in mink ( 2007 ) .
aperture photometry of m60-ucd1 gives a total integrated magnitude of @xmath25 and @xmath26 , yielding @xmath27 ( the @xmath21 and @xmath22 magnitudes in this paper are ab ) .
the measured half - light radius ( see below ) is @xmath28 pc .
the inferred total luminosities are : @xmath29 ; @xmath30 ; and @xmath31 . with @xmath32
, m60-ucd1 is the most luminous ucd known ( see 4.1 for further discussion ) .
we fit the optical imaging with one and two - component elliptical srsic models .
these fits , shown in figure 2 , were performed by fitting two - dimensional models convolved with an empirical point spread function ( psf ) using a custom software package as described in seth ( 2006 ) .
the psf ( @xmath33 subsampled ) was derived from point sources in the images .
the fits were performed on a 5@xmath345 image centered on the ucd , with the background galaxy gradient from m60 itself removed .
the fits are not very sensitive to the fitting box size . from the residual map ( figure 2 ) , it is clear that a single srsic component provides a poor fit .
in particular , the ellipticity of m60-ucd1 becomes more circular at larger radii , leaving a residual along the minor axis .
the radial shape of the surface brightness profile is also poorly fit .
however , using a two - component srsic model , a very good fit ( @xmath35=1.07 in @xmath21 ) is obtained .
table 1 gives the parameters for the single and double srsic @xmath21 band fits and the double srsic @xmath22 fits .
for the two - component fits the srsic parameters @xmath36 and @xmath37 are very similar between the filters . because the @xmath21 band provides a much better fit ( probably due to psf modeling issues in @xmath22 ) , all structural values cited are from the @xmath21 fits . for the best - fit two - component model ,
the inner component is compact ( @xmath38 pc ) , with modest ellipticity ( @xmath39 ) , and has about 58% of the total luminosity of m60-ucd1 .
the outer component is more extended ( @xmath40 pc ) , round , and with a nearly exponential profile ( @xmath41 ) .
the overall half - light radius is @xmath28 pc , empirically measured using the deconvolved @xmath21 surface brightness profile .
this value is similar to the radius derived from the single component srsic fit ; within this radius we estimate @xmath42 .
lcccccccc @xmath21 single & 2.33 & 17.51 & @xmath43 & @xmath44 & @xmath45 & @xmath46 & @xmath47 & @xmath48 + @xmath21 double ( inner ) & 1.07 & 18.11 & @xmath49 & @xmath50 & @xmath51 & @xmath52 & @xmath53 & @xmath54 + @xmath21 double ( outer ) & & 18.46 & @xmath55 & @xmath56 & @xmath57 & @xmath58 & @xmath59 & @xmath60 + @xmath22 double ( inner ) & 2.03 & 16.55 & @xmath61 & @xmath62 & @xmath63 & @xmath64 & @xmath65 & @xmath66 + @xmath22 double ( outer ) & & 16.89 & @xmath67 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath72 + using the keck / deimos spectrum , the integrated velocity dispersion of m60-ucd1 was measured by cross - correlating the region around the ca triplet with a library of templates of the same resolution and wavelength coverage , as described by strader ( 2011 ) .
this value is @xmath73 .
the radial velocity of m60-ucd1 is @xmath74 ; the systemic velocity of m60 is 1117 ( gonzlez 1993 ) .
we estimate a dynamical mass for m60-ucd1 using the virial theorem : @xmath75 .
@xmath76 is a parameter that depends on the structure of the galaxy and is smaller for more concentrated systems ; @xmath77 is the integrated velocity dispersion within @xmath37 . following the results of cappellari ( 2006 ) for a range of srsic profiles , we assume @xmath78 , intermediate between the applicable values for the @xmath79 and @xmath80 profiles ( corresponding to the inner and outer components respectively ) .
we further estimate that @xmath81 , slightly higher than the measured value of @xmath82 , by integrating over our deimos extraction window ( 1.2 @xmath34 1.0 ) and accounting for seeing .
the dynamical mass determined in this manner is @xmath83 .
the systematic uncertainties are significant : we have assumed isotropy , sphericity , and mass - follows - light .
dividing this dynamical mass by the total luminosity of m60-ucd1 yields a mass - to - light ratio of @xmath9 .
the flexible stellar population synthesis models of conroy ( 2009 ) , using padova isochrones and a kroupa initial mass function ( imf ) , predict @xmath84 ( 3.5 , 4.2 , 4.7 ) for solar metallicity and ages of ( 8,10,12 ) gyr , respectively ( 5.9 , 7.0 , 7.9 ) for these ages , an increase of nearly 70% in stellar mass over a kroupa imf . ] . if m60-ucd1 has a younger age , the dynamical @xmath10 could imply an elevation in low - mass stars or stellar remnants over kroupa imf model predictions . for older ages
there is an excellent match between the observed @xmath10 and the model predictions . as discussed in 3.3 , a modest increase in the central velocity dispersion ( and hence @xmath85 )
could also be caused by the presence of a supermassive black hole with a mass @xmath86% of that of the ucd ( mieske 2013 ) .
dark matter is not expected to contribute to the @xmath85 ( 4.2 ) .
m60-ucd1 is marginally resolved in our deimos observations , and so some spatially - resolved kinematic information is available .
the 5 slitlet was aligned close to the major axis of the object . using the sky - subtracted two - dimensional spectrum
, we determined the radial velocity and velocity dispersion on a pixel - by - pixel basis ( one pixel is @xmath87 ) . there is clear rotation present , with an amplitude of @xmath88 km s@xmath6 to a projected radius of @xmath89 .
a decline of comparable amplitude in the velocity dispersion is also observed . since the radial profiles are strongly affected by seeing ,
we do not attempt dynamical modeling . however , these data provide motivation to obtain improved kinematic maps in the future using integral - field spectroscopy .
we constrain the stellar populations of m60-ucd1 using our mmt / hectospec spectrum ( with its wide wavelength range ) and the models of conroy & van dokkum ( 2012a ; with additions from conroy & van dokkum 2012b ) .
these are stellar population synthesis models with variable abundance ratios for 11 elements .
a markov chain monte carlo method is used to simultaneously fit the entire available optical spectrum .
the derived values are listed in table 2 .
the uncertainties quoted are solely statistical , and do not include the substantial systematic uncertainties necessarily present in any integrated - light study of stellar populations .
the rms residuals in the fit were @xmath90% over most of the spectrum .
the formal age is @xmath91 gyr , indicating an old stellar population .
m60-ucd1 is of solar metallicity with a mild elevation in [ @xmath92/fe ] over solar .
the abundances for c , o , and @xmath92-elements appear very similar to the mean values for high-@xmath93 local early - type galaxies determined in a similar manner ( conroy 2013 ) .
however , the abundance of n is unusual : it is strongly enhanced , with [ n / fe@xmath94\sim+0.6 $ ] , comparable to the average value observed in globular clusters ( e.g. , briley 2004 ) .
the high abundance of n in globular clusters is generally attributed to self - enrichment by the winds of asymptotic giant branch stars ( gratton 2012 ) .
our results suggest light - element self - enrichment may also be present in ucds , presumably related to their high stellar densities .
we note [ na / fe ] varies strongly with @xmath93 in early - type galaxies , increasing from @xmath95 to @xmath96 for @xmath97 to @xmath98 .
the m60-ucd1 abundance ( [ na / fe@xmath94\sim+0.4 $ ] ) is therefore difficult to interpret : it could be expected , or could represent a large enhancement over baseline .
as a check on these values , we performed a standard lick index analysis using ez_ages ( graves & schiavon 2008 ) .
these values are also listed in table 2 , and the same caveats apply .
this analysis gave a formal age of @xmath9911 gyr and similar abundance values for most of the elements in common with the full - spectrum analysis ( c , ca , mg , fe ) .
for n the ez_ages analysis does not yield a reliable value , as the cn index strength is off the grids .
the lick cn@xmath100 index for m60-ucd1 ( 0.24 mag ) is comparable to that in most metal - rich m31 globular clusters , which are also thought to be self - enriched ( schiavon 2012 ) .
the very high abundance of n appears to be a robust conclusion of the analysis .
lcc & @xmath101 & @xmath102 + [ o / fe ] & @xmath103 & + [ c / fe ] & @xmath104 & @xmath105 + [ n / fe ] & @xmath106 & + [ na / fe ] & @xmath107 & + [ mg / fe ] & @xmath108 & @xmath109 + [ si / fe ] & @xmath110 & + [ ca / fe ] & @xmath111 & @xmath112 + [ ti / fe ] & @xmath113 & + an x - ray source at the position of m60-ucd1 is present in the _ chandra_/acis catalog of luo ( 2013 ) .
the central astrometric matching between the @xmath114 and @xmath115 data is excellent due to the large x - ray binary and globular cluster populations of m60 , with an rms scatter of 0.17 .
this x - ray source , termed xid 144 by luo ( 2013 ) , has a position consistent with the optical center of m60-ucd1 .
there is evidence that it is variable , with its x - ray luminosity ( from 0.3 to 8 kev ) ranging from @xmath116 to @xmath117 erg / s over timescales as short as a few months .
the x - ray spectrum is well - fit by a absorbed power - law with a photon index of 1.8 .
there are two reasonable possibilities for this central x - ray source : it could either be an active galactic nucleus associated with a massive black hole or a low - mass x - ray binary .
the case for a central black hole is straightforward .
if the black hole occupation fraction of dwarf galaxies is high , and if ucds are the products of tidal stripping of dwarf galaxies , then a significant fraction of ucds should have overmassive " black holes that could be detected through dynamical or accretion signatures .
if ucds have been stripped of 99% or more of their original mass ( we estimate in 4.2 that the progenitor of m60-ucd1 was @xmath118200 times more massive ) , then they could host supermassive black holes that are @xmath7% of their present - day masses ( mieske 2013 ) .
frank ( 2011 ) constrain a putative black hole to be @xmath119 of the total mass of one fornax cluster ucd through integral - field spectroscopy .
the observed x - ray luminosity would be consistent with a @xmath120 black hole accreting at @xmath121 of the eddington rate with a radiative efficiency of @xmath122 .
this eddington ratio of @xmath123 would be typical of nuclei with old stellar populations ( ho 2009 ) .
we can also estimate the odds that m60-ucd1 contains a bright x - ray binary .
sivakoff ( 2007 ) derive formulae to estimate the probability that a globular cluster contains a low - mass x - ray binary with @xmath124 erg / s .
the odds are higher for metal - rich clusters and those with high encounter rates . applying their results , but extrapolating to the fainter luminosity observed ,
suggests a @xmath125% chance of having observed a low - mass x - ray binary in m60-ucd1 .
however , these results are of uncertain relevance for an object with a different structure and formation history than a globular cluster ( since what is pertinent is the integrated not instantaneous collision rate ) .
dabringhausen ( 2012 ) suggest ucds have a lower occurrence of low - mass x - ray binaries than expected on the basis of the sivakoff ( 2007 ) results .
future observations can help clarify the nature of the x - ray source .
for example , if m60-ucd1 hosts a @xmath126 black hole that lies on the radio x - ray fundamental plane for black holes ( plotkin 2012 ) , it should be detectable with the very large array .
figure 3 shows a plot of log @xmath127 vs. log @xmath128 for dispersion - supported stellar systems .
@xmath127 is the mean surface luminosity density within @xmath37 .
globular clusters are plotted with different symbols than galaxies .
it is clear that m60-ucd1 is an unusual object : it is much denser than any other object classified as a galaxy .
it is more massive than any ucd or star cluster of comparable size , but is much more compact than other galaxies of similar luminosity .
m60-ucd1 is not the densest stellar system known .
that honor goes to any of a number of nuclear star clusters , which can reach mean surface densities of @xmath129 pc@xmath130 within @xmath37 ( walcher 2005 ; these are not plotted in figure 3 ) .
many massive globular clusters are also extremely dense .
however , m60-ucd1 is arguably the densest _ galaxy _ known in the local universe . using the @xmath10 from 3.2.1
, its mean effective surface density is @xmath131 pc@xmath130 , a factor of 2.53 higher than for m32 .
the inner component of m60-ucd1 , with @xmath132 pc , has a mean @xmath133 pc@xmath130 , comparable to that of many nuclear star clusters .
the central volume density of m60-ucd1 is not well - constrained by the present data .
the object most similar to m60-ucd1 is hucd1 , a hydra cluster ucd , which has @xmath134 pc and @xmath135 ( misgeld 2011 ) , though m59co ( chilingarian & mamon 2008 ) and several coma cluster ucds ( chiboucas 2011 ) are also similar , if less extreme .
it seems likely that ongoing surveys for ucds will turn up additional objects with properties comparable to m60-ucd1 . the extreme mass , multiple structural components , high metallicity , and possible presence of a central black hole make it unlikely that m60-ucd1 is a star cluster or merged cluster complex .
it is most plausible that the object is the tidally - stripped remnant of a more massive progenitor galaxy .
pfeffer & baumgardt ( 2013 ) present new simulations of the formation of ucds through tidal stripping of nucleated galaxies in a massive , virgo - like cluster .
they show that it is possible to reproduce the sizes and luminosities of typical ucds .
these simulations were not intended to match the most massive ucds , and hence are not directly comparable to the properties of m60-ucd1 . however , a general conclusion from this study is relevant : the nucleus is largely unaffected by the stripping process , so the inner core of the resultant ucd should have similar properties to the original nucleus .
the inner component of m60-ucd1 has @xmath136 , @xmath137 , and @xmath38 pc .
the size , luminosity , and red color are similar to nuclei in the virgo galaxies ngc 4379 and ngc 4387 ( ct 2006 ) and the fornax cluster galaxies ngc 1389 and ic 2006 ( turner 2012 ) .
these galaxies have @xmath138 to @xmath12 and stellar masses @xmath139@xmath140 .
we conclude that m60-ucd1 could have originated in the tidal stripping of a virgo galaxy in this luminosity range .
since the present luminosity of m60-ucd1 is @xmath141 , we infer that it is a factor of @xmath118200 less massive because of the stripping .
the projected distance of m60-ucd1 from the center of m60 ( @xmath23 kpc ) is consistent with the small pericenter needed for efficient stripping ( pfeffer & baumgardt 2013 ) .
galaxies in this mass range host significant globular cluster populations ( @xmath142100 ; brodie & strader 2006 ) that would also be stripped during ucd formation and might still be detectable in phase space ( e.g. , romanowsky 2012 ) .
we caution that ucds may originate in a biased subset of host galaxies that have been largely destroyed , so it is possible that there is no correspondence between ucd progenitors and a subpopulation of surviving galaxies .
the pfeffer & baumgardt ( 2013 ) simulations show that extended debris suggestive of tidal stripping becomes challenging to observe after relatively short ( @xmath139 gyr ) timescales , so the lack of evidence for such debris in figure 2 does not disfavor this scenario .
the dynamical friction timescale ( binney & tremaine 1987 ) for m60-ucd1 is @xmath143 gyr , so it is plausible that its progenitor was stripped long ago and the remnant has stalled " at its current radius . in this case
no observable tidal tails would be expected .
the stellar density of m60-ucd1 is much higher than expected for dark matter with a standard nfw profile ( a factor of @xmath144 for even for extreme , cluster - scale halo masses ; tollerud 2011 ) .
if the galaxy has undergone as much tidal stripping as inferred , this is likely to have strongly modified the dark matter profile ; nonetheless , m60-ucd1 is probably the worst ucd in which to search for dark matter . as discussed by willman & strader ( 2012
; see also hilker 2007 ) , the ucds most likely to show evidence for dark matter are the least massive and most extended ucds .
we thank l. chomiuk , s. mieske , and r. schiavon for useful discussions and an anonymous referee for a helpful report . based on observations made with the nasa / esa _ hubble space telescope _ and the hubble legacy archive .
data obtained at keck ( caltech / uc / nasa ) and mmt ( arizona / smithsonian ) .
products produced by the oir telescope data center ( sao ) .
support by arc grant dp130100388 and nsf grants ast-1109878/ast-0909237 .
abazajian , k. n. , adelman - mccarthy , j. k. , ageros , m. a. , et al .
2009 , , 182 , 543 binney , j. , & tremaine , s. 1987 , princeton , nj , princeton university press briley , m. m. , harbeck , d. , smith , g. h. , & grebel , e. k. 2004 , , 127 , 1588 blakeslee , j. p. , jordn , a. , mei , s. , et al .
2009 , , 694 , 556 brodie , j. p. , romanowsky , a. j. , strader , j. , & forbes , d. a. 2011 , , 142 , 199 brodie , j. p. , & strader , j. 2006 , , 44 , 193 brns , r. c. , kroupa , p. , fellhauer , m. , metz , m. , & assmann , p. 2011
, , 529 , a138 cappellari , m. , bacon , r. , bureau , m. , et al .
2006 , , 366 , 1126 chiboucas , k. , tully , r. b. , marzke , r. o. , et al .
2011 , , 737 , 86 chilingarian , i. v. , & mamon , g. a. 2008 , , 385 , l83 chilingarian , i. v. , mieske , s. , hilker , m. , & infante , l. 2011 , , 412 , 1627 conroy , c. , gunn , j. e. , & white , m. 2009 , , 699 , 486 conroy , c. , & van dokkum , p. 2012
, , 747 , 69 conroy , c. , & van dokkum , p. g. 2012 , , 760 , 71 conroy , c. , graves , g. , & van dokkum , p. 2013 , apj , submitted ( arxiv:1303.6629 ) cooper , m. c. , newman , j. a. , davis , m. , finkbeiner , d. p. , & gerke , b. f. 2012 , astrophysics source code library , 3003 ct , p. , blakeslee , j. p. , ferrarese , l. , et al .
2004 , , 153 , 223 ct , p. , piatek , s. , ferrarese , l. , et al . 2006 , , 165 , 57 dabringhausen , j. , kroupa , p. , pflamm - altenburg , j. , & mieske , s. 2012 , , 747 , 72 drinkwater , m. j. , jones , j. b. , gregg , m. d. , & phillipps , s. 2000 , pasa , 17 , 227 drinkwater , m. j. , gregg , m. d. , hilker , m. , et al .
2003 , , 423 , 519 faber , s. m. , phillips , a. c. , kibrick , r. i. , et al .
2003 , , 4841 , 1657 fabricant , d. , fata , r. , roll , j. , et al .
2005 , , 117 , 1411 forbes , d. , pota , v. , usher , c. , et al .
2013 , mnras , in press ( arxiv:1306.5245 ) frank , m. j. , hilker , m. , mieske , s. , et al .
2011 , , 414 , l70 gonzlez , j. j. 1993 , ph.d .
thesis , university of california , santa cruz gratton , r. g. , carretta , e. , & bragaglia , a. 2012 , , 20 , 50 graves , g. j. , & schiavon , r. p. 2008
, , 177 , 446 hilker , m. , infante , l. , vieira , g. , kissler - patig , m. , & richtler , t. 1999 , , 134 , 75 hilker , m. , baumgardt , h. , infante , l. , et al .
2007 , , 463 , 119 ho , l. c. 2009 , , 699 , 626 luo , b. , fabbiano , g. , strader , j. , et al .
2013 , , 204 , 14 mieske , s. , frank , m. , baumgardt , h. , luetzgendorf , n. , neumayer , n. , hilker , m. 2013 , , in press ( arxiv : 1308.1398 ) mieske , s. , hilker , m. , & misgeld , i. 2012 , , 537 , a3 mink , d. j. , wyatt , w. f. , caldwell , n. , et al . 2007 ,
astronomical data analysis software and systems xvi , 376 , 249 misgeld , i. , mieske , s. , hilker , m. , et al .
2011 , , 531 , a4 norris , m. a. , & kannappan , s. j. 2011 , , 414 , 739 pfeffer , j. , & baumgardt , h. 2013 , , 1504 plotkin , r. m. , markoff , s. , kelly , b. c. , krding , e. , & anderson , s. f. 2012 , , 419 , 267 romanowsky , a. j. , strader , j. , brodie , j. p. , et al .
2012 , , 748 , 29 schiavon , r. p. , caldwell , n. , morrison , h. , et al .
2012 , , 143 , 14 seth , a. c. , dalcanton , j. j. , hodge , p. w. , & debattista , v. p. 2006
, , 132 , 2539 simard , l. , mendel , j. t. , patton , d. r. , ellison , s. l. , & mcconnachie , a. w. 2011 , , 196 , 11 sivakoff , g. r. , jordn , a. , sarazin , c. l. , et al .
2007 , , 660 , 1246 strader , j. , caldwell , n. , & seth , a. c. 2011 , , 142 , 8 strader , j. , fabbiano , g. , luo , b. , et al .
2012 , , 760 , 87 tollerud , e. j. , bullock , j. s. , graves , g. j. , & wolf , j. 2011 , , 726 , 108 turner , m. l. , ct , p. , ferrarese , l. , et al .
2012 , , 203 , 5 walcher , c. j. , van der marel , r. p. , mclaughlin , d. , et al .
2005 , , 618 , 237 willman , b. , & strader , j. 2012 , , 144 , 76 | we report the discovery of a remarkable ultra - compact dwarf galaxy around the massive virgo elliptical galaxy ngc 4649 ( m60 ) , which we term m60-ucd1 . with a dynamical mass of @xmath0 but a half - light radius of only @xmath1 pc , m60-ucd1 is more massive than any ultra - compact dwarfs of comparable size , and is arguably the densest galaxy known in the local universe .
it has a two - component structure well - fit by a sum of srsic functions , with an elliptical , compact ( @xmath2 pc ; @xmath3 ) inner component and a round , exponential , extended ( @xmath4 pc ) outer component .
_ chandra _ data reveal a variable central x - ray source with @xmath5 erg s@xmath6 that could be an active galactic nucleus associated with a massive black hole or a low - mass x - ray binary .
analysis of optical spectroscopy shows the object to be old ( @xmath7 gyr ) and of solar metallicity , with elevated [ mg / fe ] and strongly enhanced [ n / fe ] that indicates light - element self - enrichment ; such self - enrichment may be generically present in dense stellar systems .
the velocity dispersion ( @xmath8 ) and resulting dynamical mass - to - light ratio ( @xmath9 ) are consistent with
but slightly higher than
expectations for an old , metal - rich stellar population with a kroupa initial mass function .
the presence of a massive black hole or a mild increase in low - mass stars or stellar remnants is therefore also consistent with this @xmath10 .
the stellar density of the galaxy is so high that no dynamical signature of dark matter is expected .
however , the properties of m60-ucd1 suggest an origin in the tidal stripping of a nucleated galaxy with @xmath11 to @xmath12 . |
all algebraic objects in this note will be defined over a field @xmath9 of characteristic zero and @xmath10 will always denote a @xmath9-vector space . we will sometimes use the formalism of operads explained , for example , in @xcite .
sections 2 , 3 and 4 containing the main results , however , do not rely on this language .
let @xmath11 denote the free pre - lie algebra generated by @xmath10 and @xmath12 the associated lie algebra .
we will focus on the lie algebra @xmath13 generated in @xmath12 by @xmath10 , called the _ subalgebra of lie elements _ in @xmath14 .
it is known @xcite that @xmath15 is ( isomorphic to ) the free lie algebra generated by @xmath10 ; we will give a new short proof of this statement in section 3 .
our main result , theorem [ .
] , describes @xmath15 as the kernel of a map @xmath16 where @xmath17 is the subspace of degree @xmath18 elements in the free graded pre - lie algebra @xmath19 generated by @xmath10 and a degree @xmath18 ` dummy ' variable @xmath8 .
the map ( [ psano_v_srni ] ) is later in the paper identified with a very simple map between spaces of trees , see proposition [ abych_neonemocnel ] and corollary [ , ] .
theorem [ . ] and corollary [ , ] have immediate applications to the analysis of natural operations on the chevalley - eilenberg complex of a lie algebra . in a future work
we also plan to prove that theorem [ . ]
implies that the only natural multilinear operations on vector fields on smooth manifolds are , in stable dimensions , iterations of the standard jacobi bracket .
there is also a possible relation of the results of this paper with loday s theory of triplettes . in the rest of this introductory section ,
we discuss some of these applications and motivations in more detail .
[ mot ] _ motivations . _ in @xcite we studied , among other things , the differential graded ( dg- ) operad @xmath20 of natural operations on the chevalley - eilenberg complex of a lie algebra with coefficients in itself , along with its homotopy version @xmath21 , the operad of natural operations on the chevalley - eilenberg complex of an @xmath22-algebra (= strongly homotopy lie algebra , see @xcite ) .
we proposed : [ -/ ] describe the homotopy types , in the non - abelian derived category , of the dg - operads @xmath20 and @xmath21 of natural operations on the chevalley - eilenberg complex .
the following conjecture was proposed by d. tamarkin .
[ tamarkin ] the operad @xmath20 has the homotopy type of the operad @xmath23 for lie algebras .
it turns out that @xmath21 , which is tied to @xmath20 by the ` forgetful ' map @xmath24 , contains a dg - sub - operad @xmath25 generated by symmetric braces @xcite such that @xmath26 ( the sub - operad of degree @xmath27 elements ) is the operad @xmath28 governing pre - lie algebras .
moreover , both @xmath26 , @xmath29 and the differential @xmath30 have very explicit descriptions in terms of planar trees .
our conviction in conjecture [ tamarkin ] made us believe that the sub - operad @xmath31 of @xmath32 equals the operad @xmath23 , @xmath33 the main result of this paper , equivalent to isomorphism ( [ . , ] ) , is therefore a step towards a solution of problem [ -/ ] .
[ gen ] _ generalizations .
_ let us slightly reformulate the above reflections and indicate possible generalizations .
let @xmath34 be a quadratic koszul operad ( * ? ? ?
* section ii.3.3 ) and @xmath35 the dg - operad of natural operations on the complex defining the operadic cohomology of @xmath36 (= strongly homotopy @xmath34-algebras ( * ? ? ?
* definition ii.3.128 ) ) with coefficients in itself . in
@xcite we conjectured that @xmath37 for each quadratic koszul operad @xmath34 .
the operad @xmath38 has a suboperad @xmath39 generated by a restricted class of operations which generalize the braces on the hochschild cohomology complex of an associative algebra @xcite .
the operad @xmath40 of degree @xmath27 elements in @xmath39 always contains the operad @xmath23 for lie algebras that represents the intrinsic brackets .
the conjectural isomorphism ( [ oooo ] ) would therefore imply : [ jak_to_dopadne ] for each quadratic koszul operad @xmath34 , @xmath41 moving from operads to free algebras ( * ? ? ?
* section ii.1.4 ) , an affirmative solution of this conjecture for a particular operad @xmath34 would immediately give a characterization of lie elements in free @xmath40-algebras . from this point of view , the main result of this paper ( theorem [ . ] ) is a combination of a solution of conjecture [ jak_to_dopadne ] for @xmath42 with the identification of @xmath43 which expresses the equivalence between symmetric brace algebras and pre - lie algebras @xcite .
conjecture [ jak_to_dopadne ] holds also for @xmath44 , the operad for associative algebras , as we know from the deligne conjecture in the form proved in @xcite .
since @xmath45 is the operad for ( ordinary , non - symmetric ) braces @xcite , one can obtain a description of _ lie elements _ in _ free brace algebras_. _ loday s triplettes .
_ theorem [ .
] can also be viewed as an analog of the characterization of lie elements in the tensor algebra @xmath46 as primitives of the bialgebra @xmath47 with @xmath48 the shuffle diagonal ; we recall this classical result as theorem [ psano_v_aute ] of section 2 .
the bialgebra @xmath49 is associative , coassociative cocommutative and its primitives @xmath50 form a lie algebra . to formalize such situations , j .- l .
loday introduced in @xcite the notion of a _ triplette _ @xmath51 , abbreviated @xmath52 , consisting of operads @xmath53 and @xmath54 , ` spin ' relations @xmath55 between @xmath53-coalgebras and @xmath54-algebras defining @xmath56-bialgebras , an operad @xmath34 describing the algebraic structure of the primitives , and a forgetful functor @xmath57 , see definition [ jeste_ji_musim_napsat_sms ] in subsection [ trip ] .
the nature of associative , cocommutative coassociative bialgebras and their primitives is captured by the triplette @xmath58 .
the classical theorem [ psano_v_aute ] then follows from the fact that the triplette @xmath58 is _ good _ , in the sense which we also recall in subsection [ trip ] .
an interesting question is whether the case of lie elements in pre - lie algebras considered in this paper is governed by a good triplette in which @xmath59 and @xmath60 .
see subsection [ trip ] for more detail .
.3em * acknowledgments .
* i would like to express my thanks to f. chapoton , m. livernet , , c. lfwall , j. stasheff and d. tamarkin for many useful comments and suggestions .
i am also indebted to m. goze and e. remm for their hospitality during my visit of the university of mulhouse in the fall of 2004 when this work was initiated .
in this section we recall some classical results about lie elements in free associative algebras in a language suitable for the purposes of this paper .
let @xmath61 be the tensor algebra generated by a vector space @xmath10 , @xmath62 where @xmath63 is the @xmath64-th tensor power @xmath65 of the space @xmath10 .
let @xmath66 denote the space @xmath61 considered as a lie algebra with the commutator bracket @xmath67 : = x\ot y - y\ot x,\hskip 1em x , y \in \t(v),\ ] ] and let @xmath68 be the lie sub - algebra of @xmath66 generated by @xmath10 .
it is well - known that @xmath69 is ( isomorphic to ) the free lie algebra generated by @xmath10 ( * ? ? ?
* , theorem 2 ) .
there are several characterizations of the subspace @xmath70 @xcite .
let us recall the one which uses the _
shuffle diagonal _ @xmath71 given , for @xmath72 , by @xmath73 \ot [ v_{\sigma(i+1 ) } \ot \cdots \ot v_{\sigma(n)}],\ ] ] where @xmath74 denotes the set of all @xmath75-shuffles , i.e. permutations @xmath76 such that @xmath77 notice that , in the right hand side of ( [ jake_bude_pocasi ] ) , the symbol @xmath78 has two different meanings , the one inside the brackets denotes the tensor product in @xmath61 , the middle one the tensor product of two copies of @xmath61 . to avoid this ambiguity
, we denote the product in @xmath61 by the dot @xmath79 , ( [ jake_bude_pocasi ] ) will then read as @xmath80 \ot [ v_{\sigma(i+1 ) } \ccdot\cdots \ccdot v_{\sigma(n)}].\ ] ] the triple @xmath81 is a standard example of a unital counital associative coassociative cocommutative hopf algebra .
we will need also the _ augmentation ideal _
@xmath82 which equals @xmath83 minus the ground field , @xmath84 and the _ reduced _ diagonal @xmath85 defined as @xmath86 or , more explicitly , @xmath87
\ot [ v_{\sigma(i+1 ) } \ccdot\cdots \ccdot v_{\sigma(n)}],\ ] ] for @xmath88 and @xmath89 .
clearly @xmath90 .
the following theorem is classical @xcite .
[ psano_v_aute ] the subspace @xmath91 equals the subspace of primitive elements , @xmath92 the diagonal @xmath71 is a homomorphism of associative algebras , that is @xmath93 where the same @xmath79 denotes both the multiplication in @xmath61 in the left hand side and the induced multiplication of @xmath94 in the right hand side .
the reduced diagonal @xmath95 is , however , of a different nature : for each @xmath96 , @xmath97 the proof is a direct verification which we leave for the reader .
we are going to reformulate ( [ 1 ] ) using an action of @xmath98 on @xmath99 defined as follows .
for @xmath100 and @xmath101 , let @xmath102 where @xmath79 denotes , as before , the tensor multiplication in @xmath99 .
observe that , while @xmath103 @xmath104 , therefore the action ( [ 2 ] ) _ does not _
make @xmath99 a bimodule over the associative algebra @xmath105 . to understand the algebraic properties of the above action better , we need to recall the following important ( @xcite ) a _ pre - lie _ algebra is a vector space @xmath106 with a bilinear product @xmath107 such that the associator @xmath108 defined by @xmath109 is symmetric in the last two variables , @xmath110 .
explicitly , @xmath111 there is an obvious graded version of this definition .
pre - lie algebras are known also under different names , such as right - symmetric algebras , vinberg algebras , & c. pre - lie algebras are particular examples of _ lie - admissible _ algebras @xcite , which means that the object @xmath112)$ ] with @xmath113 $ ] the commutator of @xmath114 , is a lie algebra .
each associative algebra is clearly pre - lie . in the following proposition
, @xmath115 denotes the augmentation ideal @xmath98 of the associative algebra @xmath61 considered as a pre - lie algebra . formulas ( [ 2 ] ) define on @xmath116 a structure of a bimodule over the pre - lie algebra @xmath115 .
this means that @xmath117 and @xmath118 for each @xmath119 and @xmath120 .
in particular , @xmath116 is a module over the lie algebra @xmath121 . to prove the first equality ,
notice that @xmath122 the second one immediately follows from ( [ 3 ] ) .
using action ( [ 2 ] ) , rule ( [ 1 ] ) can be rewritten as @xmath123 where the symmetric bilinear form @xmath124 measures the deviation of @xmath125 from being a pre - lie algebra derivation in @xmath126 on the other hand , since @xmath127 is symmetric , @xmath125 is a derivation of the associated lie algebra @xmath121 , @xmath128 which implies that @xmath129 .
the following statement is completely obvious and we formulate it only to motivate proposition [ a ] of section 3 .
[ strasne_pocasi_trva ] the map @xmath130 is uniquely determined by the rule ( [ 4 ] ) together with the requirement that @xmath131 for @xmath132 .
observe that the reduced diagonal @xmath85 is the initial differential of the cobar construction @xmath133 of the coassociative coalgebra @xmath134 .
complex ( [ 1a ] ) calculates the cohomology @xmath135 of the shuffle coalgebra .
on the other hand , by the poincar - birkhoff - witt theorem , there is an isomorphism of coalgebras @xmath136,\nabla),\ ] ] where the polynomial ring @xmath137 $ ] in the right hand side is equipped with the standard cocommutative comultiplication @xmath138 .
dualizing the proof of the classical ( * ? ? ?
* theorem vii.2.2 ) , one obtains the isomorphism @xmath139,\nabla)}(\bfk,\bfk ) \cong \ext^*(\l(v))\ ] ] where @xmath140 denotes the exterior algebra functor .
we conclude that @xmath141
in this section we show that the results reviewed in section 2 translate to pre - lie algebras .
let @xmath142 denote the free pre - lie algebra generated by a vector space @xmath10 and let @xmath12 be the associated lie algebra .
the following proposition is proved in @xcite , but we will give a shorter and more direct proof , which was kindly suggested to us by m. livernet .
the subspace @xmath143 generated by @xmath10 is isomorphic to the free lie algebra on @xmath10 . * proof * ( due to m. livernet).1.8em let us denote in this proof by @xmath144 the lie subalgebra of @xmath12 generated by @xmath145 and by @xmath146 the lie subalgebra of @xmath147 generated by @xmath148 . the canonical map @xmath149 clearly induces a map @xmath150 which restricts to a lie algebra homomorphism @xmath151 .
let @xmath15 be , as before , the free lie algebra generated by @xmath10 . since @xmath144 is also generated by @xmath10 , the canonical map @xmath152 is an epimorphism . to prove that it is a monomorphism , observe that the composition @xmath153 coincides with the canonical map induced by the inclusion @xmath154 . since @xmath146 is isomorphic to the free lie algebra generated by @xmath10 (
* , theorem 2 ) , the composition @xmath155 is an isomorphism , therefore @xmath156 must be monic .
we conclude that the canonical map @xmath157 is an isomorphism , which finishes the proof .
consider the free graded pre - lie algebra @xmath158 generated by @xmath10 and one ` dummy ' variable @xmath8 placed in degree @xmath18 .
observe that @xmath159 where @xmath160 is the subset of @xmath11 spanned by monomials with exactly @xmath64 occurrences of the dummy variable @xmath8 .
we need to consider also the graded pre - lie algebra @xmath161 ( `` r '' for `` reduced '' ) defined as the quotient @xmath162 of the free pre - lie algebra @xmath158 by the ideal @xmath163 generated by @xmath164 .
the grading ( [ 5 ] ) clearly induces a grading of @xmath161 such that @xmath165 and @xmath166 , @xmath167 the following statement , in which @xmath168 is the associator ( [ dnes_jsem_objel_polednik ] ) , is an analog of proposition [ strasne_pocasi_trva ] . [ a ] there exists precisely one degree @xmath18 map @xmath169 such that @xmath170 for @xmath132 , @xmath171 and @xmath172 where @xmath173 for @xmath174 .
moreover , @xmath175 .
the uniqueness of the map @xmath176 with the properties stated in the proposition is clear . to prove that such a map exists , we show first that there exists a degree one map @xmath177 of graded free pre - lie algebras such that @xmath178 for @xmath179 , @xmath180 and @xmath181 where @xmath182 for @xmath183 .
let us verify that the above rule is compatible with the axiom @xmath184 of graded pre - lie algebras .
applying ( [ 0 ] ) twice , we obtain @xmath185 for arbitrary @xmath186 .
let us make a small digression and observe that the associator @xmath168 behaves as a hochschild cochain , that is @xmath187 it follows from the definition of the form @xmath188 and the above equation that the last three terms of ( [ 16a ] ) equal @xmath189 , therefore ( [ 16a ] ) can be rewritten as @xmath190 since the right hand side of the above equality is graded symmetric in @xmath191 and @xmath192 , we conclude that @xmath193 which implies the existence of @xmath194 with the properties stated above .
it is easy to verify , using ( [ 0 ] ) and the assumption @xmath180 , that @xmath195 and that @xmath196 for arbitrary @xmath183 . a simple induction on the number of generators based on ( [ k ] ) together with the rule ( [ 0 ] ) shows that @xmath197 preserves the ideal generated by @xmath164 .
an equally simple induction based on ( [ l ] ) and ( [ 0 ] ) shows that @xmath198 is a subspace of the same ideal .
we easily conclude from the above facts that @xmath197 induces a map @xmath199 required by the proposition .
let us remark that each pre - lie algebra @xmath200 determines a unique _ symmetric brace algebra _
@xmath201 with @xmath202 for @xmath203 @xcite .
the bilinear form @xmath188 in ( [ 0 ] ) then can be written as @xmath204 the complex @xmath205 should be viewed as an analog of the cobar construction ( [ 1a ] ) .
we will see in section 5 that it describes natural operations on the chevalley - eilenberg cohomology of a lie algebra .
the main result of this paper reads : [ .
] the subspace @xmath206 equals the kernel of the map @xmath207 , @xmath208 in section 4 we describe the spaces @xmath11 , @xmath17 and the map @xmath209 in terms of trees .
theorem [ .
] will be proved in section 6 .
we begin by recalling a tree description of free pre - lie algebras due to f. chapoton and m. livernet @xcite . by a _ tree _ we understand a finite connected simply connected graph without loops and multiple edges .
we will always assume that our trees are _ rooted _ which , by definition , means that one of the vertices , called the _
root _ , is marked and all edges are oriented , pointing to the root .
let us denote by @xmath210 the set of all trees with @xmath64 vertices numbered @xmath211 .
the symmetric group @xmath212 act on @xmath210 by relabeling the vertices .
we define @xmath213 where @xmath214 denotes the @xmath9-vector space spanned by @xmath210 with the induced @xmath212-action and where @xmath212 acts on @xmath215 by permuting the factors .
therefore @xmath216 is the set of trees with @xmath64 vertices decorated by elements of @xmath10 .
the set @xmath217 consists of a single tree .3em @xmath1 .3em with one vertex ( which is also the root ) and no edges , thus @xmath218 .
the set @xmath219 consists of labelled trees @xmath220 where @xmath221 denotes the root and @xmath222 .
this means that @xmath10-decorated trees from @xmath223 look as @xmath224 where @xmath225 , therefore @xmath226 .
similarly , @xmath227 , where @xmath228 denotes the second symmetric power of @xmath10 .
the corresponding decorated trees are @xmath229 for @xmath230 .
finally , @xmath231 with the summands corresponding to the decorated trees @xmath232 with @xmath233 .
( chapoton - livernet @xcite ) [ jsem_zvedav_jak_povezu_tu_vosu_v_desti ] let @xmath234 then there is a natural isomorphism @xmath235 the pre - lie multiplication in the left hand side of ( [ zitra_odjizdim_na_zavody ] ) translates to the vertex insertion of decorated trees in the right hand side , see @xcite for details .
the most efficient way to identify decorated trees with elements of free pre - lie algebras is to use the formalism of symmetric brace algebras @xcite .
the trees in ( [ snad_nenachladnu ] ) then represent the following elements of @xmath11 : @xmath236 using the same tree description @xcite of the free graded pre - lie algebra @xmath158 , one can easily get a natural isomorphism @xmath237 where @xmath238 is the set of all trees with @xmath64 vertices decorated by elements of @xmath10 and one vertex decorated by the dummy variable @xmath8 .
we call the vertex decorated by @xmath8 the _ special vertex_. clearly @xmath239 while @xmath240 with the corresponding decorated trees @xmath241 where @xmath242 . similarly , @xmath243 with the corresponding trees @xmath244 for @xmath225 . in the above pictures we always placed the root on the top
. some examples of decorated trees from @xmath238 , @xmath245 , can also be found in examples [ abych_jeste_neonemocnel ] , [ uz_ctvrty_neletovy_den_v_rade ] , [ mmm ] and [ yyy ] .
let us describe the map @xmath246 of theorem [ .
] in terms of decorated trees .
we say that an edge @xmath247 of a decorated tree @xmath248 is _ special _ if it is adjacent to the special vertex of @xmath249
. given such an edge @xmath247 , we define the quotient @xmath250 by contracting the special edge of @xmath249 into a vertex and decorating this vertex by the label of the ( unique ) endpoint of @xmath247 different from the special vertex . in the following examples
, the special edge will be marked by the double line .
[ abych_jeste_neonemocnel ] if @xmath251 is the tree @xmath252 @xmath230 , then .3em
@xmath253 let @xmath254 .
we call a couple @xmath255 , where @xmath256 and @xmath247 a special edge of @xmath249 , a _ blow - up _ of @xmath257 if @xmath258 and if the arity (= the number of incoming edges ) of the special vertex of @xmath249 is @xmath259 .
we denote by @xmath260 the set of all blow - ups of @xmath257 .
[ uz_ctvrty_neletovy_den_v_rade ] the set @xmath261 is empty .
the simplest nontrivial example of a blow - up is @xmath262 .2em where the double line denotes , as in example [ abych_jeste_neonemocnel ] , the special edge .
let us give two more examples where @xmath263 , @xmath264 and @xmath265 are elements of @xmath10 : @xmath266 the last thing we need is to introduce , for @xmath267 , the sign @xmath268 as @xmath269 finally , define the map @xmath270 by @xmath271 [ mmm ] in this example , @xmath272 and @xmath265 are arbitrary elements of @xmath10 .
we stick to our convention that the root is placed on the top .
let us give first some examples of the map @xmath273 that follow immediately from the calculations in example [ uz_ctvrty_neletovy_den_v_rade ] .
we keep the double lines indicating which edges has been blown - up : @xmath274 .5em let us give some more formulas , this time without indicating the blown - up edges : @xmath275 the proof of the following proposition is a direct verification based on the induction on the number of vertices and formula ( [ 0 ] ) .
[ abych_neonemocnel ] the diagram @xmath276{$\pl(v)$ } } \put(16,16){\makebox(0,0)[cc]{$\pl^1(v)$ } } \put(6,10){\makebox(0,0)[cc]{$\tr(v)$ } } \put(16,10){\makebox(0,0)[cc]{$\tr^1(v)$ } } \put(8,16){\vector(1,0){6 } } \put(8,10){\vector(1,0){6 } } \put(6,15){\vector(0,-1){4 } } \put(16,15){\vector(0,-1){4 } } \put(11,17){\makebox(0,0)[cc]{$d$ } } \put(11,11){\makebox(0,0)[cc]{$\delta$ } } \put(7,13){\makebox(0,0)[cc]{$\cong$ } } \put(17,13){\makebox(0,0)[cc]{$\cong$ } } \end{picture}\ ] ] in which the vertical maps are isomorphism ( [ zitra_odjizdim_na_zavody ] ) and ( [ zitra_odjizdim_na_zavody - uz_tam_sem ] ) , is commutative . [ , ] there is a natural isomorphism @xmath277 [ yyy ] it follows from the formulas given in example [ mmm ] that , for each @xmath230 , # 11em .7em # 1.7em .5em @xmath278 therefore the combination @xmath279 belongs to the kernel of @xmath280 .
it is easy to see that elements of this form in fact span this kernel and that the correspondence @xmath281 $ ] defines an isomorphism @xmath282 where @xmath283 denotes the subspace of elements of monomial length @xmath284 .
in this section we show how an object closely related to the cochain complex @xmath285 of ( [ koupali - jsme - se - na - hradistku ] ) , considered in proposition [ a ] , naturally acts on the chevalley - eilenberg complex of a lie algebra with coefficients in itself . for @xmath89 ,
let @xmath286 and let @xmath287 denote the subspace of the graded vector space @xmath288 spanned by monomials which contain each basic element @xmath289 exactly once .
more formally , given an @xmath64-tuple @xmath290 , consider the map @xmath291 defined by @xmath292 let us denote by the same symbol also the induced map @xmath293 . then @xmath294 the above description immediately implies that @xmath287 is a @xmath176-stable subspace of @xmath288 , therefore @xmath295 is a chain complex for each @xmath296 . clearly @xmath297 and @xmath298 .
observe that the above reduction does not erase any information , because @xmath299 can be reconstructed as @xmath300 let us explain how each @xmath301 determines an @xmath64-multilinear degree @xmath176 operation on the chevalley - eilenberg complex @xmath302 of a lie algebra @xmath303 with coefficients in itself @xcite .
we will use the standard identification ( * ? ? ?
* definition ii.3.99 ) @xmath304 where @xmath305 denotes the cofree conilpotent lie coalgebra @xcite cogenerated by the desuspension @xmath306 of the vector space @xmath303 .
let @xmath307 be the co - extension of the desuspended lie algebra bracket @xmath308\circ ( \uparrow \land \uparrow ) : \desusp l \land \desusp l \to \desusp l\ ] ] into a coderivation .
then @xmath309 and ( [ potim_se ] ) translates the chevalley - eilenberg differential @xmath310 into the commutator with @xmath311 .
the above construction can be easily generalized to the case when @xmath303 is an @xmath22-algebra , @xmath312 @xcite .
the structure operations @xmath313 assemble again into a coderivation @xmath307 with @xmath309 ( * ? ? ?
* theorem 2.3 ) , and ( [ potim_se ] ) can be taken for a definition of the ( chevalley - eilenberg ) cohomology of @xmath22-algebras with coefficients in itself .
the last fact we need to recall here is that @xmath314 is a natural pre - lie algebra , with the product @xmath114 defined as follows ( * ? ? ?
* section ii.3.9 ) .
let @xmath315 and denote by @xmath316 the corestriction of @xmath317 .
the pre - lie product @xmath318 is then defined as the coextension of the composition @xmath319 see ( * ? ? ?
* section ii.3.9 ) for details . by the freeness of the pre - lie algebra @xmath320 , each choice @xmath321 determines a unique pre - lie algebra homomorphism @xmath322 such that @xmath323 for each @xmath324 , and @xmath325 . because @xmath326 , the map @xmath327 induces a map of the quotient @xmath328 @xmath329 define finally @xmath330 by @xmath331 one can easily verify the following formula that relates the chevalley - eilenberg differential @xmath310 with the differential @xmath176 in @xmath287 : @xmath332 for each @xmath333 and @xmath334 .
[ to_jsem_zvedav_jestli_za_jitkou_pojedu ] the collection @xmath335 forms an operad in the category of dg - vector spaces .
formula ( [ 20 ] ) determines an action that makes @xmath336 a differential graded @xmath337-algebra .
consequently , the cohomology operad @xmath338 naturally acts on the chevalley - eilenberg cohomology @xmath339 of an arbitrary lie or @xmath22 algebra .
the symmetric group @xmath212 acts on @xmath287 by permuting the basis @xmath289 of @xmath340 .
the operadic composition , induced by the vertex insertion of decorated trees representing elements of @xmath341 , is constructed by exactly the same method as the one used in the proof of ( * ? ? ?
* proposition ii.1.27 ) .
the verification that @xmath342 defines an operadic action is easy .
let @xmath20 denote , as in section 1 , the dg - operad of natural operations on the chevalley - eilenberg complex of a lie algebra with coefficients in itself and @xmath21 an analog of this operad for @xmath22-algebras . because each lie algebra is also an @xmath22-algebra
, there exists an obvious ` forgetful ' homomorphism @xmath343 . by proposition
[ to_jsem_zvedav_jestli_za_jitkou_pojedu ] , formula ( [ 20 ] ) defines maps @xmath344 and @xmath345 .
the diagram @xmath346{$c$ } } \put(0.00,10.00){\makebox(0.00,0.00){$\rpl^*$ } } \put(120.00,10.00){\makebox(0.00,0.00){$\blie^*$ } } \put(120.00,70.00){\makebox(0.00,0.00){$\bshl^*$ } } \put(20.00,10.00){\vector(1,0){80.00 } } \put(120.00,60.00){\vector(0,-1){40.00 } } \put(20.00,20.00){\vector(2,1){80.00 } } \end{picture}}\ ] ] is clearly commutative and @xmath347 is in fact an _ inclusion _ of dg - operads , compare the remarks in subsection [ mot ] .
for the purposes of the proof of theorem [ . ] , it will be convenient to reduce the complex @xmath285 constructed in proposition [ a ] as follows .
since the construction of @xmath299 is functorial in @xmath10 , one may consider the map @xmath348 induced by the map @xmath349 from @xmath10 to the trivial vector space @xmath27 .
the kernel @xmath350 is clearly a subcomplex of @xmath299 . since @xmath351 the complexes @xmath352 and
@xmath299 differ only at the second term , and , under the isomorphism ( [ zitra_odjizdim_na_zavody - uz_tam_sem ] ) , @xmath353 it is also obvious that @xmath354 the central object of this section is the commutative diagram : @xmath355{$\pl(v)$ } } \put(8,11){\makebox(0,0)[cc]{$\bt(v)$ } } \put(20,17){\makebox(0,0)[cc]{$\overline{\rpl}^1(v)$ } } \put(20,11){\makebox(0,0)[cc]{$\bt(v ) \otimes \bt(v)$ } } \put(8,23){\makebox(0,0)[cc]{$\l(v)$ } } \put(8,16){\vector(0,-1){4 } } \put(20,16){\vector(0,-1){4 } } \put(0,-1 ) { \put(1,0 ) { \put(-.5,0){\bezier{50}(8,22)(8,22.25)(7.75,22.25 ) } \bezier{50}(7.25,22.25)(7,22.25)(7,22 ) } \put(8,22){\vector(0,-1){3 } } } \put(9.8,11){\vector(1,0){6.1 } } \put(10,17){\vector(1,0){7.5 } } \put(7,20){\makebox(0,0)[cc]{$i$ } } \put(7,14){\makebox(0,0)[cr]{$p$ } } \put(14,18){\makebox(0,0)[cc]{$d$ } } \put(14,12){\makebox(0,0)[cc]{$\bdelta$ } } \put(21,14){\makebox(0,0)[lc]{$p^1 $ } } \end{picture}\ ] ] 5em in which @xmath356 is the inclusion and @xmath357 the canonical map of pre - lie algebras induced by the inclusion @xmath358 .
the definition of @xmath359 will use the following simple facts .
.5em _ fact 1_. the graded pre - lie algebra structure of @xmath299 induces on @xmath360 a structure of a @xmath11-bimodule . _
fact 2_. with the structure above , @xmath361 is the free @xmath11-bimodule generated by the dummy @xmath8 . _
fact 3_. the @xmath362-action ( [ 2 ] ) makes @xmath94 a bimodule over the pre - lie algebra @xmath363 .
therefore @xmath364 is a @xmath11-bimodule , via the canonical map @xmath365 .
.5em the above facts imply that one can define a map @xmath366 by requiring that it is a @xmath11-bimodule homomorphism satisfying @xmath367 it is clear that this @xmath368 restricts to the requisite map @xmath369 . to prove that the bottom square of ( [ d ] ) commutes , we notice that both compositions @xmath370 and @xmath371 behave in the same way with respect to the pre - lie multiplication @xmath114 on @xmath11
indeed , for @xmath372 , by ( [ 4 ] ) @xmath373 similarly , by ( [ 0 ] ) and the definition of @xmath374 , @xmath375 it remains to verify that @xmath376 . by the definitions of @xmath374 , @xmath188 , @xmath377 and the @xmath362-action ( [ 2 ] )
, @xmath378 observe finally that @xmath379 for @xmath132 .
the commutativity @xmath380 of the bottom square of ( [ d ] ) then follows from the following lemma .
let @xmath381 be a symmetric linear map such that the expression @xmath382 is symmetric in @xmath383 and @xmath384 .
then there exists precisely one linear map @xmath385 such that * @xmath386 for each @xmath387 , and * @xmath388 for each @xmath132 .
the map @xmath389 is constructed by the induction on the monomial length of elements of @xmath11 , its uniqueness is obvious .
the symmetry of the form in ( [ jsem_zvedav_jestli_budou_pastviny_fungovat ] ) in @xmath383 and @xmath384 is necessary for the compatibility of the rule ( i ) with the axiom ( [ aby_to_nezkoncilo_prusvihem ] ) .
we claim that proposition [ . ]
follows from the following [ aby_mne_nerozbolel_zub ] in diagram ( [ d ] ) , * @xmath390 , * @xmath391 and * @xmath392 .
indeed , ( i ) implies that @xmath393 while ( ii ) and ( iii ) together imply that @xmath394 maps @xmath395 monomorphically to @xmath15 .
since all these spaces are graded of finite type and their maps preserve the gradings , one concludes that @xmath396 .
* proof of lemma [ aby_mne_nerozbolel_zub ] .
* 1.8em the symmetry of @xmath188 in ( [ 0 ] ) implies that @xmath176 is a derivation of the lie algebra @xmath12 associated to @xmath11 .
this fact , together with @xmath397 , readily implies that @xmath176 annihilates lie elements in @xmath11 , which is ( i ) .
our proof of ( ii ) relies on the tree language introduced in section 4 .
we will use the following terminology .
a decorated tree @xmath398 is _ linear _ if all its vertices are of arity @xmath399 .
such a tree @xmath257 is of the form @xmath400{$v_{i}$ } } \put(5.00,10.00){\makebox(0.00,0.00)[l]{$v_{i-1}$ } } \put(5.00,30.00){\makebox(0.00,0.00)[l]{$v_3 $ } } \put(5.00,40.00){\makebox(0.00,0.00)[l]{$v_2 $ } } \put(5.00,50.00){\makebox(0.00,0.00)[l]{$v_1 $ } } \put(0.00,20.00){\makebox(0.00,0.00){$\vdots$ } } \put(0.00,0.00){\makebox(0.00,0.00){$\bullet$ } } \put(0.00,10.00){\makebox(0.00,0.00){$\bullet$ } } \put(0.00,30.00){\makebox(0.00,0.00){$\bullet$ } } \put(0.00,40.00){\makebox(0.00,0.00){$\bullet$ } } \put(0.00,50.00){\makebox(0.00,0.00){$\root$ } } \put(0.00,10.00){\line(0,-1){10.00 } } \put(0.00,50.00){\line(0,-1){20.00 } } \end{picture}}\ ] ] with some @xmath401 , @xmath402 .
each non - linear tree @xmath398 necessarily looks as @xmath403{$v_{i}$ } } \put(25.00,30.00){\makebox(0.00,0.00)[l]{$v_{i-1}$ } } \put(25.00,50.00){\makebox(0.00,0.00)[l]{$v_3 $ } } \put(25.00,60.00){\makebox(0.00,0.00)[l]{$v_2 $ } } \put(25.00,70.00){\makebox(0.00,0.00)[l]{$v_1 $ } } \put(20.00,10.00){\makebox(0.00,0.00){$s$ } } \put(20.00,40.00){\makebox(0.00,0.00){$\vdots$ } } \put(20.00,20.00){\makebox(0.00,0.00){$\bullet$ } } \put(20.00,30.00){\makebox(0.00,0.00){$\bullet$ } } \put(20.00,50.00){\makebox(0.00,0.00){$\bullet$ } } \put(20.00,60.00){\makebox(0.00,0.00){$\bullet$ } } \put(20.00,70.00){\makebox(0.00,0.00){$\bullet$ } } \put(40.00,0.00){\line(-1,1){20.00 } } \put(0.00,0.00){\line(1,0){40.00 } } \put(20.00,20.00){\line(-1,-1){20.00 } } \put(20.00,30.00){\line(0,-1){10.00 } } \put(20.00,70.00){\line(0,-1){20.00 } } } \end{picture}}\ ] ] where @xmath249 is a tree whose root vertex @xmath404 has arity @xmath259 .
we say that such a decorated tree has _
tail of length @xmath405_. these notions translate to decorated trees from @xmath406 in the obvious manner .
we leave to the reader to verify that , under identification ( [ zitra_odjizdim_na_zavody ] ) , the map @xmath407 is described as @xmath408 therefore @xmath409 consists of linear combinations of non - linear trees . before going further , we need to inspect how the map @xmath410 of ( [ nunyk ] ) , which is the differential @xmath411 written in terms of trees , acts on non - linear trees . if @xmath257 is the decorated tree ( [ 19 ] ) , then it immediately follows from the definition ( [ nunyk ] ) of @xmath412 that @xmath413 where @xmath414 is the following decorated tree with tail of length @xmath415 @xmath416{$v_i$ } } \put(25.00,50.00){\makebox(0.00,0.00)[l]{$v_3 $ } } \put(25.00,60.00){\makebox(0.00,0.00)[l]{$v_2 $ } } \put(25.00,70.00){\makebox(0.00,0.00)[l]{$v_1 $ } } \put(20.00,10.00){\makebox(0.00,0.00){$s'$ } } \put(20.00,40.00){\makebox(0.00,0.00){$\vdots$ } } \put(20.00,20.00){\makebox(0.00,0.00){$\oh$ } } \put(20.00,30.00){\makebox(0.00,0.00){$\bullet$ } } \put(20.00,50.00){\makebox(0.00,0.00){$\bullet$ } } \put(20.00,60.00){\makebox(0.00,0.00){$\bullet$ } } \put(20.00,70.00){\makebox(0.00,0.00){$\bullet$ } } \put(40.00,0.00){\line(-1,1){19.00 } } \put(0.00,0.00){\line(1,0){40.00 } } \put(19.00,19.00){\line(-1,-1){19.00 } } \put(20.00,30.80){\line(0,-1){9.00 } } \put(20.00,70.00){\line(0,-1){20.00 } } \end{picture}}\ ] ] in which @xmath417 is the tree obtained from @xmath249 by replacing the root vertex decorated by @xmath404 by the special one .
the map @xmath418 given by @xmath419 is a _
monomorphism_. let @xmath420 be a linear combination of non - linear trees and assume @xmath421 .
we must prove that then @xmath422 .
assume @xmath423 and decompose @xmath424 , where @xmath425 is , for @xmath426 , a linear combination of decorated trees with tails of length @xmath405 , and @xmath427 .
by ( [ snad - to - konecne - dopisu ] ) , the only trees with tails of length @xmath428 in @xmath429 are those spanning @xmath430 , therefore @xmath431 implies @xmath432 which in turn implies that @xmath433 , because @xmath434 is monic .
this is a contradiction , therefore @xmath422 which proves ( ii ) . to verify ( iii ) , notice that , by the commutativity of the bottom square of ( [ 19 ] ) , @xmath435 while @xmath436 by theorem [ psano_v_aute ] .
this finishes the proof of the lemma .
[ trip ] * @xmath53 and @xmath54 are operads , * @xmath55 are ` spin ' relations intertwining @xmath53-co - operations and @xmath54-operations , so that @xmath56 determines a class of bialgebras , * the operad @xmath34 governs the algebra structure of the primitive part @xmath50 of @xmath56-bialgebras , and * @xmath389 is a forgetful functor functor from the category of @xmath54-algebras to the category of @xmath34-algebras such that the inclusion @xmath437 is a morphism of @xmath34-algebras , an example is @xmath438 , with @xmath55 the usual bialgebra relation recalled in ( [ to_jsem_zvedav_jesli_budou_pastviny_fungovat ] ) .
let @xmath342 be a left adjoint to @xmath389 .
a triplette in definition [ jeste_ji_musim_napsat_sms ] is _ good _
@xcite , if the following three conditions are equivalent : the classical theorem [ psano_v_aute ] in section 2 is a consequence of the goodness of the triplette @xmath438 mentioned above , because ( [ ] ) in this case says that @xmath443 .
other , in some cases very surprising , good triplettes can be found in @xcite .
the following problem was suggested by j .-
loday : as we remarked in subsection [ gen ] , the affirmative answer to the deligne conjecture given in @xcite implies that there exist a characterization of lie elements in brace algebras @xcite similar to our theorem [ . ] .
this suggests formulating the following version of problem [ kk ] in which @xmath445 is the operad for brace algebras . _
lie elements and cobar constructions . _ in section 2 we calculated the cohomology of the cobar construction ( [ 1a ] ) of the shuffle coalgebra and observed that @xmath447 is isomorphic to the free lie algebra @xmath15 . in our characterization of lie elements in pre - lie algebras , the role of ( [ 1a ] ) is played by complex ( [ za_chvili_tam_musim_volat_a_hrozne_se_mi_nechce ! ] ) .
this leads to the following problem , which may or may not be related to problem [ kk ] , as d. tamarkin recently informed us , methods proposed in an enlarged unfinished , unpublished version of @xcite may imply that the complex ( [ za_chvili_tam_musim_volat_a_hrozne_se_mi_nechce ! ] ) is acyclic in positive dimensions , as envisaged also by some conjectures formulated in @xcite . | we give a simple characterization of lie elements in free pre - lie algebras as elements of the kernel of a map between spaces of trees .
we explain how this result is related to natural operations on the chevalley - eilenberg complex of a lie algebra .
we also indicate a possible relation to loday s theory of triplettes .
[ theorem]-2 mm # 1c__ce_^#1(l;l ) # 1#2#3#1_#2, ,#1_#3 #
1#2#1^#2 |.2em.2em # 1#2#3#4 \ { ll # 1 , & + # 3 , & .
# 1#2 .5em ( 7.03,5)(0,9 ) ( 6.,13)(0,-1)3.97 ( 6.,13)(0,0)[cc]@xmath0 ( 6.,9)(0,0)[cc]@xmath1 ( 7.,13)(0,0)[lc]@xmath2 ( 7.,9)(0,0)[lc]@xmath3 # 1#2#3 = .95pt ( 10.00,40.00)(0.00,0.00 ) ( 10.00,0.00)(0.00,0.00)[l]@xmath4 ( 10.00,20.00)(0.00,0.00)[l]@xmath3 ( 10.00,40.00)(0.00,0.00)[l]@xmath2 ( 0.00,0.00)(0.00,0.00)@xmath1 ( 0.00,20.00)(0.00,0.00)@xmath1 ( 0.00,40.00)(0.00,0.00)@xmath0 ( 0.00,40.00)(0,-1)40.00 # 1#2#3 = 1.000000pt ( 40.00,40.00)(0.00,0.00 ) ( 40.00,10.00)(0.00,0.00)@xmath1 ( 0.00,10.00)(0.00,0.00)@xmath1 ( 20.00,30.00)(0.00,0.00)@xmath0 ( 20.00,40.00)(0.00,0.00)@xmath2 ( 40.00,0.00)(0.00,0.00)@xmath4 ( 0.00,0.00)(0.00,0.00)@xmath3 ( 20.00,30.00)(1,-1)20.00 ( 0.00,10.00)(1,1)20.00 # 1#2#3#4 = 1.000000pt ( 20.00,30.00)(0.00,30.00 ) ( 10.00,0.00)(0.00,0.00)[l]@xmath5 ( 10.00,20.00)(0.00,0.00)[l]@xmath4 ( 10.00,40.00)(0.00,0.00)[l]@xmath3 ( 10.00,60.00)(0.00,0.00)[l]@xmath2 ( 0.00,60.00)(0.00,0.00)@xmath0 ( 0.00,40.00)(0.00,0.00)@xmath1 ( 0.00,20.00)(0.00,0.00)@xmath1 ( 0.00,0.00)(0.00,0.00)@xmath1 ( 0.00,60.00)(0,-1)60.00 # 1#2#3#4 = 1.000000pt ( 50.00,30.00)(0.00,30.00 ) ( 20.00,50.00)(0.00,0.00)@xmath0 ( 40.00,10.00)(0.00,0.00)@xmath1 ( 40.00,30.00)(0.00,0.00)@xmath1 ( 0.00,30.00)(0.00,0.00)@xmath1 ( 40.00,0.00)(0.00,0.00)@xmath5 ( 50.00,30.00)(0.00,0.00)[l]@xmath4 ( 0.00,20.00)(0.00,0.00)[c]@xmath3 ( 20.00,60.00)(0.00,0.00)[c]@xmath2 ( 40.00,30.00)(0,-1)20.00 ( 20.00,50.00)(1,-1)20.00 ( 0.00,30.00)(1,1)20.00 # 1#2#3#4 = 1.000000pt ( 40.00,30.00)(0.00,30.00 ) ( 40.00,0.00)(0.00,0.00)@xmath5 ( 0.00,0.00)(0.00,0.00)@xmath4 ( 30.00,30.00)(0.00,0.00)[l]@xmath3 ( 20.00,60.00)(0.00,0.00)@xmath2 ( 40.00,10.00)(0.00,0.00)@xmath1 ( 0.00,10.00)(0.00,0.00)@xmath1 ( 20.00,30.00)(0.00,0.00)@xmath1 ( 20.00,50.00)(0.00,0.00)@xmath0 ( 20.00,30.00)(1,-1)20.00 ( 20.00,30.00)(-1,-1)20.00 ( 20.00,50.00)(0,-1)20.00 # 1#2#3#4 = 1.000000pt ( 40.00,20.00)(0.00,10.00 ) ( 40.00,10.00)(0.00,0.00)@xmath1 ( 20.00,10.00)(0.00,0.00)@xmath1 ( 0.00,10.00)(0.00,0.00)@xmath1 ( 20.00,30.00)(0.00,0.00)@xmath0 ( 20.00,40.00)(0.00,0.00)@xmath2 ( 0.00,0.00)(0.00,0.00)@xmath3 ( 20.00,0.00)(0.00,0.00)@xmath4 ( 40.00,0.00)(0.00,0.00)@xmath5 ( 20.00,30.00)(0,-1)20.00 ( 20.00,30.00)(1,-1)20.00 ( 0.00,10.00)(1,1)20.00 # 1#2#3#4 = 1.000000pt ( 10.00,20.00)(0.00,0.00 ) ( 10.00,0.00)(0.00,0.00)[l]@xmath5 ( 10.00,20.00)(0.00,0.00)[l]@xmath4 ( 0.00,0.00)(0.00,0.00)@xmath3 ( 0.00,20.00)(0.00,0.00)@xmath2 ( 0.00,18.10)(0,-1)16.10 # 1#2#3#4#5#6 = .95pt ( 10.00,44.00)(0.00,0.00 ) ( 10.00,0.00)(0.00,0.00)[l]@xmath6 ( 10.00,20.00)(0.00,0.00)[l]@xmath7 ( 10.00,40.00)(0.00,0.00)[l]@xmath5 ( 0.00,0.00)(0.00,0.00)@xmath4 ( 0.00,20.00)(0.00,0.00)@xmath3 ( 0.00,40.00)(0.00,0.00)@xmath2 ( 0.00,18.00)(0,-1)16.00 ( 0.00,38.00)(0,-1)16.00 # 1#2#3#4#5#6 = 1.000000pt ( 40.00,40.00)(0.00,0.00 ) ( 40.00,0.00)(0.00,0.00)@xmath6 ( 20.00,40.00)(0.00,0.00)@xmath7 ( 0.00,0.00)(0.00,0.00)@xmath5 ( 40.00,10.00)(0.00,0.00)@xmath4 ( 20.00,30.00)(0.00,0.00)@xmath3 ( 0.00,10.00)(0.00,0.00)@xmath2 ( 38.50,11.50)(-1,1)17.00 ( 1.50,11.50)(1,1)17.00 # 1#2#3#4#5#6 = 1.000000pt ( 50.00,20.00)(-5.00,20.00 ) ( 40.00,0.00)(0.00,0.00)@xmath6 ( 20.00,40.00)(0.00,0.00)@xmath7 ( 0.00,0.00)(0.00,0.00)@xmath5 ( 40.00,10.00)(0.00,0.00)@xmath4 ( 20.00,30.00)(0.00,0.00)@xmath3 ( 0.00,10.00)(0.00,0.00)@xmath2 ( 38.50,11.50)(-1,1)17 ( 0,11.50)(1,1)18.1 ( 1.5,10)(1,1)18.1 # 1#2#3 ( 6.5,2.5)(9.8,13.5 ) ( 15,12)(0,1)2 ( 11,14)(1,1)1.85 ( 15,14)(-1,1)1.85 ( 13,16)(0,0)[cc]@xmath8 ( 11,14)(0,0)[cc]@xmath1 ( 15,14)(0,0)[cc]@xmath1 ( 15,12)(0,0)[cc]@xmath1 ( 10,14)(0,0)[cc]@xmath2 ( 16,14)(0,0)[cc]@xmath3 ( 16,12)(0,0)[cc]@xmath4 # 1#2#3 ( 5.5,2.5)(-3,0 ) ( 0,2)(0,-1)1.8 ( -2,-2)(1,1)1.85 ( 2,-2)(-1,1)1.85 ( 0,2)(0,0)[cc]@xmath1 ( -2,-2)(0,0)[cc]@xmath1 ( 2,-2)(0,0)[cc]@xmath1 ( 0,0)(0,0)[cc]@xmath8 ( 0,3)(0,0)[cc]@xmath2 ( -2,-3)(0,0)[cc]@xmath3 ( 2,-3)(0,0)[cc]@xmath4 = 100000 |
the possibility of chaotic behavior of particle production in branching processes has recently been investigated @xcite with emphasis on the search for appropriate measures of chaoticity . unlike classical nonlinear dynamics where
the lyapunov exponent @xmath0 can be defined to characterize divergent distances between trajectories , no such conventional description of chaotic behavior of branching processes exist .
one of the measures found in @xcite is the exponent @xmath1 characterizing the exponential dependence of the normalized multiplicity variance on the average multiplicity ; another is the entropy index @xmath2 . with those two measures it is possible to distinguish the properties of particle production of non - abelian from abelian dynamics to the extent that one may regard the former as chaotic and the latter not .
there are two aspects of that investigation that need further extension and exploration .
the first is an obvious one : the pure gauge theory of gluons only in the perturbative qcd branching should be extended to include quarks .
the other is less obvious .
if the qcd dynamics is indeed chaotic , there is a question rooted in the conventional chaos theory that should be asked .
that is , at what value of the control parameter can one identify the onset of chaos ? in qcd
there is no externally controllable parameter with which one can tune the dynamics .
since the nonlinearity of the dynamics that gives rise to branching can be represented perturbatively as a self - coupling term whose strength is parametrized by the strong coupling constant @xmath3 , we may regard @xmath3 as the control parameter .
thus in order to investigate the nature of the onset of chaos , we shall reconsider the qcd branching problem by treating @xmath3 as fixed , and then vary it by hand to study the dependences of @xmath1 and @xmath2 on @xmath3 . in sec . 2
we treat the full qcd problem with quarks and gluons and with running @xmath4 . in sec .
3 the onset of chaos is studied by varying the fixed @xmath3 .
as in @xcite , we follow odorico s procedure @xcite to develop the algorithm for monte carlo simulation of parton shower in qcd .
we shall not repeat here the details of the algorithm already described in @xcite .
what is new now is the inclusion of quarks in the branching processes , and the consideration of quark jets in addition to the gluon jets .
the splitting functions are @xmath5 \quad , \label{1}\end{aligned}\ ] ] @xmath6 \quad , \label{2}\end{aligned}\ ] ] @xmath7 and the running coupling constant is @xmath8 we shall choose massless quark number @xmath9 , color number @xmath10 , and set the qcd scale at @xmath11 mev . the approximations made to treat the infrared and collinear divergences result in the sudakov form factors @xcite that have the following forms for the gluon and quark ( or antiquark ) vertices , respectively , @xmath12 \right\}\quad , \label{5}\end{aligned}\ ] ] @xmath13 where @xmath14 .
we take the end of branching to occur when @xmath15 . since a gluon can now go into a number of channels , the simulation of a branching into resolvable partons requires the designation of a specific final state . to that end
we calculate the ratio @xmath16dz}\quad , \label{7}\end{aligned}\ ] ] and compare it with a random number @xmath17 $ ] . in @xmath18 , then the @xmath19 channel is chosen .
because of the divergence of ( [ 1 ] ) at @xmath20 and @xmath21 , @xmath22 is very close to @xmath21 ; it is @xmath23 even for @xmath24 .
thus the self - reproduction of gluons is a far more dominant process than @xmath25 pair creation .
by this fact alone we do not expect the gluon jet to change too much by the inclusion of quarks .
however , the quark jets are very different from the gluon jets , as we shall show .
we first show in fig
. 1 our result on the distribution of maximum number of generations , @xmath26 , for both the gluon - initiated shower ( g jet ) and quark - initiated shower ( q jet ) at two different values of initial virtuality @xmath27 .
the distributions are obtained after @xmath28 samples are simulated .
there are less generations of branching in a q jet than in a g jet because @xmath29 has divergence only at @xmath30 ( soft gluon limit ) , while @xmath31 diverges at both @xmath20 and @xmath21 .
thus the highly reproductive gluons are generally produced with low momenta ( and consequently low virtualities ) , resulting in less multiplicities for the q jet .
for the purpose of studying multiplicity fluctuations at different generations of branching , it is meaningful to consider only the subset of all showers that have the same maximum generations , @xmath32 .
we shall choose @xmath32 to be at the peaks of @xmath26 .
more specifically , we set @xmath33 and @xmath34 for q and g jets , respectively , at @xmath35 , and @xmath36 and @xmath37 at @xmath38 . for each fixed @xmath39 there is a distribution of multiplicities @xmath40 at that @xmath41 . that distribution @xmath42 is shown in fig . 2 for the case of @xmath38 and for the values @xmath43 and @xmath34 .
it is clear that @xmath44 for g jet is greater than that for q jet for both @xmath41 values .
the fluctuations from @xmath44 are measured by the normalized variance @xmath45 in fig .
3 are shown @xmath46 vs @xmath44 for the various cases of jet type and @xmath47 .
the most striking revelation one sees in that figure is that the general features of the q and g jets do not differ by very much . focusing on @xmath38
we see that their @xmath46 values reach about the same peak value even though the corresponding @xmath44 are different for the q and g jets .
that small difference , however , accounts for a difference in the value of @xmath1 defined by @xmath48 straightline fits to the linear portions for the log - log plots for the @xmath38 case yield @xmath49 this difference in the values of @xmath1 distinguishes the q jet from the g jet .
it is a quantitative measure of the differences exemplified by the thin and thick solid lines in fig . 2 . the dotted lines in fig .
3 represent the results of pure gauge theory with @xmath50 .
note that they are indistinguishable from the new results for g jet with quarks included .
the above result may be regarded as the characteristics of the temporal behaviors of the branching processes , when @xmath44 is interpreted as an analogue of the time elapsed in a classical tragectory @xcite . for a more effective description of the fluctuations ,
we now examine the spatial patterns in phase space at the end of branching and study how they fluctuate from event to event . in terms of the final - state momentum fraction @xmath51 of a parton , where @xmath52 , the inclusive distribution @xmath53 is highly nonuniform in the interval @xmath54 . even in terms of the @xmath55 variable , where @xmath56 , the distributions appear gaussian , as shown in fig .
apart from the heights of @xmath57 it appears that there is little difference between the q and g jets .
however , those are only single - particle distributions . to see spatial patterns it is necessary to study the normalized factorial moments @xmath58 in small bins . to that end
we must first use a spatial variable @xmath59 in terms of which @xmath60 is uniform . adopting the cumulative variable @xcite defined by @xmath61 where @xmath62 are the extrema of the @xmath55 range in fig . 4 , we have @xmath63 constant for @xmath64 .
it is this unit interval in @xmath59 space that we partition into @xmath65 bins of width @xmath66 . for an event @xmath67 let @xmath68 be the multiplicity of partons in bin @xmath69 at the end of branching so that the factorial moment for that event is @xmath70 the spatial pattern for the @xmath67th event at resolution @xmath71
is then described by @xmath72 the fluctuation of @xmath73 from event to event is a dynamical property of qcd branching that results in the loss of information about the final state .
thus it is important to quantify the degree of that fluctuation , which we do by taking the moments of @xmath73 : @xmath74 where @xmath75 means averaging over all events .
the exponent @xmath76 can be any real number , not necessarily an integer .
since @xmath73 deviates more from @xmath21 when @xmath71 is smaller , we search for the power - law behavior in @xmath65 , i. e. , @xmath77 the entropy index is defined by @xcite @xmath78 the simulated data we use for this part of the analysis are not restricted , as was done for the study of @xmath46 where the maximum number of generations , @xmath32 , was chosen at the peak of @xmath26 . in the study of spatial patterns that change from event to event
, we consider all events whatever @xmath32 may be . after simulating @xmath28 events we calculate @xmath79 ,
the results for which are shown in fig . 5 for q and g jets at @xmath38 .
evidently , those moments are much larger for q jet than for g jet ; thus the former has much larger spatial fluctuations .
the entropy indices provide a more efficient way of describing that property , as shown by the plots in fig .
the values of @xmath2 for the g jet are not significantly different from the ones determined in @xcite , where @xmath50 .
but they are much smaller than @xmath2 for the q jet .
this difference is one of the important findings in this work .
it could not have been inferred by examining the single - particle distributions such as those shown in fig .
because of the differences between @xmath80 and @xmath81 , the evolution , the multiplicity , and the distribution of @xmath58 are all different for the q jet , as compared to the g jet .
if the branching dynamics for the g jet is chaotic , then it is even more so for the q jet .
as we have found in @xcite , the non - abelian qcd dynamics is different from the abelian @xmath82 model in that @xmath83 and @xmath2 are larger . whether or not it means that the former is indeed chaotic is not certain , since we lack a definitive criterion for chaoticity in problems where no trajectory can be defined unambiguously and where the number of degrees of freedom increases with evolution .
one way of investigating further the nature of possible chaotic behavior in qcd is to examine the onset of chaos . since @xmath3 controls the strength of the nonlinearity in the problem , it is reasonable to propose that @xmath3 be varied in order to find the threshold of the chaotic behavior
of course , it means that we must first fix @xmath3 and calculate the fluctuation properties of the parton showers , and then examine how those properties depend on @xmath3 . with @xmath3 not running ,
the sudakov form factors in ( [ 5 ] ) and ( [ 6 ] ) lose one degree of log each and become @xmath84 \right\ } \quad , \label{17}\end{aligned}\ ] ] @xmath85 apart from this change the simulation of parton showers is just as in sec . 2 .
we shall fix @xmath47 at @xmath86 , and let @xmath3 vary over the range @xmath87 to @xmath88 , which roughly covers the running range as the virtuality degrades from @xmath27 to @xmath89 . at the low end of @xmath3 ( near @xmath87 )
the corresponding @xmath90 value is unrealistic high , but is nevertheless considered here because of our interest in the problem at the very small @xmath3 . for fixed @xmath3
the distributions of @xmath32 have shapes that are similar to those in fig . 1 for running @xmath3
. the peak of @xmath26 , denoted by @xmath91 , increases with @xmath3 for both q and g jets , as shown in fig .
it is important to recognize that the number of generations in a branching process is very small at low @xmath3 .
that is because the survivability of a parton without emitting a resolvable gluon or creating a @xmath25 pair is nearly @xmath21 at very small @xmath3 , as is evident in ( [ 17 ] ) and ( [ 18 ] ) . that is quite unlike the running @xmath92 case , where @xmath92 always gets bigger as @xmath93 evolves toward @xmath94 .
the smallness of @xmath91 at small @xmath3 is particularly acute for the q jet , since when a gluon is emitted by a quark , the gluon momentum and virtuality are suppressed , thus depriving the q jet a rich source of parton generations .
this property of low multiplicity at low @xmath3 will be recalled later to explain other features to be uncovered . from the simulated events we calculate @xmath46 as before and find the dependence on @xmath95 as shown in fig .
8 . evidently , there is a universality in how @xmath46 depends on @xmath96 .
the increase of @xmath3 merely extends the range .
there is a bend in the rise of @xmath46 for g jet , but the rise seems to persist for the q jet , a phenomenon that we have already encountered in fig . 3 . from fig .
8 we gain no insight on the development of chaos , if indeed the temporal behavior is chaotic , since there is no value of @xmath3 that we can identify as the onset of chaos .
next , for the fluctuations of spatial patterns we show @xmath97 in figs . 9 and 10 for q and g jets , respectively .
the general features are similar to fig . 5 for running @xmath92 , but there are important differences .
it helps to focus on a fixed @xmath76 and examine the dependence on @xmath3 .
while the scaling behaviors for @xmath98 and @xmath99 are rather similar , in some cases even hard to distinguish , the behavior for @xmath100 stands out markedly different .
that is , @xmath101 is much larger at small @xmath3 than at larger @xmath3 .
this feature is most pronounced for g jet at @xmath102 , where the only curves significantly different from @xmath103 are for @xmath100 . comparing the vertical scales of the two figures
, one can also see that the q jet has larger magnitudes of log @xmath104 than the g jet .
these features are both related to the fact that the parton multiplicity is low for small @xmath3 and especially so for the q jet .
fluctuations are usually large when the averages are small . the characteristics of figs .
9 and 10 are represented in capsule form by the entropy indices @xmath2 . in fig .
11 we show @xmath2 for various values of @xmath3 , while in fig .
12 the dependences on @xmath3 are shown for fixed @xmath105 .
for @xmath106 , @xmath2 approaches independence on @xmath3 , but as @xmath3 decreases toward @xmath103 , @xmath2 increases sharply
. this behavior has not been anticipated , and is opposite to what is expected if @xmath3 is to play the role of a control parameter that can turn on chaotic behavior as it is increased from a small enough value . as we have noted earlier ,
the parton multiplicities are low , when @xmath3 is small , so their fluctuations relative to their averages are very large .
that is what @xmath2 measures .
since chaos implies unpredictability of the final state of a parton shower , our results indicate that a parton shower is more chaotic at low @xmath3 than at high @xmath3 .
thus our conclusion here is that @xmath3 can not be treated as a control parameter , useful for tuning out chaos .
perturbative qcd is such an intricate nonlinear dynamics that it can not be rendered approximately linear by decreasing the strength of the nonlinear term . without the nonlinear coupling
there is no evolution , no particle production , and no linear dynamics .
our study of the full qcd branching problem that has both quarks and gluons has revealed several properties of the parton showers . the gluon jet with quarks included is not very different from the case of pure gauge theory @xcite .
but the quark jet exhibits more fluctuations ( relative to the average ) than the gluon jet . the measures used to quantify the fluctuations are @xmath46 for the temporal behavior and @xmath2 for the spatial behavior .
they both possess features that are characertistic of chaotic behavior . in the fixed coupling case
we have found that the results on fluctuations are roughly the same as in running @xmath3 problem , provided that the fixed @xmath3 is not set at a very low value , outside the range of the running @xmath3 .
however , in attempting to learn about the onset of chaos we have treated @xmath3 as a control parameter and examined the parton showers with @xmath3 being allowed to approach a very small value .
the hope was to see how the chaotic behavior would grow as @xmath3 is increased .
what we have found is that the qcd branching processes are more chaotic at very low @xmath3 . unlike classical nonlinear dynamics
, the nonlinearity in the non - abelian gauge theory can not be turned off without causing the system to lose the mechanism for temporal evolution through branching .
a small amount of nonlinearity leads to large fluctuations relative to the mean number of partons produced , resulting in more unpredictable final states .
the measure by which these properties can be described is the entropy index @xmath2 .
we have seen how @xmath2 has emerged as a highly effective description of the degree of fluctuations of spatial patterns .
it is our opinion that the method of analysis employed here is not only suitable for the study of parton showers , but also for all problems that involve changing spatial patterns , including the conventional problems in classical chaos . 1
. distributions of the maximum number of generations in qcd branching processes for quark ( @xmath27 ) and gluon ( @xmath107 ) jets . 2 .
multiplicity distributions at @xmath43 and @xmath34 for quark and gluon jets .
normalized variance vs average multiplicity as @xmath41 is increased from @xmath21 to @xmath32 .
the dotted lines are for pure gauge theory without quarks .
single - particle distribution in @xmath108 .
log - log plots of @xmath104 vs @xmath65 for various values of @xmath76 and @xmath105 .
entropy indices @xmath2 for quark and gluon jets .
7 . the dependence of @xmath91 on the fixed coupling @xmath3 , where @xmath91 is the value of @xmath32 at the peak of the distribution @xmath26 .
@xmath46 vs @xmath44 for various values of fixed @xmath3
log - log plots of @xmath104 vs @xmath65 for quark jets only when @xmath3 is set at three possible values .
same as fig .
9 for gluon jets . 11 .
@xmath2 vs @xmath105 for various @xmath3 . 12 .
@xmath2 vs @xmath3 for various @xmath105 . | the branching processes in parton showers are studied in perturbative qcd for both quark and gluon jets .
the emphasis is on the nature of fluctuations of both the parton multiplicities and the spatial patterns of the final states .
effective measures of such fluctuations are calculated from the data obtained by monte carlo simulations .
the entropy indices are used to characterize chaoticity .
both running and fixed couplings are considered .
the fixed coupling case is used to study the onset of chaos .
implications of the results are discussed .
= -.5 in = 0.2 in oits-599 + april 1996 + .5 cm * fluctuations and entropy indices + of qcd parton showers * .75 cm * zhen cao and rudolph c. hwa * .5 cm institute of theoretical science and department of physics + university of oregon , eugene , or 97403 .2 cm |
the relationship between the long term dynamics of an endomorphism of a manifold and its long term effect on the algebraic topology of the manifold can depend on the smoothness of the endomorphism .
see @xcite for a discussion of this . here
we deal with a particular case of problem 3 of that paper .
let @xmath4 be the @xmath2-sphere , oriented in the standard fashion .
fix a continuous map @xmath5 of global degree @xmath2 .
that is , the map @xmath6 is multiplication by @xmath2 .
problem 3 asks : is the growth rate inequality @xmath7 true ? here
@xmath8 is the number distinct periodic points of @xmath1 having period @xmath9 , i.e. , the number of fixed points of @xmath10 .
if @xmath1 is merely continuous the answer is `` not necessarily . '' for
as observed in @xcite , a lipschitz counterexample is @xmath11 where @xmath4 is the riemann sphere . in polar
coordinates @xmath1 sends @xmath12 to @xmath13 .
the only periodic points are the north and south poles . on the other hand , if @xmath1 is a rational map , then the answer is `` yes . '' see proposition 1 of @xcite .
so the question becomes : does there exist an @xmath14 such that if @xmath1 is @xmath15 then the growth rate inequality holds ?
perhaps @xmath16 will suffice , or even @xmath17 . from @xcite
it follows that if @xmath1 is @xmath0 then it must have infinitely many distinct periodic points , but their growth rate remains unknown . as shown in @xcite , the topological entropy of @xmath1 is better understood : if @xmath1 is @xmath0 then it is at least @xmath18 .
this implies there are invariant probability measures with measure theoretic entropy @xmath19 .
thus , if @xmath1 is @xmath20 then the sum of the lyapunov exponents is at least @xmath19 .
we expect a version of katok s theorem @xcite about diffeomorphisms to be true for endomorphisms in the case that all the lyapunov exponents are different from zero .
in fact this is already proved in the case that all the lyapunov exponents are positive .
see @xcite .
the remaining case will be where one of the lyapunov exponents is zero . at the end of section [ s.examples ]
we give three examples of smooth endomorphisms of the @xmath2-sphere , two with topological entropy @xmath21 , the other with topological entropy @xmath22 .
all have one lyapunov exponent zero and are essentially @xmath23 .
the first is of degree zero and has only one periodic point .
the second and third have degree two , are unlike the map @xmath24 , but satisfy the growth rate inequality .
all three examples preserve the latitude foliation .
we hope this elementary result might give a clue as to how homology assumptions can intervene when there are zero lyapunov exponents and also families of invariant center manifolds replacing the circles of our foliation , but that will surely go way beyond what we can accomplish here .
we say that @xmath25 preserves the latitude foliation if it carries each latitude into another latitude or to one of the poles . it need not be a homeomorphism from one latitude to another .
we assume throughout that @xmath25 is a continuous , latitude preserving endomorphism of degree @xmath2 .
[ t.fixedpoints ] if @xmath1 is @xmath0 then @xmath10 has at least @xmath26 fixed points .
[ c.grf ] if @xmath1 is @xmath0 and @xmath27 is the number of fixed points of @xmath10 then the growth rate inequality @xmath28 holds .
this is immediate from the theorem , but see the remark at the end of section [ s.counting ] for a shorter proof of the corollary .
@xmath10 is the @xmath29 iterate of @xmath1 and the fixed points are geometrically distinct .
the @xmath0 assumption is used rarely in the proof , but without it the theorem fails : as noted above , the lipschitz endomorphism @xmath11 from @xcite has degree @xmath2 but only two periodic points .
[ l.poletopole ] a latitude preserving endomorphism sends a pole to a pole , not a latitude .
obvious from continuity of the endomorphism .
parametrize the latitudes by their height @xmath30 , @xmath31 .
( the sphere @xmath4 rests on the @xmath32-plane and has center @xmath33 . )
the southpole corresponds to @xmath34 and the northpole to @xmath35 .
if @xmath36 denotes the latitude of height @xmath30 then we define the * latitude map * @xmath37 \rightarrow [ 0,1]$ ] by @xmath38 it is continuous and lemma [ l.poletopole ] implies that @xmath39 .
the map @xmath1 fibers over @xmath40 .
orient each latitude circle in @xmath4 in a counterclockwise fashion as viewed from above the northpole of @xmath4 .
if @xmath41 then the * latitude degree * @xmath42 is the degree of the map @xmath43 .
since these latitude circles are oriented , @xmath42 is well defined and locally constant as a function of @xmath44 .
corresponding to the maximal open intervals @xmath45 on which @xmath42 is well defined and constant are open bands @xmath46 , @xmath47 we denote by @xmath48 the common latitude degree of @xmath1 on latitudes in @xmath49 .
lemma [ l.poletopole ] implies that the value of @xmath40 at the endpoints of @xmath45 is @xmath50 or @xmath51 .
[ p.globaldegree ] the global degree of @xmath1 is @xmath52 where @xmath53 and the sum is taken over the bands and their band intervals @xmath54 . by continuity of @xmath40
there are at most finitely many band intervals @xmath55 with @xmath56 , so the sum makes sense .
the degree of @xmath1 is independent of homotopy , so we can deform @xmath1 on each band @xmath49 in order that @xmath40 becomes linear on the band interval @xmath45 .
then we can homotop @xmath1 further to make the intervals on which @xmath40 is constant become points .
finally we can homotop @xmath1 on each latitude so that up to homothety , it sends @xmath57 to @xmath58 or @xmath59 .
( in the case @xmath60 we homotop @xmath1 so that on each latitude , up to homothety it is the constant map @xmath61 . )
the net effect is that by homotopy , we can assume the latitude map of @xmath1 is a sawtooth as shown in figure [ f.sawtooth ] , and the map on each latitude is the simplest possible . .
we imagine @xmath1 sending the first , second , fourth , and fifth band to @xmath4 with latitude degree @xmath62 , and sending the third band to @xmath4 with latitude degree zero . ]
take a regular value @xmath63 of @xmath1 near @xmath64 .
the global degree of @xmath1 is the number of pre - images of @xmath65 , counted with multiplicity .
there are no pre - images in bands with latitude degree zero , because those bands are sent to the half - longitude through @xmath66 . in a band @xmath49 with @xmath67 and latitude degree @xmath68
there are @xmath69 pre - images .
the same is true if @xmath70 and @xmath49 has latitude degree @xmath71 .
the other bands give pre - images with corresponding negative multiplicity , so the total number of pre - images , counted with multiplicity , is the sum @xmath72 as claimed . if @xmath73 we say that the corresponding band is * directed upward * or * ascends * , while if @xmath70 it is * directed downward * or * descends*. if @xmath74 , the band is * neutral*. [ l.directedbands ] there exist directed bands .
since @xmath1 is surjective , so is @xmath40 , and @xmath40 carries some minimal interval @xmath75 \subset [ 0,1]$ ] onto @xmath76 $ ] with @xmath77 and @xmath78 or vice versa .
the interval @xmath79 corresponds to a directed band .
[ l.invariantlatitude ] if the band @xmath49 is directed and @xmath80 then @xmath49 contains an @xmath1-invariant latitude .
@xmath81 is the pair of poles , each being fixed by @xmath1 .
since @xmath1 has latitude degree @xmath2 or better and is @xmath0 at the poles , its derivative there is zero .
for let @xmath82 be a pole .
the derivative of @xmath1 at @xmath82 is a linear map of the tangent space @xmath83 to itself .
infinitesimally it preserves the latitude foliation , so it is a scalar multiple of a rotation or reflection , @xmath84 .
but if @xmath85 then @xmath1 has latitude degree @xmath86 on @xmath49 , contrary to the hypothesis that @xmath80 .
thus , @xmath87 and the poles are sinks for @xmath1 , so the latitude map @xmath88 \rightarrow [ 0,1]$ ] has @xmath89 this gives a fixed point @xmath30 of @xmath40 with @xmath90 , and @xmath36 is invariant under @xmath1 .
see figure [ f.phifixespoles ] .
@xmath81 is the pair of poles , and @xmath1 interchanges them .
differentiability of @xmath1 is irrelevant .
see figure [ f.phifixespoles ] . .
the first corresponds to case 1 and the second to case 2 . ]
@xmath81 is a pole , fixed by @xmath1 , and a latitude .
the pole is a sink and the latitude is sent the other pole .
this gives a fixed point of @xmath40 as in case 1 .
@xmath81 is a pole and a latitude , and @xmath1 sends the pole to the opposite pole .
this gives a fixed point as in case 2 .
@xmath81 is two latitudes , say @xmath91 with heights @xmath92 .
then @xmath93 \rightarrow [ 0,1]$ ] sends @xmath94 $ ] over itself , so it has a fixed point @xmath95 .
the latitude @xmath36 is invariant by @xmath1 .
the hypothesis that @xmath1 is @xmath0 is used only in cases 1 and 3 .
[ p.2tothen ] if the band @xmath49 is directed and @xmath80 then @xmath10 has at least @xmath3 fixed points . by lemma [ l.invariantlatitude ]
there is an invariant latitude on which @xmath1 has degree @xmath96 with @xmath97 .
the @xmath29 iterate of such a map of the circle has at least @xmath98 fixed points .
[ l.degreeproduct ] if @xmath1 and @xmath99 preserve the latitude foliation then the latitude degree of @xmath100 is at most the product of their latitude degrees .
this is a standard fact about maps of the circle .
[ l.numberoffixedpoints ] let @xmath25 be a continuous latitude preserving surjection , not necessarily of degree @xmath2 .
if each of its directed bands @xmath49 has @xmath101 then @xmath1 has more than @xmath102 fixed points .
we count the directed bands as follows .
* @xmath103 is the number of ascending bands with @xmath104 .
* @xmath105 is the number of ascending bands with @xmath106 .
* @xmath107 is the number of descending bands with @xmath104 .
* @xmath69 is the number of descending bands with @xmath106 .
* @xmath108 is the number of directed bands with @xmath109 .
we think of the graph of @xmath40 as composed of `` legs '' that join @xmath76 \times 0 $ ] to @xmath76 \times 1 $ ] .
formally , they are arcs in the open square @xmath110 , and they correspond to the bands @xmath111 .
see figures [ f.polesswitch ] , [ f.polestosouthpole ] , and [ f.polesfixed ] .
ascending legs correspond to ascending bands , descending to descending .
we call a leg corresponding to a band with @xmath106 a * reverse - leg * , and we call a leg corresponding to a band with @xmath109 a * zero - leg*. there are @xmath112 reverse - legs and @xmath108 zero - legs . each intersection of the diagonal @xmath113 with a reverse leg produces two fixed points of @xmath1 , since such an intersection gives an @xmath1-invariant latitude @xmath114 , and @xmath115 reverses orientation .
each intersection of @xmath113 with a zero - leg produces at least one fixed point of @xmath1 , since such an intersection gives an @xmath1-invariant latitude @xmath114 , and @xmath115 has degree zero .
intersections of @xmath113 with other legs need not produce fixed points .
the rest of the proof is a counting argument in which there are three cases to consider , concerning how @xmath1 affects the poles , and twelve subcases concerning which legs @xmath113 crosses .
let @xmath82 be the number of fixed poles , and @xmath116 the number of fixed points of @xmath1 ( including the fixed poles ) . according to proposition [ p.globaldegree ] the degree of @xmath1 is @xmath117 .
we are trying to show that @xmath118 .
naively , we imagine @xmath113 crossing all the legs , and hence generating corresponding fixed points two for each ( b)-crossing , two for each ( d)-crossing , and one for each ( e)-crossing . thus we would hope @xmath119 which leads us to ask whether @xmath120 since @xmath121 .
this inequality is in fact true , but we need a stronger one because if @xmath113 fails to cross some legs of type ( b ) , ( d ) , or ( e ) then we would have over - estimated the fixed points .
we write @xmath122 where @xmath123 is the correction term due to @xmath113 missing legs of type ( b ) ,
( d ) , ( e ) .
thus , @xmath124 @xmath40 switches the poles .
then @xmath125 , @xmath126 , @xmath127 , @xmath128 is odd , and there are @xmath129 ascending legs .
in particular @xmath103 is at most the number of ascending legs , so @xmath130 .
the diagonal meets all the legs so we have @xmath131 .
this gives @xmath132 since @xmath133 .
thus @xmath134 in case 1 .
see figure [ f.polesswitch ] . when @xmath1 interchanges the poles .
the diagonal @xmath113 meets each leg .
the graph is simplified
each leg can actually intersect @xmath113 several times ; in the case shown , the graph must intersect @xmath113 five times .
also unshown are possible arcs in the graph that join the top or bottom of the square to itself .
they correspond to neutral bands ] @xmath1 sends both poles to the same pole , say the southpole .
then @xmath135 , @xmath136 , @xmath128 is even , and there are @xmath137 ascending legs .
in particular , @xmath138 . the diagonal crosses all the legs except possibly the first . see figure [ f.polestosouthpole ]
sends both poles to the southpole .
it is also simplified .
the diagonal meets each leg except perhaps the first . ]
the first leg is of type ( a )
. then @xmath113 crosses all the legs of type ( b ) , ( d ) , ( e ) , which implies @xmath131 .
this gives @xmath139 since @xmath133 .
the first leg is of type ( e ) .
then @xmath17 .
also , @xmath140 since there are @xmath137 ascending legs , one of which ( the first one ) is of type ( e ) , not of type ( a ) .
this gives @xmath141 since @xmath142 .
the first leg is of type ( b ) .
then @xmath16 , and as in the previous case , @xmath140 .
this gives @xmath143 since @xmath144 .
thus @xmath118 in case 2 .
@xmath1 fixes both poles
. then @xmath145 , @xmath146 , @xmath147 , @xmath128 is odd , and there are @xmath148 ascending legs .
in particular , @xmath149 . the diagonal crosses all legs except possibly the first and last .
see figure [ f.polesfixed ] . leaves the poles fixed .
it is also simplified .
the diagonal meets each leg except perhaps the first and last . ]
the first and last legs are of type ( a ) .
then @xmath113 crosses all the legs of type ( b ) , ( d ) , ( e ) , so @xmath131 .
this gives @xmath150 since @xmath142 .
the first leg is of type ( a ) and the last of type ( e ) .
then @xmath17 . also , since the first and last legs ascend , and since one of them ( the last one ) is of type ( e ) , not of type ( a ) , we have @xmath151 .
this gives @xmath152 since @xmath142 .
the first leg is of type ( a ) and the last of type ( b ) .
then @xmath16 and , as in the previous case , @xmath130 .
this gives @xmath153 since @xmath144 .
the first leg is of type ( e ) and the last of type ( a ) .
this is symmetric to case 3.ae .
the first and last legs are of type ( e ) . if @xmath154 then @xmath16 and @xmath155 since two ascending legs are not of type ( a ) .
this gives @xmath156 since @xmath142 . on the other hand , if @xmath157 then there is just one ascending leg , and it is of type ( e ) .
then @xmath17 and @xmath158 .
this gives @xmath159 the first leg is of type ( e ) and the last of type ( b ) .
then @xmath154 , @xmath160 , and @xmath161 .
this gives @xmath162 since @xmath144 .
the first leg is of type ( b ) and the last of type ( a ) .
this is symmetric to case 3.ab .
the first leg is of type ( b ) and the last of type ( e ) .
this is symmetric to case 3.eb .
the first and last legs are of type ( b ) .
if @xmath154 then @xmath163 and @xmath155 .
this gives @xmath164 since @xmath165 . on the other hand , if @xmath166 then there is just one ascending leg and it is of type ( b ) .
then @xmath16 , @xmath167 , and @xmath168 .
this gives @xmath169 thus @xmath118 in case 3 , which completes the proof of lemma [ l.numberoffixedpoints ] .
the degree of @xmath1 can be negative , in which case lemma [ l.numberoffixedpoints ] says nothing .
one can re - work the preceding estimates to show that @xmath170 but we do not need this .
we assume that the @xmath0 latitude preserving map @xmath25 has degree @xmath2 and claim that @xmath10 has at least @xmath3 fixed points . by lemma [ l.directedbands ] there
exist directed bands .
if there is a directed band @xmath49 with @xmath171 then the result follows from proposition [ p.2tothen ] .
if all the directed bands have @xmath101 then by lemma [ l.degreeproduct ] the same is true of @xmath10 . applying lemma [ l.numberoffixedpoints ] to @xmath10 ,
we conclude that @xmath10 has more than @xmath172 fixed points .
the proof of corollary [ c.grf ] does not require the full force of lemma [ l.numberoffixedpoints ] .
it suffices that @xmath173 where @xmath96 is an absolute constant .
for , as just observed , if all the bands for @xmath1 have @xmath101 then the same is true for @xmath10 . thus , @xmath174 .
since @xmath175 , we get @xmath176 from which corollary [ c.grf ] is immediate .
here is the caseless proof of this weaker inequality . since all the bands of @xmath1 have @xmath101 , the number of fixed points of @xmath1 is at least @xmath177 . for the graph of the latitude map
has at least @xmath178 legs of type ( e ) that cross the diagonal , and at least @xmath179 legs of type ( b ) or ( d ) that cross the diagonal .
the former produce one fixed point each , and the latter two fixed points each .
for it is only the first and last legs of the latitude map that can fail to intersect the diagonal .
this quantity minus the degree of @xmath1 is @xmath180 where @xmath128 is the number of legs .
that is , @xmath181 since @xmath182 and @xmath133 .
it is possible to code a latitude preserving map @xmath1 as follows .
take any finite sequence of integers @xmath183 such that @xmath184 is @xmath50 or @xmath51 , and make the following interpretation .
if @xmath185 and @xmath128 is even then consider directed bands @xmath186 such that @xmath187 on the other hand , if @xmath128 is odd then the last band ascends and has latitude degree @xmath188 .
similarly , if @xmath189 then every ascending band becomes descending and vice versa . the latitude degrees remain the same .
the choice @xmath185 indicates that @xmath1 fixes the southpole , while @xmath189 indicates that @xmath1 sends the southpole to the northpole .
since ascending and descending bands alternate , such a code is well defined for any latitude preserving map @xmath25 , and it describes @xmath1 well up to non - monotonicity of the latitude maps @xmath190 , the presence of neutral bands , and latitude rotation . * the map @xmath1 preserves the latitude foliation , and up to homothety it is @xmath194 or @xmath195 on each latitude . *
@xmath4 has @xmath128 bands @xmath196 with latitude degrees @xmath197 . *
if @xmath198 then @xmath1 sends @xmath196 diffeomorphically to the sphere minus the poles . *
if @xmath199 then @xmath1 sends @xmath196 to the prime meridian @xmath200 .
( @xmath200 is the longitude arc that joins the poles and contains the point @xmath201 . ) 1 .
the code is @xmath202 .
let @xmath1 be a good representative of the code .
then @xmath4 has two bands , the southern and northern hemispheres , minus the poles and equator . *
@xmath1 wraps the southern hemisphere upward over @xmath4 , pinching the equator to the northpole , and preserving the latitude orientation .
it fixes the southpole .
* @xmath1 wraps the northern hemisphere downward over @xmath4 , pinching the equator to the northpole , and preserving the latitude orientation .
it sends the northpole to the southpole . +
the map @xmath1 has degree zero , preserves the latitude foliation , and fixes the southpole .
its latitude map @xmath40 is unimodal , so its entropy is @xmath18 , and since @xmath1 fibers over @xmath40 with diffeomorphisms in the circle fibers , it too has entropy @xmath18 . now take a latitude - preserving rotation @xmath203 of the sphere by an angle @xmath204 where @xmath205 is irrational . set @xmath206 .
the entropy is unaffected and @xmath207 preserves the latitude foliation .
the only fixed point of @xmath208 is the southpole , for the effect of @xmath208 on any invariant latitude is an irrational rotation .
thus , the growth rate inequality holds for @xmath207 , in the form @xmath209 .
the code is @xmath210 .
let @xmath99 be a good representative of the code .
again , @xmath4 has the hemisphere bands .
* @xmath99 wraps the southern hemisphere upward over @xmath4 , pinching the equator to the northpole , and preserving the latitude orientation .
it fixes the southpole .
* @xmath99 wraps the northern hemisphere downward over @xmath4 , pinching the equator to the northpole , and reversing the latitude orientation .
it sends the northpole to the southpole . +
the map @xmath99 has degree @xmath2 and fixes the southpole .
again , let @xmath203 be an irrational rotation of the sphere and set @xmath211 .
the map @xmath212 preserves the latitude foliation .
its latitude map @xmath40 is unimodal , so its entropy is @xmath18 , and since @xmath1 fibers over @xmath40 with diffeomorphisms in the circle fibers , it too has entropy @xmath18 .
the map @xmath213 has @xmath3 fixed points .
each corresponds to a @xmath214-invariant latitude @xmath114 . on half of them
, @xmath214 preserves the latitude orientation , and on half of them @xmath214 reverses it . on the latitudes with preserved orientation
, @xmath214 is an irrational rotation and has no periodic points . on the latitudes with reversed orientation
@xmath214 has two fixed points .
altogether , @xmath214 has @xmath3 fixed points , so the growth rate inequality holds for @xmath212 , in the form @xmath215 .
the code is @xmath216 .
let @xmath30 be a good representative of the code .
then @xmath4 has three bands @xmath217 .
* @xmath30 wraps @xmath218 upward over @xmath4 , pinching its boundary latitude to the northpole , and preserving the latitude orientation .
it fixes the southpole .
* @xmath30 wraps @xmath219 downward along the prime meridian @xmath200 , pinching its lower boundary latitude to the northpole and its upper boundary latitude to the southpole .
the @xmath30-image of @xmath219 equals @xmath200 . * @xmath30 wraps @xmath220 upward over @xmath4 , pinching its boundary latitude to the southpole , and preserving the latitude orientation .
it fixes the northpole .
+ the map @xmath30 has degree @xmath2 and fixes both poles .
its latitude map @xmath40 is bimodal , so its entropy is @xmath22 , and since @xmath30 fibers over @xmath40 with diffeomorphisms in the circle fibers , it too has entropy @xmath22 .
take again an irrational rotation of the sphere , @xmath203 , and set @xmath221 .
the map @xmath213 has @xmath222 fixed points . for the majority of them ,
their @xmath40-orbits contain points in @xmath223 .
( that is , the proportion of the @xmath213-fixed - points whose orbit includes points of @xmath223 tends to @xmath51 as @xmath224 .
the other orbits lie in a zero measure cantor set . ) for each such fixed point of @xmath213 we have an @xmath225-invariant latitude @xmath114 , at least one of whose @xmath226-iterates lies in @xmath219 , and therefore @xmath227 is the single point @xmath228 .
the other fixed points of @xmath213 correspond to @xmath225-invariant latitudes whose @xmath226-orbits avoid @xmath219 . on these latitudes
, @xmath225 is an irrational rotation , so they give no fixed points .
altogether , @xmath225 has @xmath222 fixed points , nearly each of which contributes one fixed point of @xmath225 , so the growth rate inequality holds in the form @xmath229 .
it is possible that lemma [ l.numberoffixedpoints ] and our theorem have proofs using a coding like this .
we would need to know how the coding is affected when we take a banding @xmath230 of @xmath4 and consider a sub - banding of each @xmath196 as @xmath231 | for a @xmath0 degree two latitude preserving endomorphism @xmath1 of the @xmath2-sphere , we show that @xmath1 has @xmath3 periodic points . |
quantum correlations plays an important role for quantum protocols which outperform those of the classical one . the typical protocols are those proposed in the literature , such as the quantum cryptography @xcite , teleportation @xcite , and quantum computation @xcite , which depend crucially on the entanglement between the considered system .
while quantum entanglement has been recognized as crucial for various quantum information processing tasks , recent investigations show that it is in fact a special kind of quantum correlation measure , and there are other forms of quantum correlations which are more fundamental than that of entanglement @xcite .
the typical and seminal one are the concept of quantum discord proposed by ollivier and zurek @xcite , and henderson and vedral @xcite , which can exist even for the separable states , and is considered to be responsible for the power of the deterministic quantum computation with one qubit @xcite . due to these reasons and its fundamentals in quantum mechanics @xcite , great efforts has been devoted to this attractive field in the pat few years . the quantum discord , although is favored for its operational interpretation and possible application , is very hard to calculate @xcite . to avoid this problem , luo and fu @xcite introduced a geometric description of quantum correlation which they termed as measurement - induced nonlocality ( min ) . as its name implies ,
the min can also be considered as a measure of nonlocality for it is defined via the maximal global disturbance of the locally invariant measurements , and it is in some sense more general than that of the bell - type nonlocality . moreover
, the geometric min can also be evaluated analytically for the two - qubit states . besides the geometric min , an entropic measure of min based on the von neumann entropy was introduced recently @xcite , which is equivalent to that defined via the relative entropy @xcite . although its calculation is harder than that of the geometric one , it was favored for its definite physical meaning , which can be interpreted as the maximal increment of one s uncertainty about a system after his local invariant measurements on one subsystem . in this work ,
we study effects of the anisotropic parameter and the external magnetic field on the above two scenarios of min for a heisenberg xy chain .
the structure of this paper is arranged as follows . in section [ sec:2 ] , we recall the concept of the geometric and entropc min , and the model of the xy chain we considered . then in section [ sec:3 ] , we examine behaviors of the min in the xy chain with different system parameters .
we summarize the main finding of this work in section [ sec:4 ] .
here we first recall the concept of min . in its original geometric quantification scheme @xcite ,
it was defined by the square of the maximum of the schatten 2-norm between two states @xmath2 and @xmath3 , which corresponds to the bipartite states before and after the projective measurements . to be explicitly ,
we can write it as follows @xmath4 with the maximum being taken over the local projective measurements @xmath5 on subsystem @xmath6 , which satisfy the condition @xmath7 .
moreover , @xmath8 is the schatten 2-norm where @xmath9 . for the @xmath10 dimensional states , solutions of the geometric min can be derived analytically . here for convenience of representation , we do not list its explicit expressions , and the readers who are interested in it can resort to the ref .
@xcite for more detail .
different from @xmath11 which measures the min from a geometric perspective , the entropic measure of min is defined as follows @xcite @xmath12,\end{aligned}\ ] ] where @xmath13 denotes the quantum mutual information , and the minimum is taken over the same projective measurements as that in the definition of @xmath11 .
this min quantifies the maximal loss of total correlations under non - disturbing local measurements , and it is equivalent to @xmath14-s(\rho)$ ] because of the definition of @xmath15 which leaves @xmath16 invariant .
the evaluation of @xmath17 is also difficult due to the optimization processor involved .
but for the special case that the bipartite states @xmath2 having nondegenerate @xmath16 , the optimal @xmath15 for obtaining the entropic min correspond to that of the spectral resolutions of @xmath18 , and thus @xmath17 can be evaluated easily .
after review of the basic formalism for min , we present the model we considered in this paper .
this is the usual two - spin heisenberg xy chain with uniform external magnetic field applies along the @xmath19 direction .
the corresponding hamiltonian is given by @xcite @xmath20 where @xmath1 denotes the strength of the external magnetic field , and @xmath0 measures the anisotropy of the system which reduces to the xx model for @xmath21 , and the ising model for @xmath22 .
after a straightforward algebra , the eigenvalues and eigenvectors of @xmath23 can be derived analytically as @xmath24 and @xmath25 where @xmath26 , and the normalization parameter @xmath27 .
based on the above eigenvalues and eigenvectors , we can now evaluate the min in the xy chain , and for this purpose , we need to introduce the notion of the thermal state @xmath28 , with @xmath29 $ ] being the partition function , @xmath30 is the boltzmann constant , and @xmath31 denotes the temperature . when @xmath32 , @xmath2 reduces to the so - called ground state .
we now begin our discussion about behaviors of min in the xy chain . to be explicitly , we will evaluate influence of the anisotropic parameter @xmath0 , strength of the external magnetic field @xmath1 , and the temperature @xmath31 on min .
we first consider the scenario of the geometric min . in fig .
[ fig:1 ] , we gave three - dimensional plots of the geometric min @xmath11 as functions of @xmath1 and @xmath33 for various values of @xmath0 .
when considering the case of ground states at absolute zero temperature , one can see clearly that for the isotropic xy chain ( i.e. , @xmath21 , for which it is also termed as the xx chain ) , @xmath11 keeps the constant value of 0.5 before the critical point @xmath34 , and disappears suddenly when @xmath35 .
this sudden death behavior occurs only for the case of @xmath21 , and for finite @xmath0 as displayed in fig .
[ fig:1](b ) , @xmath11 maintains the constant value 0.5 before @xmath36 , and decreases exponentially to zero in the infinite limit of @xmath1 .
but when the anisotropic parameter @xmath0 exceeds a critical point @xmath37 , such as that displayed in fig .
[ fig:1](c ) with @xmath38 , @xmath11 initially keeps the maximum value 0.5 , and then decreases suddenly to a finite value @xmath39 at @xmath40 , after which it revivals and then decreases gradually to zero .
this is one of the main difference between @xmath11 in the regions of @xmath41 and @xmath42 , which implies that the anisotropic parameter @xmath0 can serve as an efficient parameter for tuning magnitudes of the min .
finally , we displayed in fig .
[ fig:1](d ) behaviors of @xmath11 for the special case of @xmath22 which corresponds to the ising chain .
different from those for @xmath43 , here @xmath11 decreases gradually from its maximum 1 to zero with increasing magnetic field @xmath1 . for thermal states with finite temperatures ,
one may expects that the min @xmath11 may be decreased by increasing the temperature , and this is indeed the case for weak magnetic fields .
but there is a peculiar phenomenon needs to be pay attention to . as can be observed from fig .
[ fig:1](a ) and ( b ) , the min may be increased by increasing temperature in the region of relative strong external magnetic field .
this shows again the combined effects of temperature and the anisotropic parameter on the geometric min . from the above discussion
, one can see that the geometric min @xmath11 is sensitive to the parameter @xmath0 and @xmath1 .
we now turn to investigate behaviors of the entropic min in the xy chain .
the exemplified plots are displayed in fig .
[ fig:2 ] , where the entropic min @xmath17 are plotted as functions of magnetic field @xmath1 and the scaled temperature @xmath33 with different anisotropic parameters @xmath0 . from these plots one can observe that for relative small values of @xmath0 , e.g. , the case of @xmath21 as shown in fig .
[ fig:2](a ) and @xmath44 as shown in fig .
[ fig:2](b ) , @xmath17 exhibit similar behaviors as those for @xmath11 with the same parameters [ see , fig .
[ fig:1](a ) and ( b ) ] , and the only difference is that the constant value for @xmath17 is twice that of @xmath11 . for large values of @xmath0
, however , @xmath17 and @xmath11 exhibit completely different behaviors , which is more evident in the low temperature region . as can be seen from fig . [ fig:2](c ) and ( d ) , @xmath17 is an non - increasing function of @xmath1 for any fixed @xmath33 , and there are no revival of @xmath17 , even at zero absolute temperature .
this is the first difference between geometric and entropic description of min in the heisenberg xy chain . besides this
, one can also note that the min @xmath17 can be increased by increasing temperature in the whole region of @xmath0 with appropriate chosen magnetic fields , and this constitutes another main difference between @xmath17 and @xmath11 in describing the nonlocal property of the heisenberg xy chain . before ending this section
, we would like to point out that here the different behaviors between @xmath17 and @xmath11 for the heisenberg xy chain is caused by the different metric adopted for defining the min .
this reveals the relativity of different quantum correlation measures in quantifying a specific quantum system @xcite , and the searching of an universal scheme for quantifying correlations remains an open problem needs to be solved in future .
to summarize , we have studied influence of the anisotropic parameter @xmath0 and strength of the external magnetic field @xmath1 on behaviors of the min . through detailed calculation and comparison
, we showed that both of them can serve as efficient parameters for controlling magnitudes of min in the heisenberg xy chain .
we also showed that the geometric measure of min and the entropic measure of min may correspond to completely different behaviors of nonlocal correlations , and we interpret this phenomenon as the different metric adopted in their respective definition .
as the min is a well - defined measure of nonlocal correlations , and investigations of its properties in different quantum systems will provide us with useful information which includes its controlling methods , its potential applications and other aspects related to the fundamental problem of quantum mechanics , we hope our work presented here may shed light on understanding min for the quantum spin system , and stimulate related investigations in this field .
this work was supported by natural science foundation of china under grant nos .
11174165 and 11275099 . 50 a. k. ekert , phys . rev . lett . * 67 * , 661 ( 1991 ) . c. h. bennett , g. brassard , c. crpeau , r. jozsa , a. peres , and w. k. wooters , phys .
lett . * 70 * , 1895 ( 1993 ) .
m. l. hu , phys .
a * 375 * , 922 ( 2011 ) .
m. l. hu , ann .
( ny ) * 327 * , 2332 ( 2012 ) . c. h. bennett and d. p. divincenzo , nature * 404 * , 247 ( 2000 ) . k. modi , h. cable , t. paterek , and v. vedral , rev .
phys . * 84 * , 1655 ( 2012 ) . h. ollivier and w. h. zurek , phys .
lett . * 88 * , 017901 ( 2001 ) .
l. henderson and v. vedral , j. phys .
a * 34 * , 6899 ( 2001 ) .
e. knill and r. laflamme , phys .
* 81 * , 5672 ( 1998 ) .
a. datta , a. shaji , and c. m. caves , phys .
* 100 * , 050502 ( 2008 ) .
b. p. lanyon , m. barbieri , m. p. almeida , and a. g. white , phys .
* 101 * , 200501 ( 2008 ) .
m. l. hu and h. fan , phys . rev . a * 87 * , 022314 ( 2013 ) .
m. gu , h. m. chrzanowski , s. m. assad , t. symul , k. modi , t. c. ralph , v. vedral , and p. k. lam , nat .
phys . * 8 * , 671 ( 2012 ) .
m. l. hu and h. fan , ann .
( ny ) * 327 * , 851 ( 2012 ) .
m. l. hu and h. fan , phys .
a * 88 * , 014105 ( 2013 ) .
d. girolami and g. adesso , phys .
a * 83 * , 052108 ( 2011 ) .
s. luo and s. fu , epl * 92 * , 20004 ( 2010 ) .
s. luo and s. fu .
measurement - induced nonlocality
. phys . rev .
206 , 120401 ( 2011 ) .
m. l. hu and h. fan , ann .
( ny ) * 327 * , 2343 ( 2012 ) .
z. xi , x. wang , and y. li , phys .
a * 85 * , 042325 ( 2012 ) . | quantum correlations are essential for quantum information processing .
measurement - induced nonlocality ( min ) which is defined based on the projective measurement is a good measure of quantum correlation , and is favored for its potential applications .
we investigate here behaviors of the geometric and entropic min in the two - qubit heisenberg xy chain , and reveal effects of the anisotropic parameter @xmath0 as well as the external magnetic field @xmath1 on strength of them .
our results show that both @xmath0 and @xmath1 can serve as efficient controlling parameters for tuning the min in the xy chain . |
.. it is well known that the creation of an open homogeneous universe requires either very special potentials @xcite or fine tuned initial conditions @xcite .
thus , for example , the early mechanism suggested by coleman and de luccia @xcite , requires a special class of inflaton effective potentials which have a metastable minimum followed by a small slope region which permits a further slow - roll inflation .
the inflaton field is supposed to be initially trapped in the false vacuum leading to a period of inflation , which gives an almost de sitter space with small quantum fluctuations .
the field eventually undergoes a quantum tunnelling , nucleating bubbles within which it slowly rolls down to the true vacuum .
the interior of such a bubble is actually an open universe . to keep the quantum fluctuations small ,
a very flat potential is required while one also requires a metastable minimum .
thus this scenario can only be realised at the cost of having very special potentials and fine tunings .
an interesting alternative mechanism was suggested recently by hawking and turok ( ht ) @xcite , which aims to remove some of these shortcomings .
the ht mechanism makes use of the no boundary proposal of hartle and hawking @xcite and provides a novel technique for creating an open inflationary universe described by a singular instanton obtained in a minisuperspace model with an inflaton field .
the ht instanton provides a method of calculating the probability of the creation of a homogeneous open universe while removing the requirement of a false vacuum and at the same time giving rise to a universe with small quantum fluctuations . despite its novelty
, several aspects of this mechanism have been criticised .
for example vilenkin @xcite has criticised the use of a singular instanton and has produced a counter - example , whereby a model with a minimally coupled massless scalar field which permits a singular instanton has been shown to lead to catastrophic results .
the proposal and this controversy have led to considerable work in this field .
gratton and turok @xcite have argued that singular instantons should not be ignored , as suggested by vilenkin , as the field configurations contributing to the path integral are in any case non - differentiable .
wu @xcite has pointed out that the singular instantons should be treated as constrained solutions which are not the stationary points of the gravitational action .
turok @xcite has further suggested that a careful treatment of these instantons may remove the instability pointed out by vilenkin .
unruh @xcite has shown that the instanton creates a finite bounded universe of which the homogeneous hyperbolic geometry is a part .
linde @xcite has suggested the introduction of a change of sign in the euclidean action for an instanton in quantum cosmology , giving the probability @xmath7 , different from what one obtains with the proposal of hartle and hawking .
bousso and hawking @xcite have shown that with linde s prescription the creation of a universe with large numbers of black holes is favoured resulting in a universe without a radiation dominated era .
further work on open inflation has also been done in the context of scalar - tensor gravity theories @xcite and in higher dimensions .
garriga @xcite has given a method for obtaining ht instantons in four dimensions by dimensional reduction of higher dimensional non - singular instantons . by including a cosmological constant in the theory
, he obtains a non - singular instanton , although eleven dimensional supergravity theory does not have a cosmological constant .
such a theory has , however , been shown @xcite to permit also a singular instanton .
given the potential importance of the ht scenario , it is of interest to study its robustness with respect to additional ingredients that are expected to be present . here as a step in this direction , we shall consider the effects of including higher derivative corrections to the gravitational action , on the existence of both non - singular de sitter and singular ht - type solutions .
quadratic and higher - order terms in the riemann curvature tensor and its traces appear in the low - energy limit of superstrings @xcite , as well as when the usual perturbation expansion is applied to general relativity @xcite .
for example , it is known that with suitable counter terms viz @xmath8 added to the einstein action one may obtain a perturbation theory which is well behaved , formally renormalizable and asymptotically free . some singular as well as non - singular instanton solutions
have recently been presented in the case of quadratic lagrangians @xcite .
the renormalisation of higher loop contributions introduces terms into the effective action that are higher than quadratic order .
consequently it is important to also study the effects of these additional terms . in this paper
we shall study the effects of including higher order terms , by looking at the effects of including @xmath9-contributions to the action . by employing the conformal equivalence of higher - order gravity theories with einstein gravity coupled to matter fields
, we argue that this term is prototypical of the higher order terms than quadratic , and in this sense represents a more general perturbation to the einstein
hilbert action than the @xmath10-correction , at least within the context of four - dimensional flrw space - times .
this allows us to find the range of parameters in each theory for which the singular ht as well as non - singular instantons exist in these more general settings .
the outline of the paper is as follows . in section 2
, we obtain the conditions for the existence of de sitter instanton solutions in higher derivative theories , in particular in quadratic and cubic theories .
section 3 contains an analogous study in the presence of a scalar field , using the equivalence picture , obtained by a conformal transformation of the higher order gravitational lagrangians .
included are also the conditions for the existence of both singular and the non - singular instantons permitted in the two theories , as well a comparative study of the two .
finally section 4 contains our conclusions .
we consider a theory described by the following euclidean action @xmath11 where @xmath12 is the determinant of the 4-dimensional metric , @xmath13 is a differentiable function of the scalar curvature @xmath14 , @xmath15 is the metric induced on @xmath16 , @xmath17 is the trace of the second fundamental form , and prime denotes differentiation with respect to the argument of the function .
here we shall mainly be concerned with the case where @xmath18 is given by the general polynomial function @xmath19 choosing @xmath20 , one obtains the quadratic lagrangian which , as was mentioned above , is known to have a number interesting features .
such quadratic terms in the riemann curvature tensor and its trace also appear in the gauss - bonnet combination , as well as in the low energy limit of the superstring theory .
if one only considers the conformally flat space - times in 4 dimensions , then in the quadratic theory the gauss - bonnet combination essentially reduces to a @xmath21 term .
we shall , however , also consider the general cases , the prototype of which is the cubic case . as in the work of hawking and turok @xcite
, we are interested in @xmath22 symmetric euclidean instanton solutions described by the metric @xmath23 where @xmath24 is the metric of 2-sphere . the scalar curvature for this metric is given by @xmath25 where a dot denotes a derivative with respect to @xmath26 .
considering @xmath27 and @xmath28 as independent variables we rewrite the action ( [ action0 ] ) , including the constraint given by ( [ curv - scalar ] ) , through a lagrange multiplier @xmath29 , in the form @xmath30\ ] ] @xmath31 the variation of the action ( [ action ] ) with respect to r gives @xmath32 substituting from ( [ beta ] ) in eq.([action ] ) we obtain @xmath33 d \sigma\ ] ] @xmath34_{\sigma = 0 } .\ ] ] the variation of the action ( [ action2 ] ) with respect to @xmath27 yields @xmath35 the de sitter instanton solution has the form @xmath36 where @xmath37 is a constant which upon using ( [ curv - scalar ] ) is found to be @xmath38 . for a de sitter instanton solution , equation ( [ b - eqn ] ) therefore reduces to an equation for @xmath39 in the form , @xmath40 which in turn shows that for a polynomial action of the form ( [ poly - action ] ) , a de sitter instanton will exist for any real solution @xmath41 of the equation @xmath42 the two cases of special interest for us are the quadratic and cubic lagrangians : + * i. the quadratic case * + considering the quadratic ( @xmath43 ) truncation of the lagrangian ( [ poly - action ] ) , in the form @xmath44 the condition ( [ con1 ] ) for the existence of a de sitter instanton solution becomes @xmath45 , which implies that in this case no such solutions exist for @xmath46 . + * ii . the cubic case * + considering the cubic truncation ( @xmath47 ) of the action ( [ poly - action ] ) , in the form @xmath48 eq .
( [ con1 ] ) implies that in this case a de sitter instanton will exist for any real positive root of the equation @xmath49 note that as opposed to the quadratic case , in this case de sitter instanton solutions can also exist for @xmath50 , and are given by @xmath51 .
note also that in both the quadratic and cubic cases , the existence conditions for the instanton solutions are independent of the parameter @xmath52 , the reason being that the quadratic term in each action satisfies the equation ( [ condition ] ) identically .
the corresponding action , however , does depend on it . + * iii . the general polynomial case *
+ the existence of de sitter instanton solutions in the the case of a higher order theory based on the general lagrangian ( [ poly - action ] ) depends on the algebraic equation ( [ condition ] ) having a real solution .
this raises two important questions .
firstly , given the existence of de sitter instanton solutions in a lower order lagrangian theory , what are the effects of switching on higher order terms ? secondly , given that as @xmath53 increases the number of roots of the equation ( [ condition ] ) also increase , what is the likelihood that such an equation has a real root and hence a de sitter instanton solution .
the former question relates to the question of stability or fragility of the de sitter instanton solution with respect to addition of higher order terms to the lagrangians @xcite . in this connection
it is clear that as @xmath54 , there is a real solution for every odd @xmath53 but not necessarily for an even @xmath53 .
more generally it is known that if the series expansions are convergent ( representing some analytic function say on @xmath55 ) and if @xmath56 , where @xmath57 , then for large @xmath53 , the number of real zeros of the infinite series , representing the function in @xmath58 is the same as the number of real zeros of any partial sum of size greater than @xmath53 , counting zeros by multiplicity @xcite .
regarding the latter , the results by kac @xcite concerning the probability of n - th order polynomials to have real roots demonstrates that this is in fact rare , thus demonstrating that such theories are likely to have only few de sitter instanton solutions ( see also @xcite , @xcite ) .
the above two points are important to bear in mind when considering such solutions in more general settings where higher order lagrangian terms are present .
+ now the primary criterion for deciding whether an instanton solution is physically favoured is to compute the euclidean action @xmath59 @xcite .
the wave function for the system to the leading approximation is then given by @xmath60 .
this allows the calculation of the probability of creation of a de sitter universe , described by these instantons .
the euclidean action , which is obtained by integrating over half of the @xmath61 is , for the general action ( [ action0 ] ) considered here , given by @xmath62 for the quadratic and cubic cases considered here , these actions take the explicit forms @xmath63,\ ] ] @xmath64\ ] ] respectively , where @xmath65 is given by @xmath66 and @xmath39 is a real positive root of the equation ( [ condition ] ) .
note that an increase in the values of @xmath67 and @xmath29 enhances the probability of the creation of the de sitter instanton .
although the solution ( 2.9 ) does not depend on @xmath67 , the action can not remain negative unless @xmath68 .
thus we obtain a lower bound on @xmath69 for the quadratic case . in the cubic case , if @xmath70 , @xmath71 , and hence a negative @xmath72 is not allowed in this case . since @xmath37 is independent of @xmath69 , it is obvious that a larger @xmath73 , will give greater probability of creation . to obtain an open inflationary universe ,
one considers the @xmath22 symmetric metric ( [ o4 ] ) and substitutes @xmath74 to obtain @xmath75 which is a spatially inhomogeneous de sitter like metric .
now setting @xmath76 and @xmath77 , the metric ( [ metric ] ) becomes @xmath78 with @xmath79 , which is a singular expanding open model .
+ our aim here is to find out the precise conditions under which the ht singular instantons @xcite are permitted in each case considered above .
we shall do this in the next section by employing a conformal transformation which converts both the quadratic and cubic lagrangian theories into einstein gravity with a minimally coupled , self interacting scalar field @xmath80 .
it is well known that the theories with higher order lagrangians are conformally equivalent to einstein gravity with a matter sector containing a minimally coupled , self interacting scalar field @xmath80 , with an effective potential @xmath81 @xcite .
the precise form of the self interaction of the scalar field is determined by the higher derivative terms in the action . for a general polynomial lagrangian of order @xmath82 of the form ( [ poly - action ] ) ,
the form of the potential is extremely complicated .
however , one can obtain the qualitative behavior of the potentials for all values of @xmath82 at small and large values of the field @xmath83 .
the asymptotic behavior of potentials at infinity depends on the combination of the highest degree of the lagrangian polynomial and the dimensionality @xmath84 of space - time . in particular , for @xmath85
the potential is unbounded from above , for @xmath86 it flattens into a plateau and for @xmath87 it has an exponentially decaying tail @xcite .
importantly , for @xmath88 , @xmath81 remains qualitatively the same as @xmath53 increases . as a result , in the 4 dimensional case
, the qualitative behavior of @xmath81 does not change relative to the @xmath89 case even when higher order terms with @xmath90 are considered .
this implies that the @xmath91 contribution is rather special in four dimensions , and that the @xmath9-term is in this sense more generic , being prototypical of higher order perturbations to the action .
thus , it is important to check if the results of the quadratic theory are robust with respect to the more general @xmath9 perturbations .
it is also worth mentioning in this connection the results of @xcite which show that in the case of the cubic theories , there exists a region of parameter space in which neither the tunnelling nor the hartle - hawking boundary conditions predict a suitable inflationary scenario , good enough to solve the horizon and flatness problems , contrary to what is obtained in the case of quadratic theories .
we note that another cubic term , @xmath92 , which may also be reduced to a theory of einstein gravity , minimally coupled to two scalar fields , after a suitable conformal transformation , is not considered here . as was mentioned above , the theory represented by the lagrangian @xmath18 can be written as a scalar field theory minimally coupled to einstein gravity .
more precisely , considering a conformal transformation of the form @xmath93 where @xmath94 , allows such a general action to be transformed into @xmath95,\ ] ] with @xmath96 $ ] . thus knowing the precise form of @xmath18 , allows the corresponding potential to be determined , in principle .
+ the einstein field equations corresponding to the action ( [ conf - action ] ) is derived using the equation @xmath97 and the scalar field equation is obtained from @xmath98 .
choosing @xmath99 to be given by the metric @xmath100 the corresponding field equations become @xmath101 = \left [ \frac{1}{2 } { \phi'}^{2 } - v ( \phi ) \right ] , \ ] ] @xmath102 where primes denote derivatives with respect to @xmath103 .
we now look for non - singular de sitter as well as singular ht instanton solutions of these equations .
the two cases of interest to us here are the theories with quadratic and cubic lagrangians .
* i. the quadratic case * + in this case the lagrangian takes the form @xmath104 , with the corresponding scalar field potential given by @xmath105^{2 } + \lambda e^{- 2 \sqrt{\frac{2}{3 } } \ ; \phi } .\ ] ] with a non - zero cosmological constant , the de sitter instanton solution exists for @xmath73 , @xmath106 as well as for @xmath107 , @xmath108 , and is given by @xmath109 @xmath110 the corresponding euclidean action evaluated in this case becomes @xmath111 which is the same as in the original quadratic theory .
considering the conditions for the singular ht instanton solutions , we note that such solutions can not be obtained for all values of the parameters @xmath67 and @xmath112 in the quadratic theory .
the existence conditions can be obtained by recalling the general features of the potential @xmath81 in this case ( with @xmath113 ) , namely : 0.2 in @xmath114 @xmath81 has two zeros , @xmath115 and @xmath116 , given by @xmath117 @xmath114 @xmath81 has a maximum at @xmath118 , with @xmath119 .
@xmath114 @xmath120 as @xmath121 .
0.2 in the choice of initial conditions play a dominant role in the evolution of these solutions .
let us assume that at @xmath122 , @xmath123 and @xmath124 .
it then follows that @xmath125 and @xmath126 .
since the point @xmath122 is a non - singular point , the manifold looks locally like @xmath127 in spherical polar coordinates and we may assume @xmath128 where @xmath129 is the initial velocity , i.e. , @xmath130 .
the potential has a negative gradient at @xmath131 .
the initial conditions along with the field eqs .
( [ a - eq ] ) and ( [ phi - eq ] ) determine the evolutions of @xmath27 and @xmath80 . to find conditions for the existence of singular ht in this case , we assume that close to the singularity @xmath132 , ( @xmath133 ) behave as @xmath134 @xmath135 eq .
( [ a - eq ] ) then determines @xmath136 for @xmath137 and eq.([phi - eq ] ) gives @xmath138.\ ] ] this in turn determines @xmath139 in terms of @xmath140 @xmath141 and hence results in the bound @xmath142 on the coupling constant of the quadratic term .
thus the conditions for the existence of a singular ht instanton ( a singularity at which @xmath143 vanishes as @xmath144 and @xmath80 diverges logarithmically ) are in this case given by : @xmath145 and @xmath146 .
there are therefore the following possibilities in this case : 0.2 in @xmath147 the scalar field may move uphill and become stabilised at @xmath148 , giving the @xmath149non - singular instanton solution .
@xmath150 the universe may end up in a singularity at @xmath151 , giving the singular ht @xmath149 instanton .
+ the important point here is that the existence of singular ht instantons in this case requires @xmath152 , which makes them rather special .
we also note that in this model , @xmath81 can not be neglected even close to the singularity , in contradiction to the case considered by @xcite . + * ii .
the cubic case * + in this case the lagrangian takes the form @xmath153 , with the corresponding scalar field potential given by @xmath154\ ] ] @xmath155 with a non - zero cosmological constant , there exist de sitter instanton solutions in this case of the form @xmath156 @xmath157 where @xmath158 is the solution of @xmath159 for the potential ( [ pot - cubic ] ) and @xmath37 is given by @xmath160 . in the case of @xmath46 , the euclidean action can be evaluated to be @xmath161 which is real and negative for @xmath162 .
+ we now look for consitions for the existence of singular ht instanton in the cubic theory .
we recall that the scalar field potential in this case has a maximum for @xmath73 and @xmath163 , but becomes flat at large @xmath80 for @xmath164 , as in the quadratic theory .
1 shows the potential @xmath81 ( [ pot - cubic ] ) for a number of values of @xmath29 with @xmath165 and @xmath70 .
given that in this case @xmath107 makes @xmath166 , we shall restrict ourselves to @xmath167 .
now in the case of the cubic theory with @xmath70 , one obtains ht instantons corresponding to the solution of the equations ( 3.28 ) and ( 3.29 ) with @xmath168 on the other hand , when @xmath169 , ht instantons are allowed if @xmath170 satisfies @xmath171.\ ] ] thus @xmath139 in this case is determined in terms of the dimensional constants @xmath69 and @xmath29 satisfying the constraint @xmath172 .
note that @xmath69 can assume some small negative values if @xmath169 .
the conditions for the existence of ht instantons in this case are : 0.2 in @xmath114 @xmath81 has a maximum .
@xmath114 @xmath173 .
@xmath174 @xmath139 satisfies the constraint given by ( 3.35 ) and could be positive , zero or negative , @xmath149 depending upon the values of @xmath69 and @xmath29 .
we have studied the existence of singular and non - singular instanton solutions in theories of gravity with quadratic and cubic lagrangians .
we find that in the quadratic case , the existence of both types of instantons requires a non
zero cosmological constant ; whereas in the cubic case instanton solutions can also exist with a vanishing cosmological constant . to study the existence of ht instantons and clarify the special features of the two types of solutions , expressed the higher order lagrangian theories as einstein gravity minimally coupled to a scalar field , using a conformal transformation of the metric .
in the case of the quadratic theory , the non - singular instanton corresponds to the scalar field sitting at the maximum of the potential .
this gives an indication that the lorentzian continuation of this solution ( with @xmath107 ) is stable , as is the case with the de sitter solution in the quadratic theory for @xmath107 @xcite .
the fact that ht instantons can not be obtained with @xmath73 shows that they are rather special in the quadratic theory .
there is also no ht instantons for @xmath175 in this case . in the case of the cubic theory
, ht instantons can exist for both @xmath107 and @xmath73 subject to a constraint which relates the cosmological constant to the dimensional constants @xmath69 and @xmath29 in the gravitational action .
finally , in view of our discussion of higher order polynomial lagrangians in section 2 , we would expect our general results regarding the behaviour of the ht instantons , obtained using the cubic theory , to hold in higher order lagrangian settings , although the algebraic relations among the dimensional constants , @xmath69 , @xmath29 _ etc . _ and @xmath170 would be different at different orders . _
acknowledgement : _ bcp and sm would like to thank iucca and iucca reference centre at north bengal university for the use of facilities .
rt would like to thank the iucca reference centre at north bengal university for warm hospitality . | we study the hawking - turok ( ht ) instanton solutions which have been employed to describe the creation of an open inflationary universe , in the context of higher derivative theories .
we consider the effects of adding quadratic and cubic terms of the forms @xmath0 and @xmath1 to the gravitational action .
using a conformal transformation to convert the higher derivative theories into theories of self interacting scalar fields minimally coupled to einstein gravity , we argue that the cubic term represents a generic perturbation of the polynomial type to the action and obtain the conditions on the parameters of these theories for the existence of singular and non - singular instanton solutions .
we find that , relative to the quadratic case , there are significant changes in the nature of the constraints on the parameters for the existence of these instantons , once cubic ( and higher order perturbations ) are added to the action . 8.5 in -30pt c # 1/#2 0_0 @xmath2 _ physics department , + p. o. north bengal university , + siliguri , dist .
: darjeeling , pin : 734 430 , india _
@xmath3 _ astronomy unit , + school of mathematical sciences , + queen mary , university of london , + mile end road , london , e1 4ns , england _
pacs numbers : 04.20.jb , 98.80.hw @xmath4electronic address : [email protected] @xmath5electronic address : [email protected] @xmath6electronic address : [email protected] |
lyman break galaxies ( lbgs ) form the main galaxy population observed in the rest - frame ultraviolet ( uv ) at high redshift .
lbgs are similar to local starbursts and it was suggested that these objects are forming stars at a high rate ( pettini et al .
2000 ; teplitz et al .
2000 ) . observed colors of lbgs are redder than dust - free star - forming objects .
however , the actual amount of dust in lbgs and the total star formation rate ( sfr ) is still insecure because high redshift lbgs are mainly undetected in the fir / sub - mm range and only small lbg samples are available ( chapman et al . 2000 ;
huang et al . 2005 ) .
the morphology of lbgs is also a matter of debate ( see giavalisco 2002 ) .
early works suggested that lbgs could be ellipsoidal and perhaps the progenitor of ellipticals or bulges of today galaxies .
the problem is that we can hardly detect low surface brightness areas above @xmath0 ( burgarella et al .
. however , papovich et al .
( 2005 ) found that the morphology of a nir - selected sample of galaxies at @xmath7 have regular and symmetric morphologies while a @xmath8 sample present irregular and/or compact morphologies .
observations in the sub - mm range introduced fir - bright galaxies which might be similar to ( ultra ) luminous ir galaxies ( ( u)lirgs ) ( blain et al . 1999 ) with @xmath9 for lirgs and @xmath10 for ulirgs that are likely to dominate the cosmic infrared background ( elbaz & cesarsky 2003 ) at high redshift .
the link between lbgs and lirgs is still an open question .
are they two different classes of objects ?
are they two facets of the same population ?
could we correct uv fluxes for the dust attenuation to recover the full star formation density ? in this paper , we combine the detection of true ( i.e. with a detected break ) lbgs at @xmath0 thanks to @xmath1 ( e.g. morrissey et al . 2005 ) with 24 @xmath4 @xmath11 ( rieke et al .
2004 ) data to estimate the total dust emission and the dust - to - uv flux ratio .
we also use goods images to analyse the morphology . a cosmology with @xmath12 , @xmath13 and @xmath14
is assumed .
to build the sample , we use @xmath1 data that are cross - correlated with combo-17 to get redshifts and @xmath11 to get @xmath15 fluxes .
we down - select objects that combo-17 ( wolf et al .
2004 ) clearly puts in the `` galaxy '' class .
we extract sources with redshifts @xmath16 .
we keep only objects down to galex nuv = 24.5 corresponding to galex 80 % completeness level ( 456 objects ) .
then , we select 344 fuv - dropout objects with @xmath17 .
the observed fuv and nuv filters are in the same rest - frame wavelength ranges than @xmath18 and @xmath19 and this selection is consistent with high redshift works ( e.g. giavalisco et al .
2002 for a review ) . of this list , 90 objects ( 23.3 % )
have a measured flux above the 80 % completeness limit of spitzer 24 @xmath4 data ( e.g. prez - gonzlez et al .
2005 ) and 264 ( 76.7 % ) only have upper limits .
this sample of 344 lbgs in the redshift range @xmath16 constitutes the database that we will study in this work .
this is the first lbg sample in this redshift range that can be used as a basis to which higher redshift lbgs can be compared .
1 shows two lbgs at several wavelengths including the two @xmath1 bands , the b and r eis bands and the @xmath20 @xmath11 band .
we find a large range of observed ( i.e. uncorrected for dust attenuations ) fuv luminosities with @xmath21 corresponding to @xmath22 . the luminosity range ( assuming @xmath23 ) is wider than either the @xmath0 or the @xmath24 one in adelberger & steidel ( 2000 ; a&s00 ) .
our lbgs reach luminosities as faint as a&s00 balmer break sample at @xmath0 but as bright as their lbg sample at @xmath24 .
total ir luminosities @xmath25 are estimated from @xmath26 ( i.e. @xmath27 a t @xmath7 from takeuchi et al .
2005 ) . in our lbg sample ,
20 - 30 % are uv luminous galaxies ( uvlgs ; heckman et al .
the ir luminosity range of the sample is about @xmath28 , which means that essentially all lbgs are compatible with a classification into lirgs .
an association between uvlgs and lirgs was already proposed by burgarella et al .
we confirm here that uv - luminous galaxies are also ir - luminous objects and extend it to lbgs .
the cdfs was observed by goods and we can deduce a morphological information in the rest - frame b band .
the advantage of observing lbgs at low redshift is obviously to have a better indication of their true morphology and not only the high surface brightness regions .
we used asymmetries and concentrations from lauger et al .
( 2005 ) to build fig . 2 . from @xmath29 observations
, it was found that lbgs at @xmath30 can have a wide dispersion of morphology but they generally appear to be smaller , irregular - like and compact ( papovich et al . 2005 ) .
however , we must be careful in the interpretation because of the cosmological dimming ( e.g. burgarella et al .
2001 but see papovich et al .
note , however , that in some rare deep spectroscopic observations ( mooorwood et al . 2000 ; pettini et al .
2001 ) , the profile of optical nebular lines suggests the presence of disks in some high - z lbgs .
3 leads us to the three following conclusions : i ) all but one lbgs are disk - dominated galaxies , ii ) one of them might be associated to bulge - dominated galaxies within the uncertainties and iii ) part of them ( 21 % ) are in the top part of the diagram , i.e. with an asymmetry larger than 0.25 and could be interpreted as mergers .
a number of studies have been devoted to the morphology of galaxy in the redshift range @xmath31 ( e.g. wolf et al .
2005 ; lauger et al .
2005 ; zheng et al .
2004 ; papovich et al .
the figure emerging from these papers is that most of the star formation at @xmath32 resides in disks and about half in spirals .
we find that @xmath33 % of the lbgs are likely mergers ( e.g. fig .
1b ) , @xmath34 % are disks ( e.g. fig .
1a ) and only @xmath35 % ( i.e. 1 galaxy ) is a spheroid .
given that almost all our lbgs can also be classified as lirgs from their ir luminosity , our classification is also consistent with the lirg morphology carried out by zheng et al .
( 2004 ) .
[ [ ultraviolet - dust - attenuations - and - implications - on - the - cosmic - star - formation - density ] ] ultraviolet dust attenuations and implications on the cosmic star formation density ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ so far , it was difficult to estimate the validity of dust attenuations of lbgs , because we had no clear idea of @xmath25 .
a&s00 tried to estimate the @xmath36 fluxes of their lbg sample from the @xmath37 method ( e.g. calzetti et al .
1994 ) and compared them to observations .
but , only the most extreme ones can be detected and used to safely estimate a calibration . in this paper
, we use total ir luminosities @xmath25 , and @xmath38 to compute the fir / uv ratio which is calibrated into fuv dust attenuation @xmath39 ( e.g. burgarella et al .
this method has been shown to provide better dust attenuations than the uv slope @xmath37 .
4a shows that the total ir luminosity correlates with the uv dust attenuation ( e.g. a&s00 ) .
on the other hand , we observe an apparent anti - correlation of @xmath39 with the uv luminosity which might be an observational bias .
it is very interesting to note that we do not detect lbgs with both a high uv luminosity and a high uv dust attenuation , and this can not be caused by observational limits .
in other words , we seem to observe a population of high-@xmath38 lbgs ( which qualify as uvlgs ) with dust attenuations similar to uv - selected galaxies in the local universe ( e.g. buat et al . 2005 ) . using a&s00 equations we estimate the ir luminosity ( fig .
globally , points estimated from the @xmath37 method are in the same region but shifted towards the top - right part of the diagram .
this method seems to overestimate @xmath25 ( i.e. the dust attenuation ) as compared to observed fir luminosities .
the @xmath37-based average dust attenuation estimated for our sample at @xmath0 is @xmath40 while the fir - to - uv average dust attenuation amounts to @xmath41 ( which is is fully consistent with takeuchi , buat & burgarella 2005 ) .
both of them are marginally consistent but the net effect is that total @xmath37-based luminosities , are larger than the actual values by a factor of about 2 in average and therefore provide higher sfrs .
this means that the influence of lbgs in the evolution of the star formation density might be overestimated , at least at @xmath0 and the cosmic star formation density estimated from lbgs should be decreased by a factor of 2 .
adelberger k.l . &
steidel c.c .
2000 , , 544 , 218 ( a&s00 ) blain a. w. , smail i. , ivison r.j .
, kneib j .-
1999 , , 302 , 632 buat v. et al .
2005 , , 619 , l51 burgarella d. et al .
2001 , , 369 , 421 burgarella d. et al .
2005 , , 360 , 1413 calzetti d. , kinney a.l . & storchi - bergmann t. 1994 , 429 , 582 chapman s.c .
2000 , , 319 , 318 elbaz d. & cesarsky c.j .
2003 , science 300 , 270 giavalisco m. 2002 , , 40 , 579 giavalisco m. , ferguson h.c . , koekemoer a.m. et al .
2004 600 , l93 heckman t.m .
2005 , 619 , l35 huang j .- s .
2005 , in press ( astro - ph/0507685 ) lauger s. et al .
2005 , submitted morrissey p. et al .
2005 . , 619 , l7 rieke g.h .
2004 154 , 25 moorwood a.f.m . , van der werf p.p .
, cuby j.g . , oliva e. 2000 362 , 9 prez - gonzlez , p.g .
2005 in press , ( astro - ph/0505101 ) pettini m. , steidel c.c .
, adelberger k.l . , dickinson m. , giavalisco m. 2000 528 , 96 pettini m. et al . , 2001 554 , 981 teplitz h.i .
et al . , 2000 , 533 , 65 takeuchi t.t . , buat v. , burgarella d. , 2005 in press ( astro - ph/0508124 ) wolf c. et al . , 2004 421 , 913 wolf c. et al . ,
2005 submitted ( astro - ph/0408289 ) zheng c. et al . , 2004 421 , 847 | the first large sample of lyman break galaxies ( lbgs ) at @xmath0 is selected from @xmath1 far - uv and near - uv images of the chandra deep field south .
this multi-@xmath2 study is based on the wealth of data available in this field : @xmath3 to estimate total ir luminosities and dust attenuations , redshifts from combo-17 , eis and goods to check the validity of dropout identifications and goods to derive the morphology of about 40 of our lbgs .
the advantages of this `` nearby '' lbg sample are that images reach fainter fluxes and surface brightnesses , allowing a deeper detection at all wavelengths and a better morphological analysis .
moreover , the 24 @xmath4 fluxes ( i.e. 12@xmath5 rest - frame ) permit a good estimation of dust attenuations and total ir luminosities .
the main results are that the vast majority of our lbgs are also luminous infrared galaxies ( lirgs ) .
the morphology of 75 % of our lbgs is consistent with a disk .
previous estimates of dust attenuations based on the ultraviolet slope are likely to be too large by a factor of @xmath6 2 , which implies that estimated star formation rates were too large by the same amount .
this sample in the @xmath0 universe provides us with a high quality reference sample of lbgs . |
the deformation properties of bulk metallic glasses are characterised by two temperature regimes @xcite . at high temperatures , close to the glass transition temperature ,
bmgs deform homogeneously with the strain rate properties being reasonably well understood by thermally activated plastic flow . at lower temperatures ,
plasticity becomes heterogeneous and highly localized to shear - bands , resulting in limited ductility in tension and a few percent plastic strain prior to failure for some bmgs in compression . in this regime of temperatures both athermal @xcite and thermal @xcite theories of plasticity have been proposed .
in the early work of turnbull and cohen @xcite these different regimes emerged as a result of two competing processes , one diffusional and one stress induced .
later , spaepen @xcite attributed the homogeneous regime as resulting from a steady state of free volume creation and diffusional related free volume annihilation .
heterogeneous flow on the other hand was dominated by free volume creation since annihilation became diffusion limited at low temperatures . that two distinct temperature scales exist
was also recognised by argon @xcite , who from the perspective of thermal activation developed two microscopic theories where at high temperature , plasticity was mediated by small atomic reconfigurations that resulted in a localized shear strain event , and at lower temperature , a less local structural excitation that was akin to the nucleation of a dislocation loop occurs .
such mechanisms could give insight into the observed high temperature homogeneous plasticity and the low temperature heterogeneous plasticity .
some time after this , high - strain rate atomistic simulation revealed the generality of such local structural excitations , where the transition from homogeneous to heterogeneous plasticity ( as a function of decreasing temperature or increasing strain rate ) emerged from the degree of correlation between such localized activity @xcite . due to their high strain rates and stresses , both dynamic and static atomistic simulations
primarily probe the athermal region of plasticity . out of the earliest of this work by falk and
langer @xcite arose the concept of the shear transformation zone ( stz ) which was assumed to exist in regions of the material not so well relaxed the so - called liquid like regions @xcite .
the corresponding structural transformations were assumed to be activated _ athermally _ once a critical stress had been reached .
such ideas developed into what is now called the effective temperature theories in which the quenched disorder well below the glass transition temperature is characterised by an effective temperature that describes the structural fluctuations within the material , see for example @xcite .
this approach has been quite successful in developing a quantitative understanding of experimental low temperature deformation data .
it was recognised by bouchbinder , langer and falk @xcite that if one were to assume a thermal activation picture within the stz framework @xcite , the density of such local structural excitations would have to be `` improbably '' large if the corresponding activation energies are in the realistic range of hundreds of kilojoule per mole ( or electron volts per excitation ) . in going to an athermal picture ,
this difficulty was avoided and indeed stzs could now occur at quite dilute densities @xcite , a picture that is compatible with the distinct liquid like regions assumed to exist within the glass .
understanding the nature of such stzs , or more generally localized structural excitations , has also been influenced by the physics of the under - cooled liquid regime , in which two distinct relaxation time - scales are evidenced : that of the slow @xmath0-relaxation and the fast @xmath1-relaxation processes first introduced by goldstein @xcite . at temperatures near the glass transition and below
, the @xmath0-relaxation processes rapidly increase their time - scale with the traditional viewpoint being that they freeze out of the structural dynamics in the amorphous solid regime @xcite . despite such an assumed freezing ,
the terminology of @xmath0- and @xmath1-relaxation has also been applied to the amorphous solid regime where the microscopic @xmath1 processes mediate escape from the mega - basins of the elastic energy landscape of the @xmath0 processes @xcite .
the present work also exploits this terminology and uses an additional fundamental feature of the under - cooled liquid regime , that the number of structural transformations available will scale exponentially with respect to a fundamental volume measure @xcite .
such ideas have been used in the early thermodynamic theories of adam and gibbs @xcite and kirkpatrick _ et al _ @xcite see also the review article by heuer @xcite and provides a sufficient number of structural transformations to justify a thermal activation picture for temperatures that are well below the glass transition .
in particular the present work develops further the ideas of refs .
@xcite . in these works ,
a thermal activation picture was developed in which the characteristic plastic rate with respect to an internal heterogeneous volume scale was determined from a distribution of @xmath0-relaxation barriers whose number scales exponentially with the size of the heterogeneous volume and whose statistics is described by a distribution with extensive first and second cumulants . by doing this , the average time scale associated with plasticity at a particular temperature becomes intimately connected to the @xmath0-relaxation potential energy landscape ( mediated by @xmath1 relaxation processes ) which is thermally accessible at that temperature . at zero loading or
any given shear stress , two critical temperature scales emerge : 1 ) the plastic transition temperature which indicates a significant drop in the amount of plastic activity as the temperature is reduced , and 2 ) the ( lower ) kinetic freezing temperature where ( on average ) a single structural transformation becomes available per heterogeneous volume .
the transition between the higher temperature regime of 1 ) and the lower temperature region of 2 ) corresponds to a change of statistics , from that of the most probable to that of the extreme .
importantly , before the critical freezing temperature is reached , there will exist a low temperature regime ( below the glass transition ) where a sufficient number of thermally activated events exist to mediate a non - negligible macroscopic plasticity .
thus , in ref .
@xcite , the low temperature heterogeneous and high temperature homogeneous deformation regimes are distinguished by a transition in the nature of the statistics of the @xmath0-relaxation energy landscape . put in other words ,
the freezing out of @xmath0-relaxation processes is only complete at temperatures far below that of the glass transition temperature , and until this extreme regime is reached thermal activation remains the underlying phenomenon controlling macroscopic plasticity a fact that has been established experimentally for a variety of deformation geometries @xcite including most recently that of shear band nucleation , propagation and arrest @xcite .
the present work extends on the work done in ref .
@xcite ( which will be referred to as paper i ) by using a log - normal distribution to describe the statistics of the @xmath0-relaxation barrier energy landscape to develop a mean - field description of thermally activated plasticity for the high temperature / low strain rate regime of homogeneous deformation .
[ secsummary ] gives a brief overview of the current theory , secs .
[ seclnd]-[sectheory ] develops the theory for the positive valued log - normal distribution of barrier energies , sec .
[ secmfp ] develops a mean field description of plasticity suitable for the high temperature homogeneous regime of deformation , and sec .
[ secexp ] applies the theory to well known experimental data for the well studied bmg vitroley-1 . in the discussion and conclusion , sec .
[ secconc ] , two results are summarized : 1 ) when comparing with the usual arrhenius description of high temperature homogeneous plasticity , the corresponding activation volume parameter is found to be equivalent to the variance of the internal slipped volume distribution , and 2 ) the fragility of the material is a measure of the number of structural transformations available at the glass transition temperature .
paper i considered a characteristic time scale of irreversible structural transformation activity , @xmath2 , whose inverse is seen as a plastic transition rate below the glass transition temperature .
studying the resulting temperature and stress dependence of @xmath3^{-1}$ ] gave insight into the transition from elasticity to micro - plasticity .
to determine @xmath3^{-1}$ ] it was recognised that , at a large enough length scale , bmgs are considered structurally homogeneous indicating that a sufficient amount of self - averaging occurs with respect to a shorter heterogeneous length scale .
thus two length - scales naturally emerge which are realised through the homogeneous and heterogeneous volume elements .
the plastic rate for a homogeneous representative volume element ( rve ) of volume @xmath4 is @xmath5^{-1}(t)=\sum_{n=1}^{n'}\left[\tau_{\mathrm{p , n}}\right]^{-1}(t)=n'\times\left[\tau_{\mathrm{p}}\right]^{-1}(t)\ ] ] in which @xmath6^{-1}(t)=\frac{1}{n'}\sum_{n=1}^{n'}\left[\tau_{\mathrm{p},n}\right]^{-1}(t ) .
\label{eqntaup}\ ] ] in the above @xmath7^{-1}$ ] is the particular plastic rate for the @xmath8th heterogeneous volume element and @xmath9 is the number of such heterogeneous volume elements leading to a well - converged self - averaged @xmath3^{-1}$ ] . for the current work ,
the characteristic volume of the heterogeneous volume is labelled as @xmath10 giving @xmath11 . under the assumption of thermally activated plasticity ,
the plastic rate associated with one particular heterogeneity volume is written as a linear sum of the @xmath12 thermally active transition rates available to that volume element : @xmath13^{-1}(t)=\sum_{i=1}^{m}\left[\tau_{\mathrm{p0},ni}\right]^{-1}(t)\exp\left(-\frac{e_{\mathrm{p0},ni}}{k_{\mathrm{b}}t}\right ) , \label{eqntaupn}\ ] ] where @xmath14^{-1}(t)$ ] and @xmath15 are the attempt rate and barrier energy for the @xmath16th irreversible structural transformation within the @xmath8th heterogeneous volume element .
[ eqntaup ] then becomes @xmath6^{-1}(t)=\left\langle\left[\tau_{\mathrm{p0}}\right]^{-1}(t)\exp\left(-\frac{e_{\mathrm{p0}}}{k_{\mathrm{b}}t}\right)\right\rangle \label{eqnavetaup}\ ] ] in which the average is performed with respect to @xmath17^{-1}(t)$ ] and @xmath18 . in terms of the underlying potential energy landscape ( pel ) ,
the simple ( first order ) expression of eqn .
[ eqntaupn ] is only valid for thermally activated processes that do not multiply recross their energy barrier .
this naturally leads to a coarse graining of the pel , in which the barriers and ( now diffusive ) pre - factors entering into eqn .
[ eqntaupn ] underlie the collective microscopic activity that results in a lasting escape from a characteristic energy valley . by analogy to the under - cooled liquid pel framework @xcite where two distinct time scales occur the `` slow '' @xmath0-relaxation and `` fast '' microscopic @xmath1-relaxation modes the self - averaging of eqn .
[ eqnavetaup ] reduces ( see paper i ) to @xmath6^{-1}=\left[\tau_{\mathrm{p00}}\right]^{-1}\exp\left(-\frac{e_{\mathrm{p00}}}{k_{\mathrm{b}}t}\right)\times m\left\langle\exp\left(-\frac{e}{k_{\mathrm{b}}t}\right)\right\rangle .
\label{eqnplasticrate}\ ] ] eqn .
[ eqnplasticrate ] consists of a diffusive attempt rate with a simple arrhenius temperature dependence , representing the mediating @xmath1 mode relaxation dynamics multiplied by a thermal factor whose temperature dependence will be derived from the statistical properties of the @xmath0-relaxation mode coarse grained pel @xcite . in paper
i the number of structural transformations available to each heterogeneous volume was assumed to scale exponentially with the number of atoms within @xmath10 , giving @xmath19 where @xmath20 and @xmath21 is the characteristic number of atoms within a heterogeneous volume .
the right most average in eqn .
[ eqnplasticrate ] is then evaluated via a distribution of barrier energies defined by an extensive first and second cumulant , and written as @xmath22=\exp\left[-\frac{f(t)}{k_{\mathrm{b}}t}\right ] .
\label{eqnfe}\ ] ] here @xmath23 is interpreted as the temperature dependent internal barrier energy and @xmath24 as the temperature dependent barrier entropy . within this context
, @xmath25 is referred to as the free barrier energy .
all of these quantities are extensive and @xmath26 gives the apparent number of structural transformations available at temperature @xmath27 . in paper
i , a gaussian distribution was used giving closed form expressions for both @xmath23 and @xmath24 .
an important property of the above is that for a large enough @xmath21 there exists a sharp crossover in behaviour where for , @xmath28 , eqn .
[ eqnfe ] is negligible and for , @xmath29 , it is exponentially large .
thus @xmath30 , where @xmath31 , defines a critical temperature at which the plastic transition rate rapidly rises . by assuming the application of a pure shear stress broadens the distribution via a quadratic increase in the variance ( with respect to stress ) , all mentioned quantities , including @xmath30 , gain a shear stress dependence . via the assumption that at zero applied shear
stress @xmath30 corresponds approximately to the glass transition temperature , the stress dependence of @xmath30 rises rapidly from zero at @xmath32 as the temperature is lowered .
at a temperature of approximately 0.8 - 0.9@xmath32 this temperature dependence weakens to an approximately linear behaviour upon further reduction in temperature . interpreting @xmath33 ( where @xmath34 is the temperature at which the deformation experiment is performed ) as being the stress at which plasticity begins
, an analytical form of yield stress versus temperature that is valid in both the low and high temperature regimes could be developed , and could be fitted well to the available experimental data ( see fig . 8 of @xcite ) of lu _ et al _ @xcite and that shown in johnson and samwer @xcite . whilst a successful fit to low temperature yield data is not new , the current model has the virtue that the yield stress rapidly decreases as the temperature approaches @xmath32 due to a change in barrier energy statistics rather than a change in mechanism .
in paper i a normal distribution of barrier energies was considered and also a modified version such that the distribution would limit to zero at zero barrier energy .
the present work will consider a log - normal distribution of barrier energies .
this distribution has the desirable features of being a positive valued distribution with independent first and second cumulants ( average and variance ) a requirement of the current model .
the log - normal distribution is given as @xmath35 \label{eqlnd0}\ ] ] where the first moment , @xmath36 , is @xmath37\ ] ] and the second moment , @xmath38 , is defined via @xmath39.\ ] ] in terms of @xmath36 and @xmath38 , the distribution can be written as @xmath40}{4\log a}\right ] \label{eqlnd}\ ] ] with @xmath41 in paper i it was argued that the barrier energy distribution shall have extensive first and second cumulants , that is , the barrier energy mean @xmath42 and variance @xmath43 will be linear functions of the number of atoms , @xmath21 , within the heterogeneous volume element . as in paper
i , an over - lined quantity will represent the corresponding intensive variable giving @xmath44 and @xmath45 .
thus , the first and second moments of eqn . [ eqlnd ] will be @xmath46 resulting in @xmath47 the log - normal distribution in the form of eqn .
[ eqlnd ] will now be used to obtain the average plastic rate per heterogeneous volume ( sec .
[ sectheory ] ) and , through a mean - field description of the internal stress , the average plastic strain rate per heterogeneous volume ( sec . [ secmfp ] ) which can be considered as the flow equation for the high temperature homogeneous deformation regime .
more generally , it will be shown that the same central results of paper i are also obtained using a log - normal distribution of barrier energies , instead of a normal distribution that was originally employed .
the central mathematical challenge is to evaluate the final factor of eqn .
[ eqnplasticrate ] : @xmath48 or , with respect to the barrier energy distribution , @xmath49 the integral in eqn .
[ eqngf ] is the generating function of the distribution , and for the gaussian distribution considered in paper i , has a simple closed - form representation . for the log - normal distribution ,
such a straight forward avenue of calculation appears not to be available , however , the analogy to statistical mechanics derived in paper i provides a procedure for the evaluation of eqn .
[ eqngf ] .
indeed , eqn .
[ eqnplasticrate1 ] can be viewed as @xmath50 where @xmath51 is an average over heterogeneous volumes .
the right - hand - side term has the structure of an environmentally averaged partition function average allowing the mathematical problem to exploit the full apparatus of equilibrium statistical mechanics . to obtain a closed form expression of such a partition function
, the micro - canonical approach of derrida @xcite can be used to construct the corresponding free ( barrier ) energy .
this procedure is now applied when the barrier energy distribution is taken as a log - normal distribution . the average number of barrier energies between @xmath52 and @xmath53 , @xmath54 , is given by @xmath55 where @xmath56 must be small enough to ensure a well defined barrier energy but also large enough so that @xmath54 is a smooth function of @xmath52 ( see ref .
@xcite for a related discussion ) . for a sufficiently large heterogeneous volume , fluctuations in eqn .
[ eqnnumden ] become small and @xmath54 becomes a statistically meaningful quantity , allowing for a corresponding barrier entropy to be defined via @xmath57 . doing this for the log - normal distribution , eqn .
[ eqlnd ] , results in the barrier entropy : @xmath58}{4\log a}+\log\left[\frac{1}{\sqrt{2\log a}}\right]+\log\left[\frac{de}{\sqrt{2\pi}e}\right]\right ) \label{eqnse}.\ ] ] to obtain the internal barrier energy and entropy as functions of temperature , an analogy to the thermodynamic definition , @xmath59 is applied to eqn .
[ eqnse ] , giving @xmath60=-2\log a\left(1+\frac{e(t)}{k_{\mathrm{b}}t}\right).\ ] ] a solution to the above with respect to @xmath23 is obtained via the lambert w - function giving @xmath61 .
\label{eqnet}\ ] ] eqn .
[ eqnet ] gives the internal barrier energy as a function of temperature and its substitution into eqn .
[ eqnse ] gives the barrier entropy as a function of temperature : @xmath62 + 3\log\left[\frac{e(t)}{\left\langle e\right\rangle}\right]+ \frac{1}{2\log a}\log\left[\frac{e(t)}{\left\langle e(t)\right\rangle}\right]^{2}\right)\right).\label{eqnst}\ ] ] in the above , the term containing @xmath56 in eqn .
[ eqnse ] has been taken as @xmath63=\ln[\left\langle e\right\rangle / e]+\ln[de/(\sqrt{2\pi}\left\langle e\right\rangle)]\sim\ln[\left\langle e\right\rangle/(\sqrt{2\pi}e)]$ ] .
this approximation will be discussed in the proceeding paragraphs .
[ eqnet ] and [ eqnst ] give the free barrier entropy , @xmath64 , and therefore the final form of eqn .
[ eqnpfapp ] as @xmath65.\label{eqnpfapp1}\ ] ] subsitution of this back into eqn .
[ eqnplasticrate ] gives @xmath6^{-1}=\left[\tau_{\mathrm{p00}}\right]^{-1}\exp\left(\frac{s(t)}{k_{\mathrm{b}}}\right)\exp\left(-\frac{e_{\mathrm{p00}}+e(t)}{k_{\mathrm{b}}t}\right ) \label{eqnplasticrateapp}\ ] ] where @xmath66 can be viewed as the apparent barrier energy and the exponential involving @xmath24 as the number of thermally accessible structural transformations available to the heterogeneous volume .
using eqns .
[ eqne0 ] to [ eqna ] , the @xmath67 limit of @xmath25 , gives @xmath68\right]^{2}+2w\left[\frac{\delta\overline{e}_{0}^{2}}{k_{\mathrm{b}}t\overline{e}_{0}}\right]\right)-\overline{\alpha}\right)\label{eqnfben}\ ] ] demonstrating that for large enough heterogeneous volumes the free barrier energy is an extensive quantity as is the case of the a gaussian distribution . indeed by expanding those terms involving the lambert w - function as a taylor series to second order in @xmath69 recovers a central result of paper i
: @xmath70 that is , for a narrow enough distribution ( set by @xmath71 ) the free barrier energy limits to that obtained for a gaussian distribution of barrier energies as derived in paper i. this is compatible with the fact that for a small enough @xmath71 , the log normal distribution limits to a gaussian distribution .
[ eqnfbenlimit ] has the virtue that it leads to a simple analytic solution to @xmath31 for the critical temperature at which significant plastic activity can be expected ( see eqn .
11 of ref .
@xcite ) .
[ eqnfben ] will be used for further development since it is only at this limit that the right most term in eqn .
[ eqnse ] becomes negligible and can be ignored .
indeed derrida @xcite has argued that @xmath72 where @xmath73
. moreover eqn .
[ eqnfben ] affords a simpler mathematical formalism which is more easily comparable to the normal distribution used in paper i. reabsorbing the factor @xmath21 into the free barrier energy gives @xmath64 as @xmath74\right]^{2}+2w\left[\frac{\delta e_{0}^{2}}{k_{\mathrm{b}}te_{0}}\right]\right)-\alpha\right)\label{eqnfben1}\ ] ] with @xmath75 \label{eqnben1}\ ] ] and @xmath76\right]^{2}\right ) .
\label{eqnbsn1}\ ] ] thus the plastic strain rate ( eqn .
[ eqnplasticrate ] ) becomes @xmath6^{-1}=\left[\tau_{\mathrm{p00}}\right]^{-1}\exp\left(-\frac{e_{\mathrm{p00}}+f(t)}{k_{\mathrm{b}}t}\right ) \label{eqnplasticrate2}\ ] ] withe @xmath25 being given by eqn .
[ eqnfben1 ] .
the above expression gains its external shear stress dependence via the assumed quadratic dependence of the variance of the barrier energy distribution @xcite : @xmath77 where @xmath78 becomes a fitting parameter .
presently , the alternative form @xmath79 will be used . here
@xmath80 is an appropriate value for the bulk shear modulus and @xmath81 is the corresponding fitting parameter with units of barrier energy .
paper i considered only the characteristic escape rate from a mega - basin . for plastic strain evolution the new mega - basin into which the material enters is of importance .
such a structural transformation from one mega basin to another will contribute to the bulk plastic strain .
a natural consequence of a distribution of barrier energies is that any such event will have an associated plastic strain signature which , like the barrier energy , is a random variable . under zero load conditions ,
the material traverses ( at a long enough time - scale and high enough temperature ) a series of mega - basins and the internal plastic strain fluctuates accordingly . assuming the heterogeneous volumes interact via elasticity , this strain fluctuation corresponds to an internal stress fluctuation and therefore an elastic potential energy landscape .
it is such a coarse grained potential energy landscape which underlies the alpha energy landscape and therefore the heterogeneity of the material . due to the assumed exponential number of available structural transformations
there will also be an exponential number of mega - basins that the material can enter .
hence upon exciting one mega - basin and entering another , the statistics will be negligibly affected , with the consequence being that the system has again access to all possible strain states .
which one it chooses will be stochastically biased towards those that minimise the local elastic energy .
this situation does not change upon application of a load , however those structural transformations that can reduce the elastic energy of a particular deformation geometry will become more likely .
so far , the average properties of the non - interacting heterogeneous volumes have been considered , giving @xmath82 where @xmath83 is the externally applied pure shear stress tensor . from the above discussion
, the primary interaction between heterogeneous volumes will be of elastic origin .
the inclusion of such interactions may be done via the eshelby construction . here
the goal is to calculate the elastic energy contribution arising from a local plastic strain increment within a particular heterogeneous volume ( the @xmath8th ) of volume @xmath10 . in the eshelby construction @xcite , this is done by removing from the the amorphous matrix the heterogeneous volume in which the plastic event has occurred , and allowing it to relax to a stress free state given by @xmath84 .
the heterogeneous volume is then deformed elastically to return it to its original shape via the corresponding elastic strain @xmath85 and re - inserted into the matrix . at this point
, the particular heterogeneous volume ( the inclusion ) has a stress derived from hookes law equal to @xmath86 whilst the surrounding matrix is stress free .
the combined system is then allowed to relax to an equilibrium internal stress configuration reducing the inclusion stress by @xmath87 .
eshelby has then shown that the total change in elastic energy is given by @xmath88 where @xmath89 is an additional external stress ( to the @xmath8th heterogeneous volume ) . here , and throughout the text , repeated indices are summed . assuming that this change in elastic energy modifies linearly the intrinsic barrier energy corresponding to this plastic event , one has @xmath90 the strain rate for the @xmath8th heterogeneous volume then becomes @xmath91^{-1}\exp\left[-\frac{e_{n , i}+\frac{1}{2}v_{0}\varepsilon_{\mathrm{t},ni}^{\mu\nu}\left(\sigma_{\mathrm{t},ni}^{\mu\nu}-\sigma_{\mathrm{c},ni}^{\mu\nu}-\sigma_{\mathrm{e},n}^{\mu\nu}\right)}{k_{\mathrm{b}}t}\right ] , \label{eqnpsr}\ ] ] where @xmath92 is the strain increment associated with the barrier energy @xmath93 and attempt rate @xmath94 .
thus each barrier energy is modified from its ( local ) zero stress value by the total elastic strain energy associated with the plastic event occurring
. the strain rate for the rve is therefore @xmath95 here @xmath4 is the volume of the rve volume giving @xmath96 and @xmath97^{-1}\exp\left[-\frac{e_{n , i}+\frac{1}{2}v_{0}\varepsilon_{\mathrm{t},ni}^{\mu\nu}\left(\sigma_{\mathrm{t},ni}^{\mu\nu}-\sigma_{\mathrm{c},ni}^{\mu\nu}-\sigma_{\mathrm{e},n}^{\mu\nu}\right)}{k_{\mathrm{b}}t}\right ] .
\label{eqnavepsr0}\ ] ] the elastic relaxation associated with the @xmath16th plastic event , within and surrounding the @xmath8th heterogeneous volume , results in @xmath98 and an extended stress field field felt by other nearby and distant heterogeneous volumes ( through their own @xmath99 ) . from this latter perspective @xmath98 can be viewed as a self stress contribution to the elastic energy of the plastic event .
the spatial dependence of these stress contributions depend in a non - trivial way on @xmath86 and must be calculated numerically ( see for example the book of t. mura @xcite and refs
. @xcite ) .
it is in this way , that elastic interactions between heterogeneous volumes may be formally introduced .
in addition , the @xmath89 may contain a global applied external stress component written as @xmath83 giving @xmath100 .
thus for each heterogeneous volume , @xmath89 arises from an external applied stress plus an internal stress arising from the elastic interaction between the plastic events from all other heterogeneous volumes .
such a formalism has been considered by bulatov and argon @xcite in their development of a thermal activation model for the plasticity of glasses , and more recently using the finite - element - method by homer and schuh @xcite . in the present work
a mean field approach will be taken which exploits the fact that the local contribution to @xmath89 arising from other heterogeneous volumes will on average equal zero , giving @xmath101 .
for an assumed isotropic elasticity , @xmath102 will equal @xmath103 where @xmath104 will be the shear strain magnitude of the plastic event and @xmath80 is the appropriate bulk isotropic shear modulus for the system .
@xmath105 embodies the elastic relaxation of the plastic strain event and will be different for each plastic event . within the current mean field approach
this will be absorbed into @xmath80 resulting in the @xmath80 being an effective bulk shear modulus of the system .
these approximations afford a simple average with respect to heterogeneous volumes resulting in eqn .
[ eqnavepsr0 ] becoming @xmath106^{-1}\exp\left[-\frac{e_{\mathrm{p0}}+2v_{0}g\gamma_{\mathrm{t}}^{2}-\frac{1}{2}v_{0}\varepsilon_{\mathrm{t}}^{\mu\nu}\sigma^{\mu\nu}}{k_{\mathrm{b}}t}\right]\right\rangle,\ ] ] in which the average is performed with respect to @xmath84 , @xmath107 and @xmath18 .
since there should be no correlation between the strain increment @xmath84 and the parameters @xmath107 and @xmath18 associated with the escape of a particular mega - basin ( since each strain state is always accessible ) , the above average with respect to @xmath84 may be performed independently from @xmath107 and @xmath18 , giving , @xmath108\right\rangle_{\varepsilon_{\mathrm{t}}^{\mu\nu } } m\left\langle\left[\tau_{\mathrm{p0}}\right]^{-1}\exp\left[-\frac{e_{\mathrm{p0}}}{k_{\mathrm{b}}t}\right]\right\rangle_{\tau_{\mathrm{p0}},e_{\mathrm{p0 } } } \nonumber \\ & = & \left\langle\varepsilon_{\mathrm{t}}^{\mu\nu}\exp\left[\frac{-2v_{0}g\gamma_{\mathrm{t}}^{2}+\frac{1}{2}v_{0}\varepsilon_{\mathrm{t}}^{\mu\nu}\sigma^{\mu\nu}}{k_{\mathrm{b}}t}\right]\right\rangle_{\varepsilon_{\mathrm{t}}^{\mu\nu}}\left[\tau_{\mathrm{p}}\right]^{-1}\left(t,\sigma^{\mu\nu}\right).\label{eqnavepsr1}\end{aligned}\ ] ] in the last equality , @xmath3^{-1}\left(t,\sigma^{\mu\nu}\right)$ ] is given by eqn .
[ eqnplasticrate ] .
a consequence of such a lack of correlation is that the magnitude of a barrier energy is , on average , not correlated with the magnitude of the resulting plastic strain increment .
this scenario , again , appeals to the notion of an exponentially large diversity of possible structural excitation .
moreover , numerical evidence in simple lennard jones structural glasses exists , demonstrating that the strain associated in traversing a nearby local pel minima does not at all correlate with the height of the intermediate saddle - point ( barrier ) energy @xcite . the plastic strain average in eqn .
[ eqnavepsr1 ] may be written as @xmath109\right\rangle_{\varepsilon_{\mathrm{t}}^{\mu\nu}}=\frac{\partial}{\partial\left[\frac{\frac{1}{2}v_{0}\sigma^{\mu\nu}}{k_{\mathrm{b}}t}\right]}\left\langle\exp\left[\frac{-2v_{0}g\gamma_{\mathrm{t}}^{2}+\frac{1}{2}v_{0}\varepsilon_{\mathrm{t}}^{\mu\nu}\sigma^{\mu\nu}}{k_{\mathrm{b}}t}\right]\right\rangle_{\varepsilon_{\mathrm{t}}^{\mu\nu}}\label{eqngfidentity0}\ ] ] and therefore the goal is to calculate @xmath110\right\rangle_{\varepsilon_{\mathrm{t}}^{\mu\nu}}. \label{eqnstrainav}\ ] ] to evaluate this average further , the two dimensional case of plane strain will be considered involving only an external pure shear stress component : @xmath111 under these constraints , the characteristic shear strain matrix takes the form @xmath112 giving @xmath113 thus all possible plastic strain increments are defined by an orientation angle @xmath114 and a plastic shear strain magnitude @xmath115 . due to the exponentially large number of plastic strain increments available and the lack of a preferred direction ,
the random variable @xmath114 is assumed to be derived from a uniform distribution ranging from @xmath116 to @xmath117 . on the other hand , for reasonable values of the shear strain magnitude
, the distribution for @xmath115 is expected to be bounded .
presently ( and for mathematical simplicity ) the distribution is assumed to be a normal distribution centred on zero with standard deviation @xmath118 . it is acknowledged that other distributions are also feasible . with these assumptions eqn
. [ eqnstrainav ] becomes @xmath110\right\rangle_{\theta_{\mathrm{t}},\gamma_{\mathrm{t}}}=\exp\left[\left(\frac{v_{0}\delta\gamma'\sigma}{2k_{\mathrm{b}}t}\right)^{2}\right]\mathrm{i}_{0}\left[\left(\frac{v_{0}\delta\gamma'\sigma}{2k_{\mathrm{b}}t}\right)^{2}\right ] .
\label{eqngfidentity1}\ ] ] where @xmath119 is the modified bessel function of order @xmath8 and @xmath120 in sec .
[ secexp ] , the latter approximation will be shown to be valid for temperatures close to the glass transition temperature .
using eqn .
[ eqngfidentity0 ] this finally gives the strain rate per heterogeneous volume as @xmath121\nonumber \\ & & \times\left(\mathrm{i}_{0}\left[\left(\frac{v_{0}\delta\gamma\sigma}{2k_{\mathrm{b}}t}\right)^{2}\right]+\mathrm{i}_{1}\left[\left(\frac{v_{0}\delta\gamma\sigma}{2k_{\mathrm{b}}t}\right)^{2}\right]\right)\nonumber \\ & & \times\left[\tau_{\mathrm{p}}\right]^{-1}\left(t,\sigma^{\mu\nu}\right)\end{aligned}\ ] ] using eqn .
[ eqnplasticrate2 ] and writing , @xmath122 , this reduces to @xmath123\left(\mathrm{i}_{0}\left[\left(\frac{\delta\omega\sigma}{2k_{\mathrm{b}}t}\right)^{2}\right]+\mathrm{i}_{1}\left[\left(\frac{\delta\omega\sigma}{2k_{\mathrm{b}}t}\right)^{2}\right]\right)\nonumber \\ & & \times\exp\left[-\frac{e_{\mathrm{p00}}+f(t,\sigma)}{k_{\mathrm{b}}t}\right ] \label{eqnavepsr3}\end{aligned}\ ] ] where @xmath124 and @xmath125 eqn .
[ eqnavepsr3 ] , written explicitly in terms of the internal barrier energy and barrier entropy , becomes @xmath126\left(\mathrm{i}_{0}\left[\left(\frac{\delta\omega\sigma}{2k_{\mathrm{b}}t}\right)^{2}\right]+\mathrm{i}_{1}\left[\left(\frac{\delta\omega\sigma}{2k_{\mathrm{b}}t}\right)^{2}\right]\right ) \nonumber \\ & & \times\exp\left[-\frac{e_{\mathrm{p00}}+e(t,\sigma)}{k_{\mathrm{b}}t}\right].\label{eqnavepsr4}\end{aligned}\ ] ] the above equation should be compared to the early thermal activation models grounded on the work of spaepen @xcite and argon @xcite , which for homogeneous plastic deformation in the vicinity of the glass transition temperature is usually taken as @xcite @xmath127\sinh\left[\frac{v_{\mathrm{ta}}\sigma}{k_{\mathrm{b}}t}\right ] \label{eqnsimplet}\ ] ] where @xmath128 contains various quantities including the fraction of volume available to deformation , @xmath129 is the attempt frequency , @xmath130 is the characteristic strain , @xmath131 is the activation energy and @xmath132 is the activation volume .
comparison of eqn .
[ eqnavepsr4 ] to [ eqnsimplet ] suggests that @xmath133 can be viewed as a characteristic strain rate ( @xmath134 of eqn .
[ eqnsimplet ] ) , @xmath135 $ ] as a temperature and stress dependent number of available structural transformations , @xmath136 as a temperature and stress dependent activation energy , and @xmath137 as an activation volume .
this latter interpretation is motivated by the term @xmath138+\mathrm{i}_{1}[x^{2}])$ ] being qualitatively similar to @xmath139 , with both limiting to a linear function for small @xmath140 .
thus , although the present theory has a number of quite different physical assumptions when compared to earlier high - temperature thermal activation models , the mean - field prediction of the average strain rate as a function of temperature and stress in the high temperature / low strain rate regime does not operationally differ from these early models .
this observation will be shown in more detail in the next section , when the model is fitted to actual experimental data .
inspection of eqn . [ eqnavepsr3 ] reveals the free parameters of the model are @xmath133 , @xmath137 , and those associated with @xmath141 which are @xmath142 , @xmath42 , @xmath143 , @xmath0 and @xmath81 .
the latter shear stress dependence gets its definition via eqn .
[ eqsd ] where here @xmath143 is equal to @xmath144 . as in paper
i , @xmath42 and @xmath143 , which define the barrier energy distribution can be entirely determined from experimentally accessible parameters , in particular via the viscosity @xmath145 and fragility @xmath146 at the glass transition temperature @xmath32 .
linear ( newtonian ) viscosity may be formally given as @xmath147.\label{eqnviscocity}\ ] ] one common method to define the glass transition temperature is when the viscosity reaches @xmath148 pa - sec at a cooling rate of 20 k / min .
that is , @xmath149.\label{eqnviscocity1}\ ] ] in addition to the value of the viscosity at @xmath32 , a common material parameter to structural glasses is the fragility the rate at which viscosity changes with respect to temperature at @xmath32 , @xmath150 substitution of eqn .
[ eqnviscocity ] into eqn .
[ eqnfragility ] gives @xmath151 or the internal barrier energy at @xmath32 @xmath152 this physically appealing result is general and arises because eqn .
[ eqnter ] requires @xmath153 or equivalently the expression , @xmath154 , which is analogous to the gibbs - helmholtz equation in thermodynamics .
further substitution of eqn .
[ eqnconstraint1 ] into [ eqnviscocity1 ] , gives the barrier entropy at @xmath32 @xmath155\right ) .
\label{eqnconstraint2}\ ] ] thus the experimental parameters @xmath145 , @xmath146 define the free barrier energy at @xmath32 for a given choice of @xmath137 , @xmath133 and @xmath0 .
[ eqnconstraint1 ] and [ eqnconstraint2 ] in conjunction with eqns .
[ eqnben1 ] and [ eqnbsn1 ] also uniquely determine @xmath42 and @xmath143 , @xmath156\ ] ] and @xmath157}.\ ] ] it is noted that the time rate of change of viscosity at the glass transition is given as @xmath158 where cr is the cooling rate from the under - cooled liquid regime .
thus the cooling rate does not give additional information about the theory and emphasizes the fact that the current thermal activation approach is not applicable to the physics of the under - cooled liquid regime .
the remaining free parameters are @xmath133 , @xmath137 , @xmath142 , @xmath0 and @xmath81 .
@xmath142 can be determined from dynamical mechanical analysis or differential scanning calorimetry experiments @xcite and the rest from steady state strain rate versus stress data as a function of temperature , a common data set used to fit to the arrhenius form of eqn .
[ eqnsimplet ] . for vitreloy-1 such
a data set is found in the work of lu _ et al _ @xcite . for this structural glass
initial estimates of @xmath159 k , @xmath160 and @xmath161 ev are taken from both lu _ et al _
@xcite and wang @xcite .
[ figexpcurves]a plots the experimental steady state strain rate versus stress for a number of temperatures close to and above the glass transition temperature . following schuh _
@xcite and wang @xcite it is this data to which the current model ( eqn . [ eqnavepsr3 ] ) will be fitted with all parameters allowed to vary .
the experimental data is taken from uniaxial compression tests and for conversion between the strain rate and uni - axial stress to the appropriate shear quantities , the former is multiplied by @xmath162 and the latter is divided by @xmath162 .
this is motivated by fig .
[ figexpcurves]b , which plots lu s linear viscosity ( fig . 8 of ref .
@xcite ) data multiplied by the shear strain rate and divided by the shear stress .
such numerical factors are common and represent an effective angle of dominant slip activity in a uniaxial deformation experiment , see for example kawamura _
et al _ @xcite . as expected , for a wide range of low shear stresses , the experimental data in fig .
[ figexpcurves]b has converged to approximately unity indicating a stress independent newtonian viscosity coefficient .
the solid curves in figs .
[ figexpcurves]a - b indicate a fit of the model obtained via the simulated annealing global minimisation algorithm @xcite .
the corresponding parameters of this fit are @xmath163 k , @xmath164 , @xmath165 ev , @xmath166 sec@xmath167 , @xmath168 m@xmath169 , @xmath170 and @xmath171 ev , with @xmath145 fixed at @xmath172 pa - sec . it is noted that for these parameters , the numerator of eqn .
[ eqnapprox ] evaluated at @xmath32 equals @xmath173 for @xmath174 nm@xmath169 , justifiying the ensuing approximation of eqn .
[ eqnapprox ] . when interpreted as an activation volume , the parameter @xmath137 has a numerical value that is similar to that found in refs .
@xcite when using the arrhenius form of eqn .
[ eqnsimplet ] . when assuming that the volume of the heterogeneous volumes is one to three orders of magnitude larger than this `` activation volume '' , eqn .
[ eqnactivationvolume ] and [ eqncharacteristicsr ] together with the numerical value of @xmath133 indicate a value for @xmath175 ranging somewhere between @xmath176 to @xmath172 a range of values compatible with it being some multiple of the debye frequency of the system .
the numerical value of @xmath81 indicates a weak stress dependence of the distribution of barrier energies , a consequence of which is that the apparent barrier energy has a weak stress dependence and therefore a value ( @xmath177 ev ) differing little from the activation energy obtained when using eqn .
[ eqnsimplet ] , which is @xmath178 ev @xcite .
the corresponding log - normal barrier energy parameters have values of @xmath179 ev and @xmath180 ev indicating that in the high temperature regime , the relevant part of the distribution ( @xmath181 remains that of low barrier energy tail .
[ figexpcurves]c displays experimental uniaxial deformation curves at a temperature of 643k for a number of different strain rates . the data is taken directly from fig . 2 of ref .
the higher strain rate curves are characterised by an initial rise in stress that may be loosely associated with the elastic deformation regime , a peak stress indicating the onset of a material instability and an eventual stead state flow regime .
appendix b of paper i used a simple method to obtain a constant strain rate stress strain curve from eqn .
[ eqnavepsr3 ] , resulting in the stress at time @xmath182 being given by @xmath183.\ ] ] for a small enough time interval this may be approximated as @xmath184\ ] ] and iterated to generate a stress - strain curve . in the above the plastic strain rate
, @xmath185 is given by @xmath186 ( eqn .
[ eqnavepsr3 ] ) , @xmath187 is the appropriate bulk youngs modulus and @xmath188 the chosen constant total strain rate . using this model , fig .
[ figexpcurves]c also displays the resulting model stress - strain curves . whilst the steady state stress regime is well described by the model , particularly for low values of strain rate , the peak stress regime associated with the emergence of a stress overshoot is not at all present
a result due to the mean field nature of the current plasticity model , which necessarily can only describe homogeneous plasticity . fig .
[ figexpcurves]d now shows the experimental peak stress values of lu _ et al _
@xcite for different strain rates as a function of temperature ( shown as open circles in the figure ) . with @xmath32 at approximately 623k , such data clearly shows the transition from a high temperature / low strain rate homogeneous plasticity to a low temperature / high strain rate regime of deformation .
whilst the model is unable to reproduce the peak stress behaviour it is revealing to plot on this curve the temperature dependence of the steady state stresses for comparable strain rates ( solid curves ) .
for the lowest strain rate , good agreement is seen in the high temperature regime where experimentally the peak stress regime is absent . for larger strain rates
, the quantitative agreement reduces due to the mean field nature of the current plasticity model however qualitatively the trends are similar . for this very same reason ,
the rapid change in experimental behaviour seen as the temperature reduces is not evident in the model , although at higher stresses ( beyond the range of the figure ) the predicted steady state stress does limit to a plateau with a weak temperature dependence a regime that was investigated in paper i ( see fig .
8 of @xcite ) .
having said the above , fig .
[ figexpcurves]d also includes ( the dashed lines ) the consequences of the model when fitted to the peak stress data , as in paper i , rather than the steady state data . the emerging plateau region at low temperatures now becomes evident , although the sharper transition to this temperature regime seen when using the modified gaussian of paper i is no longer evident when using the more restrictive log - normal distribution of barrier energies .
the parameters of this fit are @xmath189 , @xmath190 , @xmath165 ev , @xmath191 sec@xmath167 , @xmath192 m@xmath169 , @xmath193 and @xmath194 ev .
these values do not differ substantially from those associated with the fit to steady state flow stress data .
the corresponding log - normal barrier energy parameters have values of @xmath195 ev and @xmath196 ev which represents a distribution that is closer in position to the the fitted modified gaussian distribution used in paper i ( see fig . 7c of @xcite ) .
whilst such an approach is certainly not justified in the temperature regime where there is a cross - over from homogeneous to heterogeneous plasticity , and a strong difference between the peak stress and flow stress , well below the glass transition temperature there is little experimental distinction between these stresses .
thus one could argue ( as was done in paper i ) that mean field would be applicable both to the quite low temperature and high temperature deformation modes for the onset of plastic deformation ( that is yield ) , with the intermediate cross - over regime not being well described .
the dashed curves in fig .
[ figexpcurves]d tend to support this conclusion .
the present thermal activation theory provides a entirely kinetic interpretation of the fragility index at the glass transition temperature .
this may be seen by writing the barrier entropy , eqn .
[ eqnconstraint2 ] , as @xmath197=k_{\mathrm{b}}\left(m\ln10 + 1\right).\ ] ] or rather as @xmath198=k_{\mathrm{b}}\left(m\ln10 + 1\right).\label{eqnconstraint3}\ ] ] with @xmath199 being viewed as the characteristic experimental time scale associated with the deformation experiment ( see paper i ) . if @xmath200 where equal to the fundamental time scale of the @xmath1-relaxation modes , @xmath175 , then @xmath201 would equal @xmath202 where @xmath146 is the correspondingly measured fragility . in this limit ,
the fragility is therefore a direct measure of the apparent number of structural transformations available to each heterogeneous volume element at @xmath32 , that is , @xmath203 .
an analogous identification has also been proposed between fragility and the configurational entropy of the under - cooled liquid pel by sastry @xcite .
this present result immediately implies that @xmath0 can not be less than @xmath202 since @xmath204 ( eqn .
[ eqnst ] ) .
note that , like the barrier entropy , the fragility is an extensive quantity . for a realistic deformation experiment , @xmath205 , resulting in the above equality between @xmath201 and @xmath146 being an over estimation of the barrier entropy .
this is because over the period of time , @xmath200 , significant _ experimentally unresolvable _ activity occurs which also contributes to the correspondingly measured fragility . through the term
@xmath206 $ ] this activity contributes to an effective reduction of the needed barrier entropy .
thus when the glass transition is viewed as an entirely kinetic phenomenon ( as is done here ) , the fragility becomes a direct measure of the apparent number of available @xmath0-mode structural transformations .
thus eqn .
[ eqnconstraint3 ] ( and eqn .
[ eqnconstraint2 ] ) is considered more fundamental than the more familiar eqn .
[ eqnconstraint1 ] which ( here ) arises from the requirement that the temperature @xmath27 defined via eqn .
[ eqnter ] corresponds to the thermodynamic temperature of the system . using the fitted parameters of viteroly-1 obtained in sec .
[ secexp ] , the value of the ratio @xmath207 suggests a reasonable choice of @xmath175 ( in the range of @xmath208 to @xmath209 sec ) will give a @xmath200 that is within the domain of a typical deformation experiment .
the current mean - field description of macroscopic plasticity is only suited to purely homogeneous plasticity , where within one rve , it is valid to describe the plasticity as arising from the average response of a heterogeneous volume . whilst this might be valid for a broad range of temperatures in the flow stress regime , figs .
[ figexpcurves]c - d demonstrate that for high enough strain rate or low enough temperature , the mean field approximation is unable to describe the presumably heterogeneous transition from elasticity to macroscopic plasticity seen in experiment . indeed at quite low temperatures , where the available number of structural transformations per heterogeneous volume approaches unity a new regime of strongly heterogeneous statistics emerges associated with that of the extreme value .
the average temperature at which this occurs is referred to as the kinetic freezing temperature and may be found as the solution to the barrier entropy equalling zero , @xmath210 : the case when there exists , on average , one available structural transformation per heterogeneous volume . using eqn .
[ eqnbsn1 ] and the known properties of the lambert w - function , this temperature is found to be @xmath211 .
\label{eqnft}\ ] ] for the log - normal distribution .
a gaussian distribution gives a similar result , but without the exponential factor .
for @xmath212 the free barrier energy becomes independent of temperature equalling @xmath213 .
when sampling the @xmath9 heterogeneous volumes of the rve , the kinetic freezing temperature will fluctuate around this value due to fluctuations in the corresponding kinetic freezing barrier energy , the statistics of which is set by the extreme value weibull distribution @xcite . in this temperature regime
, a heterogeneous volume element may contain no accessible barriers below a particular energy threshold allowing for the possibility that no thermally activated structural transformation occurs .
it is only in this temperature regime that a single structural transformation associated with a single barrier energy is formally valid .
this scenario is quite different from that of the statistics of the most probable , occurring at higher temperatures , where on average there exists a large number of accessible barrier energies resulting in a statistically meaningful average plastic transition rate for each heterogeneous volume . for viteroly-1 , the parameters of sec .
[ secexp ] give @xmath214 k indicating that below room temperature this strongly heterogeneous regime of statistics begins to dominate . when using the numerical values of the model parameters derived from a fit to the peak stress , the freezing temperature rises to approximately 203 k , which is somewhat smaller than that obtained when using a gaussian distribution fitted also to the peak stress ( as done in paper i ) .
[ figfreeenergy ] plots the temperature dependence of the free barrier energy , apparent barrier energy and barrier entropy for the flow stress parameters of viteroly-1 . both the apparent barrier entropy and barrier entropy reduce with temperature , until the latter becomes zero at the freezing temperature , @xmath215 .
below this temperature range both quantities become constants independent of temperature .
the corresponding free barrier energy shown in fig .
[ figfreeenergy]a is negative above a temperature regime associated with the glass transition , and rises when the temperature is lowered . according to eqn .
[ eqnfe ] , in the temperature regime where the free barrier energy is positive , the plastic rate reduces many orders of magnitude when compared to the value associated with the glass transition temperature regime . at @xmath216
the free barrier energy also becomes constant , and it is in this regime a simple temperature independent value for both the number of available structural and the energy barrier is formally justified . despite this , the work of sec .
[ secexp ] demonstrates that temperature independent quantities can be used to describe reasonably well the high temperature / homogeneous deformation regime .
indeed fig .
[ figfreeenergy ] demonstrates that both barrier energy and entropy do not vary greatly over the 550 k to 650 k temperature range .
, below which the barrier entropy is zero and on average there exists one available structural transformation within each heterogeneous volume , resulting in the free energy and barrier energy becoming temperature independent quantities for @xmath216.,scaledwidth=90.0% ] the mean - field form for the shear strain rate , eqn .
[ eqnavepsr4 ] , is in practice quite similar to the well known arrhenius form , eqn .
[ eqnsimplet ] .
indeed it is operationally similar justifying the so - called critical barrier energy alluded to by argon @xcite and johnson and samwer @xcite . indeed , in paper i , the fluctuations around the apparent barrier energy where shown to scale as @xmath217 .
despite this similarity , there exist some fundamental differences between the models . the arrhenius shear strain rate is assumed to be proportional to the fraction of volume in which structural transformations can take place @xcite .
the present theory and eqn . [ eqnavepsr4 ] also gives a plastic shear rate that is proportional to the number of available structural transitions , @xmath26 , however this number scales exponentially with atom number and therefore volume .
since this number can be rather large , the traditional athermal concept of a dilute density of ( liquid like ) regions existing within the structural glass which are amenable to plastic deformation has been abandoned in the present work , any particular region of the material can admit a very large number and variety of structural transformations . of course , at low enough @xmath27 the apparent number of structural transformations drastically reduces as the entire @xmath0-relaxation pel gradually freezes out and the kinetic freezing regime is entered .
another similarity between eqns .
[ eqnavepsr4 ] and [ eqnsimplet ] is a shear stress dependent term which for the present work has the form @xmath218+\mathrm{i}_{1}[x^{2}])$ ] and for the arrhenius form is @xmath139 . in both cases @xmath219 . for the arrhenius form ,
the volume term is seen as an activation volume associated with the barrier enthalpy of the volume distortion needed for the structural transformation to occur . in the present work
, this quantity originates from the variance of the distribution of available local plastic strains , @xmath118 , via @xmath220 .
the right - hand - side of this equality may be seen as a measure of the variance of the available slipped volume within the system , which itself is coming from a characteristic internal slipped area multiplied by a characteristic slip distance . in eqn .
[ eqnsimplet ] , the @xmath139 factor arises from reversibility of the single characteristic structural transformation of the system and that the sign of the enthalpy term changes for the inverse process . for the present work , the factor @xmath218+\mathrm{i}_{1}[x^{2}])$ ] in eqn .
[ eqnavepsr4 ] is due to a quite different reason where each transition to a new mega - basin results in the material on average having once again access to all possible plastic transitions .
thus although the reverse transition is allowed , it will occur with negligible probability , since it must compete with all other accessible @xmath0 structural transformations .
the factor @xmath218+\mathrm{i}_{1}[x^{2}])$ ] then arises from the mean - field average ( integration ) over all possible plastic transitions .
whilst the currently developed mean - field picture affords some insight into the consequences of the present theory to macro - plasticity , it certainly is limited to the low strain rate and high temperature range , a regime that is experimentally insensitive to fluctuations away from homogeneity , and also to the thermal and loading history of the material .
indeed , inherent to the above picture is that the material has no memory of its past state .
this is direct result of the mean field theory variant presently used since the elastic interaction between heterogeneous volumes is not at all considered .
when such interactions are taken into account the pre - history of the material will play an integral role in its response to an external condition . in this regard the internal stress field
may be viewed as the state - variable of the model . to address these aspects of low temperature , high strain rate and material history , the mean field approach must be abandoned and an approach which considers correlated spatial variations in plastic activity must be considered . how to develop a theory of macroscopic plasticity beyond mean - field , and which spans the regime of @xmath215 up to @xmath32 for experimentally accessible strain rates , will be the subject of paper iii in this series of work .
the authors wish to thank d. rodney and k. samwer for helpful discussions . 99 c.a .
schuh , t.c .
hufnagel , u. ramamurty , acta . mater .
55 ( 2007 ) p.4067 .
m.l . falk and j.s .
langer , annual review of condensed matter physics 2 ( 2011 ) p.353 .
f. spaepen , acta metall . 25 ( 1977 )
a. argon , acta metall . 27
( 1979 ) p.47 .
w.l . johnson and k.a .
samwer , phys .
95 ( 2005 ) p.195501 .
m. h. cohen and d. turnbull , j. phys .
( 1959 ) p. 16
d. turnbull and m. h. cohen , j. phys .
34 ( 1961 ) p. 120
d. turnbull and m. h. cohen , j. phys .
52 ( 1970 ) p.
3038 m.l . falk and j.s .
langer , phys .
e 57 ( 1998 ) p.7192 .
schuh and a.c .
lund , nat .
. 2 p.449 .
maloney and a. lematre , phys .
93 ( 2004 ) p.016001 .
m.j . demkowicz and a.s .
argon , phys .
91 ( 2004 ) p.025505 .
y. shi and m.l .
falk , phys .
b 73 ( 2006 ) p.214201 .
p. guan , m. chen and t. egami , phys .
104 ( 2010 ) p.205701 .
m.j . demkowicz and a.s .
argon , phys .
b 72 ( 2005 ) p.245205 .
m.j . demkowicz and a.s .
argon , phys .
b 72 ( 2005 ) p.245206 .
langer , phys .
e 77 ( 2008 ) p.021502 .
bouchbinder , j.s .
langer and i. procaccia , phys .
e 75 ( 2007 ) p.036107 .
falk , j.s .
langer and l. pechenik , phys . rev .
b 70 ( 2004 ) p.011507 j.s .
langer , phys .
e 70 ( 2004 ) p.041502 .
m. goldstein , j. chem .
( 1969 ) p.3728 .
angell , k.l .
ngai , g.b .
mckenna , p.f .
mcmillan and s.w .
martin , j. appl .
88 ( 2000 ) p.3113 .
harmon , m.d .
demetriou , w.l .
johnson and k. samwer , phys .
99 ( 2007 ) p.135502 .
wang , j. appl .
( 2011 ) p.053521 .
m. scott shell , p.g .
debenedetti and a.z .
panagiotopoulous , phys .
92 ( 2004 ) p.035506 .
fyodorov , phys .
92 ( 2004 ) p.240601 .
g. adam and j.h .
gibbs , j. chem .
43 ( 1965 ) p.139 .
kirkpatrick , d. thirumalai and p.g .
wolynes , phys .
a 40 ( 1989 ) p.1045 .
a. heuer , j. phys .
: condens .
matter 20 ( 2008 ) p.373101 .
p.m. derlet and r. maa , phys .
b ( rc ) 84 ( 2011 ) p.220201 .
p.m. derlet and r. maa , mater .
proc . vol . 1520 ,
doi : 10.1557/opl.2012.1689 ( 2012 ) .
p.m. derlet and r. maa , phil . mag .
, doi : 10.1080/14786435.2013.826396 ( 2013 ) .
d. klaumnzer , r. maa , f. h. dalla torre and j. f. lffler , appl .
96 ( 2010 ) p.061901 .
r. maa , d. klaumzer and j. f. lffler , acta mater .
59 ( 2011 ) p.3205 .
r. maa , d. klaumzer , g. villard , p. m. derlet and j. f. lffler , appl .
100 ( 2012 ) p. 071904 .
j. lu , g. ravichandran and w.l .
johnson , acta mater .
51 ( 2003 ) p.3429 .
b. derrida , phys .
( 1980 ) p.79 .
eshelby , proc .
soc . a 241 ( 1957 )
v.v . bulatov and a.s .
argon , model .
eng . 2 ( 1994 ) p.167 .
v.v . bulatov and a.s .
argon , model .
eng . 2 ( 1994 )
v.v . bulatov and a.s .
argon , model .
eng . 2 ( 1994 )
t. mura , _ micromehanics of defects in solids _ , martinus hijhoff publishers , the hauge , 1982 .
homer and c.a .
schuh , acta mater .
57 ( 2009 ) p.2823 .
e.r . homer and c.a .
schuh , modelling simul .
( 2010 ) p.065009 .
l. li , e.r .
homer and c.a .
schuh , acta mater .
61 ( 2013 ) p.3347 .
d. rodney and c.a .
schuh , phys .
102 ( 2009 ) p.235503 .
d. rodney and c.a .
schuh , phys .
b. 80 ( 2009 ) p.184203 .
y. kawamura , t. nakamura and a. inoue , scripta mat . 39 ( 1988 )
corana a , marchesi m , martini c and ridella s acm trans . math .
( 1987 ) p.262 .
s. sastry , nature 409 ( 2001 ) p.164 .
gumbel , statistics of extremes ( dover , 2004 ) | a thermal activation model to describe the plasticity of bulk metallic glasses ( derlet and maa , phil . mag .
2013 , doi : 10.1080/14786435.2013.826396 ) which uses a distribution of barrier energies and some aspects of under - cooled liquid physics is developed further .
in particular , a log - normal distribution is now employed to describe the statistics of barrier energies . a high temperature mean - field description of homogeneous macro - plasticity is then developed and is shown to be similar to a thermal activation picture employing a single characteristic activation energy and activation volume . in making this comparison , the activation volume is interpreted as being proportional to the average mean - square - value of the plastic shear strain magnitude within the material .
also , the kinetic fragility at the glass transition temperature is shown to represent the effective number of irreversible structural transformations available at that temperature . |
let @xmath0 be a digraph . for any vertex @xmath15 , the sets @xmath16 and @xmath17 are called the inset and outset of @xmath18 .
the indegree and outdegree of @xmath18 are defined by @xmath19 and @xmath20 .
let @xmath0 be a digraph .
a subset @xmath1 of @xmath2 is called a twin dominating set of @xmath3 if for every vertex @xmath4 , there exists vertices @xmath5 ( @xmath21 may coincide ) such that @xmath6 and @xmath7 are arcs in @xmath3 .
the minimum cardinality of a twin dominating set in @xmath3 is called the twin domination number of @xmath3 and is denoted by @xmath8 . for different orientations @xmath22 and @xmath23 of a graph @xmath9 ,
it is possible to have @xmath24 . in @xcite
, the authors defined the concept of upper orientable twin domination number of a graph @xmath9 , @xmath25 in @xcite , it is conjectured that for @xmath12 and the complete graph @xmath11 , @xmath13 .
we will prove in section @xmath26 that for @xmath27 , @xmath28 . in section @xmath29 , we prove that @xmath14 for all even number @xmath12 .
the following observation is a consequence of the results given in @xcite .
[ sink - source ] let @xmath30 be a tournament ( one orientation of the complete graph ) of order @xmath31 from _ theorem 2.6 _ @xcite , if @xmath30 contains at least one vertex @xmath32 such that @xmath33 or @xmath34 , then @xmath35 . in the case of @xmath12 , @xmath36 .
[ thm : eight_two ] let @xmath30 be an orientation of @xmath37 . if there exists @xmath38 such that @xmath39 or @xmath40 , then @xmath41 .
suppose there exists a vertex @xmath42 such that @xmath43 , @xmath44 and @xmath45 .
without loss of generality we can assume that the arcs @xmath46 . 1 .
if the arcs @xmath47 or @xmath48 are in @xmath49 , then @xmath50 is a twin dominating set of @xmath30 .
so we can assume that the arcs @xmath51 .
see figure [ fig : example8_two ] .
if the arc @xmath52 , then the set @xmath53 is a twin dominating set of @xmath30 .
if the arc @xmath54 , then the set @xmath55 is a twin dominating set of @xmath30 .
therefore , if @xmath43 , @xmath41 . the case @xmath40 , is symmetric . for the tournament @xmath30 of order @xmath56 , which has a vertex @xmath18 of @xmath43
; _ dashed arrows _ denote new added edges in step 1 . ]
[ thm : eight_one ] let @xmath30 be an orientation of @xmath37 . if there exists @xmath38 such that @xmath57 or @xmath58 , then @xmath41 .
let @xmath38 such that @xmath59 with @xmath60 and @xmath61 .
without loss of generality we can assume that the arcs @xmath62 . 1 .
if @xmath63 or @xmath64 are in @xmath49 , then @xmath65 is a twin dominating set of @xmath30 .
so @xmath66 .
2 . by the observation
[ sink - source ] , @xmath67 , in other case , @xmath68 .
assume , without lost of generality , that @xmath69 .
if one of the arcs @xmath70 or @xmath71 are in @xmath30 , then @xmath72 is a twin dominating set of @xmath30 .
so let us assume that @xmath73 , see figure [ fig : example8_one ] .
3 . if @xmath74 and @xmath75 are arcs in @xmath30 , then @xmath76 and from theorem [ thm : eight_two ] , @xmath68 . on the other hand , if @xmath77 ( resp .
@xmath78 ) is an arc in @xmath30 , then @xmath79 ( resp .
@xmath80 ) is a @xmath81-set of @xmath30 . the case @xmath82 is symmetric . for the tournament @xmath30 of order @xmath56 , which has a vertex @xmath18 of @xmath59 ; _ dashed arrows _ denote new added edges in step 1 and _ gray arrows _ denote new added edges in step 2 . ]
[ obs : more_than_four ] denote by @xmath83 the tournament of order @xmath84 of the figure @xmath26 . in @xcite
is proved that @xmath85 .
so for any @xmath86 , @xmath87 and @xmath88 . ;
_ boxes _ denote the @xmath81-set of @xmath83 of size @xmath29 , for @xmath86 . ]
[ case8 ] the upper orientable twin domination number of @xmath37 , @xmath28 .
let @xmath30 be an orientation of @xmath37 .
by the previous observation , if there exist @xmath38 such that @xmath89 or @xmath90 , then @xmath91 .
so , we can suppose that for every @xmath38 , @xmath92 or @xmath93 .
let @xmath38 such that @xmath94 and @xmath95 . without lost of generality
, we can suppose that the arcs @xmath96 .
if one of the arcs @xmath97 or @xmath98 are in @xmath49 , then @xmath99 is a twin dominating set of @xmath30 .
so , we can suppose that @xmath100 and @xmath101 . 1 . if @xmath102 , then @xmath103 is a twin dominating set of @xmath30 .
if @xmath104 , then @xmath105 is a twin dominating set of @xmath30 .
the case of @xmath38 such that @xmath106 is symmetric .
therefore , @xmath107 . by the previous observation
, we can conclude that @xmath108
in this section we proof that for any even number @xmath12 , the conjecture given in @xcite is false .
suppose the theorem is true for any tournament @xmath115 of order @xmath116 .
let @xmath117 to be a tournament of order @xmath118 .
consider @xmath119 and @xmath115 the subtournament of @xmath117 induced by @xmath120 . by our hypothesis induction ,
there exist a @xmath81-set @xmath121 of @xmath115 , such that @xmath122 . if @xmath123 , then @xmath124 is a twin dominating set of @xmath117 with @xmath125 .
so , suppose @xmath126 1 .
if for one vertex @xmath127 , @xmath128 or @xmath129 , then the set @xmath130 or @xmath131 , respectively , is a twin dominating set of @xmath117 with cardinality @xmath132 . therefore we can assume that @xmath133 is a source and @xmath134 a sink with respect to the set @xmath121 .
notice that @xmath135 is a twin dominating set of @xmath121 .
if one of the vertices in @xmath117 is source or sink the proof is finished according to observation [ sink - source ] .
so there exists @xmath136 such that the arcs @xmath137 .
if @xmath138 , @xmath139 or @xmath140 is an oriented cycle in @xmath117 , then @xmath141 with @xmath142 is a twin dominating set of @xmath117 with @xmath132 .
thus , the arcs @xmath143 .
it is clear , that the set @xmath144 is a twin dominating set of @xmath117 with cardinality @xmath145 .
we will prove that @xmath146 it is not minimum .
let @xmath147 .
1 . if the arc @xmath148 , then @xmath149 is a twin dominating set of @xmath117 with cardinality @xmath150 .
so , we can suppose we have the arc @xmath151 , see figure [ fig : even_proof ] .
2 . if the arc @xmath152 or @xmath153 is in @xmath154 , then @xmath155 or @xmath149 are twin dominating set of @xmath117 , respectively .
so we can assume that in @xmath117 , we have the arcs @xmath156 , which implies that @xmath157 is a twin dominating set of @xmath117 with cardinality @xmath150 . | let @xmath0 be a digraph .
a subset @xmath1 of @xmath2 is called a twin dominating set of @xmath3 if for every vertex @xmath4 , there exists vertices @xmath5 such that @xmath6 and @xmath7 are arcs in @xmath3 .
the minimum cardinality of a twin dominating set in @xmath3 is called the twin domination number of @xmath3 and is denoted by @xmath8 . in @xcite , is defined the concept of upper orientable twin domination number of a graph @xmath9 , @xmath10 in @xcite , it is conjectured that for the complete graph @xmath11 with @xmath12 , @xmath13 . in this work we prove that @xmath14 for all even number @xmath12 . _
keywords : _ twin domination ; upper orientable twin domination , tournaments .
_ ams subject classification numbers : _ 05c69 ; 05c05 ; 05c99 . |
two dimensional gravity naturally appears in the string functional integral in subcritical dimension , where it represents an effective theory of quantum fluctuations of matter fields coupled to the metric of the string world sheet @xcite .
the induced , effective action is closely related to the weyl anomaly of the original string theory , and represents a gravitational analogue of the usual wess
zumino action in gauge theories .
dynamical structure of 2d gravity is , therefore , an important aspect of string theory , but it also represents a useful model for the theory of gravitational phenomena in four dimensions .
polyakov and his collaborators @xcite demonstrated that in the light
cone gauge the @xmath1point functions of the effective 2d gravity can be explicitly found .
although the gauge is fixed , these solutions display a hidden chiral @xmath0 symmetry , which turned out to be very important for the analysis of quantum dynamics .
these results motivated the investigation of the structure of 2d gravity in the conformal gauge , where it becomes the standard liouville theory @xcite . although the @xmath0 symmetry is naturally connected to the light
cone gauge , there exists a canonical formulation of the theory in terms of gauge independent variables , the @xmath0 currents , which demonstrates the importance of this symmetry for the general structure of the theory @xcite . dynamical significance of the @xmath0 symmetry , and strong analogy between the induced gravity and the usual wess
zumino action , inspired detailed investigations of the relation between the @xmath0 wess zumino novikov witten ( wznw ) theory and the induced gravity .
polyakov found the connection between the @xmath0 wznw theory and the induced gravity in the _ light
cone gauge _ @xcite .
similar results in the light cone gauge have been also obtained in refs .
the same problem was discussed in the _ conformal _ gauge in ref.@xcite , where it was shown that liouville theory may be regarded as the wznw theory , reduced by certain conformally invariant constraints .
these constraints can be automatically produced if one considers gauge extension of the @xmath0 wznw model , based on two gauge fields @xcite . in a recent letter @xcite
we used a general method of constructing canonical gauge invariant actions to establish the connection between 2d induced gravity and a wznw system , defined by a difference of two simple wznw actions for @xmath0 group : i(g_1,g_2)=i(g_1)-i(g_2 ) , g_1,g_2sl(2,r ) . in this paper we set up the lagrangian framework for this connection , starting from a gauge invariant extension of the wznw system
the connection is established in a _ covariant _ way , fully respecting the diffeomorphism invariance of both theories .
the approach will be very useful for constructing and studying properties of general solutions of the induced gravity , in terms of the related simpler solutions of the wznw system .
after recalling some basic properties of the wznw theory in section 2 , we introduce in section 3 a consistent formulation of the gauge invariant extension of our basic object , the wznw system . by taking the difference of _ two _ wznw actions we are able to overcome the usual difficulties which one encounters in the process of gauging a _ single _ wznw theory @xcite . in section 4
we explicitly choose the gauge group , a four parameter subgroup of @xmath2 leading to four gauge fields , which is sufficient for generating an effective transition to the induced gravity .
in section 5 we show that new gauge invariance can be fixed in such a way that , after integrating out some variables , one arrives at the induced gravity action : i(,g_)=d^2 . in this process , the original symmetry of the action under conformal rescalings is also fixed .
appendices a , b and c are devoted to some details concerning geometrical properties of spacetime @xmath3 and the group manifold @xmath0 , and gauge properties of the wznw model .
basic properties of two dimensional wznw model are defined by the action i(g)=i_0(v ) + n(v ) = _ ( ^*v , v ) + _ m ( v , v^2 ) , vg^-1dg , where @xmath1 is an integer , @xmath4 , and @xmath5 is a normalization constant .
the first term is a @xmath6model action which provides dynamics for a group valued field @xmath7 , defined over a two dimensional , riemann manifold @xmath3 , and taking values in a semi - simple lie group @xmath8 , while the second term is the topological wess
zumino term , defined on a three
manifold @xmath9 whose boundary is @xmath10 . here
, @xmath11 is the maurer
cartan ( lie algebra valued ) 1form , @xmath12 is the dual of @xmath11 , and @xmath13 is the cartan killing bilinear form on the lie algebra of g ( @xmath14 denotes the ordinary matrix trace operation in the adjoint representation of @xmath8 ) .
the normalization factor @xmath5 is chosen in such a way that the wess
zumino term is well defined modulo a multiple of @xmath15 , which is irrelevant in the functional integral @xmath16 $ ] .
let us now parametrize the group elements by some local coordinates @xmath17 , @xmath18 , so that @xmath19 where @xmath20 are the generators of @xmath8 , satisfying the lie algebra @xmath21=f_{ab}{^c}t_c$ ] . then , & & ( ^*v , v)=^*dq^dq^ _ , _ ( q)e^a_e^b__ab , + & & ( v , v^2)=e^ae^be^cf_abc=-6d , where @xmath22 is the cartan metric on @xmath8 , and @xmath23 . the last equation is based on the theorem that any closed form is locally exact . therefore , the wznw action on the group manifold takes the form [ 2.2 ] i(q)=(^*dq^dq^ _ - dq^dq^ _ ) , [ 2.2-a ] where we used @xmath24 .
next , we introduce local coordinates @xmath25 @xmath26 on @xmath3 , and rewrite the action as @xmath27 where @xmath28 is the inverse metric on @xmath3 .
it is convenient to define an orthonormal basis of tangent vectors @xmath29 @xmath30 , in which the metric @xmath31 takes the light cone form : @xmath32 . in this basis , i(q)=_ d^2 ( 1 2 ^ij_iq^_j q^ _ - ^ij_iq^_jq^ _ ) .
[ 2.2-b ] note that the action @xmath33 is invariant under conformal transformations @xmath34 ( which implies @xmath35 ) .
we now turn our attention to @xmath36 . using the fact that any element g of @xmath0 in a neighborhood of identity admits the gauss decomposition @xmath37 , where @xmath38 , @xmath39 and @xmath40 are defined in terms of group coordinates @xmath41 as in eqs.([b.3 ] ) , one can find explicit expressions for @xmath42 and @xmath43 , eqs . and , and obtain i(q ) & & = _ d^2 + & & = _ d^2 ( _ + _ -+4_+x_-ye^- ) .
we shall now discuss how one can gauge the wznw theory starting from the existence of global symmetries , as usual .
the action @xmath44 , where @xmath7 belongs to @xmath0 , is invariant under the _ global _ transformations on the @xmath0 manifold : @xmath45 where @xmath46 is an element of @xmath2 .
we want to introduce the corresponding gauge theory , having the following properties : it should be invariant under the _ local _ transformations g=g^-1 , = (^-,^+ ) , = ( ^+,^- ) , where @xmath46 belongs to a subgroup @xmath47 of @xmath2 ( which may be equal to the whole group ) , it should be defined as a field theory on @xmath3 .
it is well known that the second requirement can not be fulfilled for every gauge group @xmath47 @xcite , since the wz term @xmath48 , originally defined on @xmath9 , does not have a gauge invariant extension that can be reduced to an integral over @xmath49 .
possible solutions of this problem will be discussed after clarifying the meaning of the first requirement @xmath50 .
the transformation law of @xmath51 under gauge transformation is changed , but the change can be compensated by introducing the _ covariant derivative _ : dgdg+ag - gb , ( dg)=(dg)^-1 , where @xmath52 are _ gauge fields _ ( lie algebra valued 1forms ) .
the covariant derivative @xmath53 transforms homogeneously under local transformations , provided the gauge fields transform according to a=(a + d)^-1 , b=(b + d)^-1 . having defined @xmath53 , one can try to gauge the wznw action by replacing @xmath54 , i.e. by replacing 1form @xmath55 with the corresponding covariant 1form @xmath56 : vg^-1dg = v + g^-1ag -b , v=v^-1 .
it is also useful to define the field strengths , @xmath57 and @xmath58 , which transform as follows : @xmath59 , @xmath60 .
now , we apply this procedure to _ formally _ define a gauge invariant extension of the wznw action : i(g , a , b)=i_0(v)+n(v ) = _ ( ^*v , v)+_m ( v , v^2 ) .
the first term @xmath61 is both @xmath50 gauge invariant , and @xmath62 defined over @xmath3 , so that it represents an acceptable gauge invariant action .
it can be written as [ 3.6 ] & & i_0(v)=i_0(v ) + _ 0 , + & & _ 0 = _ , [ 3.6-a ] where @xmath63 . the second term @xmath64 is defined as an integral of a three form on @xmath9 , which is gauge invariant
however , this form is in general not exact , so that @xmath64 can not be expressed as an integral over @xmath3 ; therefore , it can not be used as part of the @xmath6model action on @xmath3 .
we shall now analyze some additional restrictions under which an acceptable gauge extension of @xmath44 _ can _ be defined .
first we note that , after some algebra , the second term can be rewritten as & & n(v)=n(v)+_1+_2+_3 , + & & _ 1=_ , + & & _ 2=_m , + & & _ 3=_m , [ 3.6-b ] where @xmath65 is the chern simons three form .
formal extension of @xmath44 , obtained in eqs . and ( [ 3.6 ] ) , can be written as & & i(g , a , b)=i^r(g , a , b)+_2(a , b)+_3(g , a , b ) , + & & i^r(g , a , b)i(g)+_0(g , a , b)+_1(g , a , b ) , where @xmath66 and @xmath67 are defined not on @xmath3 but on @xmath9 , violating thereby the basic requirement @xmath62 .
now , one should observe that the term @xmath67 is gauge invariant , therefore it can be removed from @xmath68 , leaving us with the gauge invariant combination @xmath69 . since @xmath66 is a three
form on @xmath9 , only @xmath70 can be included as part of the action for the @xmath6model on @xmath3 , but it is not gauge invariant : @xmath71\ne 0 \ , .\ ] ] the action @xmath70 is very close to what we want : it is defined on @xmath3 , and its variation under gauge transformations gives an expression which depends on gauge fields @xmath52 , but not on @xmath7 .
can one find a mechanism that compensates this non invariance , and yields an acceptable gauge invariant extension of the wznw action ? in the analogous four
dimensional model witten @xcite solved the problem by requiring the constraint @xmath72 on the gauge group @xmath47 , sufficient for gauge invariance . in string models
one can simply remove @xmath66 without assuming any constraint on @xmath47 , while the gauge invariance of the theory is ensured by the presence of some additional field in the action , with anomalous " transformation law @xcite . in this paper
we shall solve the @xmath66problem by considering a gauge extension of the action , describing a system of _ two _ simple wznw models , in which the problematic @xmath66 term in the first sector will cancel the corresponding term in the second sector , leading to the theory which is both @xmath50 gauge invariant and @xmath62 defined on @xmath3 .
the construction goes as follows .
we start with the formal extension of @xmath68 , as obtained in eq .. next , using gauge invariance of @xmath73 we define a simpler gauge invariant
action : @xmath74 it is now easy to see that an acceptable gauge extension of the action for the wznw system can be defined by i(1,2)=i(g_1,a , b)-i(g_2,a , b ) , where @xmath75 and @xmath76 are different fields , belonging to the same representation of @xmath0 . indeed , since @xmath77 does not depend on @xmath7 , the contribution of two @xmath66 terms to @xmath78 vanishes .
thus , the gauge invariant action can be written in the simpler form [ 3.9 ] i(1,2)=i^r(g_1,a , b)-i^r(g_2,a , b ) , [ 3.9-a ] where we clearly see that it is an action defined on @xmath3 .
the reduced action can be written as i^r(g , a , b ) = i(g ) + _ , [ 3.9-b ] where the @xmath7independent term @xmath79 in @xmath80 is ignored , as its contribution to sectors 1 and 2 in ( [ 3.9-a ] ) is canceled .
the absence of this term implies that @xmath78 does not depend on @xmath81 and @xmath82 [ see eq . ] , i.e. selfdual and anti - selfdual parts of the gauge fields @xmath83 and @xmath84 , respectively .
as we shall see , the absence of these parts will greatly influence the dynamical structure of the gauged wznw system .
in this section we shall specify the gauge group , and derive an explicit expression for the gauge action ( [ 3.9 ] ) in terms of the group coordinates @xmath86 and @xmath87 .
using now the matrix @xmath88 , that defines the adjoint representation of the gauge group , @xmath89 , where @xmath90 ( appendix b ) , the reduced action takes the form i^r(g , a , b)=i(g ) + 2_d^2_ab .
as we mentioned , the action @xmath78 does not contain variables @xmath91 , which implies the existence of an _ extra _ gauge symmetry , allowing an arbitrary change of the absent components .
this is a specific feature of the action for the wznw system . to simplify further considerations
we shall fix this symmetry by imposing the following gauge conditions : a_-=0 , b_+=0 .
up to now we did not specify the gauge group @xmath47 .
we could take @xmath47 to be the whole @xmath2 , but for our purposes this is not necessary .
we assume that @xmath47 is a subgroup of @xmath2 , defined by h= h_+h_- , where @xmath92 and @xmath93 are subgroups of @xmath0 defined by the generators @xmath94 and @xmath95 , respectively .
when compared to @xmath2 , our choice means that the gauge fields should be restricted as follows : a^=0 , b^=0 .
the gauge symmetry @xmath85 is defined in terms of the following gauge fields and gauge parameters : @xmath96 gauge transformations of dynamical variables are given in eqs.(c.4 ) and ( c.5 ) .
using the general relations @xmath97 and assuming the restrictions ( 4.2 ) and ( 4.4 ) , one obtains ^a_- a^b_+_ab & & = 2^_-a^_+ + ^_-a^_+ + & & = -2e^-_- y -_-a^_+ , + v^a_+ b^b_-_ab & & = 2 v^_+b^_- + v^_+b^_- + & & = 2e^-_+ x + _ + b^_- .next , with the help of the expression ( b.8 ) for @xmath98 we find b^a_-r_aba^b_+=&&-2e^- -b^_-a^_+
.the final result for the reduced action takes the form i^r(g , a , b ) = d^2 & & , where @xmath99x + a^\+_+ \ , , \qquad d_- y = \bigl[\pd_- -b^\0_-\bigr]y -b^\-_- \ , , \ ] ] are covariant derivatives on the group manifold , eqs.([c.2 ] ) .
the last , @xmath7independent term in will be canceled in the complete action @xmath78 , eq.([3.9-a ] ) .
in this section we shall consider the functional integral of the theory defined by the action ( [ 3.9 ] ) , and show , by performing a suitable gauge fixing and integrating out some dynamical variables , that this theory leads to the induced gravity action .
it is useful to introduce auxiliary fields @xmath100 , and rewrite the part of the action @xmath78 given by 4d_+x_1d_-y_1 e^-_1 -4d_+x_2d_-y_2 e^-_2 , in the form = & & f_1-d_+x_1 + f_1+d_-y_1 -f_1-f_1+e^_1 + & & -f_2-d_+x_2 -f_2+d_-y_2 + f_1-f_1+e^_2 , or , more explicitly , = & & -b^_-(f_1 + -f_2 + ) + a^_+(f_1 - -f_2- ) + & & + f_1-x_1 + f_1+y_1 -f_1-f_1+e^_1 + & & -f_2-x_2 -f_2+y_2 + f_2-f_2+e^_2 .
gauge transformations determined by @xmath101 and @xmath102 have the form & & x_1=^ , x_2=^ , a^_+=-^ , + & & y_1=-^ , y_2=-^ , b^_-=-^ .this part of the complete gauge symmetry can be fixed by imposing the following gauge conditions : @xmath103 integrations @xmath104 and @xmath105 in the functional integral lead to the elimination of @xmath106 from the action : @xmath107 . writing @xmath108 instead of @xmath109 for simplicity , one obtains @xmath110x_1 -f_{+}\bigl[\pd_-
-b^\0_-\bigr]y_1 -\fr{1}{4}f_{-}f_{+}\bigl(e^{\vphi_1 } -e^{\vphi_2}\bigr ) \ , .\ ] ] after that the integration @xmath111 produces @xmath112\cdot \d \bigl[(\nabla_- + b^\0_-)f_{+}\bigr ] \ , , \ ] ] where @xmath113 is the covariant derivative on @xmath3 ( appendix a ) . then , @xmath114 yields a^_+=-_+ + _ + f_- , b^_-=-_- -_-f_+ , where @xmath115 is the connection on @xmath3 ( appendix a ) , and an additional factor @xmath116^{-1}$ ] appears in the functional measure .
consequently , we find = -f_-f_+(e^_1-e^_2 ) , and the reduced action becomes & & i^r ( , f_-,f_+)=d^2 .
the complete action @xmath78 is invariant under the remaining piece of gauge transformations : & & f_- = -^ f_- , f_+ = ^ f_+ , + & & _ 1 = ^ -^ , _ 2 = ^-^ .
now , we introduce gauge invariant variables , @xmath117 in terms of which , after some cancellation of @xmath118independent pieces , we obtain & & i(1,2)=i^r(_1)-i^r(_2 ) , + & & i^r()=_d^2 .
note that the part of the functional integral depending on @xmath119 is decoupled from the rest , and can be absorbed into the normalization factor .
thus , using gauge invariant variables effectively restricts the space of dynamical variables ; it is , essentially , equivalent to a gauge fixing corresponding to @xmath120 transformations , e.g. @xmath121 , @xmath122 , followed by the integration @xmath123 .
thus , the final form of the effective action for the gauged wznw system is given by ( 5.5 ) . to show that the effective theory ( 5.5 ) is equivalent to the induced gravity ( 1.2 ) ,
let us observe that ( 5.5 ) is invariant under the following conformal rescalings : g_e^2fg _ , , _ 1_1 - 2f , _
2_2 -2f , which imply @xmath124 , and @xmath125 .
this symmetry is directly connected to the invariance of the original wznw theory under conformal rescalings .
it can be gauge fixed by demanding @xmath126 whereafter
the effective action becomes i(,g_)=_d^2 , where we introduced @xmath127 .
now , partial integrations together with equation ( a.3 ) , and the replacement @xmath128 , lead to the induced gravity action ( 1.2 ) , where @xmath129 , and @xmath130 .
we presented here the connection between the gauge extension of the wznw system ( 1.1 ) and the induced gravity action ( 1.2 ) , fully respecting the diffeomorphism invariance of both theories .
it is well known that an acceptable gauge extension of the simple wznw model does not exist unless one requires specific constraints on the gauge group @xcite .
the reason for this unusual behaviour steams from the fact that gauged wz term @xmath48 does not represent , in general , a field theory on a 2d manifold @xmath3 . if one tries to select an action defined on @xmath3 , one looses gauge invariance , and vice versa . in string models
one can overcome these problems with the help of an additional field @xcite .
following the ideas developed in ref .
@xcite , we introduced in this paper an acceptable action , which is both gauge invariant and defined on @xmath3 , by considering a dynamical system described by a difference of two simple wznw models .
our gauge group is @xmath85 , a four parameter subgroup of @xmath2 .
two of the gauge fields , @xmath131 and @xmath132 , ensure the equality of currents in two sectors , in accordance with the results of the hamiltonian analysis of the wznw system @xcite .
the remaining two fields , @xmath133 and @xmath134 , become components of the connection of the induced gravity action . in this way , the riemannian structure on @xmath3 is seen to be closely related to the @xmath0 gauge fields . the results obtained here can be used to clarify the connection between globally regular solutions of the equations of motion for the wznw system , and the related singular solutions ( in the coordinate sense ) of the induced gravity @xcite . in particular
, it will be interesting to improve our understanding of the wznw black hole solutions in the context of induced gravity @xcite .
in this appendix we present some formulae on the riemannian structure of 2d manifold @xmath3 , which are used in the paper .
coordinates of points in two dimensional manifold @xmath3 are denoted by @xmath25 @xmath26 .
basic tensorial objects in the coordinate basis are : vectors : @xmath135 ; 1forms : @xmath136 ; metric : @xmath137 ; @xmath138 ; another useful basis is the local light
cone basis : vectors : @xmath139 ; 1forms : @xmath140 @xmath141 ; metric : @xmath142 , @xmath143 ; @xmath144 ; riemannian connection on @xmath3 , @xmath145 , is defined by the first structural equation : d^i+^i_j^j=0 . the exterior derivative of a 1form @xmath146 can be written as @xmath147 where @xmath148 is the covariant derivative of a 1form : @xmath149 by noting that @xmath150 , one easily finds the covariant derivative of a vector .
the curvature is defined by the second structural equation : d^i_j = r^i_jkl^k^l .
using @xmath151 , one finds & & r^i_jkl=^i_j(_k_l-_l_k ) , + & & r=2r_-+ = 2(_-_+ -_+_- ) .
let us now derive the transformation law of the connection under conformal rescaling of the metric .
the relation @xmath152 implies @xmath153 .
replacing this into the first structural equation gives @xmath154 since @xmath155 , we easily obtain _ + -_+f = _ + e^f , _ - + _ -f = _ - e^f . acting on these equations with @xmath156 and @xmath157 , respectively , and using ( a.4 ) again , one finds & & _
-_+ -_-_+f = e^2f_-_+ , + & & _ + _ - + _ + _ -f = e^2f_+_- . subtracting two equations
one finds the effect of conformal rescaling on the curvature : r()-2^ij_i_jf = e^2fr(g ) .
we display here some useful formulae : & & ^-^+=d^2 , ^*^i=^i_k^k , + & & ^i^j=^ij d^2 , ^*^i^j=^ij d^2 , + & & a^*a=2a_^ , ( ^*a a)b= 2a_b _ d^2 .
here we present some useful results concerning the riemannian structure of the group manifold @xmath0 . if the generators of the group @xmath0 are chosen as @xmath158 , @xmath159 , where @xmath160 are the pauli matrices , the lie algebra @xmath21=f_{ab}{^c}t_c$ ] takes the form = 2 t _ , = t _ ( ) . then , the calculation of the cartan metric @xmath161 yields _
ab= , a , b= , , . raising and lowering of the tangent space indices
@xmath162 are performed with @xmath163 and its inverse @xmath164 .
the group @xmath0 has the property that any element @xmath7 in a neighborhood of identity admits the gauss decomposition : [ b.3 ] & & g = g_+(x)g_0()g_-(y ) , + & & g_+=e^xt_=1+x t _ ,
g_-=e^yt_=1+yt _ , + & & g_0=e^t_= ( /2)+2t_(/2 ) . [ b.3-a ] where @xmath41 are
group coordinates . in this parametrization
we have g= .
[ b.3-b ] now , we can write @xmath165 , where the quantities @xmath166 serve as the vielbein on the group manifold .
the above expression for @xmath7 leads to & & e^=e^-dx , + & & e^=2ye^-dx + d , + & & e^=-y^2e^-dx -yd+dy , so that the vielbein @xmath166 and its inverse @xmath167 are given as e^a _ = , e^ = .
the cartan metric in the coordinate basis , @xmath168 , has the form _
= , , = x,,y . from the relation
@xmath169 we obtain d = e^ e^ e^ = d ( e^-dx dy ) .
similarly , the calculation of @xmath170 leads to & & ^=-dx + xd+ x^2e^-dy , + & & ^= -d-2xe^-dy , + & & ^= -e^-dy , or ^a_= , ^= .
the metric @xmath171 is the same as @xmath42 .
also , we shall be making use of the matrix @xmath172 , defined by @xmath173 . starting from the identity @xmath174 , which can be written in the form @xmath175
, one finds @xmath176 .
the calculation of @xmath177 yields r_ab(g)= .
in this appendix we exhibit gauge properties of the wznw system in some detail . writing the expression ( 3.4 ) for the covariant 1form @xmath56 in group coordinates @xmath17
, one can obtain coordinate expression for the covariant derivative @xmath178 on the group manifold : & & t_ae^a_dq^=t_ae^a_dq^-t_a r^a_b a^b -t_ab^a , + & & dq^=dq^-^a^a - e^b^a .
effectively , the components @xmath179 and @xmath180 are absent , eq.(4.2 ) , so that @xmath181 taking into account additional conditions @xmath182 , eq.(4.4 ) , one finds [ c.2 ] & & d_+ x = x + a^_+ , + & & d_+=_++a^_+ , + & & d_+ y = _ + y , [ c.2-a ] and & & d_- x = _ - x , + & & d_-=_--b^_- , + & & d_- y = y -b^_- .
[ c.2-b ] let us now consider @xmath2 gauge transformations .
group elements transform according to @xmath183 , where @xmath184 , @xmath185 , and @xmath186 , @xmath187 .
infinitesimal transformations are & & g^-1g = g^-1g- , + & & q^=-^^a -e^^a , or , in components : & & x=^ + x^ -x^2^ -e^^ , + & & = ^ -2x^ + 2y^ -^ , + & & y= e^^ + y^2^ -y^ -^ .
now , restriction to @xmath85 is achieved by demanding @xmath190 .
the restricted transformations take the form : & & x=^ + x^ , + & & = ^ -^ , + & & y= -y^ -^ , and & & a^_+= -_+^ , a^_+= -[_+ + a^_+]^ + a^_+^ , + & & b^_-= -_-^ , b^_-= -[_-
-b^_-^ . using @xmath191 , where @xmath192 , one finds @xmath193dq^\b \ , .\ ] ] restriction to @xmath85 @xmath194 $ ] yields @xmath195 then , gauge transformations of auxiliary fields @xmath108 are : @xmath196 , @xmath197 . from this
it follows @xmath198 . 99 a. m. polyakov , phys .
lett . 103b ( 1981 ) 207 .
a. m. polyakov , mod .
lett . a2 ( 1987 )
893 ; v. g. knizhnik , a. m. polyakov and a. b. zamolodchikov , mod .
a3 ( 1988 ) 819 .
o. alvarez , nucl .
b216 ( 1983 ) 125 ; d. friedan , in recent advances in field theory and statistical mechanics , les houches , 1982 , eds .
j. b. zuber and r. stora ( north holland , 1984 ) , p. 839 . e. abdalla , m. c. b. abdalla , j. gamboa and a. zadra , phys .
lett . 273b ( 1991 ) 222 ; m. blagojevi , m. vasili and t. vukainac , class .
. 11 ( 1994 ) 1155 .
a. m. polyakov , int .
a5 ( 1983 ) 125 .
m. bershadsky and h. ooguri , comm .
( 1989 ) a. alekseev and s. shatashvili , nucl .
b323 ( 1989 ) 719 .
p. forgcs , a. wipf , j. balog , l. fehr and l. oraifeartaigh , phys .
lett . b227 ( 1989 ) 214 .
j. balog , l. fehr , l. oraifeartaigh , p. forgcs and a. wipf , annals of physics , 203 ( 1990 ) 76 .
m. blagojevi , d. s. popovi and b. sazdovi , mod .
lett . a ( 1998 ) , to be published ( hep - th 9709173 ) ; e. witten , comm .
( 1992 ) 189 .
r. i. nepomechie , phys .
d33 ( 1986 ) 3670 .
e. witten , nucl .
phys . b223 ( 1983 ) 422 ; b223 ( 1983 ) 433 .
e. witten , phys .
d44 ( 1991 ) 314 . | we introduced a dynamical system given by a difference of two simple @xmath0 wess zumino novikov witten ( wznw ) actions in 2d , and defined the related gauge theory in a consistent way .
it is shown that gauge symmetry can be fixed in such a way that , after integrating out some dynamical variables in the functional integral , one obtains the induced gravity action .
# 1#2 # 1.3cm*#1 *
= .7 cm = .7 cm = -.3 cm # 1[eq:#1 ] # 1([eq:#1 ] ) : 04.60 kz , 11.10 kk , 11.15 q : two dimensional , wess
zumino , gauge , induced gravity |
perhaps the most basic estimation problem in statistics is the canonical problem of estimating a multivariate normal mean .
based on the observation of a @xmath0-dimensional multivariate normal random @xmath1 the problem is to find a suitable estimator @xmath2 of @xmath3 .
the celebrated result of stein ( @xcite ) dethroned @xmath4 , the maximum likelihood and best location invariant estimator for this problem , by showing that , when @xmath5 , @xmath6 is inadmissible under quadratic loss @xmath7 from a decision theory point of view , an important part of the appeal of @xmath6 was the protection offered by its minimax property .
the worst possible risk @xmath8 incurred by @xmath6 was no worse than the worst possible risk of any other estimator .
stein s result implied the existence of even better estimators that offered the same minimax protection .
he had begun the hunt for these better minimax estimators . in a remarkable series of follow - up papers
stein proceeded to set the stage for this hunt .
james and stein ( @xcite ) proposed a new closed - form minimax shrinkage estimator @xmath9 the now well - known james stein estimator , andshowed explicitly that its risk was less than @xmath10 for every value of @xmath3 when @xmath5 , that is , it uniformly dominated @xmath11 .
the appeal of @xmath12 under @xmath8 was compelling .
it offered the same guaranteed minimax protection as @xmath6 while also offering the possibility of doing much better .
stein ( @xcite ) , though primarily concerned with improved confidence regions , described a parametric empirical bayes motivation for ( [ jsestimator ] ) , describinghow @xmath13 could be seen as a data - based approximation to the posterior mean @xmath14 the bayes rule which minimizes the average risk @xmath15 when @xmath16 .
he here also proposed the positive - part james stein estimator@xmath17 , a dominating improvementover @xmath13 , and commented that `` it would be even better to use the bayes estimate with respect to a reasonable prior distribution . ''
these observations served as a clear indication that the bayesian paradigm was to play a major role in the hunt for these new shrinkage estimators , opening up a new direction that was to be ultimately successful for establishing large new classes of shrinkage estimators .
dominating fully bayes shrinkage estimators soon emerged .
strawderman ( @xcite ) proposed @xmath18 , a class of bayes shrinkage estimators obtained as posterior means under priors @xmath19 for which @xmath20 strawderman explicitly showed that @xmath21 uniformly dominated @xmath6 and was proper bayes , when @xmath22 and @xmath23 or when @xmath24 and @xmath25 .
this was especially interesting because any proper bayes was necessarily admissible and so could not be improved upon .
then , stein ( @xcite ) showed that @xmath26 , the bayes estimator under the harmonic prior @xmath27 dominated @xmath6 when @xmath5 . a special case of @xmath28 when @xmath29 , @xmath30 was only formal bayes because @xmath31 is improper .
undeterred , stein pointed out that the admissibility of @xmath30 followed immediately from the general conditions for the admissibility of generalized bayes estimators laid out by brown ( @xcite ) .
a further key element of the story was brown s ( @xcite ) powerful result that all such generalized bayes rules ( including the proper ones of course ) constituted a complete class for the problem of estimating multivariate normal mean under quadratic loss .
it was now clear that the hunt for new minimax shrinkage estimators was to focus on procedures with at least some bayesian motivation .
perhaps even more impressive than the factthat @xmath30 dominated @xmath6 was the way stein proved it . making further use of the rich results in brown ( @xcite ) , the key to his proof was the fact that any posterior mean bayes estimator under a prior @xmath32 can be expressed as @xmath33 where @xmath34 is the marginal distribution of @xmath35 under @xmath32 .
[ here @xmath36 is the familiar gradient . ] at first glance it would appear that ( [ keyrep1 ] ) has little to do with the risk . however , stein noted that insertion of ( [ keyrep1 ] ) into @xmath8 , followed by expansion and an integration - by - parts identity , now known as one of stein s lemmas , yields the following general expression for the difference between the risks of @xmath37 and @xmath6 : @xmath38 \\[-8.5pt ] & & \quad = e_\mu \biggl[\|\nabla\log m_\pi(x)\|^2 - 2 \frac{\nabla^2 m_\pi(x)}{m_\pi ( x ) } \biggr ] \nonumber\\[-0.5pt ] \label{uber1b } & & \quad = e_\mu \bigl[-4\nabla^2 \sqrt{m_\pi(x)}\big/\sqrt{m_\pi ( x ) } \bigr].\end{aligned}\ ] ] ( here @xmath39 is the familiar laplacian . ) because the bracketed terms in ( [ uber1a ] ) and ( [ uber1b ] ) do not depend on @xmath3 ( they are unbiased estimators of the risk difference ) , the domination of @xmath6 by @xmath37 would follow whenever @xmath40 was such that these bracketed terms were nonnegative .
as stein noted , this would be the case in ( [ uber1a ] ) whenever @xmath40 was superharmonic , @xmath41 , and in ( [ uber1b ] ) whenever @xmath42 was superharmonic , @xmath43 , a weaker condition .
the domination of @xmath6 by @xmath30 was seen now to be attributable directly to the fact that the marginal ( [ margx ] ) under @xmath44 , a mixture of harmonic functions , is superharmonic when @xmath5 .
however , such an explanation would not work for the domination of @xmath6 by @xmath21 , because the marginal ( [ margx ] ) under @xmath45 in ( [ pi_a ] ) is not superharmonic for any @xmath46 .
indeed , as was shown later by fourdrinier , strawderman and wells ( @xcite ) , a superharmonic marginal can not be obtained with any proper prior .
more importantly , however , they were able to establish that the domination by @xmath21 was attributable to the superharmonicity of @xmath47 under @xmath45 when @xmath48 ( and strawderman s conditions on @xmath49 ) .
in fact , it also followed from their results that @xmath47 is superharmonic when @xmath50 and @xmath5 , further broadening the class of minimax improper bayes estimators .
prior to the appearance of ( [ uber1a ] ) and ( [ uber1b ] ) , minimaxity proofs , though ingenious , had all been tailored to suit the specific estimators at hand .
the sheer generality of this new approach was daunting in its scope . by restricting attention to priors that gave rise to marginal distributions with particular properties ,
the minimax properties of the implied bayes rules would be guaranteed .
[ sec : pred - emerges ] the seminal work of stein concerned the canonical problem of how to estimate @xmath3 based on an observation of @xmath51 .
a more ambitious problem is how to use such an @xmath35 to estimate the entire probability distribution of a future @xmath52 from a normal distribution with this same unknown mean @xmath3 , the so - called predictive density of @xmath52 .
such a predictive density offers a complete description of predictive uncertainty . to conveniently treat the possibility of different variances for @xmath35 and @xmath52
, we formulate the predictive problem as follows .
suppose @xmath53 and @xmath54 are independent @xmath0-dimensional multivariate normal vectors with common unknownmean @xmath3 but known variances @xmath55 and @xmath56 .
letting @xmath57 denote the density of @xmath52 , the problem is to find an estimator @xmath58 of @xmath59 based on the observation of @xmath60 only .
such a problem arises naturally , for example , for predicting @xmath61 based on the observation of @xmath62 which is equivalent to observing @xmath63 .
this is exactly our formulation with @xmath64 and @xmath65 . for the evaluation of @xmath58 as an estimator of @xmath66 ,
the analogue of quadratic risk @xmath8 for the mean estimation problem is the kullback
leibler ( kl ) risk @xmath67 where @xmath68 denotes the density of @xmath35 , and @xmath69 is the familiar kl loss . for a ( possibly improper ) prior distribution @xmath70 on @xmath3 ,
the average risk @xmath71 is minimized by the bayes rule @xmath72\nonumber \\[-8pt ] \\[-8pt ] & = & \int p(y { |}\mu ) \pi(\mu{|}x)\,d\mu , \nonumber\end{aligned}\ ] ] the posterior mean of @xmath59 under @xmath70 ( aitchison , @xcite ) .
it follows from ( [ eq : bayes ] ) that @xmath73 is a proper probability distribution over @xmath74 whenever the marginal density of @xmath75 is finite for all @xmath76 ( integrate w.r.t .
@xmath74 and switch the order of integration ) .
furthermore , the mean of @xmath77 ( when it exists ) is equal to @xmath78 , the bayes rule for estimating @xmath3 under quadratic loss , namely the posterior mean of @xmath3 .
thus , @xmath79 also carries the necessary information for that estimation problem .
note also that unless @xmath70 is a trivial point prior , such @xmath77 will not be of the form of @xmath59 for any @xmath3 .
the range of the bayes rules here falls outside the target space of the densities which are being estimated .
a tempting initial approach to this predictive density estimation problem is to use the simple plug - in estimator @xmath80 to estimate @xmath59 , the so - called estimative approach .
this was the conventional wisdom until the appearance of aitchison ( @xcite ) .
he showed that the plug - in estimator @xmath81 is uniformly dominated under @xmath82 by @xmath83\hspace*{-17pt}\nonumber \\[-7pt ] \\[-8pt ] & = & \frac{1}{\ { 2\pi(v_x + v_y)\}^{{{p}/{2 } } } } \exp \biggl\ { -\frac{\|y - x\| ^2}{2(v_x + v_y ) } \biggr\},\hspace*{-17pt } \nonumber\end{aligned}\ ] ] the posterior mean of @xmath59 with respect to the uniform prior @xmath84 , the so - called predictive approach . in a related vein ,
akaike ( @xcite ) pointed out that , by jensen s inequality , the bayes rule @xmath77 would dominate the random plug - in estimator@xmath85 when @xmath86 is a random draw from @xmath70 .
strategies for averaging over @xmath3 were looking better than plug - in strategies .
the hunt for predictive shrinkage estimators had turned to bayes procedures .
distinct from @xmath81 , @xmath87 was soon shown to be the best location invariant predictive density estimator ; see murray ( @xcite ) and ng ( @xcite ) . that @xmath87 is best invariant and minimax also follows from the more recent general results of liang and barron ( @xcite ) , who also showed that @xmath87 is admissible when @xmath88 .
the minimaxity of @xmath87 was also shown directly by george , liang and xu ( @xcite ) .
thus , @xmath87 , rather than @xmath81 , here plays the role played by @xmath6 in the mean estimation context .
not surprisingly , @xmath89 , the posterior mean under the uniform prior @xmath90 is identical to @xmath6 in that context .
the parallels between the mean estimation problem and the predictive estimation problem came into sharp focus with the stunning breakthrough result of komaki ( @xcite ) .
he proved that when @xmath5 , @xmath91 itself is dominated by the bayes rule @xmath92,\ ] ] under the harmonic prior @xmath93 in ( [ pi_h ] ) used by stein ( @xcite ) . shortly thereafter
liang ( @xcite ) showedthat @xmath94 is dominated by the proper bayes rule @xmath95 under @xmath96 for which @xmath97 when @xmath98 , and when @xmath22 and @xmath99 or and @xmath25 , the same conditions that strawderman had obtained for his estimator .
notethat @xmath19 in ( [ eq : pa ] ) is an extension of ( [ pi_a ] ) which depends on the constant @xmath100 .
as before , @xmath93 is the special case of @xmath19 when @xmath101 .
note that @xmath87 is now playing the `` straw - man '' role that was played by @xmath6 in the mean estimation problem .
[ sec : theory ] the proofs of the domination of @xmath87 by @xmath102 in komaki ( @xcite ) and by @xmath103 in liang ( @xcite ) were both tailored to the specific forms of the dominating estimators .
they did not make direct use of the properties of the induced marginal distributions of @xmath35 and @xmath52 . from the theory developed by brown ( @xcite ) and stein ( @xcite ) for the mean estimation problem
, it was natural to ask if there was a theory analogous to ( [ keyrep1])([uber1b ] ) which would similarly unify the domination results in the predictive density estimation problem .
as it turned out , just such a theory was established in george , liang and xu ( @xcite ) , the main results of which we now proceed to describe .
the story begins with a representation , analogous to brown s representation @xmath104 in ( [ keyrep1 ] ) , that is available for posterior mean bayes rules in the predictive density estimation problem .
a key element of the representation is the form of the marginal distributions for our context which we denote by @xmath105 for @xmath106 and a prior @xmath32 . in terms of our previous notation ( [ margx ] ) , @xmath107 .
[ thm : pform ] the bayes rule @xmath77 in ( [ eq : bayes ] ) can be expressed as @xmath108 where @xmath94 is the bayes rule under @xmath109 given by ( [ eq : piu ] ) , @xmath110 is the marginal distribution of @xmath35 , and @xmath111 , where @xmath112 , is the marginal distribution of @xmath113 for independent @xmath53 and @xmath54 .
lemma [ thm : pform ] shows how the form of @xmath77 is determined entirely by @xmath94 and the form of @xmath114 and @xmath111 .
the essential step in its derivation is to factor the joint distribution of @xmath75 and @xmath74 into terms including a function of the sufficient statistic @xmath115 . inserting the representation ( [ eq : mform ] ) into the risk
@xmath82 leads immediately to the following unbiased estimate for the @xmath116 risk difference between @xmath117 and @xmath77 : @xmath118 as one can see from ( [ eq : uber3 ] ) and the fact that @xmath119 , @xmath94 would be uniformly dominated by @xmath77 whenever @xmath120 is decreasing in @xmath121 . as if by magic , the sign of @xmath122 turned out to be directly linked to the same unbiased risk difference estimates ( [ uber1a ] ) and ( [ uber1b ] ) of stein ( @xcite ) .
[ thm : rderiv ] @xmath123 \\[-8pt ] & & \quad \quad = e_{\mu , v } \biggl[\frac { \nabla^2 m_\pi(z ; v)}{m_\pi(z ; v ) } - \frac{1}{2 } \|\nabla\log m_\pi(z ; v)\|^2 \biggr ] \nonumber \\ \label{uber2a } & & \quad \quad = e_{\mu , v } \bigl [ 2\nabla^2 \sqrt{m_\pi(z ; v)}\big/\sqrt{m_\pi(z ; v ) } \bigr].\end{aligned}\ ] ] the proof of lemma [ thm : rderiv ] relies on brown s representation , stein s lemma , and the fact that any normal marginal distribution @xmath124 satisfies @xmath125 the well - known heat equation which has a long history in science and engineering ; for example , see steele ( @xcite )
. combining ( [ eq : uber3 ] ) and lemma [ thm : rderiv ] with the fact that @xmath117 is minimax yields the following general conditions for the minimaxity of a predictive density estimator , conditions analogous to those obtained by stein for the minimaxity of a normal mean estimator .
[ theo1 ] if @xmath126 is finite for all @xmath76 , then @xmath73 will be minimax if either of the following hold for all @xmath127 : @xmath128 is superharmonic .
@xmath129 is superharmonic .
although condition ( i ) implies the weaker condition ( ii ) above , it is included because of its convenience when it is available . since a superharmonic prior always yields a superharmonic @xmath128 for all @xmath121 , the following corollary is immediate .
[ cor1 ] if @xmath126 is finite for all @xmath76 , then @xmath73 will be minimax if @xmath32 is superharmonic . because @xmath130 is superharmonic , it is immediate from corollary [ cor1 ] that @xmath102 is minimax . because @xmath131 is superharmonic for all @xmath121 ( under suitable conditions on @xmath49 ) , it is immediate from theorem [ theo1 ] that @xmath103 is minimax .
it similarly follows that any of the improper superharmonic @xmath132-priors of faith ( @xcite ) or any of the proper generalized @xmath132-priors of fourdrinier , strawderman and wells ( @xcite ) yield minimax bayes rules .
the connections between the unbiased risk difference estimates for the kl risk and quadratic risk problems ultimately yields the following identity : @xmath133 \\[-8pt ] & & \quad \quad= \frac{1}{2 } \int_{v_w}^{v_x } \frac{1}{v^2 } [ r_q(\mu , \hat\mu _ u ) - r_q(\mu , \hat\mu_\pi ) ] _ v \,dv , \nonumber\end{aligned}\ ] ] explaining the parallel minimax conditions in both problems .
brown , george and xu ( @xcite ) used this identity to further draw out connections to establish sufficient conditions for the admissibility of bayes rules under kl loss , conditions analogous to those of brown ( @xcite ) and brown and hwang ( @xcite ) , and to show that all admissible procedures for the kl risk problems are bayes rules , a direct parallel of the complete class theorem of brown ( @xcite ) for quadratic risk .
the james stein estimator @xmath13 in ( [ jsestimator ] ) provided an explicit example of how risk improvements for estimating @xmath3 are obtained by shrinking @xmath35 toward 0 by the adaptive multiplicative factor @xmath134 .
similarly , under unimodal priors , posterior mean bayes rules @xmath135 shrink @xmath75 toward the center of @xmath136 , the mean of @xmath32 when it exists .
( section [ sec : multshrink ] will describe how multimodal priors yield multiple shrinkage estimators . ) as we saw earlier , @xmath75 here plays the role both of @xmath137 and of the formal bayes estimator @xmath138 .
the representation ( [ eq : mform ] ) reveals how @xmath77 analogously `` shrinks '' the formal bayes estimator @xmath94 , but not @xmath139 , by an adaptive multiplicative factor @xmath140 however , because @xmath77 must be a proper probability distribution ( whenever @xmath40 is always finite ) , it can not be the case that @xmath141 for all @xmath74 at any @xmath75 .
thus , `` shrinkage '' here really refers to a reconcentration of the probability distribution of @xmath94 .
furthermore , since the mean of @xmath142 is @xmath78 , this reconcentration , under unimodal priors , is toward the center of @xmath32 , as in the mean estimation case .
consider , for example , what happens under @xmath130 which is symmetric and unimodal about 0 .
figure [ fig1 ] illustrates how this shrinkage occurs for @xmath143 for various values of @xmath75 when @xmath22 . figure [ fig1 ] plots @xmath117 and @xmath144 as functions of @xmath145 when @xmath146 and @xmath147 .
note first that @xmath117 is always the same symmetric shape centered at @xmath75 .
when @xmath148 , shrinkage occurs by pushing the concentration of @xmath144 = @xmath149 toward 0 . as @xmath75 moves further from @xmath150 to @xmath151 and @xmath152 this shrinkage diminishes as @xmath144 becomes more and more similar to @xmath94 .
as in the problem of mean estimation , the shrinkage by @xmath102 manifests itself in risk reduction over @xmath87 . to illustrate this , figure [ fig2 ] displays the risk difference
@xmath153 $ ] at @xmath154 , @xmath155 when @xmath146 and @xmath147 for dimensions @xmath156 . paralleling the risk reduction offered by @xmath157 in the mean estimation problem ,
the largest risk reduction offered by @xmath102 occurs close to @xmath158 and decreases rapidly to 0 as @xmath159 increases .
[ @xmath160 is constant as a function of @xmath3 .
] at the same time , the risk reduction by @xmath102 is larger for larger @xmath0 at each fixed @xmath159 .
[ sec : targets ] by a simple shift of coordinates , the modified james
stein estimator , @xmath161 remains minimax , but now shrinks @xmath75 toward @xmath162 where its risk function is smallest .
similarly , minimax bayes shrinkage estimators of a mean or of a predictive density can be shifted to shrink toward @xmath163 , by recentering the prior @xmath32 to @xmath164 .
these shifted estimators are easily obtained by inserting the and @xmath165 when @xmath166 , @xmath167 , @xmath168 . ]
corresponding translated marginal @xmath169 into ( [ keyrep1 ] ) to obtain @xmath170 and into ( [ eq : mform ] ) to obtain @xmath171 recentered unimodal priors such as @xmath172 and @xmath173 yield estimators that now shrink @xmath75 and @xmath94 toward @xmath163 rather than toward 0 .
since the superharmonic properties of @xmath40 are inherited by @xmath174 , the minimaxity of such estimators will be preserved . in his discussion of stein ( @xcite ) ,
lindley ( @xcite ) noted that the james stein estimator could be modified to shrink toward @xmath175 ( @xmath176 is the mean of the components of @xmath75 ) , by replacing @xmath163 and in ( [ jsb ] ) by @xmath177 and @xmath178 , respectively .
the resulting estimator remains minimax as long as and offers smallest risk when @xmath179 is close to the subspace of @xmath3 with identical coordinates , the subspace spanned by the vector @xmath180 .
note that @xmath177 is the projection of @xmath75 into this subspace .
more generally , minimax bayes shrinkage estimators of a mean or of a predictive density can be similarly modified to obtain shrinkage toward any ( possibly affine ) subspace @xmath181 , whenever they correspond to spherically symmetric priors .
such priors , which include @xmath130 and @xmath45 , are functions of @xmath3 only through @xmath159 .
such a modification is obtained by recentering the prior @xmath32 around @xmath182 via @xmath183 where @xmath184 is the projection of @xmath3 onto @xmath182 .
effectively , @xmath185 puts a uniform prior on @xmath186 and applies a suitably modified version of @xmath70 to @xmath187 .
note that the dimension of @xmath187 , namely @xmath188 , must be taken into account when determining the appropriate modification for @xmath70 .
for example , recentering the harmonic prior @xmath189 around the subspace spanned by @xmath190 yields @xmath191 where @xmath192 . here ,
the uniform prior is put on @xmath193 , and the harmonic prior in dimension @xmath194 ( which is different from the harmonic prior in @xmath195 ) is put on @xmath196 , the orthogonal complement of @xmath182 .
the marginal @xmath197 corresponding to the recentered @xmath198 in ( [ eq : phb ] ) can be directly obtained by recentering the spherically symmetric marginal @xmath40 corresponding to @xmath70 , that is , @xmath199 where @xmath200 is the projection of @xmath76 onto @xmath182 .
analogously to @xmath201 , @xmath202 is uniform on @xmath200 and applies a suitably modified version of @xmath40 to @xmath203 . here ,
too , the dimension of @xmath203 , namely @xmath188 , must be taken into account when determining the appropriate modification for @xmath40 .
for example , recentering the marginal @xmath40 around the subspace spanned by @xmath190 would entail replacing @xmath204 by @xmath205 , where @xmath206 , and appropriately modifying @xmath40 to apply to @xmath207 . applying the recentering ( [ eq : phb ] ) to priors such as @xmath130 and @xmath45 ,
which are unimodal around 0 , yields priors @xmath208 and @xmath209 and hence marginals @xmath210 and @xmath211 , which are unimodal around @xmath182 . such recentered marginals yield mean estimators @xmath212 and predictive density estimators @xmath213 that now shrink @xmath75 and @xmath94 toward @xmath182 rather than toward 0 .
shrinkage will be largest when @xmath214 , and will diminish as @xmath75 moves away from @xmath182 .
these estimators offer smallest risk when @xmath215 , but do not improve in any important way over @xmath75 and @xmath216 when @xmath3 is far from @xmath182 .
a superharmonic @xmath40 will lead to a superharmonic @xmath197 as long as @xmath188 is large enough .
for example , the recentered marginal @xmath210 will be superharmonic only when @xmath217 . in such cases ,
the minimaxity of both @xmath218 and @xmath219 will be preserved .
[ sec : multshrink ] stein s discovery of the existence of minimax shrinkage estimators such as @xmath220 in ( [ jsb ] ) demonstrated that costless improvements over the minimax @xmath6 were available near any target preselected by the statistician .
as stein ( @xcite ) put it when referring to the use of such an estimator to center a confidence region , the target `` should be chosen as one s
best guess '' of @xmath179 . that frequentist considerations had demonstrated the folly of ignoring subjective input was quite a shock to the perceived `` objectivity '' of the frequentist perspective .
although the advent of minimax shrinkage estimators of the form @xmath221 in ( [ keyrepb ] ) and @xmath219 in ( [ eq : bform ] ) opened up the possibility of small risk near any preselected ( affine ) subspace @xmath181 ( this includes the possibility that @xmath182 is a single point ) , it also opened up a challenging new problem , how to best choose such a @xmath182 . from the vast number of possible choices , the goal was to choose @xmath182 close to the unknown @xmath3 , otherwise risk reduction would be negligible . to add to the difficulties , low - dimensional @xmath182 , which offered the greatest risk reduction ,
were also the most difficult to get close to @xmath3 . when faced with a number of potentially good target choices ,
say @xmath222 , rather than choose one of them and proceed with @xmath218 or @xmath219 , an attractive alternative is to use a minimax multiple shrinkage estimator ; see george ( @xcite ) .
such estimators incorporate all the potential targets by combining them into an adaptive convex combination of @xmath223 for mean estimation , and of @xmath224 for predictive density estimation . by adaptively shrinking toward the more promising targets , the region of potential risk reduction
is vastly enlarged while at the same time retaining the safety of minimaxity .
the construction of these minimax multiple shrinkage estimators proceeds as follows , again making fundamental use of the bayesian formulation . for a spherically symmetric prior @xmath32 , a set of subspaces @xmath222 of @xmath195 , and a set of nonnegative weights @xmath225 such that @xmath226 , consider the mixture prior @xmath227 where each @xmath228 is a recentered prior as in ( [ eq : phb ] ) . to simplify notation
, we consider the case whereeach @xmath228 is a recentering of the same @xmath70 , although in principle such a construction could be applied with different priors .
the marginal @xmath229 corresponding to the mixture prior @xmath230 in ( [ eq : priorstar ] ) is then simply @xmath231 where @xmath232 are the recentered marginals corresponding to the @xmath228 as given by ( [ eq : mb ] ) .
applying brown s representation @xmath233 from ( [ keyrep1 ] ) with @xmath229 in ( [ eq : mstar ] ) immediately yields the multiple shrinkage estimator of @xmath3 , @xmath234 where @xmath235 similarly , applying the representation @xmath236 from ( [ eq : mform ] ) with @xmath229 immediatelyyields the multiple shrinkage estimator of @xmath59 , @xmath237 where @xmath238 the forms ( [ pstar1 ] ) and ( [ pstar2 ] ) reveal @xmath239 and @xmath240 to be adaptive convex combination of the individual posterior mean estimators @xmath241 and @xmath242 , respectively .
the adaptive weights @xmath243 in ( [ pprob1 ] ) and ( [ pprob2 ] ) are the posterior probabilities that @xmath179 is contained in each of the @xmath244 , effectively putting increased weight on those individual estimators which are shrinking most .
note that the uniform prior estimates @xmath245and @xmath87 are here doubly shrunk by @xmath239 and @xmath246 ; in addition to the individual estimator shrinkage they are further shrunk by the posterior probability @xmath247 .
the key to obtaining @xmath239 and @xmath246 which are minimax is simply to use priors which yield superharmonic @xmath248 .
if such is the case , then trivially from ( [ eq : mstar ] ) @xmath249 so that @xmath229 will be superharmonic , and the minimaxity of @xmath250 and @xmath246 will follow immediately . note that marginals whose square root is superharmonic will not be adequate , as this argument will fail . and
multiple shrinkage @xmath251 when @xmath252 , @xmath167 , @xmath168 , @xmath253 , @xmath254 , and @xmath255 . ]
the adaptive shrinkage behavior of @xmath239 and @xmath240 manifests itself as substantial risk reduction whenever @xmath3 is near any of @xmath256 .
let us illustrate how that happens for the predictive density estimator @xmath257 , the multiple shrinkage version of @xmath258 .
figure [ fig3 ] illustrates the risk reduction @xmath259 $ ] at various @xmath154 obtained by @xmath257 which adaptively shrinks @xmath94 toward the closer of the two points @xmath260 and @xmath261 using equal weights @xmath262 .
as in figure [ fig2 ] , we considered the case @xmath263 for @xmath156 .
as the plot shows , maximum risk reduction occurs when @xmath3 is close to @xmath264 or @xmath265 , and goes to 0 as @xmath3 moves away from either of these points . at the same time , for each fixed @xmath266 , risk reduction by @xmath257 is larger for larger @xmath0 .
it is impressive that the size of the risk reduction offered by @xmath267 is nearly the same as each of its single target counterparts .
the cost of multiple shrinkage enhancement seems negligible , especially compared to the benefits .
beyond their attractive risk properties , the james stein estimator @xmath12 and its positive - part counterpart @xmath268 are especially appealing because of their simple closed forms which are easy to compute .
as shown by xu and zhou ( @xcite ) , similarly appealing simple closed - form predictive density shrinkage estimators can be obtained by the same empirical bayes considerations that motivate @xmath269 and @xmath268 .
the empirical bayes motivation of @xmath12 , alluded to in section [ sec1 ] , simply entails replacing @xmath270 in ( [ conjugate - coef ] ) by @xmath271 , its unbiased estimate under the marginal distribution of @xmath272 when @xmath273 .
the positive - part @xmath268 is obtained by using the truncated estimate @xmath274 which avoids an implicitly negative estimate of the prior variance @xmath275 . proceeding analogously , xu and zhou considered the bayesian predictive density estimate , @xmath276 \\[-8pt ] & & \hspace*{65pt}\frac{v_x}{v_x + \nu}v_y + \biggl(1-\frac{v_x}{v_x + \nu } \biggr)(v_x+v_y ) \biggr ) , \nonumber\end{aligned}\ ] ] when @xmath53 and @xmath277 are independent , and @xmath278 . replacing @xmath279 by its truncated unbiased estimate @xmath280 under the marginal distribution of @xmath35
, they obtained the empirical bayes predictive density estimate @xmath281 \\[-8pt ] & & \hphantom{\hat p_{p-2}(y { |}x ) \sim
n_p \biggl ( } v_y + \biggl(1-\frac{(p-2)v_x}{\|x\|^2 } \biggr)_+ v_x \biggr ) \nonumber\end{aligned}\ ] ] where @xmath282 , an appealing simple closed form .
centered at @xmath268 , @xmath283 converges to the best invariant procedure @xmath284 as @xmath285 , and converges to @xmath286 as @xmath287 .
thus , @xmath283 can be viewed as a shrinkage predictive density estimator that `` pulls '' @xmath288 toward @xmath289 , its shrinkage adaptively determined by the data . to assess the kl risk properties of such empirical bayes estimators
, xu and zhou considered the class of estimators @xmath290 of the form ( [ eq : positive_js_hp ] ) with @xmath291 replaced by a constant @xmath292 , a class of simple normal forms centered at shrinkage estimators of @xmath3 with data - dependent variances to incorporate estimation uncertainty .
for this class , they provided general sufficient conditions on @xmath292 and the dimension @xmath0 for @xmath290 to dominate the best invariant predictive density @xmath288 and thus be minimax .
going further , they also established an `` oracle '' inequality which suggests that the empirical bayes predictive density estimator is asymptotically minimax in infinite - dimensional parameter spaces and can potentially be used to construct adaptive minimax estimators .
it appears that these minimax empirical bayes predictive densities may play the same role as the james stein estimator in such problems .
it may be of interest to note that a particular pseudo - marginal empirical bayes construction that works fine for the mean estimation problem appears not to work for the predictive density estimation problem .
for instance , the positive - part james stein estimator @xmath293 can be expressed as @xmath294 , where @xmath295 is the function @xmath296 with @xmath297 ( see stein , @xcite ) .
we refer to @xmath298 as a pseudo - marginal because it is not a bona fide marginal obtained by a real prior .
nonetheless , it plays the formal role of a marginal in the mean estimation problem , and can be used to generate further innovations such as minimax multiple shrinkage james stein estimators ( see george , @xcite ) .
proceeding by analogy , it would seem that @xmath298 could be inserted into the representation ( [ eq : mform ] ) from lemma [ thm : pform ] to obtain similar results under kl loss .
unfortunately , this does not yield a suitable minimax predictive estimator because @xmath299 is not a proper probability distribution .
indeed , @xmath300 and varies with @xmath75 .
what has gone wrong ? because they do not correspond to real priors , such pseudo - marginals are ultimately at odds with the probabilistic coherence of a valid bayesian approach .
in contrast to the mean estimation framework , the predictive density estimation framework apparently requires stronger fidelity to the bayesian paradigm .
moving into the multiple regression setting , stein ( @xcite ) considered the estimation of a @xmath0-dimensional coefficient vector under suitably rescaled quadratic loss .
he there established the minimaxity of the maximum likelihood estimators , and then proved its inadmissibility when @xmath301 , by demonstrating the existence of a dominating shrinkage estimator . in a similar vein , as one might expect , the theory of predictive density estimation presented in sections [ sec : pred - emerges ] and [ sec : theory ] can also be extended to the multiple regression framework .
we here describe the main ideas of the development of this extension which appeared in george and xu ( @xcite ) .
similar results , developed independently from a slightly different perspective , appeared at the same time in kobayashi and komaki ( @xcite ) .
consider the canonical normal linear regression setup : @xmath302 where @xmath303 is a full rank , fixed @xmath304 , @xmath182 is a fixed @xmath305 matrix , and @xmath306 is a common @xmath307 unknown regression coefficient .
the error variance @xmath308 is assumed to be known , and set to be @xmath309 without loss of generality .
the problem is to find an estimator of @xmath58 of the predictive density @xmath310 , evaluating its performance by kl risk @xmath311 where @xmath312 is the kl loss between the density @xmath313 and its estimator @xmath314 the story begins with the result , analogous to aitchison s ( @xcite ) for the normal mean problem , that the plug - in estimator @xmath315 , where @xmath316 is the least squares estimate of @xmath306 based on @xmath75 , is dominated under kl risk by the posterior mean of @xmath313 , the bayes rule under the uniform prior @xmath317 \\[-8pt ] & & { } \times\exp \biggl\ { - \frac { { \mathit{rss}}_{x , y } - { \mathit{rss}}_{x}}{2 } \biggr\}. \nonumber\end{aligned}\ ] ] here , too , @xmath87 is minimax ( liang , @xcite ; liang and barron , @xcite ) and plays the straw - man role of the estimator to beat .
the challenge was to determine which priors @xmath70 would lead to bayes rules which dominated @xmath87 , and hence would be minimax .
analogously to the representation ( [ eq : mform ] ) in lemma [ thm : pform ] for the normal mean problem , the following representation for a bayes rule @xmath77 here , was the key to meeting this challenge .
[ thm : pform : reg ] the bayes rule @xmath318 can be expressed as @xmath319 where @xmath320 @xmath321 @xmath322 , @xmath316 is the least squares estimates of @xmath306 based on @xmath75 , and @xmath323 based on @xmath75 and @xmath74 , and @xmath324 is the marginal distribution of @xmath325 under @xmath326 .
the representation ( [ eq : mform : reg ] ) leads immediately to the following analogue of ( [ eq : uber3 ] ) for the kl risk difference between @xmath94 and @xmath77 : @xmath327 the challenge thus became that of finding conditions on @xmath40 to make this difference positive , a challenge made more difficult than the previous one for ( [ eq : uber3 ] ) because of the complexity of @xmath328 and @xmath329 .
fortunately this could be resolved by rotating the problem as follows to obtain diagonal forms .
since @xmath330 and @xmath329 are both symmetric and positive definite , there exists a full rank @xmath331 matrix @xmath332 , such that @xmath333 \\[-8pt ] d&= & \operatorname{diag}(d_1,\ldots , d_p ) .
\nonumber\end{aligned}\ ] ] because @xmath334 where @xmath335 is nonnegative definite , it follows that @xmath336 $ ] for all @xmath337 with at least one @xmath338 .
thus , the parameters for the rotated problem become @xmath339 \\[-8pt]\hat{\mu}_{x , y } & = & w^{-1 } \hat{\beta}_{x , y } \sim n_p(\mu , d ) .
\nonumber\end{aligned}\ ] ] letting @xmath340 for @xmath341 $ ] , the risk difference ( [ eq : uber3:reg ] ) could be reexpressed as @xmath342 \\[-8pt ] & & { } \qquad - e_{\mu , i } \log m_{\pi_w } ( \hat{\mu}_{x } ; i ) \nonumber\\ & & \quad = h_{\mu}(v_0 ) - h_{\mu}(v_1 ) , \nonumber\end{aligned}\ ] ] where @xmath343 and @xmath344 .
the minimaxity of @xmath79 would now follow from conditions on @xmath345 such that @xmath346 for all @xmath3 and @xmath341.$ ] the following substantial generalizations of theorem [ theo1 ] and corollary [ cor1 ] provide exactly those conditions .
suppose @xmath347 is finite for all @xmath76 with the invertible matrix @xmath332 defined as in ( [ eq : w : reg ] ) .
let @xmath348 be the hessian matrix of @xmath349 . if @xmath350 \ } \le0$]for all @xmath341 $ ] , then @xmath351 is minimax . if @xmath352 \ } \le0$]for all @xmath341 $ ] , then @xmath353 is minimax .
suppose @xmath347 is finitefor all @xmath76 .
then @xmath351 is minimax if @xmath354 \ } \le0 \quad a.e.\ ] ] as a consequence of corollary 2 , the scaled harmonic prior @xmath355 can be shown to yield minimax predictive density estimators for the regression setting .
going further , george and xu ( @xcite ) went on to show that the minimax bayes estimators here can be modified to shrink toward different points and subspaces as in section [ sec5 ] , and that the minimax multiple shrinkage constructions of section [ sec6 ] apply as well .
in particular , they obtained minimax multiple shrinkage estimators that naturally accommodate variable selection uncertainty .
moving in another direction , xu and liang ( @xcite ) considered predictive density estimation in the context of modern nonparametric regression , a context in which the james stein estimator has turned out to play an important asymptotic minimaxity role ; see wasserman ( @xcite ) .
their results pertain to the canonical setup for nonparametric regression : @xmath356 where @xmath349 is an unknown smooth function in @xmath357 $ ] , @xmath358 , and @xmath359 s are i.i.d .
@xmath360 . a central problem here is to estimate @xmath349 or various functionals of @xmath349 based on observing @xmath361 . transforming the problem with an orthonormal basis , ( [ eq : x : nonpa ] )
is equivalent to estimating the @xmath362 s in @xmath363 known as the gaussian sequence model .
the model above is different from the ordinary multivariate normal model in two aspects : ( 1 ) the model is increasing with the sample size , and ( 2 ) under function space assumptions on @xmath349 , the @xmath362 s lie in a constrained space , for example , an ellipsoid @xmath364 .
a large body of literature has been devoted to minimax estimation of @xmath349 under @xmath365 risk over certain function spaces ; see , for example , johnstone ( @xcite ) , efromovich ( @xcite ) , and the references therein . as opposed to the ordinary multivariate normal mean problem
, exact minimax analysis is difficult for the gaussian sequence model ( [ seq : model : nonpa ] ) when a constraint on the parameters is considered .
this difficulty has been overcome by first obtaining the minimax risk of a subclass of estimators of a simple form , and then showing that the overall minimax risk is asymptotically equivalent to the minimax risk of the subclass .
for example , an important result from pinsker ( @xcite ) is that when the parameter space is constrained to an ellipsoid , the nonlinear minimax risk is asymptotically equivalent to the linear minimax risk , namely the minimax risk of the subclass of linear estimators of the form @xmath366 . for nonparametric regression , the following analogue between estimation under @xmath365 risk and predictive density estimation under kl risk was established in xu and liang ( @xcite ) .
the prediction problem for nonparametric regression is formulated as follows .
let @xmath367 be future observations arising at a set of dense ( @xmath368 ) and equally spaced locations @xmath369 .
given @xmath349 , the predictive density @xmath370 is just a product of gaussians .
the problem is to find an estimator @xmath371 of @xmath372 , where performance is measured by the averaged kl risk @xmath373 in this formulation , densities are estimated at the @xmath374 locations simultaneously by @xmath371 . as it turned out
, the kl risk based on the simultaneous formula - tion ( [ risk : simultaneous : nonpa ] ) is the analog of the @xmath365 risk for estimation .
indeed , under the kl risk ( [ risk : simultaneous : nonpa ] ) , the prediction problem for a nonparametric regression model can be converted to the one for a gaussian sequence model .
based on this formulation of the problem , minimax analysis proceeds as in the general framework for the minimax study of function estimation used by , for example , pinsker ( @xcite ) and belitser and levit ( @xcite , @xcite ) .
the linear estimators there , which play a central role in their minimax analysis , take the same form as posterior means under normal priors .
analogously , predictive density estimates under the same normal priors turned out to play the corresponding role in the minimax analysis for prediction .
( the same family of bayes rules arises from the empirical bayes approach in section [ sec7 ] . )
thus , xu and liang ( @xcite ) were ultimately able to show that the overall minimax kl risk is asymptotically equivalent to the minimax kl risk of this subclass of bayes rules , a direct analogue of pinker s theorem for predictive density estimation in nonparametric regression .
stein s ( @xcite ) discovery of the existence of shrinkage estimators that uniformly dominate the minimax maximum likelihood estimator of the mean of a multivariate normal distribution under quadratic risk when @xmath375 was the beginning of a major research effort to develop improved minimax shrinkage estimation . in subsequent papers
stein guided this effort toward the bayesian paradigm by providing explicit examples of minimax empirical bayes and fully bayes rules .
making use of the fundamental results of brown ( @xcite ) , he developed a general theory for establishing minimaxity based on the superharmonic properties of the marginal distributions induced by the priors .
the problem of predictive density estimation of a multivariate normal distribution under kl risk has more recently seen a series of remarkably parallel developments . with a focus on bayes rules catalyzed by aitchison ( @xcite ) , komaki ( @xcite ) provided a fundamental breakthrough by demonstrating that the harmonic prior bayes rule dominated the best invariant uniform prior bayes rule .
these results suggested the existence of a theory for minimax estimation based on the superharmonic properties of marginals , a theory that was then established in george , liang and xu ( @xcite ) .
further developments of new minimax shrinkage predictive density estimators now abound , including , as described in this article , multiple shrinkage estimators , empirical bayes estimators , normal linear model regression estimators , and nonparametric regression estimators .
examples of promising further new directions for predictive density estimation can be found in the work of komaki ( @xcite ) which included results for poisson distributions , for general location - scale models and for wishart distributions , in the work of ghosh , mergel and datta ( @xcite ) which developed estimation under alternative divergence losses , and in the work of kato ( @xcite ) which established improved minimax predictive domination for the multivariate normal distribution under kl risk when both the mean and the variance are unknown .
minimax predictive density estimation is now beginning to flourish .
this work was supported by nsf grants dms-07 - 32276 and dms-09 - 07070 .
the authors are grateful for the helpful comments and clarifications of an anonymous referee . | in a remarkable series of papers beginning in 1956 , charles stein set the stage for the future development of minimax shrinkage estimators of a multivariate normal mean under quadratic loss .
more recently , parallel developments have seen the emergence of minimax shrinkage estimators of multivariate normal predictive densities under kullback leibler risk .
we here describe these parallels emphasizing the focus on bayes procedures and the derivation of the superharmonic conditions for minimaxity as well as further developments of new minimax shrinkage predictive density estimators including multiple shrinkage estimators , empirical bayes estimators , normal linear model regression estimators and nonparametric regression estimators . , . |
fig.1 ( a ) distribution @xmath107 of @xmath32-level population in ion velocity @xmath108 within the first order of the perturbation theory in the field intensity .
different curves correspond to distinct values of detuning @xmath109 ( from left to right @xmath110 ) .
parameters are assumed to be @xmath111 , @xmath112 , @xmath113 . | the profile of a bennett hole induced by laser field in ionic distribution in collisional plasma is calculated .
influence of chandrasekhar s dependence of coefficients of velocity space transport on the profile is included into the calculation for the first time .
it is found that the hole narrows down as the field detuning frequency increases .
physical cause of the effect is the falling dependence of coulomb collision frequency on the ionic velocity .
estimations show that the effect is quite observable under conditions of high - current gas - discharge plasma .
the nonlinear spectroscopic technique as a plasma diagnostic tool is a subject of considerable recent interest because of its novelty and potential in basic and applied plasma physics . of particular interest are the spectroscopic effects of coulomb scattering @xcite , since it differs radically in appearance from binary collisions in gases @xcite . in linear absorption spectra doppler width
usually exceeds collisional broadening , so it masks effects of the scattering .
however , the velocity distribution of particles interacting with a monochromatic field near the resonance is changed significantly by the scattering .
if the frequency @xmath0 of the field is near - resonant with bohr s frequency @xmath1 between levels @xmath2 and @xmath3 of the atoms with velocity @xmath4 that satisfies the condition @xmath5 , then the induced transitions @xmath6 occur ( @xmath7 denotes the wavevector ) . as a result
, the narrow peak or dip ( bennett hole @xcite ) arises in the velocity distribution of level populations .
the steady - state profiles of holes are controlled by both the homogeneous width of transition @xmath6 and the scattering processes with a change in velocity .
measuring the width , shift and asymmetry of nonequilibrium structures in the distribution with the help of probe - field technique one can extract information about the scattering .
previous calculations of coulomb broadening @xcite were based on the assumption of the constancy of the diffusion coefficient in the velocity space .
this assumption is applicable when the detuning frequency @xmath8 is less than the doppler width @xmath9 , where @xmath10 is the thermal velocity of ions and @xmath2 is the ion mass .
the constant diffusion model was enough to interpret lamb dip measurement in argon laser plasma or another with near- resonance pumping .
in contrast , farther away from the resonance , at @xmath11 , one must take chandrasekhar s velocity dependence of the diffusion tensor that complicates the evaluation .
one way around this problem was shown in linear theory of dicke narrowing in the ionic spectrum , which had been built up for low intensity of the incident wave @xcite . in the present letter
we study a first - order nonlinear problem , the influence of collisions in plasma on the profile of the bennett hole .
the calculation presented below for nonlinear structure in the velocity distribution is performed in the basis of density matrix formalism @xcite .
first we derive the formula for the distribution , then analyze it under some simple but realistic conditions , and at the end estimate whether the effect obtained be observable in experiment .
in particular , we show that the width of hole decreases with the detuning of the electromagnetic field from the resonance . by observing the effect , one could directly measure the diffusion tensor .
it would be helpful to confirm our understanding and offer scope for new diagnostic methods .
the diffusion coefficient as a function of the velocity was measured for q - machine plasma at low density by bowles et al . @xcite .
the effect of transient spread of a bennett hole in the distribution of tagged ion population in a metastable state was exploited .
however , that effect can not be used for denser plasma , because the diffusion time becomes smaller and the measurement requires a faster registration technique . in this paper
we shall examine the stationary hole shape .
analysis of its width is useful for diagnostics at high as well as at low density . to develop the theory of nonequilibrium structures in the velocity distribution ,
let us describe the two - level subsystem of a probe ion in an ideal nondegenerate plasma by the spectroscopic density matrix @xmath12 .
this matrix over internal states @xmath13 is at the same time the wigner function of position @xmath14 and velocity @xmath4 .
it has been shown @xcite that elements of the matrix satisfy the quantum kinetic equation with the classical landau collision term .
this equation makes it possible to analyze resonant transitions between levels in the field of the traveling wave @xmath15 c.c . denoting the relaxation constants of level populations as @xmath16 and that of the coherence as @xmath17 , write steady - state equations for off - diagonal @xmath18 and diagonal @xmath19 elements of the density matrix @xmath20 where the upper / lower sign corresponds to upper / lower level .
here @xmath21 , @xmath22 , @xmath23 is the matrix element of the dipole moment , @xmath24 is the coulomb collision frequency , @xmath25 is the charge of a probe ion , @xmath26 is the coulomb logarithm .
we assume that there are only singly charged ions in plasm ; @xmath27 is their concentration .
the excitation of ionic levels @xmath28 usually occurs from the ground or metastable state , so we are justified in assuming the shape of the excitation function @xmath29 to be maxwellian @xmath30 where @xmath31 is the excitation rate of level @xmath32 . for plasma in a state close to the thermodynamic equilibrium with ionic temperature @xmath33
, the distribution of buffer particles has the maxwellian shape .
then it is possible to write an explicit expression for collision operator @xmath34 involving the dynamic friction and the diffusion in the velocity space @xmath35 where @xmath36 is the @xmath37 - component of the dimensionless velocity , the differential operator @xmath38 acts on the right .
functions @xmath39 and @xmath40 , occurring in the expression for the diffusion tensor @xmath41 can be written as single integrals @xmath42 the velocity distribution of populations @xmath43 of both levels @xmath28 needs to be found . to simplify analysis
, we assume the electromagnetic field to be weak @xmath44 and will neglect the effects of strong saturation .
the right sides of eqs .
( [ qke ] ) can therefore be regarded as perturbation .
the formal solution of ( [ qke ] ) in terms of operator exponents is @xmath45 @xmath46 it remains to write out the explicit solution and analyze its limiting cases . for this purpose
we must calculate the commutators of operators @xmath47 and @xmath34 and get , in the general case , the infinite series .
fortunately , the relation @xmath48,\hat a],\hat a]=0 $ ] allows us to break the series within the first order in @xmath49 : @xmath50 , \label{bennett}\end{aligned}\ ] ] where the axis @xmath51 is chosen along the wavevector @xmath7 .
function @xmath52 is the @xmath53 - component of the diffusion tensor .
this function depends on two variables @xmath54 and @xmath55 .
the integrand in ( [ bennett ] ) , obtained by the expansion , is inappropriate at long times @xmath56 .
consequently , for application of this expansion it is necessary to have the main contribution to integral ( [ bennett ] ) in the domain where the quadratic terms in @xmath49 remain small . since @xmath57 , estimation of those terms gives the following applicability conditions of the expansion ( [ bennett ] ) @xmath58 the latter inequality means the doppler limit
; the former one coincides with the condition of smallness of the diffusion broadening compared to the doppler width .
the collision frequency in high - current gas - discharge plasma is about @xmath59
@xmath60 s@xmath61 ( see @xcite ) .
it is by a few orders of magnitude less than the doppler width @xmath62 and , as a rule , less than the relaxation rates @xmath63 , @xmath64 .
that is why we may use small parameters @xmath37 , @xmath65 and it is enough to hold only terms linear in @xmath49 .
taking an integral over @xmath66 we obtain the correction to the distribution function as a single integral over @xmath67 @xmath68 \over \gamma _ j+\phi _ { zz}(\vec\xi)\nu k^2v_t^2t^2/2 - 2i\nu kv_z\phi _
l(\xi ) t}\,dt.\ ] ] the integrand in ( [ nb ] ) regarded as a function of @xmath67 at @xmath69 is maximal at @xmath70 and decreases with @xmath67 .
the shape of the distribution depends on a few parameters .
considering conditions ( [ condition ] ) and the inequality @xmath71 we can set apart two limiting cases of parameters : @xmath72 @xmath73 for the domain ( [ d1 ] ) integral ( [ nb ] ) is accumulated at @xmath74 .
the diffusion and friction cause the bennett hole to somewhat change its width and position . for the limit ( [ d2 ] ) integral ( [ nb ] )
is gathered at @xmath75 .
diffusion and friction effects control the width and shift of the hole .
the width is defined by the maximum among the homogeneous width @xmath76 and diffusion width @xmath77 .
thus , the scattering has stronger impact on shape of the hole , while the relaxation constants of selected levels satisfy inequality ( [ d2 ] ) .
let us find the width as a function of detuning in this limit @xmath78 .
it is convenient to write the expression for the nonlinear correction to the maxwellian distribution as a convolution of two functions @xmath79 where @xmath80.\label{f}\end{aligned}\ ] ] in the limit @xmath81 , the width of function @xmath82 is much less than that of @xmath83 .
accordingly , the central part of the distribution is given by expression ( [ g ] ) @xmath84 at @xmath85 .
meanwhile function @xmath83 drops exponentially with @xmath86 , so the asymptotic of the distribution @xmath87 at @xmath88 is given by a narrow but slowly falling function @xmath82 and has the lorentzian shape .
the most interesting feature of the distribution obtained is the narrowing of its central part as the detuning @xmath8 increases , due to the falling velocity dependence of the collision frequency .
one can estimate the half - width by substituting the extreme value of the longitudinal velocity @xmath89 into all components of the diffusion tensor : @xmath90 this quantity is proportional to the square root of the @xmath53- component of the diffusion tensor .
it falls off with detuning @xmath8 .
the effect is absent in the model of the constant collision frequency @xcite . to describe this effect quantitatively
we calculate the @xmath8- dependence of @xmath91 averaged over the distribution ( [ cb ] ) . keeping only the terms linear in parameters @xmath37 and @xmath65 we obtain @xmath92 @xmath93.\ ] ] the first term on the right - hand side of ( [ width ] )
is independent of @xmath8 and corresponds to the contribution of the lorentzian wings of the distribution .
the second term describes the collisional broadening and decreases with @xmath94 . under the approximations
the line shape is voigt s contour .
the dependence of the variance @xmath95 on detuning @xmath8 consists of two terms : the contribution of the wings , which increases with detuning , and the decreasing contribution of the diffusion broadening . at small detuning @xmath96
we have the explicit expression @xmath97 summarizing the results , we conclude that the bennett hole narrows down as detuning increases .
the physical reason of the effect is that the width of the hole is not affected by the homogeneous width @xmath98 , but depends on the change in velocity due to diffusion @xmath99 , while @xmath100 .
the excited state lifetime @xmath101 determines the stationary width of the hole .
while the detuning @xmath8 of the field remains inside the doppler contour , the field interacts only with a group of slow ions @xmath102 .
when we turn the field off the resonance , the light interacts with faster excited ions . collision frequency of those ions with ions in the ground state @xmath103 is smaller , therefore the width decreases . it would be more proper to call this effect decrease of the broadening instead of the narrowing . to observe the narrowing experimentally
, the probe - field spectroscopy can be used .
one should measure the near - resonant absorption or gain of a weak probe wave as a function of its detuning on the same or adjacent transition .
another possibility is recording the spectrum of spontaneous emission in the presence of a strong resonant continuous field .
let us estimate two alternative effects independent of the detuning , namely , stark broadening and change of velocity due to the interaction with electrons .
the stark effect is quadratic in the arii spectrum .
its value @xcite is about 100 mhz , which is small compared with both the homogeneous width @xmath17 and the diffusion one @xmath104 .
the velocity change owing to scattering on electrons is as small as the ratio of the ion and electron thermal velocities @xmath105 @xcite .
the ion - ion collision frequency decreases no more than by a factor of 23 at the detuning about @xmath9 ; therefore , neither alternative broadening mechanism can compensate the effect in question .
the measured diffusion broadening in low - temperature argon plasma at zero detuning is by a factor of 34 ( see @xcite ) . at the detuning @xmath106 the narrowing factor
expected is more than 10% , so this effect seems observable , see fig . 1 .
we thank s.a.babin and s.g.rautian for stimulating discussions , e.g.shapiro for excellent assistance in computation , and a.v .
shafarenko for valuable comments .
the research described in this publication was made possible in part by grant no .
rcn000 from the international science foundation .
s.a .
babin , d.a .
shapiro , phys . rep . *
241 * , 119 ( 1994 ) .
p.r .
berman , adv . at .
mol .
phys .
* 13 * , 57 ( 1977 ) . s.g .
rautian , a.m. shalagin , kinetic problems of non - linear spectroscopy ( elsevier , amsterdam , 1991 ) .
w.r .
bennett , jr .
, phys . rev . *
126 * , 580 ( 1962 ) ; appl .
opt .
, supplement no 1 , 24 ( 1962 ) .
g.i .
smirnov , d.a .
shapiro , zh .
eksp .
teor .
fiz .
* 76 * , 2084 ( 1979 ) [ sov .
phys .
jetp * 49 * , 1054 ( 1979 ) ] .
e.v .
podivilov , a.i .
chernykh , d.a .
shapiro , zh .
eksp .
. fiz . * 105 * , 1214 ( 1994 ) [ sov .
phys .
jetp * 78 * , 653(1994 ) ] .
s. stenholm , phys .
rep .
* 43 * , 151 ( 1978 ) .
j. bowles , r. mcwilliams , n. rynn , phys . rev .
lett . * 68 * , 1144 ( 1992 ) .
s.g .
rautian , zh .
eksp .
. fiz . * 51 * , 1176 ( 1966 ) [ sov .
phys .
jetp * 24 * , 788 ( 1967 ) ] . l.a .
gelmedova , d.a .
shapiro , j. mod .
opt . * 38 * , 573 ( 1991 ) . |
the _ covering spectrum _ of a geodesic space , introduced by christina sormani and guofang wei ( @xcite , @xcite ) , provides a way to understand the topology of the space at a given scale . their construction uses a classical method of spanier ( @xcite ) that produces a covering map for a given open cover ; in particular , they use the covering map @xmath5 of a geodesic space @xmath1 corresponding to the open cover by @xmath3-balls .
such a cover is constructed for every @xmath6 , yielding a parameterized collection @xmath7 .
the covering spectrum consists of those @xmath6 such that for all @xmath8 , the covering maps @xmath9 and @xmath10 are not equivalent ( see also remark [ swr ] below ) .
sormani and wei used these covering maps , and the covering spectrum , to obtain interesting results concerning limits of compact riemannian manifolds with ricci curvature uniformly bounded below .
valera berestovskii and conrad plaut independently developed a method to construct covering maps determined by geometric information - first for topological groups ( @xcite ) and later for uniform spaces ( @xcite , @xcite ) , hence metric spaces .
unlike spanier s construction , which uses classical ( continuous ) paths and homotopies , the berestovskii - plaut construction uses discrete chains and homotopies , which we will discuss in more detail in the following section .
one consequence is that the construction works for an arbitrary connected metric space and yields , for each @xmath11 , a covering map @xmath12 and corresponding deck group @xmath13 , called the @xmath4-cover and @xmath14-group , respectively .
plaut and wilkins show in @xcite that for compact geodesic spaces the sormani - wei cover @xmath10 is equivalent to @xmath15 when @xmath16 . at the same time , in @xcite , @xcite , and @xcite , they give several topological and geometric applications of the discrete methods .
for example , in @xcite the authors strengthen a theorem of m. gromov ( @xcite ) that the fundamental group of any compact riemannian manifold of diameter at most @xmath17 has a set of generators @xmath18 of length at most @xmath19 and relators of the form @xmath20 .
in fact , they obtain an explicit bound for the number @xmath21 of generators in terms of the number of equivalence classes of short loops at every point and the number of balls required to cover a given semi - locally simply connected geodesic space . as a corollary
they obtain a curvature free fundamental group finiteness theorem ( new even for riemannian manifolds ) that generalizes the finiteness theorems of anderson ( @xcite ) and shen - wei ( @xcite ) . in @xcite
, they investigate the covers of gromov - hausdorff convergent sequences , where discrete methods are particularly useful due to the well - established principle that many properties of the limit of a convergent sequence of compact metric spaces @xmath22 can be formulated in terms of the limits of finite point sets ( cf .
chapter 7 of @xcite ) .
they establish a type of weak closure result for a collection of covers called _ circle covers _ that generalize @xmath4-covers , allowing them to complete the investigation of sormani - wei into the structure of limits of @xmath3-covers . in @xcite , wilkins employs these discrete methods to relate the existence of universal covers of peano continua to the geometry and topology of various generalized fundamental groups .
for example , he shows that such a space has a universal cover if and only if the revised and uniform fundamental groups are finitely presented , which , in turn , holds if and only both groups are discrete in a specific topology called the _ covering topology_. in @xcite , plaut and wilkins ask to what degree these concepts are topological invariants .
the present paper represents a start on that program by moving beyond the geodesic case .
there are many reasons why one would want to pursue such a program , including the existence of new topological invariants . as an example , given a peano continuum @xmath1 , one can consider the collection of all covering maps of @xmath1 ( with connected domain ) that are @xmath4-covers for some geodesic metric ( or more general metric ) on @xmath1 . for instance , up to rescaling , there is only one geodesic metric on a topological circle , and the only @xmath4-covers for that metric are the trivial cover and the universal cover ; the @xmath23-fold covers of the circle are not @xmath14-covers for the geodesic metric .
we conjecture that the same is true for any metric on the circle ( assuming the covers have connected domain ) . on the other hand , @xmath24 with an open disc removed
is homeomorphic to the moebius band , and so there are geodesic metrics on the moebius band that gromov - hausdorff approximate @xmath24 .
it follows from the convergence results of @xcite that the 2-fold cover of the moebius band occurs as an @xmath4-cover for some geodesic metrics . in other words
, one can distinguish between the circle and the moebius band on this basis , despite the fact that the circle is a deformation retraction of the moebius band .
however , other than directly classifying all of the @xmath14-covers in examples like the circle , at this point we have few good methods to show that a given cover is _ not _ an @xmath4-cover for some geodesic metric , and it may be easier to consider more broadly the collection of covering maps that are @xmath4-covers for some compatible metric in general . in this case one
has to consider the possibility that @xmath25 is not connected , and also that the analog of the covering spectrum may not be discrete in @xmath26 .
the latter possibility is one of the main complications of the non - geodesic case , and understanding it is a main thrust of this paper . note also that these discrete methods - unlike geodesic methods - can be applied to fractals with no rectifiable curves like the sierpinski gasket with its resistance metric , which is not geodesic and different from the metric induced by the plane ( cf .
@xcite and @xcite ) .
one may also consider more exotic spaces such as metrized solenoids , and - as was already observed in @xcite - the topologist s sine curve .
finally , one should recall that gromov - hausdorff convergence of non - compact proper geodesic spaces is generally reduced to consideration of metric balls which , even in geodesic spaces , need not themselves be geodesic spaces when equipped with the subspace metric .
the results of this paper are built around a natural defintion of the _ critical spectrum _ for an arbitrary metric space .
the precise definition ( definition [ crit value def ] ) requires some technical background given in section 2 , but the critical spectrum of a metric space @xmath1 - denoted by @xmath0 - can be intuitively thought of as the set of all positive real numbers @xmath4 at which the equivalence type of @xmath4-covers changes as @xmath4 decreases toward @xmath27 .
this spectrum extends both the covering spectrum of sormani - wei and the homotopy critical spectrum of plaut - wilkins ( @xcite ) to the non - geodesic case .
the latter spectra are known to be discrete in @xmath26 for any compact geodesic metric ( see section 3 ) , but for non - geodesic spaces the situation is more complicated .
first of all , we have there exist compact , connected metric spaces with critical spectra that are not discrete in @xmath26 .
in section 3 we give a formal classification of the ways in which discreteness may fail . in section 4 , we give examples of many of the types of critical values deduced in section 3 using spaces we call rapunzel s combs . to begin with ,
one may identify two types of fundamental critical values - homotopy and refinement critical values .
the resulting spectra are denoted by @xmath28 and @xmath29 , respectively .
homotopy critical values are defined for metric spaces in the same way that they are defined for geodesic spaces , and are more topologically significant than refinement critical values , although the latter do indicate a lack of connectivity of certain metric balls ( proposition [ local conn thm ] ) .
in particular , @xmath29 is empty for geodesic spaces , and refinement critical values can often be removed by changing the metric : [ eintrinsic metric ] let @xmath30 be a compact , connected metric space .
then for every @xmath11 there is a metric @xmath31 on @xmath1 such that 1 .
@xmath32 , and @xmath33 if and only if @xmath34 .
2 . @xmath35 and @xmath36 are bi - lipschitz equivalent .
@xmath36 has no refinement critical values greater than @xmath4 .
the metric @xmath31 is called the _
@xmath4-intrinsic metric induced by @xmath35_. note that the only obstruction in theorem [ eintrinsic metric ] to possibly removing _ all _ refinement critical values is that one may have @xmath37 ( see theorem [ ulcc ] ) .
rapunzel s combs show that not all critical values are refinement or homotopy critical values , but these two types determine the entire spectrum : [ spec equals closure hrcv ] let @xmath1 be a compact , connected metric space . if @xmath38 denotes the closure of the homotopy critical spectrum @xmath28 in @xmath39 , then @xmath40 since a geodesic space @xmath1 has no refinement critical values , the discreteness of @xmath28 in @xmath41 when @xmath1 is compact immediately yields the following .
[ geodesic crs ] if @xmath1 is geodesic , then @xmath42 .
if @xmath1 is also compact , then @xmath43 .
we are also able to classify isolated critical values : [ characterize isolated cv ] if @xmath1 is a compact metric space , and if @xmath11 is an isolated critical value of @xmath1 , then @xmath4 is a homotopy or a refinement critical value . finally , it is interesting to consider the limit of the metrics @xmath31 described in theorem [ eintrinsic metric ] , which are monotone increasing as @xmath44 . in general the limit
need not be finite , or if it is finite , topologically equivalent to the original metric .
however , we do have : [ ulcc ] let @xmath30 be a compact , connected metric space .
if @xmath45 is finite , then @xmath46 has no refinement critical values .
if , in addition , @xmath30 and @xmath46 are topologically equivalent then @xmath45 is a geodesic metric and hence @xmath1 is locally path connected . some of the research contained in this paper was carried out by the undergraduate authors during an reu at the university of tennessee in summer 2009 and during the subsequent year .
the quality of contributions by undergraduate co - authors in published papers is known to vary from paper to paper . in the case of the present work ,
the contributions of the undergraduate authors are substantial , including the discovery of rapunzel s combs and insights into the distinct types of critical values in general metric spaces .
omitted proofs of statements in this section may be found in @xcite or @xcite . given a metric space @xmath30
, we will denote the open metric balls of radius @xmath14 centered at @xmath47 by @xmath48 .
if multiple metrics on the same space are being considered , we will distinguish them by subscripts : @xmath49 , @xmath50 , etc .
a map @xmath51 between metric spaces is an _ isometry _ if @xmath52 is surjective and preserves distances ( i.e. @xmath53 for all @xmath54 ) . if @xmath51 is surjective and for some @xmath11 the restriction of @xmath52 to any @xmath55 is an isometry onto @xmath56 , we call @xmath52 an @xmath14-local isometry
. it follows easily that an @xmath14-local isometry is in fact a covering map such that @xmath14-balls are evenly covered ; when @xmath14 is unimportant we may simply call @xmath52 a _ metric covering map_. recall that @xmath1 is a _ geodesic space _ if every pair of points in @xmath1 is joined by a _ geodesic _ , namely an arclength parameterized curve having length equal to the distance between its endpoints . note that when @xmath1 is a geodesic space , @xmath57 is a connected topological space , and @xmath51 is any covering map , the well - known lifted metric on @xmath57 is the unique , topologically equivalent , geodesic metric on @xmath57 such that @xmath52 is a metric covering map .
an @xmath4__-chain _ _ @xmath58 in @xmath1 is a finite sequence of points @xmath59 such that @xmath60 for @xmath61 . for any @xmath4-chain @xmath62
, the _ reversal _ of @xmath58 is denoted by @xmath63 .
a _ basic move _ on an @xmath14-chain is the addition or removal of a single point with the conditions that the endpoints remain fixed and the resulting chain is still an @xmath14-chain . if two @xmath14-chains @xmath64 and @xmath65 have the same initial and terminal points , then we say they are @xmath14__-homotopic _ _ if there is a finite sequence of @xmath4-chains , @xmath66 - called an @xmath4-homotopy - such that each @xmath67 differs from @xmath68 by a basic move . for a fixed basepoint @xmath69 ,
let @xmath25 be the set of all @xmath4-equivalence classes @xmath70_{\varepsilon } $ ] of @xmath14-chains in @xmath1 beginning at @xmath71 , which has a natural metric ( definition [ covermetric ] ) .
one may then define @xmath72 to be the endpoint map taking @xmath73_{\varepsilon } $ ] to @xmath74 . if one assumes that @xmath1 is _ chain connected _ in the sense that every pair of points is joined by @xmath4-chains in @xmath1 for all @xmath11 , then @xmath75 is surjective , and in fact is a covering map and local isometry .
given @xmath76 , there is a well - defined map ( cf .
@xcite ) @xmath77 that simply treats a @xmath3-chain like an @xmath4-chain , i.e. @xmath78_{\delta } ) = [ \alpha ] _ { \varepsilon }
these maps satisfy the composition relation @xmath79 for @xmath80 .
basic arguments show that connected spaces are chain connected , and compact chain connected spaces are connected - although , for example , the rationals with the subspace metric from @xmath81 are chain connected but not connected . to ensure that the maps @xmath82 are always covering maps , we assume from now on that all metric spaces are connected unless otherwise noted .
we are concerned primarily with compact connected metric spaces .
although it is important to keep in mind that all chains technically have both initial and terminal points , for simplicity we will often use the notation @xmath83 instead of @xmath84 for constant or trivial chains .
the notation @xmath85 will denote the ( equivalence ) relation @xmath58 is @xmath4-homotopic to @xmath86 .
we will often write @xmath87_{\varepsilon } $ ] rather than the technically correct @xmath88_{\varepsilon } $ ] .
there is a well - defined , associative concatenation operation on equivalence classes of @xmath4-chains : if the initial point of @xmath65 equals the terminal point of @xmath58 , then we can define @xmath89_{\varepsilon } [ \beta ] _ { \varepsilon }
= [ \alpha \beta ] _ { \varepsilon }
the following properties are easily checked : let @xmath64 and @xmath65 be @xmath14-chains . if the initial point of @xmath65 is the terminal point of @xmath64 , and if @xmath90 and @xmath91 , then @xmath92 . furthermore , @xmath93 is @xmath14-null , and if @xmath94 , then @xmath95 .
an @xmath4-_loop _ is an @xmath4-chain with equal initial and terminal points .
if an @xmath4-loop @xmath96 is @xmath4-homotopic to the trivial loop @xmath97 then we say that @xmath58 is @xmath14-_null_. if @xmath58 is an @xmath4-chain , @xmath98 , and @xmath64 is @xmath4-homotopic to a @xmath3-chain @xmath86 , then we say that @xmath64 _ can be _
@xmath14__-refined to _
_ @xmath65 .
we call such a homotopy an @xmath4__-refinement _ _ , and refer to @xmath86 as a @xmath3-__refinement _ _ of @xmath58 .
when no confusion will result , we will often drop the @xmath4 s and @xmath3 s in our notation .
note that if @xmath58 is an @xmath4-chain in a _
space then we can refine @xmath58 to any desired degree of fineness by successively adding midpoints of geodesics joining each pair of consecutive points in the chain .
refinements of this sort play important roles in @xcite and @xcite , and the ability ( or inability ) to refine chains is an important possible feature of non - geodesic spaces .
the set of classes @xmath99_{\varepsilon } $ ] , where @xmath100 is an @xmath4-loop at @xmath101 , forms a group @xmath102 with operation induced by concatenation .
this group acts on @xmath25 by preconcatenation : for any @xmath4-loop @xmath100 , define @xmath103_{\varepsilon } ) : = [ \lambda \alpha ] _
{ \varepsilon } $ ] .
then @xmath104 is a bijection that depends only on the @xmath4-equivalence class of @xmath100 and , as was shown in @xcite , @xmath105 is a well - defined regular covering map with deck group naturally identified with @xmath102 .
one can think of @xmath106 , roughly , as a fundamental group at a specific metric scale .
for instance , for a geodesic space @xmath1 there is always a natural surjective homomorphism @xmath107 , the kernel of which is generated by classes of loops that are , roughly , small on the scale of @xmath4 .
we define a metric @xmath108 on @xmath109 as follows : [ covermetric ] the length of an @xmath14-chain , @xmath110 , in @xmath30 is @xmath111 the length of the equivalence class @xmath89_{\varepsilon}\in x_{\varepsilon}$ ] is @xmath112_{\varepsilon}):=\inf\{l(\beta):\beta\in \lbrack\alpha]_{\varepsilon } \}$ ] , and for @xmath89_{\varepsilon},[\beta]_{\varepsilon } \in x_{\varepsilon}$ ] , we define @xmath113_{\varepsilon},[\beta]_{\varepsilon}\bigr)=l\bigl(\lbrack \alpha^{-1}\beta]_{\varepsilon}\bigr)$ ] .
note that if the initial point of @xmath86 is the terminal point of @xmath58 , then @xmath114 . we also have @xmath115 and , hence , @xmath116_{\varepsilon})=l([\alpha]_{\varepsilon } ) $ ] .
it is straightforward to check that @xmath117_{\varepsilon } ) \leq l([\alpha ] _ { \varepsilon }
) + l([\beta ] _ { \varepsilon } ) $ ] , from which the triangle inequality for @xmath118 follows .
symmetry is obvious , and positive definiteness is implied by the otherwise useful fact that if @xmath119_{\varepsilon } ) < \varepsilon $ ] , where @xmath62 is an @xmath4-chain , then @xmath120 , @xmath58 is @xmath4-homotopic to the two - point @xmath4-chain @xmath121 , and @xmath112_{\varepsilon } ) = d(x_{0},x_{n})$ ] .
more generally one also sees that for @xmath122 and @xmath4-chain @xmath123 , @xmath124_{\varepsilon } , \delta ) \subset x_{\varepsilon } $ ] consists precisely of those @xmath125_{\varepsilon } \in x_{\varepsilon } $ ] such that @xmath86 is @xmath4-homotopic to an @xmath4-chain of the form @xmath126 , where @xmath127 .
it also holds that @xmath15 is @xmath128-lipschitz ( or distance non - increasing ) and an @xmath129-local isometry .
in fact , every @xmath129-ball in @xmath1 is evenly covered by a union of @xmath129-balls in @xmath109 . as with the case of classical covering space theory , change of basepoint in a chain connected space
induces a natural equivalence of covering spaces , which with the present metric is an isometry .
therefore we will treat base points very informally , but assuming , when needed , that maps are base point preserving .
in particular , if @xmath71 is the base point in @xmath1 we will always use @xmath130_{\varepsilon } $ ] as the base point in @xmath109 . for brevity
we will denote @xmath131_{\varepsilon}$ ] by @xmath132 .
finally , the maps @xmath133 defining the action of @xmath134 on @xmath109 are isometries .
there is an identification of fundamental importance , @xmath135 , that is defined by @xmath136_{\delta } \bigr)=\bigl[\lbrack \ast ] _ { \varepsilon},[\ast , x_{1}]_{\varepsilon } , \dots,[\ast , x_{1},\dots , x_{n}]_{\varepsilon } \bigr]_{\delta } \ ] ] and was shown in @xcite to be a well - defined uniform homeomorphism . according to the metrics we have defined in the spaces @xmath25 and @xmath137 , @xmath138_{\varepsilon},[\ast,\dots , x_{i},x_{i+1}]_{\varepsilon } \bigr)=d(x_{i},x_{i+1}),\ ] ] and , therefore , @xmath139_{\varepsilon } , [ \ast , x_{1}]_{\varepsilon},\dots,[\ast , x_{1},\dots , x_{n}]_{\varepsilon } \bigr\}\bigr).\ ] ] this implies that @xmath140 is in fact an isometry in the case of metric spaces .
for one thing , this fact implies that the basic results that we have proved about @xmath141 hold equally well for the function @xmath142 , including the properties of being lipschitz and a @xmath143-local isometry when it is surjective .
we will denote the restriction of @xmath144 to @xmath145 by @xmath146 .
these maps are homomorphisms ( again , invoke @xmath147 ) , and @xmath148 is injective ( respectively , surjective ) if and only if @xmath149 is injective ( respectively , surjective ) . in the case of geodesic spaces , the functions @xmath148 are always surjective .
in fact , when @xmath1 is geodesic , the metric @xmath108 coincides with the lifted geodesic metric ( proposition 24 , @xcite ) , so that @xmath109 is path connected and , hence , chain connected .
the identification @xmath147 then tells us that @xmath150 is surjective .
an @xmath14-loop of the form @xmath151 will be called an @xmath14__-triangle__. note that an @xmath14-triangle is necessarily @xmath14-null .
if @xmath51 is a map between metric spaces , a lift of a chain , @xmath110 , from @xmath1 to a point @xmath152 is a chain @xmath153 such that @xmath154 for each @xmath155 .
the next lemma shows that @xmath109 largely inherits the local topology and metric properties of @xmath1 .
the first part follows directly from the fact that @xmath156 is an isometry from any @xmath129-ball onto its image .
the last parts were proved in @xcite in the more general setting of uniform spaces .
[ ecover complete ] let @xmath1 be a connected metric space . if @xmath1 is locally compact _
_ ( _ _ respectively , complete _ _ ) _ _ , then @xmath25 is locally compact _ _ ( _ _ respectively , complete__)__. furthermore , suppose the @xmath14-balls in @xmath1 possess any one of the following properties : connected , chain connected , path connected
. then the whole space @xmath25 is , respectively , connected , chain connected , path connected .
[ chain lifting ] let @xmath51 be a surjective map between metric spaces that is a bijection from @xmath4-balls in @xmath57 onto @xmath4-balls in @xmath1 .
let @xmath157 be an @xmath4-chain in @xmath1 , and let @xmath158 be any point in @xmath159 .
then @xmath58 lifts uniquely to an @xmath4-chain @xmath160 beginning at @xmath158 .
if , in addition , @xmath52 has the property that the lift of any @xmath4-triangle in @xmath1 is an @xmath4-triangle in @xmath57 , and if @xmath86 is an @xmath4-chain that begins at @xmath161 and is @xmath4-homotopic to @xmath58 , then the lifts of @xmath58 and @xmath86 to @xmath158 end at the same point and are @xmath4-homotopic .
the first part is proved by induction on the number of points in @xmath64 . if @xmath64 contains one point , the result is trivial .
so , assume the result holds for all chains with @xmath23 or fewer points for some @xmath162 .
let @xmath110 be an @xmath14-chain with @xmath163 points , and let @xmath164 .
using the inductive hypothesis , let @xmath165 be the unique lift of @xmath166 to @xmath158
. then @xmath167 , @xmath52 is a bijection from @xmath168 onto @xmath169 , and @xmath170 .
let @xmath171 be the unique point in @xmath168 mapping to @xmath74 under @xmath52 , and let @xmath172 .
then @xmath160 is an @xmath14-chain and @xmath173 , proving existence .
if there were another @xmath14-chain , @xmath174 , beginning at @xmath158 and projecting to @xmath64 , then by the uniqueness part of the inductive hypothesis , the first @xmath23 points of @xmath175 and @xmath176 must coincide : @xmath177 for @xmath178 .
the bijectivity of @xmath52 on @xmath168 then implies that @xmath171 and @xmath179 must be the same , proving uniqueness . to prove the second part
, it suffices to consider the case in which @xmath58 and @xmath86 differ by only a basic move , since an @xmath14-homotopy is just a finite sequence of basic moves .
suppose @xmath86 is obtained by removing a point from @xmath58 , say @xmath180 and @xmath181 .
let @xmath160 and @xmath182 denote the unique lifts of @xmath58 and @xmath86 to @xmath158 . by uniqueness
, @xmath160 and @xmath182 must agree for their first @xmath183 points .
denote @xmath160 by @xmath184 and @xmath182 by @xmath185 .
since we can remove @xmath186 from @xmath58 , the loop @xmath187 is an @xmath4-triangle .
the lift of this @xmath4-triangle to @xmath188 is an @xmath4-triangle by hypothesis , and , by uniqueness of lifts , the first three points of that triangle must be @xmath188 , @xmath189 , and @xmath190 . since a triangle is a loop , the fourth point of this lift must be @xmath188 .
in other words , we have @xmath191 .
so , @xmath190 and @xmath192 both lie in @xmath193 and project under @xmath52 to @xmath194 , implying that @xmath195 .
finally , by uniqueness of lifts , the rest of @xmath182 agrees with @xmath160 .
thus , @xmath182 is obtained by removing a point from @xmath160 , so @xmath196 . the case for
the other basic move is similar .
[ homotopy lifts ] let @xmath1 be a connected metric space , and let @xmath58 and @xmath86 be @xmath4-chains beginning at a common point @xmath47 that are @xmath4-homotopic .
then their lifts , @xmath160 and @xmath182 , to any @xmath197 are @xmath4-homotopic in @xmath25 .
let @xmath198 be an @xmath4-triangle in @xmath1 , and let @xmath199_{\varepsilon } $ ] be any point in @xmath200 , where @xmath201 .
let @xmath202 and @xmath203 .
then @xmath204_{\varepsilon } , [ \beta ] _ { \varepsilon } )
$ ] and @xmath204_{\varepsilon } , [ \lambda ] _ { \varepsilon } ) $ ] are less than @xmath4 .
moreover , @xmath205 , and since @xmath206 , we can remove @xmath207 from this chain to conclude that @xmath208 .
thus , @xmath209_{\varepsilon } , [ \lambda]_{\varepsilon})<\varepsilon $ ] .
it follows that @xmath210_{\varepsilon } , [ \beta]_{\varepsilon } , [ \lambda ] _ { \varepsilon},[\alpha ] _ { \varepsilon } \}$ ] is an @xmath4-triangle , and it projects under @xmath82 to @xmath198 . uniqueness of lifts now implies that @xmath4-triangles lift to @xmath4-triangles , so lemma [ chain lifting ] then applies . for @xmath14-covers , we can precisely characterize the lifts of @xmath14-chains .
[ chain lifting ecover ] let @xmath1 be a connected metric space , and let @xmath11 be given .
if @xmath211 is an @xmath4-chain beginning at the base point @xmath69 then the unique lift of @xmath58 to @xmath212_{\varepsilon } \in x_{\varepsilon } $ ] is given by @xmath213_{\varepsilon},[x_{0},x_{1}]_{\varepsilon},\dots,[x_{0},\dots , x_{n-1}]_{\varepsilon } , [ x_{0},\dots , x_{n}]_{\varepsilon } = [ \alpha]_{\varepsilon } \bigr\}.\ ] ] in particular , the endpoint of the lift of @xmath58 is @xmath89_{\varepsilon } $ ] and the distances between consecutive points , as well as the chain length , are preserved in the lift .
first , note that , for @xmath214 , @xmath215_{\varepsilon},[x_{0},\dots , x_{i-1},x_{i}]_{\varepsilon}\bigr)=l\bigl(\lbrack x_{i-1},\dots , x_{1},x_{0},x_{1},\dots , x_{i-1},x_{i}]_{\varepsilon}\bigr)\ ] ] @xmath216_{\varepsilon}\bigr)=d(x_{i-1},x_{i } ) < \varepsilon\text{,}\ ] ] since we can successively remove from @xmath217 the point @xmath161 , then each @xmath218 , and so on via an @xmath14-homotopy .
thus , @xmath160 is an @xmath14-chain , and clearly @xmath219 .
the previous lemma immediately yields the following .
[ trivial loop lifts ] if @xmath1 is a connected metric space and @xmath11 , then an @xmath4-loop @xmath220 based at @xmath71 lifts to an @xmath4-loop at @xmath221 if and only if @xmath220 is @xmath4-null .
thus , any representative of a nontrivial element of @xmath102 lifts open _ _ ( _ _ i.e . to a non - loop__)__. [ xe simply connected ] for a connected metric space @xmath1 and any @xmath11
, @xmath109 is @xmath14-connected and @xmath14-simply connected , i.e. every @xmath4-loop based at @xmath221 is @xmath14-null or , equivalently , @xmath222 is trivial .
the @xmath14-connectivity follows from lemma [ chain lifting ecover ] .
given an @xmath14-loop @xmath223 at @xmath224 , it will project to an @xmath14-loop , @xmath225 , at @xmath71 . since @xmath226 lifts to a closed loop
, it is @xmath14-null . by corollary
[ homotopy lifts ] , this @xmath14-nullhomotopy will lift to @xmath109 .
we will postpone some concrete examples until the next section when we introduce the critical spectrum .
[ swr ] in their definition of the covering spectrum , sormani - wei use the condition @xmath227 @xmath228 for all @xmath8 , but from their proofs it is clear that they take this to mean non - equivalence of the corresponding covering maps @xmath229 and @xmath10 . in fact , it seems to be an interesting open question ( in our terminology ) whether it is possible for @xmath25 and @xmath230 to be homeomorphic when @xmath105 and @xmath231 are not equivalent .
recall that it is possible in general for non - equivalent covers to involve homeomorphic spaces - for example with @xmath23-fold covers of the circle .
however , these specific covers may not be @xmath4-covers of the geodesic circle with any compatible metric , as we conjectured in the introduction .
we begin by defining the two aforementioned fundamental types of critical values , the first of which was originally introduced in @xcite .
[ homotopy cv def ] a number @xmath11 is a _ homotopy critical value _ of @xmath1 if there is an @xmath4-loop @xmath220 in @xmath1 that is not @xmath4-null but is @xmath232-null for all @xmath233 ( when @xmath220 is considered as a @xmath3-chain ) .
such a @xmath220 is an _
essential @xmath4-loop_. the _ homotopy critical spectrum _ of @xmath1 is the set @xmath28 of all homotopy critial values .
[ homotopy refinement cv definition ] let @xmath234 be a two - point chain in a metric space @xmath1 such that the following hold : _ 1 ) _
@xmath235 ; _ 2 ) _ for all @xmath3 greater than but sufficiently close to @xmath4 , @xmath64 is not @xmath3-homotopic to an @xmath4-chain .
then we say that @xmath14 is a refinement critical value of @xmath1 and that @xmath236 is an essential @xmath4-gap , or just an essential gap when @xmath14 is clear .
the set of all refinement critical values of @xmath1 is denoted by @xmath29 .
of course , since @xmath237 , an essential @xmath14-gap @xmath236 is not an @xmath4-chain , and the point of the above definition is that , in a sense , we can not refine it to one either .
the existence ( or non - existence ) of essential gaps is closely connected to the surjectivity of the maps @xmath238 ; see lemma [ characterize surjective ]
. even adding refinement critical values to the mix , however , does not tell the full story , and to completely understand the behavior of the @xmath4-covers we need the following general definition of a critical value .
[ crit value def ] let @xmath1 be a chain connected metric space .
a non - critical interval of @xmath1 is a non - empty open interval @xmath239 such that for each @xmath98 in @xmath240 , the map @xmath241 is bijective .
we call @xmath242 a critical value of @xmath1 if and only if it does not lie in a non - critical interval
. the set @xmath0 of all critical values of @xmath1 is called the critical spectrum of @xmath1 .
there are two immediate consequences of the definition .
first , @xmath0 is bounded above by @xmath243 when @xmath1 is compact ( or simply bounded ) .
second , the set of non - critical values , @xmath244 , is open in @xmath26 .
this means that @xmath0 is closed in @xmath41 , although @xmath27 may be a limit point of the critical spectrum . even though we are interested largely in non - geodesic spaces , it is useful to sketch out what happens in some simple but illustrative cases by starting with the geodesic case .
two non - geodesic examples follow the first two , and more are given in section 4 .
_ consider @xmath245 , the geodesic circle of circumference @xmath128 .
the details of this example are given in @xcite . if @xmath246 , then all @xmath14-loops are @xmath14-null , meaning that the @xmath14-cover @xmath109 is isometric to @xmath1 and is , therefore , the trivial cover . for any @xmath247 , however , the @xmath14-loop that wraps around the circle one time ( in either direction ) is no longer @xmath14-null .
the @xmath14-cover for any @xmath248 is now @xmath81 , the universal cover , and these covers `` unravel '' the hole in @xmath1 . since there is a @xmath249-loop in @xmath1 that is not @xmath249-null but is @xmath232-null for all @xmath250 , it follows that @xmath249 is a homotopy critical value of @xmath1 .
it is , in fact , the only critical value of @xmath1 , since @xmath109 is universal for @xmath247 .
@xmath251 _ while basic , this first example illustrates some important phenomena .
first , as noted in corollary [ geodesic crs ] , the critical spectrum of a compact geodesic space @xmath1 is simply its homotopy critical spectrum .
furthermore , as we mentioned previously , one of the fundamental results concerning the covering / homotopy critical spectra of a compact geodesic space @xmath1 is that this set is discrete in @xmath252 when @xmath1 is compact .
these properties do not hold for general metric spaces .
the discreteness was first proved by sormani and wei for their covering spectrum in @xcite , and a stronger result was later established by plaut and wilkins for the homotopy critical spectrum in @xcite . thus , the situation in the previous example is , by extension , typical of compact geodesic spaces .
one can imagine @xmath11 sliding continuously along the positive real axis from the diameter of @xmath1 towards @xmath27 and consider the corresponding covers @xmath253 .
for @xmath254 , the covering map @xmath15 is trivial .
as @xmath4 decreases to @xmath27 , the covering map remains trivial for a while until @xmath4 is the first critical value of @xmath1 , if one exists . at that point , the equivalence class of @xmath156 and the @xmath14-group change from the trivial cover and group , respectively , to the next cover and group in the sequence .
the covering map @xmath15 remains the same for some non - trivial interval , then changes again at the next critical value - again , if one exists .
the common discreteness of both the critical and covering spectra in the compact geodesic case is actually a consequence of a more general result .
plaut and wilkins also showed in @xcite that despite the apparent difference in methods , the berestovskii - plaut construction , when applied to geodesic spaces , yields the same covers as the @xmath232-covers used by sormani - wei , for @xmath16 .
it follows that the covering spectrum and homotopy critical spectrum of a compact geodesic space differ only by a multiplicative factor of @xmath255 . however , the requirement of chain connectedness is in general weaker than connectedness , and is even satisfied by some totally disconnected spaces .
therefore , the requirement of local path connectedness in the spanier construction used by sormani - wei may be dropped when using discrete chains and homotopies . as was noted in the introduction , this opens the way to investigate all of these topics for non - geodesic spaces .
_ for the geodesic hawaiian earring determined by circles with distinct circumferences @xmath256 , the homotopy critical spectrum is @xmath257 . at each of the critical values ,
another circle is unrolled , and @xmath25 consists of an infinite tree with successively smaller hawaiian earrings attached to each vertex .
@xmath251 _
the previous example shows that for a compact geodesic space that is not semilocally simply connected , the critical spectrum is still discrete but may not have a positive lower bound .
this is not true in general , though ; there are compact geodesic spaces that are not semilocally simply connected but have finite or even empty critical spectra ( cf .
@xcite ) . in general , for a compact , semilocally simply connected geodesic space , @xmath258 is the universal cover of @xmath1 for all sufficiently small @xmath11 ; in fact , a stronger statement holds in the more general setting of uniform spaces ( theorem 88 , @xcite ) .
note that in this case it is interesting to consider the so - called _ uniform universal cover of @xmath1 _
( @xcite ) , which is the inverse limit of the covers @xmath253 as @xmath259 .
the bonding maps are the functions @xmath148 mentioned earlier .
the uniform universal cover - which exists for all compact geodesic spaces , even if they are not semilocally simply connected - has the same lifting , regularity , and categorical universality properties as the traditional universal cover , though it is not generally a classical covering space defined by evenly covered neighborhoods . in this paper , however , we are only interested in the maps @xmath15 and @xmath260 and not the inverse limit space .
the next two examples are non - geodesic , and example [ geodesic circle with gap ] gives a simple example of a refinement critical value .
[ euclidean square ] _ let @xmath1 be the square of side - length @xmath261 with its ( non - geodesic ! ) euclidean metric inherited from @xmath262 .
it is not hard to see that , for @xmath263 , @xmath109 is the trivial cover - while for @xmath264 , @xmath109 is the universal cover , which is homeomorphic to @xmath81 .
@xmath251 _
[ geodesic circle with gap ] _ let @xmath265 be the geodesic circle of circumference @xmath128 .
fix @xmath266 so that @xmath267 , and remove the open geodesic segment from @xmath268 to @xmath269 .
let @xmath270 be the resulting set with the inherited subspace metric , not the induced geodesic metric , which would just make @xmath1 a line segment .
choose as the base point @xmath271 the midpoint of the longer segment from @xmath268 to @xmath269 .
see figure [ fig : gap_circle ] ( showing the full sequence of @xmath14-covers ) . as in the case of @xmath265
, @xmath109 is the trivial cover for @xmath246 .
since @xmath4 is larger than the gap we created by removing the segment , an @xmath14-loop can cross over the gap ; informally speaking , @xmath109 does not `` see '' the gap . for @xmath272 , one can show that @xmath273 . in this case
, @xmath109 is the subspace of @xmath81 with the open segments @xmath274 , @xmath275 , removed .
the @xmath14-group , @xmath276 , acts by shifts , as in the case of @xmath265 itself .
intuitively , we still unravel the circle as in the standard geodesic case , but the missing segment of the circle gets unraveled along with it .
note that @xmath25 is not connected in this case .
_ -covers of the geodesic circle with a gap . ]
_ now , suppose @xmath277 .
since the removed segment has length @xmath278 , it becomes the case at @xmath279 that no @xmath4-loop can cross the gap .
thus , the @xmath14-covers now recognize the gap , and it is now impossible to travel around the circle via an @xmath4-loop .
in other words , all @xmath4-loops are necessarily trivial , and @xmath102 is the trivial group .
the pair @xmath280 is , in fact , an essential gap , and @xmath278 is a refinement critical value . in this case ,
@xmath281 , @xmath282 , and @xmath283 .
_ _ a subtle point deserves mention : @xmath156 is topologically trivial for @xmath284 and @xmath277 , but it is only an isometry only for @xmath284 .
the reason for this is that , for @xmath285 , the metric that we have defined makes @xmath25 isometric to a euclidean segment .
in fact , the changes in the topology of the covering space are measuring geometric features of the space rather than topological features .
this example illustrates two things that can occur in non - geodesic spaces but not in geodesic ones : the covering space @xmath25 may not be connected even when @xmath1 is path connected and locally path connected , and @xmath106 may be non - trivial even when @xmath286 is trivial .
@xmath251 _
we now continue to analyze the general critical spectrum .
[ characterize surjective ] let @xmath1 be a metric space , and let @xmath76 be given .
the following are equivalent : _ 1 ) _
@xmath287 is surjective ; _ 2 ) _ every @xmath4-chain in @xmath1 can be @xmath4-refined to a @xmath3-chain ; _ 3 ) _ every two - point @xmath14-chain in @xmath1 can be @xmath4-refined to a @xmath3-chain .
the equivalence of @xmath288 and @xmath289 is obvious , since an @xmath14-chain is the concatenation of two - point @xmath14-chains .
the equivalence of 1 and parts 2 and 3 follows from the definition of @xmath290 .
specifically , given @xmath89_{\varepsilon } \in x_{\varepsilon}$ ] , the existence of an element @xmath291_{\delta } \in x_{\delta}$ ] such that @xmath291_{\varepsilon } = \varphi_{\varepsilon \delta}([\beta]_{\delta } ) = [ \alpha]_{\varepsilon}$ ] means precisely that @xmath64 is @xmath14-homotopic to the @xmath232-chain @xmath65 . in the following , we may occasionally need to refer to the bonding map between two covers @xmath292 and @xmath109 when either @xmath293 or vice versa is possible . in this case
, we will simply refer to `` the map between @xmath230 and @xmath25 , '' and this will mean @xmath150 if @xmath294 and @xmath295 if @xmath296 .
[ critical value seq ] a positive number @xmath4 is in @xmath0 if and only if there is a sequence @xmath297 such that @xmath298 for all @xmath23 , @xmath299 , and the map between @xmath25 and @xmath300 is not bijective for all @xmath23 .
if such a sequence exists , then @xmath4 can not lie in a non - critical interval , for if it did then @xmath301 would also lie in that interval for all large @xmath23 , contradicting that the map between @xmath109 and @xmath302 is not bijective .
conversely , suppose @xmath303 .
then there is no non - critical interval containing @xmath4 .
so there are two positive numbers , @xmath304 and @xmath305 , in @xmath306 such that @xmath307 and the map @xmath308 is not bijective .
we will show that at least one of the maps between @xmath25 and @xmath309 , @xmath310 must be non - bijective .
in other words , we can choose @xmath311 to be @xmath312 or @xmath305 , and the proof is complete by iterating this process to construct @xmath313 . if @xmath4 equals @xmath312 or @xmath305 , we are done . if @xmath314 , then @xmath315 . if both @xmath316 and @xmath317 were bijective , then @xmath318 would be also , a contradiction .
the proofs for the other two possible orderings are similar . in light of this result , there are four possible ways in which a critical value might occur .
that is , @xmath303 if and only if one or more of the following hold ( the sloped arrows indicate strictly increasing or decreasing sequences ) : 1 .
there is a sequence , @xmath319 , such that the map @xmath320 is non - injective ( resp .
non - surjective ) for all @xmath23 . in these cases
, we say that @xmath1 is @xmath14__-upper non - injective _ _ ( resp .
@xmath14__-upper non - surjective _
_ ) , and we call @xmath14 an _ upper non - injective critical value _ ( resp . _ upper non - surjective critical value _ ) .
there is a sequence , @xmath321 , such that the map @xmath322 is non - injective ( resp .
non - surjective ) for all @xmath23 . in these cases , we say that @xmath1 is @xmath14__-lower non - injective _ _ ( resp .
@xmath4__-lower non - surjective _ _ ) , and we call @xmath14 a _ lower non - injective critical value _ ( resp .
_ lower non - surjective critical value _ ) .
as we will see in the last section , each of these possibilities may occur in a compact metric space , and , in fact , two or more of these cases may simultaneously hold for a given @xmath14 .
[ refinable ] spaces for which all maps @xmath144 are surjective exhibit a much stronger connection between critical values and topology . such spaces - which include , for example , all geodesic spaces , all convex subsets of euclidean space , and all peano continua - are called refinable . by definition [ homotopy refinement cv definition ] ,
such spaces have no esssential gaps and , thus , no refinement critical values .
if @xmath4 is a homotopy critical value of @xmath1 , then the maps @xmath323 are non - injective for all @xmath324 , meaning that @xmath1 is @xmath14-upper non - injective .
the converse of this statement is true when @xmath1 is compact geodesic ; that is , every upper non - injective critical value is a homotopy critical value . if @xmath14 is a refinement critical value of @xmath1 then the maps @xmath325 are non - surjective for all sufficiently small @xmath326 , and @xmath4 is an upper non - surjective critical value .
the converse of this statement is not true in general .
example [ geodesic circle with gap ] shows that the two notions may coincide , but it may also be the case that @xmath14 is an upper non - surjective critical value when there is no essential @xmath4-gap , as illustrated in examples [ noncompact non - gap ] and [ rapunzel comb four ] . these examples also show that the second part of lemma [ indicating cv ] below can not be improved to the existence of a refinement critical value . [ indicating cv ] let @xmath76 be given .
if @xmath241 is not injective , then there is a homotopy critical value of @xmath1 in the interval @xmath327 .
if @xmath328 is not surjective , then there is an upper non - surjective critical value in @xmath327 .
if @xmath329 is not injective , then there is a non - trivial @xmath3-loop @xmath220 that is @xmath4-null .
since an @xmath4-chain is also an @xmath330-chain for sufficiently small @xmath326 , and since an @xmath4-homotopy is just a finite sequence of @xmath4-chains , it follows that an @xmath4-homotopy is also an @xmath330-homotopy for sufficiently small @xmath324 .
hence , @xmath226 is also @xmath330-null for small @xmath326 .
let @xmath331 .
note that @xmath332 can not be in this set , for if @xmath220 were @xmath332-null , it would be @xmath333-null for small @xmath326 , contradicting that @xmath332 is the infimum .
thus @xmath220 is @xmath332-nontrivial but @xmath334-null for all @xmath324 , making @xmath335 a homotopy critical value . if @xmath329 is not surjective then there is a two - point @xmath4-chain @xmath336 that can not be @xmath4-refined to a @xmath3-chain .
let @xmath337 and @xmath332 be the infimum of the set @xmath338 of all @xmath339 such that @xmath220 can be @xmath4-refined to a @xmath340-chain .
clearly we have @xmath341 .
moreover , @xmath220 can not be @xmath4-refined to an @xmath332-chain , for such a chain would also be an @xmath342-chain for sufficiently small @xmath343 , contradicting @xmath344 .
suppose @xmath345 were not an upper non - surjective critical value .
then there would be some interval @xmath346 such that @xmath347 and , for every @xmath348 , every @xmath334-chain can be @xmath334-refined to an @xmath332-chain .
but by definition of @xmath335 , we can @xmath4-refine @xmath220 to an @xmath334-chain for some @xmath349
. then we could @xmath334-refine that chain to an @xmath332-chain , yielding an @xmath4-refinement of @xmath220 to an @xmath332-chain , a contradiction .
[ lower non inject ] for a connected metric space @xmath1 the following hold : 1 .
@xmath4 is a lower non - injective critical value of @xmath1 if and only if it is the upper limit of a strictly increasing sequence of homotopy critical values ; 2 .
if @xmath14 is an upper non - injective critical value , then it is either a homotopy critical value or the limit of a strictly decreasing sequence of homotopy critical values .
if @xmath4 is a lower non - injective critical value of @xmath1 then there exists a sequence @xmath321 such that each map @xmath350 is non - injective . by lemma [ indicating cv ] , for each @xmath23 there is a homotopy critical value @xmath351 in the interval @xmath352 . from the sequence
@xmath353 we may choose any strictly increasing subsequence .
conversely , suppose there is a sequence of homotopy critical values , @xmath297 , such that @xmath354 .
then for each @xmath23 , there is some @xmath324 such that @xmath355 and the map @xmath356 is not injective .
the map @xmath357 is equal to the composition @xmath358 . since @xmath359 is not injective , neither is @xmath360 , completing the proof of the first statement .
for the second statement suppose @xmath4 is not a homotopy critical value and consider a sequence @xmath319 such that each map @xmath361 is non - injective .
then there is a homotopy critical value @xmath362 in @xmath363 , and since @xmath364 , we may choose any strictly decreasing subsequence of @xmath353 .
[ lower non surj ] let @xmath1 be a metric space and @xmath11 a lower non - surjective critical value .
then there exists a sequence of upper non - surjective critical values converging up to @xmath4 . if in addition @xmath1 is compact , then @xmath4 is also an upper non - injective critical value .
the proof of the first part is an application of lemma [ indicating cv ] , similar to part 1 of lemma [ lower non inject ] .
going further , by hypothesis , there is an @xmath4-chain @xmath365 that can not be @xmath4-refined to an @xmath366-chain .
let @xmath367 , and note that @xmath368 .
now , let @xmath369 and @xmath370 .
then let @xmath371 be an @xmath14-chain that can not be @xmath14-refined to a @xmath372-chain .
let @xmath373 , noting that @xmath374 and @xmath375 . continuing this process inductively , we obtain a sequence of @xmath4-chains , @xmath376 with @xmath377 such that 1 ) @xmath378 for each @xmath379 , and 2 ) for each @xmath380 , there is some @xmath340 strictly between @xmath381 such that @xmath382 can not be @xmath4-refined to a @xmath340-chain .
in particular , @xmath383 .
if @xmath1 is compact , by choosing subsequences if necessary we may suppose that : @xmath384 and @xmath385 for some @xmath386 such that @xmath387 , and @xmath388 , @xmath389 for all @xmath23 . for each @xmath23
, let @xmath390 , and let @xmath391 denote the loop @xmath392 . note that @xmath393 is an @xmath14-chain and
@xmath394 is an @xmath14-loop . for any @xmath395
, @xmath396 is a @xmath100-chain , since @xmath388 , @xmath397 and @xmath398 . in addition
, the chain @xmath399 can not be @xmath14-homotopic to @xmath396 ; if this were true , then @xmath399 would be @xmath4-homotopic to a chain in which all distances are at most @xmath400 , contradicting the fact that there is some @xmath340 strictly between @xmath401 and @xmath301 such that @xmath399 is not @xmath4-homotopic to a @xmath340-chain .
it follows that @xmath391 is not @xmath4-null .
we next observe that that @xmath402 and @xmath403 .
in fact , if the first inequality did not hold , we would have the following @xmath4-nullhomotopy between @xmath394 and @xmath404 : @xmath405 where the first step is allowed because @xmath406 a similar conclusion follows if the other inequality does not hold .
finally , define @xmath407 then @xmath391 is @xmath408-null , via the same sequence of steps in homotopy [ nullhomotopy ] .
this means that the map @xmath409 is not injective . since @xmath410
, it now follows that @xmath4 is an upper non - injective critical value .
[ refex]suppose that @xmath1 is a compact metric space , @xmath4 is an upper non - surjective critical value , and there exists @xmath411 such that whenever @xmath412 , the map @xmath413 is injective .
then @xmath4 is a refinement critical value . by assumption
, there is a sequence @xmath319 , with @xmath414 for all @xmath23 , such that each map @xmath415 is not surjective .
that is , for each @xmath23 there is an @xmath301-chain @xmath416 that can not be @xmath301-refined to an @xmath4-chain .
we claim that for any @xmath417 @xmath391 can not be @xmath311 -refined to an @xmath4-chain either . if it could be then , equivalently , @xmath418_{\varepsilon
_ { 1}}$ ] would lie in the image of @xmath419 .
but by assumpion , @xmath420 is an injection , and since @xmath421 , that would place @xmath422_{\varepsilon _ { n}}$ ] in the image of @xmath423 , a contradiction . since @xmath1 is compact , by choosing subsequences
if necessary we may suppose @xmath384 and @xmath385 for some @xmath386 with @xmath387 .
we finish by proving that @xmath424 is an essential gap .
suppose , to the contrary , that we can find values @xmath232 strictly greater than but arbitrarily close to @xmath4 such that @xmath226 can be @xmath3-refined to an @xmath4-chain .
we may choose @xmath425 and @xmath23 large enough that @xmath426 and @xmath427 , @xmath428 .
noting that @xmath429 , there is a @xmath3-homotopy @xmath430 combining this with a @xmath3-refinement of @xmath236 to an @xmath4-chain , we obtain a @xmath3-refinement of @xmath391 to an @xmath4-chain .
since @xmath431 , this @xmath232-refinement would also be an @xmath311-refinement , which we showed previously does not exist .
lemmas [ lower non inject ] and [ lower non surj ] together imply that @xmath4 must be upper non - injective or upper non - surjective .
if the former holds , then lemma [ lower non inject ] and the fact that this critical value is isolated imply that @xmath4 is a homotopy critical value .
if @xmath4 is upper non - surjective then since @xmath4 is an isolated critical value , the hypotheses of lemma [ refex ] are satisfied and @xmath4 is a refinement critical value .
compactness is only required for the second part of the proof of theorem [ characterize isolated cv ] , but the theorem is false without it : [ noncompact non - gap ] _ let @xmath1 be the union of the graphs of @xmath432 and @xmath433 , for @xmath434 , and the vertical segment connecting the right endpoints of each graph . give @xmath1 the subspace metric inherited from @xmath262 .
pictured as a subset of @xmath262 , as @xmath435 the tails of this space asymptotically approach a distance of @xmath288 from each other .
in fact , @xmath288 is an upper non - surjective critical value . to see this ,
take any pair of vertically aligned points , @xmath436 , for any @xmath437 .
the distance between these points is @xmath438 .
for all @xmath4 greater than but sufficiently close to this value , this chain can be @xmath4-refined to a @xmath439 chain for any @xmath440 , but it can not be refined to a @xmath288-chain .
( if @xmath4 is large enough we can simply refine it around the other end .
_ ) _ however , there is no essential @xmath288-gap in this space . @xmath251 _ the containment @xmath441 is clear since @xmath442 is closed in @xmath41 .
conversely , let @xmath11 be a critical value of @xmath1 .
lemma [ lower non inject ] immediately handles the cases when @xmath14 is upper or lower non - injective .
if @xmath4 is lower non - surjective then we may use lemma [ lower non surj ] followed by lemma [ lower non inject ] .
if @xmath4 is upper non - surjective then by lemma [ refex ] we need only consider the following case : there exist , for each @xmath380 , @xmath443 and @xmath444 such that @xmath445 and the map @xmath446 is non - injective .
but then lemma [ indicating cv ] implies there is a homotopy critical value in each interval @xmath447 , and the proof is finished .
the next few results further examine the utility and topological significance of refinement critical values .
[ local conn thm ] if @xmath236 is an essential @xmath14-gap in a metric space @xmath1 , then the balls @xmath448 and @xmath449 are disconnected for all @xmath232 greater than but sufficiently close to @xmath14 .
let @xmath450 be an essential @xmath14-gap in @xmath1 .
there is a @xmath451 such that for all @xmath452 , @xmath64 is not @xmath232-homotopic to an @xmath14-chain .
we claim that @xmath448 is not connected for all such @xmath232 ; a parallel argument holds for @xmath449 .
in fact , @xmath448 is not even @xmath14-connected .
if @xmath448 were @xmath14-connected , there would be an @xmath14-chain @xmath453 lying in @xmath454 .
then we could construct a @xmath232-homotopy between @xmath65 and @xmath64 by just successively removing @xmath455 , @xmath456 .
but this contradicts that @xmath236 is an essential @xmath14-gap . the topologist s sine curve with its euclidean metric has no refinement critical values but has a continuum of points at which small balls are not connected .
thus , the converse of the previous proposition is not true .
[ eintrinsic metric def ] let @xmath30 be a connected metric space with @xmath242 .
the @xmath14-intrinsic metric determined or induced by @xmath35 is the metric @xmath36 defined by @xmath457 the fact that @xmath36 is a metric follows from the same type of argument that shows @xmath108 is a metric on the @xmath14-cover , and it follows immediately from the triangle inequality that @xmath458 .
this inequality and the fact that @xmath236 is the shortest @xmath14-chain between @xmath459 and @xmath460 when @xmath461 further imply that @xmath462 if and only if @xmath461 , in which case the two metrics agree .
this also shows that not only are @xmath30 and @xmath463 locally isometric but the topologies induced by @xmath35 and @xmath36 are equivalent .
finally , for fixed @xmath464 , it is easy to see that @xmath465 for @xmath294 , since any @xmath232-chain from @xmath459 to @xmath460 is also an @xmath14-chain .
thus , the function @xmath466 is monotone decreasing in @xmath14 , or increasing as @xmath259 .
the @xmath14-intrinsic metric induced by a given metric can be thought of as a discretized or coarse analog of an induced length metric . if @xmath30 is already a length or geodesic space , then @xmath36 is equal to @xmath35 for every @xmath14 .
thus , one can roughly think of the difference between a given metric @xmath35 and the induced metric @xmath36 as a measure - at a particular metric scale - of how far @xmath35 is from being intrinsic in some sense .
it is also interesting to note that the @xmath4-cover , @xmath109 , with its natural metric @xmath108 , is _ always _
@xmath14-intrinsic , whether @xmath30 is geodesic or not .
one of the results of theorem [ eintrinsic metric ] is that , for each @xmath11 , one can replace the given metric @xmath35 on a compact , connected metric space @xmath1 with a topologically equivalent metric that eliminates all refinement critical values greater than @xmath4 . in particular
, we have the following immediate corollary .
[ eliminate rcv ] if @xmath30 is a compact , connected metric space such that @xmath467 , then there is a metric @xmath468 on @xmath1 that is topologically equivalent to @xmath35 and is such that @xmath469 has no refinement critical values .
this result also supports the idea put forth in the introduction that refinement critical values arise not as much as a result of the underlying topology of the given space as they do as a result of the particular metric one imposes on the space .
part 1 has already been noted .
consider the metric space @xmath470 .
we claim that for any @xmath471 , any two - point @xmath232-chain can be @xmath232-refined to an @xmath14-chain , showing that there are no refinement critical values greater than @xmath14 . but
this essentially follows from the definition of @xmath36 . if @xmath472 , then there is - with respect to @xmath35 - an @xmath14-chain @xmath473 such that @xmath474 .
since the metrics agree on @xmath14-balls , @xmath64 is also an @xmath14-chain in @xmath463 . to see that @xmath64 is @xmath232-homotopic to @xmath236 in @xmath463
, we construct a @xmath232-homotopy as follows .
since @xmath475 is an @xmath14-chain with respect to @xmath35 , we have @xmath476 .
thus , we can remove @xmath455 from @xmath64 via @xmath3-homotopy to obtain @xmath477 .
similarly , @xmath478 is an @xmath14-chain in @xmath30 from @xmath479 to @xmath480 , and @xmath481 .
so , we can then remove @xmath482 from the previous chain to obtain @xmath483 .
continuing in the obvious way , we obtain a @xmath232-homotopy in @xmath463 from @xmath64 to @xmath484 . now , assume @xmath30 is compact , which implies that there is a natural number @xmath485 such that any two points in @xmath30 can be joined by an @xmath14-chain having at most @xmath485 points . if @xmath459 , @xmath486 and @xmath461 , then @xmath487 and we trivially have @xmath488 .
suppose @xmath489 , and let @xmath64 be an @xmath14-chain in @xmath30 from @xmath459 to @xmath460 with @xmath485 or fewer points .
it follows that @xmath490 .
thus , @xmath491 , and the two metrics are bi - lipschitz equivalent .
since @xmath492 is monotone increasing as @xmath14 decreases , it is natural to consider what happens when we let @xmath14 go to @xmath27 .
if we set @xmath493 then it is easy to see that @xmath494 is also a metric on @xmath1 when @xmath495 is finite for all @xmath459 and @xmath460 .
furthermore , we obviously have @xmath496 for all @xmath464 and @xmath242 . in general , however , @xmath494 need not be finite even for a compact connected metric space .
for instance , if we take the usual koch snowflake with its subspace metric in @xmath497 , then @xmath495 will be infinite for all pairs of points .
moreover , even if @xmath494 is finite and uniformly bounded , the resulting metric space @xmath498 need not be topologically equivalent to the original space @xmath30 .
if @xmath1 is the boundary of the unit square in @xmath497 with the vertical segments from the top boundary to the bottom attached at the points @xmath499 , then @xmath1 with the subspace metric is compact and path connected , though not locally connected . in this case
, @xmath495 is finite for every @xmath500 , but the resulting space @xmath498 is not compact .
the issue that arises here is that the while @xmath494 is bounded , the relative distortion @xmath501 is unbounded .
the proof that @xmath46 has no refinement critical values follows as in the proof of theorem [ eintrinsic metric ] .
in fact , if @xmath502 and @xmath236 is a @xmath3-chain with respect to @xmath45 , then @xmath503 , and the preceding proof goes through without change . for the second part , it suffices to prove that for every @xmath504 there is a midpoint between @xmath459 and @xmath460 ( cf .
@xcite or @xcite ) , i.e. a point @xmath505 such that @xmath506 . by definition of @xmath45 for every natural number @xmath183
there is a @xmath507-chain @xmath508 such that @xmath509 .
let @xmath510 be the largest index such that @xmath511 . by the triangle inequality , @xmath512 setting @xmath513 we have by definition @xmath514 . on the other hand , @xmath515 combining these inequalites we obtain @xmath516 by choosing a subsequence
if necessary we may assume that @xmath517 .
since @xmath1 is compact , it follows from dini s theorem ( cf .
@xcite ) that @xmath518 actually converges to @xmath45 uniformly .
thus , given any @xmath519 , we have @xmath520 and @xmath521 for all large @xmath183 . for a fixed @xmath21
, it follows that @xmath522 for all sufficiently large @xmath183 .
letting @xmath523 in these inequalities , we obtain @xmath524 finally , we let @xmath525 , and this is the desired result .
note that an immediate corollary of the above proof is that @xmath463 converges in the uniform sense ( cf .
@xcite ) - hence , in the gromov - hausdorff sense - to @xmath498 as @xmath259 .
in this section we will present several examples illustrating some of the phenomena that can occur concerning the critical spectrum of a general compact metric space .
we will first prove some technical results that facilitate identifying refinement critical values .
roughly speaking , the following construction yields a method for detecting or constructing essential gaps .
indeed , the following discussion and definition should make it clear why the name ` essential gap ' is appropriate for the structure that induces a refinement critical value .
let @xmath1 be a connected metric space .
assume there are points , @xmath504 , with @xmath526 , and a number @xmath527 such that the following holds : for each @xmath4 in the interval @xmath528 $ ] , if we let @xmath529 and @xmath530 , then we can express @xmath1 as a disjoint union , @xmath531 , such that 1 .
@xmath532 and @xmath533 ( hence @xmath534 ) , 2 .
the only points in @xmath535 that are strictly within @xmath4 of a point in @xmath536 lie in @xmath537 , and the only points of @xmath57 that are strictly within @xmath4 of a point in @xmath537 lie in @xmath536 .
if these conditions hold , we call @xmath236 a _ pre - essential gap_. a pre - essential gap need _ not _ be an essential gap . note , also , that though it behaves like one locally around @xmath459 and @xmath460 , @xmath538 need not be a disconnection of @xmath1 . as we have already seen , connected spaces can have refinement critical values . given a pre - essential gap , @xmath236 , with @xmath539 as above ,
let @xmath540 be any @xmath14-chain in @xmath1 .
a pair of consecutive points , @xmath541 , @xmath542 , will be said to _ contain _ or _ cross the @xmath386-gap _ if and only if @xmath543 lies in either @xmath537 or @xmath536 and @xmath186 lies in the other ball .
assign each pair of consecutive points a value @xmath544 as follows : @xmath545 note that the order of the points in the notation @xmath544 does matter ; the second case in this definition , for instance , occurs when the first point of the pair lies in @xmath537 and the second point lies in @xmath536 , while the third case occurs when the opposite holds .
now , define @xmath546 .
we call this the _ @xmath547-gap number of @xmath220 _ ; it measures the net number of times @xmath220 crosses the @xmath386-gap .
[ essential gap homotopy invariance ] assume that the above conditions hold for some @xmath548 , so that @xmath236 is a pre - essential gap . given @xmath4 such that @xmath549 , the integer @xmath550 is an @xmath4-homotopy invariant .
that is , for fixed @xmath551 $ ] , if @xmath58 and @xmath220 are @xmath4-chains such that @xmath552 , then @xmath553 .
since any @xmath4-homotopy taking @xmath220 to @xmath58 will consist of a finite sequence of basic moves , it suffices to prove the result in the case where @xmath58 is obtained by adding or removing a single point to / from @xmath220 .
the proof is not difficult , but it is a tedious process in working through all the possible cases .
we will prove one case to illustrate the reasoning used .
the rest of the cases follow in exactly the same manner .
let @xmath554 , and assume that @xmath58 is obtained by adding @xmath555 between @xmath543 and @xmath186 .
since this basic move only affects three different pairs of points in the sums defining the @xmath547-gap numbers of @xmath220 and @xmath58 , we only need to show that @xmath556 .
assume that @xmath557 .
if @xmath558 , then the result is clear .
if @xmath559 and @xmath560 ( or @xmath561 and @xmath562 ) , the result is also clear . the subcase @xmath563 can not occur , for the first equality would imply that @xmath564 and @xmath565 , while the second would imply that @xmath566 and @xmath567 , which would further imply that @xmath568 , a contradiction .
the case @xmath569 also can not occur , for the first equality would imply @xmath570 and @xmath566 , while the second would imply that @xmath565 and @xmath571 , another contradiction .
suppose @xmath559 and @xmath572 .
then @xmath564 , @xmath565 , and @xmath186 can not be in @xmath537 or @xmath536 ( or else we would have @xmath573 in the latter case and @xmath560 in the former ) .
but @xmath186 must lie in @xmath535 or @xmath57 , and @xmath186 is strictly within @xmath4 of @xmath543 , a point in @xmath537 , and strictly within @xmath4 of @xmath555 , a point in @xmath536 .
if @xmath574 , then , since @xmath575 , condition 2 above implies that @xmath571 , a contradiction . if @xmath576 , then @xmath577 implies that @xmath567 , another contradiction
similar reasoning applies to the cases @xmath561 and @xmath572 , @xmath578 and @xmath562 , and @xmath578 and @xmath560 .
thus , given that @xmath557 , the only possible cases that can occur result in the equality @xmath556 .
proceeding , one would argue simliarly for the cases @xmath579 , and then work through the same procedure in the case where a point is removed from @xmath226 to obtain @xmath64 .
all cases that can occur lead to the desired equality , thus proving the result .
[ essential gap lemma ] let @xmath1 be a chain connected metric space , and suppose @xmath236 is a pre - essential gap with @xmath580 .
if @xmath581 for all sufficiently small @xmath582 , then @xmath236 is an essential @xmath583-gap .
let @xmath527 be as in the definition of a pre - essential gap , and we may assume that @xmath584 is small enough that @xmath581 for all @xmath585 . fix @xmath4 so that @xmath549
. then @xmath586 is an @xmath4-chain , and @xmath587 .
no @xmath583-chain can cross the @xmath386-gap .
in fact , if @xmath588 is an @xmath583-chain , and if we had @xmath589 and @xmath590 , then we would have @xmath591 , contradicting the fact that @xmath592 .
thus , the @xmath547-gap number of any @xmath583-chain must be @xmath27 .
the @xmath4-homotopy invariance of this value then implies that @xmath336 is not @xmath4-homotopic to an @xmath583-chain .
since @xmath593 was arbitrary , it follows that @xmath236 is an essential @xmath583-gap .
suppose the conclusion does not hold .
let @xmath595 , and let @xmath596 be given .
choose @xmath597 such that there are points @xmath598 , @xmath599 satisfying @xmath600 . we can transform @xmath236 via @xmath14-homotopy as follows : @xmath601 .
the second step is valid since @xmath602 . moreover
, this last chain is an @xmath583-chain , since @xmath603 and @xmath604 .
in other words , @xmath236 can be @xmath14-refined to an @xmath583-chain for all @xmath14 sufficiently close to @xmath583 , contradicting the hypothesis .
[ essential gap example ] _ let @xmath605 , @xmath606 , @xmath607 , and @xmath608 be positive real numbers such that @xmath605 is significantly larger than @xmath606 ( say , @xmath609 ) , @xmath610 , and @xmath611
. let @xmath1 be the metric subspace of @xmath262 shown in figure [ fig : egl_example ] , and let @xmath459 , @xmath460 , @xmath612 , and @xmath613 be the points @xmath614 , @xmath615 , @xmath616 , and @xmath617 , respectively .
let @xmath35 be the diagonal distance from @xmath459 to @xmath613 ( or @xmath460 to @xmath612 , by symmetry ) .
_ _ the condition @xmath611 implies that @xmath618 .
let @xmath332 be such that @xmath619 .
now , fix any @xmath4 such that @xmath620 .
let @xmath621 be the left half of @xmath622 , and let @xmath57 be the right half .
the conditions @xmath623 and @xmath624 ensure that the balls @xmath625 and @xmath626 do not intersect or extend to the vertical sides of @xmath1 .
_ _ suppose that @xmath627 and is strictly within @xmath4 of a point in @xmath536 .
clearly , @xmath555 can not lie on the vertical segment of @xmath535 .
if @xmath555 lies on the lower boundary of @xmath535 , then the closest it could be to any point in @xmath536 is @xmath35 , which occurs when @xmath628 .
but @xmath629 , so @xmath555 can not , in fact , lie on this lower boundary .
if @xmath555 lies on the upper boundary of @xmath535 but outside of @xmath537 , then it is at least @xmath630 away from any point of @xmath536 .
thus , it must hold that @xmath555 lies in @xmath537 . likewise , by symmetry , if @xmath631 and is strictly within @xmath4 of a point in @xmath537 , then @xmath565 .
moreover , we clearly have that @xmath581 for sufficiently small @xmath582 .
therefore , @xmath236 is an essential @xmath583-gap , and @xmath583 is a refinement critical value . _
_ to see what makes this essential gap phenomenon occur , consider the trapezoid @xmath632 . the diagonals of this trapezoid are longer than the longest base of the trapezoid .
this , essentially , is why @xmath236 can not be @xmath4-refined to an @xmath606-chain for @xmath633 .
adding @xmath613 in between @xmath459 and @xmath460 is not allowed , because @xmath634 .
likewise , one can not jump from @xmath459 to @xmath612 to @xmath460 for the same reason .
in other words , because the diagonal is too long , one can not overcome the @xmath236-gap by going around it , at least via hops that are sufficiently close to @xmath606 in length .
note that if @xmath635 then we could , in fact , go around the @xmath236-gap .
indeed , for any @xmath636 , we could then transform the @xmath4-chain , @xmath236 , via @xmath4-homotopy by adding @xmath613 and then @xmath612 .
so , it is the diagonal length that makes this gap essential .
_ now , we will use the essential gap lemma to produce examples of compact metric spaces having critical spectra with positive limit points .
moreover , these examples will show that critical values of one type can converge to critical values of the other type .
they will also illustrate some of the other properties mentioned in the previous section .
_ define a metric subspace of @xmath497 by @xmath643 . for @xmath637 ,
let @xmath644 and @xmath645 , and let @xmath646 , @xmath647 , @xmath648 .
let @xmath649 , and , for @xmath380 , let @xmath650 .
note that @xmath651 .
for @xmath652 , let @xmath653 , and note that @xmath654 for @xmath380 .
see figure [ fig : rapcomp1 ] below .
we call @xmath1 a `` rapunzel s comb . '' _ _ the following results can be easily verified : 1 ) @xmath655 , and @xmath656 ; 2 ) @xmath657 @xmath658 , and , for fixed @xmath23 , @xmath659 is minimized when @xmath660 .
now , fix @xmath380 .
it is evident that @xmath661 for sufficiently small @xmath582 .
fix any @xmath4 such that @xmath662 .
let @xmath535 be the left half of @xmath1 , including @xmath207 ( so @xmath535 is closed ) , and let @xmath57 be the rest of the space .
let @xmath663 and @xmath664 .
suppose @xmath665 is strictly within @xmath14 of a point in @xmath666 .
clearly @xmath667 , and @xmath668 .
in fact , the closest any point of @xmath669 can be to @xmath666 is the distance from @xmath207 to @xmath670 , which is greater than @xmath671 .
but @xmath672 , so @xmath668 .
thus , @xmath555 is in the left half of one of the sets , @xmath673 .
however , @xmath555 can not be in @xmath673 for @xmath674 , since - in that case - the distance between @xmath555 and @xmath666 would be at least @xmath675 , which , in turn , is at least as great as @xmath676 . since @xmath677
, this shows that this case can not occur .
hence , @xmath555 must lie on @xmath678 for some @xmath679 .
if @xmath555 were in @xmath678 for @xmath680 , the distance between @xmath555 and @xmath666 would be at least @xmath681 . for fixed @xmath23 ,
@xmath681 is minimized when @xmath660 , so the distance between @xmath555 and any point of @xmath666 is at least @xmath682 .
this contradicts that @xmath555 is strictly within @xmath4 of a point of @xmath666 .
therefore , @xmath683 , and since it is within @xmath4 of a point of @xmath666 , it must lie in @xmath684 . by symmetry , if @xmath631 and is strictly within @xmath14 of a point of @xmath684 , then @xmath685 .
hence , @xmath686 is an essential @xmath128-gap , and @xmath128 is a refinement critical value . _
_ fix @xmath380 , and let @xmath687 . for @xmath688
, @xmath394 is an @xmath4-loop with @xmath689-gap number @xmath690 .
an @xmath4-null loop has @xmath689-gap number @xmath27 , implying that @xmath691 is @xmath4-nontrivial .
this holds for all @xmath688 .
however , for @xmath692 , @xmath391 is easily seen to be trivial .
thus , @xmath693 , and @xmath694 , showing that @xmath442 is not discrete .
note that @xmath1 is compact , path connected , and even simply connected .
_ _ this example illustrates some other interesting properties .
there are infinitely many essential @xmath128-gaps , but there are no non - trivial @xmath128-loops in @xmath1 .
thus , @xmath286 is trivial , and the map @xmath695 is injective for all @xmath4 sufficiently close to @xmath128 .
so , there is a sequence , @xmath696 , of critical values of one type converging to a critical value , @xmath128 , of an entirely different type .
@xmath251 _
there are many different variations on rapunzel s comb that one can use to illustrate critical value limiting behavior .
all of them use the essential gap lemma in some form , and the details follow much as before .
[ rapunzel comb two ] _ for @xmath380 , let @xmath697 and @xmath698 .
let @xmath1 be the subspace of @xmath497 shown in figure [ fig : rapcomb2 ] below ( purposefully not drawn exactly to scale to show detail ) . here ,
the gaps increase in length to a limiting gap of length @xmath128 .
we also define the following : @xmath699 , @xmath700 , @xmath701 @xmath702 @xmath417 , @xmath703 , @xmath704 , and _
@xmath705 _ reasoning as in the previous example , one can show that each @xmath706 , for @xmath380 , is an essential @xmath707-gap .
these values converge up to @xmath128 , but @xmath128 is not a refinement critical value or even an upper non - surjective critical value .
in fact , for every @xmath708 , every @xmath4-chain can be @xmath4-refined to a @xmath128-chain , even the chain @xmath709 . given any @xmath710 ,
the distances @xmath711 and @xmath712 are eventually less than @xmath14 , thus making this refinement possible .
so , what type of critical value is @xmath128 ? _ _ fix @xmath379 , and let @xmath391 be the loop @xmath713 . for all @xmath4 greater than @xmath714 , @xmath391 is an @xmath4-loop .
since @xmath686 is an essential gap , we also know that - for each @xmath4 greater than but sufficiently close to @xmath714 - the @xmath689-gap number is an @xmath4-homotopy invariant .
fixing any such @xmath4 , we see that the @xmath14-chain @xmath715 has non - zero @xmath716-gap number , while the @xmath4-chain @xmath717 does not cross the @xmath718-gap at all .
hence , @xmath719 can not be @xmath4-null for such an @xmath4 .
@xmath391 is @xmath4-nontrivial for all @xmath4 greater than but sufficiently close to @xmath714 . on the other hand
, @xmath391 is @xmath128-null .
in fact , since the diagonals between @xmath74 and @xmath720 are less than @xmath128 in length ( this is easily verified ) , we can successively remove @xmath720 , @xmath721 , and @xmath670 from @xmath391 , giving us the trivial chain .
therefore , @xmath128 is a lower non - injective critical value that is an upper limit of sequences of refinement critical values and homotopy critical values .
note that , as before , @xmath1 is compact , path connected , and simply - connected .
@xmath251 _
[ rapunzel comb three ] _ the construction of this example is very similar to the previous case .
in fact , the lengths of the gaps will be the same .
the key difference will be changing the heights between the gaps in the comb .
we increase them just enough so that the diagonal lengths between @xmath722 and @xmath723 are greater than @xmath128 for each @xmath23 , but still small enough so that the sum of the heights is finite . _ _
so , for @xmath417 , let @xmath724 , and let @xmath725 .
we define @xmath1 exactly as in the previous example except for the different values @xmath726 , and we similarly define the points @xmath722 , @xmath727 , @xmath728 , @xmath729 , and @xmath730 .
also as before , we let @xmath731 , so that @xmath696 is the length of the diagonal between @xmath722 and @xmath723 .
since the construction is the same , figure [ fig : rapcomb2 ] holds equally well for this example .
we just need to keep in mind that the heights , @xmath726 , and , therefore , the diagonals , @xmath696 , are larger in this case .
one can verify by direct computation the following : _ @xmath732 @xmath733 _ in addition , for fixed @xmath23 and @xmath734 , the diagonal lengths , @xmath735 , increase as @xmath736 increases .
_ _ now , fix @xmath379 , and , recalling that @xmath737 , let @xmath4 be such that _ @xmath738 _ the condition that @xmath4 be less than or equal to @xmath739 is to ensure that the ball of radius @xmath740 centered at @xmath74 ( or @xmath670 ) does not intersect any nearby teeth of the comb or either of the vertical sides of @xmath1
. that is , these balls are just segments of the teeth of the comb formed by @xmath741 .
as before , we let @xmath535 be the left half of @xmath1 , and we let @xmath57 be the right half .
suppose @xmath627 and lies within @xmath14 of a point of @xmath742 .
clearly , @xmath555 can not lie in @xmath743 or @xmath744 .
if @xmath555 were in @xmath678 for some @xmath745 , the distance between @xmath555 and any point of @xmath666 would be at least @xmath746 .
but @xmath747 for such @xmath736 , so this can not occur . if @xmath555 were in @xmath748 , then the closest @xmath555 could be to any point of @xmath666 is @xmath749 , but , again , we have @xmath750 .
so , this can not occur either .
niether can @xmath555 be in @xmath678 for @xmath751 , since the diagonal lengths , @xmath746 , are greater than @xmath752 for @xmath751 .
hence , @xmath555 must lie in @xmath741 , and , in fact , it must lie in @xmath753 . by symmetry ,
the same result holds if @xmath631 and lies within @xmath14 of a point in @xmath753 .
we also have @xmath661 for sufficiently small @xmath582 , so @xmath686 is an essential @xmath707-gap .
it follows that @xmath754 is an upper non - surjective critical value ; for all @xmath4 greater than but sufficiently close to @xmath714 , @xmath686 is an @xmath4-chain that can not be @xmath4-refined to a @xmath707-chain . _ _ finally , since @xmath755 , we know that @xmath128 is a critical value .
note that _ @xmath756 _ because all diagonals have length greater than @xmath128 and _ @xmath757 _
thus , for @xmath758 , the @xmath689-gap number of an @xmath4-chain is an @xmath4-homotopy invariant .
now , @xmath686 is a @xmath128-chain , and its @xmath759-gap number is @xmath128 .
however , no @xmath707-chain can cross the @xmath718-gap .
thus , @xmath686 can not be @xmath128-homotopic to a @xmath760-chain . in other words ,
the map @xmath761 is not surjective , and this holds for all @xmath380 .
hence , @xmath128 is a lower non - surjective critical value .
as we have shown , such a critical value can only occur as the upper limit of upper non - surjective critical values .
@xmath251 _
[ rapunzel comb four ] _ working as in the previous examples , let @xmath1 be the variation of rapunzel s comb shown in figure [ fig : rapcomb3 ] . here
, we have a sequence of gaps of length @xmath762 converging down to a gap of length @xmath128 .
moreover , we have attached an extra single set of teeth below the limiting gap , which adds a gap of length @xmath763 .
_ _ we choose the heights , @xmath764 , to be @xmath765 , and we choose @xmath608 so that @xmath766 and @xmath767 , @xmath768 .
a straightforward computation shows that this can be done .
it can then be shown as in the previous examples that @xmath769 is an essential @xmath770-gap for each @xmath23 . thus , for each @xmath23 and all @xmath14 greater than but sufficiently close to @xmath771 , @xmath769 can not be @xmath14-refined to a @xmath770-chain and , thus , to a @xmath128-chain , either . hence , @xmath128 is an upper non - surjective critical value .
however , there is no essential @xmath128-gap in @xmath1 . because we have added the gap @xmath280 below the gap , @xmath772 , and chosen @xmath608 so that @xmath773 , the chain @xmath709 can be @xmath4-refined to a @xmath128-chain for all @xmath4 greater than @xmath128 .
the homotopy , itself , is simply _
@xmath774 _ since @xmath775 , this is a @xmath128-chain .
this example shows that - even in a compact space - an upper non - surjective critical value need not correspond to an essential gap .
as we have already shown , the two only correspond for certain when the critical value in question is isolated .
@xmath251 _
m. gromov , _ metric structures for riemannian and non - riemannian spaces _ , with appendices by m. katz , p. pansu , and s. semmes .
translated from the french by sean michael bates .
progress in mathematics , 152 , birkhuser boston , inc . ,
boston , ma , 1999 . | utilizing the discrete homotopy methods developed for uniform spaces by berestovskii - plaut , we define the critical spectrum @xmath0 of a metric space , generalizing to the non - geodesic case the covering spectrum defined by sormani - wei and the homotopy critical spectrum defined by plaut - wilkins .
if @xmath1 is geodesic , @xmath0 is the same as the homotopy critical spectrum , which differs from the covering spectrum by a factor of @xmath2 .
the latter two spectra are known to be discrete for compact geodesic spaces , and correspond to the values at which certain special covering maps , called @xmath3-covers ( sormani - wei ) or @xmath4-covers ( plaut - wilkins ) , change equivalence type . in this paper
we initiate the study of these ideas for non - geodesic spaces , motivated by the need to understand the extent to which the accompanying covering maps are topological invariants .
we show that discreteness of the critical spectrum for general metric spaces can fail in several ways , which we classify .
the newcomer critical values for compact , non - geodesic spaces are completely determined by the homotopy critical values and _ refinement _ critical values , the latter of which can , in many cases , be removed by changing the metric in a bi - lipschitz way . _
key words and phrases _ : metric space , discrete homotopy , critical spectrum , homotopy critical value , refinement critical value . 2010 _ mathematics subject classification _ : primary 54e45 ; secondary 54g20 , 53c23 . |
recently , much attention has been paid to the exactly solvable models of one - dimensional magnetism , consisting of small quantum spin clusters connected to each other via ising spins.@xcite for the sake of simplicity we will call them heisenberg ising chains ( hic ) .
the interest toward such spin systems stems from several features they possess .
first of all , they allow one to obtain an exact statistical - mechanical solution of the problem of strongly interacting spins in terms of classical transfer matrix@xcite .
in addition to that , such systems combine the quantum and classical properties and also , to some extent , are connected to the problems and models of molecular magnetism@xcite , as the classical transfer matrix is constructed by the diagonalization of the small quantum spin cluster . and
finally , very recently the family of trimetallic coordination polymer compounds has been synthesized@xcite .
these compounds in the magnetic sense are indeed a one - dimensional magnets with both classical ( ising ) and quantum ( heisenberg ) bonds . the single - chain magnet reported in ref .
is an example of a heisenberg - ising chain of triangles with three - spin linear quantum clusters ; another interesting molecular magnet ( spin ring ) based on the highly anisotropic properties of the dy@xmath0 ions has been reported in ref . .
both systems have been analyzed by the transfer - matrix solution of hic chains in ref . .
another class of examples is given by chain magnets with two alternating magnetic ions , thus giving rise to alternating quantum and classical spins ( see , e.g. , refs . ) .
let us describe the general scheme of the hic : + * there is a small cluster of a few quantum spins with a certain topology of bonds which could contain various types of spin - spin interaction as well as interactions with external fields .
the hamiltonian of the single small cluster is , thus , a finite dimensional matrix . + * these clusters are assembled to the chain or any other one - dimensional structure by alternating with the single ising spins . +
* the interaction between the single ising spins and any quantum spin from a small cluster must contain only @xmath1-components ( ising interaction ) . for the mixed quantum
classical spin system with the above mentioned properties the exact solution is obtained in the following way .
the idea is to apply a classical transfer - matrix formalism in which the entries of the classical transfer matrix are obtained from the diagonalization of a small quantum spin cluster .
although the number of papers devoted to the exact solutions of the hic and to their magneto - thermal properties is rather impressive , only a few models have been solved within the classical transfer matrix formalism@xcite , while the other method of the so - called decoration iteration transformation has been also used@xcite .
particularly , within the direct classical transfer matrix technique the following models have been solved exactly : the double - tetrahedral chain with quantum triangles of spins@xcite , the sawtooth chain with an asymmetric pair of quantum spins@xcite , diamond chain@xcite and orthogonal - dimer chain with anisotropic and asymmetric quantum triangles@xcite . despite the complete and exact description of the ground states and ground states phase diagrams , as well of all thermodynamic functions , finite - temperature entropy and magnetization , magnetocaloric effect , etc , the expressions for the correlation function are still missing .
it is worth mentioning , that the correlation functions can be in principle calculated by another method which is used to handle the hic , the technique of the decoration iteration transformation@xcite .
however , the method of the classical transfer matrix is more straightforward and is based on a direct application of the canonical expectation values@xcite , which can be calculated exactly . in the present paper we are going to develop an exact technique for the calculation of any finite - temperature correlation function for any hic by means of the direct application of the classical transfer - matrix method . beside the pure academic interest , the knowledge about the exact correlation functions of the hic recently became important from the experimental point of view as well .
as mentioned above , a few years ago it was reported about the synthesis of a novel class of trimetallic coordination polymer compounds@xcite which are the real examples of the hic ( at least one of them ) .
as known , the neutron scattering is one of the most powerful and straightforward methods of investigation of the magnetic structure , spin textures , magnetic excitations , etc . in the leading order
the cross - section of the neutron scattering is given by the two - point spin - spin correlation functions ( their fourier - transform ) .
this fact makes the topic of the present paper a practical issue .
the paper is organized as follows . in the second section
we remind the reader of the main features of the classical transfer - matrix formalism for the hic , in the next section we present the formalism for calculating the correlation function of hic in terms of traces of certain matrix products and prove that for all kinds of correlation functions the correlation length is the same . in section
iv we apply the general formalism for several models .
the paper ends with a conclusion .
the hamiltonian of a hic has the form of a sum of local block hamiltonians @xmath2 which correspond to the n - th small quantum cluster and its interaction with left and right ising spins : @xmath3 here @xmath4 is the ising spin situated between the n - th and ( n+1)-th small cluster and @xmath5 is the magnetic field acting on ising spins .
the block hamiltonian can contain various terms with quantum and classical spin variables which are determined by the topology of the elementary block and the character of interactions .
let us denote the quantum spins by @xmath6 ; then one can suppose that the block hamiltonian depends on @xmath6 , @xmath7 and may be on other classical spins @xmath8 , @xmath9 due to the properties listed above the block hamiltonians commute to each other , @xmath10=0.\ ] ] this fact allows one to expand the exponential in the expression for the partition function of the model and obtain an expression which by its structure corresponds to the classical transfer matrix scheme : @xmath11 here the cyclic boundary conditions are supposed , @xmath12 are the eigenvalues of the block hamiltonian @xmath2(which is supposed to be an @xmath13 by @xmath13 matrix ) depending on the values of classical spins @xmath14 and classical spins @xmath4 which alternate with the blocks , @xmath15 means the trace over the states of the @xmath16-th quantum block , and @xmath17 stands for the trace of the 2 by 2 transfer matrix with entries @xmath18 .
after summation over the values of @xmath14 for each block one gets the expression depending only on @xmath4 and @xmath19 which is the classical 2 by 2 transfer matrix @xmath20 where @xmath21 is the partition function of one block , depending on the values of two neighboring ising spins , signs `` @xmath22 '' and `` @xmath23 '' in indexes correspond to @xmath24 and @xmath25 respectively .
the formalism of classical transfer - matrix allows one to obtain exact analytic expressions for almost any expectation values of the function of local variables ( spins ) and , of course , for any thermodynamic functions , as the expression for the free energy per block can be obtained in a straightforward way by finding the maximal eigenvalue of the transfer matrix ( [ t ] ) : @xmath26 from this expression any thermodynamic function , like magnetization , entropy , specific head etc can be obtained exactly
. however , to obtain correlation functions of the spins demands more efforts and further development of the classical transfer - matrix technique .
let us concentrate on the scheme of calculation of the expectation values of combinations of spins operators .
first of all , let us start from the simplest case , the expectation values of @xmath7-s , which is quite similar to that of the ising chain@xcite .
thus , for the thermodynamic average of just one @xmath7 one obtains @xmath27 this expression is the trace of the product of @xmath28 transfer - matrices with the matrix @xmath29 inserted before the @xmath30-th transfer matrix in the trace formula .
thus , we have @xmath31 where we used the matrix identity @xmath32 , with the diagonal matrix @xmath33 and the orthogonal matrix @xmath34 .
thus , @xmath35 where @xmath36 are the eigenvalues of the transfer - matrix , and @xmath37 is a parameter . having all these notations we now can express the canonical expectation values from eq .
( [ m_sig ] ) in terms of the entries of the orthogonal matrix diagonalizing the transfer - matrix . using the properties of the trace ( cyclic
permutation inside the trace does not change it ) one can get @xmath38 now , we should take care about the thermodynamic limit ( @xmath39 ) .
it is easy to see , that in this limit eq .
( [ m_sig3 ] ) leads to ( suppose that @xmath40 is greater than @xmath41 ) @xmath42 where @xmath43 is the partial magnetization corresponding to the subsystem with @xmath7-spins , @xmath44 thus , we can express the entries of the orthogonal matrix diagonalizing the transfer matrix in terms of the subsystem magnetization @xmath43 : @xmath45 these quantities are the building blocks for the calculation of all correlation function for the hic . as will be clear from further consideration , any correlation function of hic has the form of the trace of a certain product of transfer matrices and special matrices representing the spin operators .
moreover , these matrices are constructed from the local expectation values of spin operators with the aid of the matrix @xmath46 in the standard way @xmath47 .
let us write down the entries of a general matrix of that type @xmath48 , \\ & & { \widetilde}{p}_{+-}=\frac{1}{2}\left[\sqrt{1-m_{\sigma}^2}\left(p_{--}-p_{++}\right ) + \left(1+m_{\sigma}\right)p_{+-}-\left(1-m_{\sigma}\right)p_{-+}\right],\nonumber \\ & & { \widetilde}{p}_{-+}=\frac{1}{2}\left[\sqrt{1-m_{\sigma}^2}\left(p_{--}-p_{++}\right ) + \left(1+m_{\sigma}\right)p_{-+}-\left(1-m_{\sigma}\right)p_{+-}\right],\nonumber \\ & & { \widetilde}{p}_{--}=\frac{1}{2}\left[\left(1-m_{\sigma}\right)p_{++}-\sqrt{1-m_{\sigma}^2}\left ( p_{+-}+p_{-+}\right)+ \left(1+m_{\sigma}\right)p_{--}\right].\nonumber\end{aligned}\ ] ] for instance , the matrix @xmath49 which is the building block for constructing various correlation functions of @xmath50 has a particulary simple form : @xmath51 the simple two - site correlation function is just a thermal average of the product of two @xmath7 spins situated at a distance @xmath52 from each other : @xmath53 thus , using the matrix form of @xmath54 and taking a thermodynamic limit one obtains : @xmath55 where @xmath56 is a correlation length .
so far , all results have the same form as those for a simple ising chain , the difference is only in the form of the magnetization for @xmath7-spins , which in our case is given by eqs .
( [ m_via_t ] ) and ( [ t ] ) .
now , let us turn to the more complicated correlation functions of the quantum spins with each other and mixed correlation functions , @xmath57 where the new 2 by 2 matrices @xmath58 and @xmath59 are the classical transfer - matrix formalism representation of the @xmath60-component of the @xmath61-th quantum spin and classical @xmath61-th spins from the corresponding quantum clusters .
their entries are found according to @xmath62 the correlation functions containing the quantum spins @xmath63 have the unusual overall power of the @xmath33-matrix smaller than the number of the blocks @xmath28 .
the point is that the inclusion of each quantum spin into the function reduces the overall power of the @xmath33-matrix by one .
this technical difference from the standard classical technique@xcite stems out from the following feature .
one does not have the spin operators @xmath64 by themselves under the trace in the canonical expectation value expression corresponding to the correlation function . instead , there are some complicated matrices given by eq .
( [ sigma , tau ] ) and representing the quantum spin operators in the generalized classical transfer - matrix formalism . as it is seen from eq .
( [ sigma , tau ] ) , due to the quantum nature of the variables @xmath63 , the corresponding matrix representation for them adsorbs the transfer matrix and it is an issue to find a representation where the resulting matrix under the trace into the partition function would be just a product of the transfer - matrix and some other matrix . on the contrary , for the classical spin chains for arbitrary spin
the expression @xmath65 is the analog of the eq .
( [ sigma , tau ] ) which appears under the trace formula in the corresponding correlation function .
thus , the corresponding matrix to be inserted into the trace formula can be easily and naturally split into the product of the transfer matrix and a diagonal matrix with the possible values of the classical variable @xmath63 on its diagonal@xcite .
this fact leads also to another consequence concerning the properties of the correlation functions from eq .
( [ cor_ss_ssig ] ) .
the expressions are no longer valid for @xmath66 .
all these cases should be considered separately out of the framework of the technique developed here . however , the correlation functions at @xmath66 for the same spin operator are trivial because they are just a expectation values of the square of spin , while correlation functions for the different spins belonging to the same block can be easily calculated by taking a derivatives of the free energy with respect to the corresponding coupling constants .
as the expressions under the trace in eq.([cor_ss_ssig ] ) have the same structure , let us write down a general expression for such a correlation function @xmath67 where the correlation length @xmath56 is the same as in eq.([cor_sig_sig ] ) . obviously , the part independent of @xmath52 is the product of expectation values of quantities corresponding to matrices @xmath68 and @xmath69 , @xmath70 the product of two eigenvalues of the 2 by 2 transfer matrix is its determinant , thus , eventually one can write down the following general form of pair - correlation function for hic : @xmath71 if one of the matrices is @xmath7 the result is different , because the corresponding expression contains one more transfer matrix . in general , it also depends on the position of the @xmath7 , unless the matrix @xmath72 is symmetric .
thus , the corresponding expressions for the correlation functions are @xmath73 the matrix @xmath72 , in its turn , is symmetric when the corresponding matrix @xmath68 is symmetric .
thus , the order of @xmath74 and @xmath75 in the correlation function does not matter in this case : @xmath76 . from this general consideration
one can see that there exists only a single correlation length for any kind of correlation functions of the hic .
this is rather expected result because the appearance of a single correlation length even in the inhomogeneous situation is the common feature of all one - dimensional systems which do not become critical at finite temperature .
this is , however , true only for a long range correlation with respect to the block structure of the system ; inside one block short range correlations are non - uniform .
in this section we are going to illustrate the technique of calculation of correlation functions for hic by several examples . as the technique , despite its clearness and simplicity , leads to rather cumbersome expressions , we will present explicit calculations for the simplest models which have been investigated earlier in a context of magnetic and thermodynamic properties@xcite .
one can distinguish at least two classes of hic : the models where each block has a left - right symmetry with respect to the interaction with @xmath7 spins@xcite , which implies that the term describing the interaction between block spins and their left and right @xmath7-neighbors has the form @xmath77 , which is completely symmetric with respect to the permutation of the operators @xmath78 , and the models which are non symmetric with respect to left and right @xmath7-spins on each block@xcite .
it is worth mentioning , that the experimentally obtained single chain magnet compound with ising and heisenberg bonds , which is an existing example of hic , belongs to the second type@xcite .
magnetic properties of the asymmetric model are vast and complicated , they can exhibit a large number of ground states with a breaking of translational symmetry or doubling of the unit cell ( block)@xcite . in the most symmetric case the block hamiltonian is completely symmetric with respect to the permutation of all spins in it .
such models are the simplest ones .
for instance , the various variants of the heisenberg ising diamond chain@xcite and heisenebrg - ising tetrahedral@xcite and double - tetrahedral chains@xcite belong to this class of hic .
it is obvious , that in this case the matrices @xmath79 are the same for all spins inside one block , or , in other words , in this case one does not need the additional index @xmath61 . heisenberg - ising diamond chain .
solid ( dashed ) lines denote quantum xxz ( classical ising ) interaction bonds .
all dashed bonds correspond to the coupling constant @xmath80 , the interaction between quantum spins is characterized by the coupling constant @xmath81 and anisotropy @xmath82 . ]
the symmetric heisenberg - ising diamond chain is the simplest example of hic ( see figure [ fig1 ] ) .
its block hamiltonian has the form : @xmath83 the corresponding transfer - matrix is @xmath84+q^{dc}\right ) & { \operatorname{ch}}\left(\beta h \right)+q^{dc } \\ { \operatorname{ch}}\left(\beta h \right)+q^{dc } & e^{-\beta\frac{h}{2}}\left({\operatorname{ch}}\left[\beta\left(h+k \right)\right]+q^{dc}\right ) \end{array}\right),\end{aligned}\ ] ] where @xmath85
as the block hamiltonian is so(2)-invariant with respect to the @xmath1-axis the corresponding @xmath86-matrix is non - zero only for @xmath87 and , as discussed above , in virtue of the symmetry with respect to permutation of @xmath6-operators for one block @xmath88 .
the matrix @xmath89 has a rather simple form : @xmath90 & { \operatorname{sh}}\left(\beta h\right ) \\ { \operatorname{sh}}\left(\beta h\right ) & e^{-\beta\frac{h}{2}}{\operatorname{sh}}\left[\beta\left ( h+k\right)\right ] \end{array}\right)\nonumber\end{aligned}\ ] ] according to eqs.([tild_p_ent])-([corr_psig ] ) , the non - zero two - site correlation functions for this model are @xmath91\sqrt{1-m_{\sigma}^2 } \right)^2}{16 \left({\operatorname{sh}}^2\left ( \beta k\right)+2 q^{dc } \left ( { \operatorname{ch}}\left(\beta k\right)-1\right){\operatorname{ch}}\left(\beta h\right ) \right)}e^{-\frac{r}{\xi}},\\ & & \langle s_{j , a}^z \sigma_{j+r}^z\rangle-\langle s_{j , a}^z\rangle\langle \sigma_j\rangle=-\frac{e^{-\beta\frac{j}{4}}\left\{2{\operatorname{sh}}\left(\beta h\right)m_{\sigma}\sqrt{1-m_{\sigma}^2}+\left[e^{\beta k}{\operatorname{ch}}\left(\beta\frac{h}{2}\right)-e^{-\beta k}{\operatorname{ch}}\left(\beta\frac{3h}{2}\right)\right]\left(1-m_{\sigma}^2\right)^2 \right\}}{4\lambda_1^{dc}}e^{-\frac{r}{\xi } } \nonumber\end{aligned}\ ] ] where @xmath92 is the partial magnetization of the sublattice of quantum spins .
@xmath43 is the magnetization of @xmath7-sublattice obtained according to eq.([m ] ) from transfer - matrix ( [ tdc ] ) and @xmath93 is the largest eigenvalue of the transfer - matrix ( [ tdc ] ) .
the eigenvalues have the following form : @xmath94 ^ 2 + 4\left ( { \operatorname{ch}}\left(\beta h\right)+q^{dc}\right)^2}\right\}.\nonumber\end{aligned}\ ] ] heisenberg - ising sawtooth chain with a pair of quantum spins in one block .
the unit cell consists of 4 spins , @xmath95 and @xmath96 .
all bonds denoted by dashed lines are of ising type and the corresponding coupling constant is @xmath80 , while the @xmath97-interaction corresponding to the quantum bond ( solid line ) is given by the coupling and anisotropy @xmath81 and @xmath82 . ] the next example we are going to consider is the heisenberg - ising sawtooth chain introduced in refs . and .
the lattice is depicted in figure [ fig2 ] .
this model is quite convenient for illustrating the asymmetric role of quantum spins from one block as well as for the demonstration of the role of the additional classical spin included in the block . here
the block hamiltonian contains two parts , one with quantum operators and another one - without them .
thus , one can write down the block hamiltonian for the sawtooth hic with 2 quantum spins in one block in the following way : @xmath98 here all notations are explained in figure [ fig2 ] .
in contrast to the case considered previously in refs . and , here , for the sake of simplicity , we consider all ising bonds with the same coupling constant @xmath80 . as
the block hamiltonian contains also the ising spin @xmath14 , in order to obtain the entries of transfer matrix one needs , after taking the trace over quantum states of the pair of @xmath6-spin , to sum over the two possible values of the classical spin @xmath14 .
the entries of the resulting classical transfer matrix are : @xmath99+e^{\beta\frac{k}{4}}\left[{\operatorname{ch}}\left(\beta\left(h-\frac{k}{2}\right)\right ) + q_1^{sc}\right]\right\},\\ & & t_{+-}^{sc}=2e^{-\beta\frac{j}{4}}\left\{e^{\frac{\beta}{2}\left(h-\frac{k}{2}\right)}\left[{\operatorname{ch}}\left(\beta h\right)+q_2^{sc}\right]+e^{-\frac{\beta}{2}\left(h-\frac{k}{2}\right)}\left[{\operatorname{ch}}\left(\beta\left(h+\frac{k}{2}\right)\right ) + q_1^{sc}\right]\right\},\nonumber \\ & & t_{-+}^{sc}=2e^{-\frac{\beta}{4}}\left\{e^{-\frac{\beta}{2}\left(h+\frac{k}{2}\right)}\left[{\operatorname{ch}}\left(\beta h\right ) + q_2^{sc}\right]+e^{\frac{\beta}{2}\left(h+\frac{k}{2}\right)}\left[{\operatorname{ch}}\left(\beta\left ( h-\frac{k}{2}\right)\right ) + q_1^{sc}\right]\right\},\nonumber \\ & & t_{--}^{sc}=2e^{-\beta\frac{j}{4}}\left\{e^{-\beta\left(h+\frac{k}{4}\right)}\left[{\operatorname{ch}}\left(\beta\left(h+k\right)\right ) + q_2^{sc}\right]+e^{\beta\frac{k}{4}}\left[{\operatorname{ch}}\left(\beta\left(h+\frac{k}{2}\right)\right ) + q_1^{sc}\right]\right\ } , \nonumber\end{aligned}\ ] ] where @xmath100 the block hamiltonian ( [ ham_sc ] ) is not symmetric with respect the to permutation of the quantum spin operators @xmath101 and @xmath102 , thus , there are two matrices representing the @xmath1-component of the corresponding spin - operators using for the calculation of the correlation functions . in virtue of the so(2)-invariance of eq.([ham_sc ] ) the corresponding matrices for other spin components are identically equal to zero .
let us write down the entries of the @xmath103 and @xmath104 matrices for the heisenberg - ising sawtooth chain .
@xmath105 where @xmath106 and upper(lower ) signs correspond to @xmath103(@xmath104 ) .
the matrix @xmath107 in terms of which the expectation values and correlation function of the classical spin from block hamiltonian are expressed has the following entries : @xmath108-e^{\beta\frac{k}{4}}\left[{\operatorname{ch}}\left(\beta\left(h-\frac{k}{2}\right)\right ) + q_1^{sc}\right]\right\},\\ & & \tau_{+-}=e^{-\beta\frac{j}{4}}\left\{e^{\frac{\beta}{2}\left(h-\frac{k}{2}\right)}\left[{\operatorname{ch}}\left(\beta h\right)+q_2^{sc}\right]-e^{-\frac{\beta}{2}\left(h-\frac{k}{2}\right)}\left[{\operatorname{ch}}\left(\beta\left(h+\frac{k}{2}\right)\right ) + q_1^{sc}\right]\right\},\nonumber \\ & & \tau_{-+}=e^{-\frac{\beta}{4}}\left\{-e^{-\frac{\beta}{2}\left(h+\frac{k}{2}\right)}\left[{\operatorname{ch}}\left(\beta h\right ) + q_2^{sc}\right]+e^{\frac{\beta}{2}\left(h+\frac{k}{2}\right)}\left[{\operatorname{ch}}\left(\beta\left ( h-\frac{k}{2}\right)\right ) + q_1^{sc}\right]\right\},\nonumber \\ & & \tau_{--}=e^{-\beta\frac{j}{4}}\left\{-e^{-\beta\left(h+\frac{k}{4}\right)}\left[{\operatorname{ch}}\left(\beta\left(h+k\right)\right ) + q_2^{sc}\right]+e^{\beta\frac{k}{4}}\left[{\operatorname{ch}}\left(\beta\left(h+\frac{k}{2}\right)\right ) + q_1^{sc}\right]\right\}. \nonumber\end{aligned}\ ] ] all possible two - spin correlation functions ( except the @xmath109 , which is uniform for all hic and is given by eq .
( [ cor_sigma_3_xi ] ) ) of the system are listed below , @xmath110\right)\left(\sum_{k=1}^4\left[a_k^3{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^3{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi}},\\ & & \langle s_{j,2}^z s_{j+r,2}^z\rangle-\langle s_{j,2}^z\rangle^2=\frac{\left(\sum_{k=1}^4\left[a_k^2{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^2{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)\left(\sum_{k=1}^4\left[a_k^4{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^4{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle s_{j,1}^z s_{j+r,2}^z\rangle-\langle s_{j,1}^z\rangle\langle s_{j,2}^z\rangle=\frac{\left(\sum_{k=1}^4\left[a_k^1{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^1{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)\left(\sum_{k=1}^4\left[a_k^4{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^4{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle s_{j,2}^z s_{j+r,1}^z\rangle-\langle s_{j,1}^z\rangle \langle s_{j,2}^z\rangle=\frac{\left(\sum_{k=1}^4\left[a_k^2{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^2{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)\left(\sum_{k=1}^4\left[a_k^3{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^3{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi}},\nonumber \\ & & \langle s_{j,1}^z \sigma_{j+r}\rangle-\langle s_{j,1}^z\rangle\langle \sigma_j\rangle=-\frac{e^{-\beta\frac{j}{4}}\sqrt{1-m_{\sigma}^2}\sum_{k=1}^4\left(a_k^1{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^1{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right)}{4\lambda_1^{sc}}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle s_{j,2}^z \sigma_{j+r}\rangle-\langle s_{j,2}^z\rangle\langle \sigma_j\rangle=-\frac{e^{-\beta\frac{j}{4}}\sqrt{1-m_{\sigma}^2}\sum_{k=1}^4\left(a_k^2{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^2{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right)}{4\lambda_1^{sc}}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle s_{j,1}^z \tau_{j+r}\rangle-\langle s_{j,1}^z\rangle\langle \tau_j\rangle=\frac{\left(\sum_{k=1}^4\left[a_k^1{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^1{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)\left(\sum_{k=0}^4\left[a_k^6{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^6{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle s_{j,2}^z \tau_{j+r}\rangle-\langle s_{j,2}^z\rangle\langle \tau_j\rangle=\frac{\left(\sum_{k=1}^4\left[a_k^2{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^2{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)\left(\sum_{k=0}^4\left[a_k^6{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^6{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle \sigma_{j}s_{j+r,1}^z\rangle-\langle s_{j,1}^z\rangle\langle \sigma_j\rangle=-\frac{e^{-\beta\frac{j}{4}}\sqrt{1-m_{\sigma}^2}\sum_{k=1}^4\left(a_k^3{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^3{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right)}{4\lambda_1^{sc}}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle \sigma_{j}s_{j+r,2}^z\rangle-\langle s_{j,2}^z\rangle\langle \sigma_j\rangle=-\frac{e^{-\beta\frac{j}{4}}\sqrt{1-m_{\sigma}^2}\sum_{k=1}^4\left(a_k^4{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^4{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right)}{4\lambda_1^{sc}}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle \sigma_j \tau_{j+r}\rangle-\langle \sigma_j\rangle\langle \tau_j\rangle=-\frac{e^{-\beta\frac{j}{4}}\sqrt{1-m_{\sigma}^2}\sum_{k=0}^4\left[a_k^6{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^6{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]}{4\lambda_1^{sc}}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle \tau_{j } s_{j+r,1}^z\rangle-\langle s_{j,1}^z\rangle\langle \tau_j\rangle=\frac{\left(\sum_{k=0}^4\left[a_k^5{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^5{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)\left(\sum_{k=1}^4\left[a_k^3{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^3{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle \tau_{j } s_{j+r,2}^z\rangle-\langle s_{j,2}^z\rangle\langle \tau_j\rangle=\frac{\left(\sum_{k=0}^4\left[a_k^5{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^5{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)\left(\sum_{k=1}^4\left[a_k^4{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^4{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle \tau_{j } \sigma_{j+r}\rangle-\langle \sigma_j\rangle\langle \tau_j\rangle=-\frac{e^{-\beta\frac{j}{4}}\sqrt{1-m_{\sigma}^2}\sum_{k=0}^4\left[a_k^5{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^5{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]}{4\lambda_1^{sc}}e^{-\frac{r}{\xi } } , \nonumber\\ & & \langle \tau_j \tau_{j+r}\rangle-\langle \tau_j\rangle^2=\frac{\left(\sum_{k=0}^4\left[a_k^5{\operatorname{sh}}\left(\beta\frac{k
h}{2}\right)+b_k^5{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)\left(\sum_{k=0}^4\left[a_k^6{\operatorname{sh}}\left(\beta\frac{k h}{2}\right)+b_k^6{\operatorname{ch}}\left(\beta\frac{k h}{2}\right)\right]\right)}{16\sum_{k=0}^2 d_k{\operatorname{ch}}\left(\beta k h\right)}e^{-\frac{r}{\xi } } , \nonumber\end{aligned}\ ] ] where the eigenvalues of the transfer - matrix ( [ t_sc ] ) are given by the following expressions : @xmath111 the coefficients in eqs.([sc_all_corr ] ) and ( [ lambda_sc ] ) are @xmath112 to illustrate the spatial and temperature dependence of various correlation functions of eq.([sc_all_corr ] ) it is helpful to look at their plots .
first of all , the general remarks about correlations in hic are noteworthy .
the general structure of the hic exclude any possibility to have an entangled non trivial macroscopic ground state , on the contrary , any macroscopic ground state is just a tensor product of the local ground states , which , in their turn , could be entangled .
thus , the zero - temperature long - range correlations between the spins from different blocks , calculated with respect to their expectation values , are trivial which implies the zero value of all correlation function from eq .
( [ sc_all_corr ] ) at @xmath113 , as for the arbitrary ground state @xmath114 one has @xmath115 however , the thermal fluctuations as well as the effect of the magnetic field lead to non - zero correlations within a narrow interval of low temperatures for short distances . the plots of the temperature dependence of all correlation functions for the asymmetric sawtooth chain are presented in figure [ fig3 ] for the distance equal to one block , @xmath116 . here ,
the values of the parameters correspond to the spin - modulated zero - temperature ground state of the asymmetric heisenberg - ising sawtooth chain@xcite .
it is rather expectable that the finite - temperature correlations are mostly of short range with respect to blocks , but as one block contains four spins , the non - zero , though weak , correlations even for @xmath116 are the non - trivial result .
one can see from the plots in figure [ fig3 ] the vanishing of all correlations at high as well as at low temperatures as was mentioned above .
as expected , the closer to each other are situated the spins from different blocks , the larger is the magnitude of the correlation between them .
for instance , the correlation between @xmath96 and @xmath117 at the temperature corresponding to the peak is four times larger in absolute value than the correlation between @xmath96 and @xmath118 and more that seventy times larger than the correlation between @xmath119 and @xmath120 . in the same time , the correlation between @xmath121 are the strongest ones , though , the are separated by the dependence of the correlation functions against the distance for the same values of coupling constants and anisotropy and for temperatures corresponding to the strongest correlation of the thermal fluctuations are presented in figure [ fig4 ] . for all correlation functions one can see the usual exponential decay in their absolute value .
the sign of the correlations is defined by the mutual orientation of the corresponding spins which can be seen in the @xmath113 ground state@xcite for the given set of the parameters . in figure [ fig5 ]
we presented the plots of the same correlation functions @xmath122 and @xmath123 against the distance for different temperatures .
the non - monotonous behavior of the correlation functions described above can be clearly seen from the plots .
in the present paper we elaborated a technique for calculating correlation functions in hic .
the technique is based on the classical transfer - matrix formalism which proves itself as the most efficient and straightforward method of treating the hic , allowing one to obtain exact expressions for the partition function and , thus , for all thermodynamic functions in a quite simple way@xcite . however , the classical transfer - matrix formalism is also very convenient for the calculation of the correlation functions of various kinds , because all canonical expectation values are expressed in terms of traces of certain products of the transfer - matrices with matrices representing the microscopic variables in the transfer - matrix formalism@xcite .
we presented a detailed scheme of calculation of the correlation functions for hic , obtained general expressions for the corresponding trace formulas and presented the general expressions for the matrix representations of the main microscopic variables for the typical hic , quantum spin operators , additional classical spins from block hamiltonians and intermediate spin . as it is expected from general principles , the spatial decay of any correlation between any pair of spin
is characterized by a single correlation length obtained in a standard way as the reciprocal of the logarithm of the ratio of the largest and second largest eigenvalues of the classical transfer matrix . in the simplest case considered here when intermediate classical @xmath7-spins are 1/2 spins the transfer matrix has dimension 2 by 2 and , thus , only two eigenvalues . however , the prefactors are strongly affected by the microscopic variables .
we illustrated our general formalism by two examples , the simplest hic , symmetric heisenberg - ising diamond - chain and asymmetric heisenberg - ising sawtooth chain with a pair of quantum spins in the unit cell .
we presented the exact form of the transfer - matrix , its eigenvalues , matrix representation corresponding to the classical transfer - matrix formalism for quantum spin operators , matrix representation for additional classical spin from block hamiltonian and all possible pair correlation functions .
a very important feature of the correlations function of the hic considered in the present paper is the absence of correlations between transverse components of quantum spins from different blocks .
the corresponding matrix representation of the @xmath124 and @xmath125 operators are identically zero .
this fact is an obvious consequence of the structure of the hic where blocks interact to each other only through @xmath1-components of the spins .
it is also worth mentioning that the general idea of using the classical transfer - matrix for describing the thermodynamic properties of decorated spin lattices can be applied also for two - dimensional systems .
although the existing examples of the papers devoted to the topic deal with the decoration - iteration transformation@xcite , the direct generalization of the existing transfer - matrix method is possible .
as known , two - dimensional systems can exhibit phase transitions at finite temperature in contrast to the one - dimensional models where only zero - temperature transitions are possible . in principle
, two - dimensional spin systems can possess a complicated structure of the order parameters , and multiple orderings , which , particularly , can be different in different spatial directions .
for these reasons , the statement about the unique correlation length is no longer valid here . [ cols="^,^ " , ]
v.o . expresses his gratitude to the lnf - infn in frascati for warm hospitality during the work on the paper .
he also acknowledges partial financial support from volkswagen foundation ( grant no.i/84 496 ) and from the project scs - bfbr 11rb-001 . | a general technique of exact calculation of any correlation functions for the special class of one - dimensional spin models containing small clusters of quantum spins assembled to a chain by alternating with the single ising spins is proposed .
the technique is a natural generalization of that in the models solved by a classical transfer matrix .
the general expressions for corresponding matrix operators which are the key components of the technique are obtained . as it is clear from the general principles , the decay of the correlation functions of various types is explicitly shown to be governed by a single correlation length .
the technique is illustrated by two examples : symmetric diamond chain and asymmetric sawtooth chain . |
the @xmath3 spectrum exhibits many features that are emblematic of qcd ; that they contain three valence quarks is a manifestation of su(3 ) symmetry , while the linearly rising regge trajectories are evidence for a flux - tube - like confining force .
finally , the fine and hyperfine structure reveals details of the interquark interactions .
thus the study of the spectrum of excited nucleon resonances has always been a vital component of the experimental hadronic physics program , as typified by the hall b experimental effort at jefferson laboratory . the motivation for studying the @xmath3 spectrum in many ways mirrors that for the hybrids , which also provide an arena in which to explore flux tubes .
one distinction is that , in the case of nucleons , all spin - parity combinations are accessible to the quark model , and thus there are no nucleon `` spin exotics '' .
however , both hybrids and excited nucleons share the feature that they are sensitive to the presence of excited glue .
the observed excited nucleon spectrum already poses many questions .
there are `` missing resonances '' , predicted by the quark model but not yet observed experimentally .
most strikingly , there is the so - called roper resonance , and the lightest negative - parity @xmath4 state , the @xmath5 with anomalously light masses suggesting that they may be `` molecular '' states , rather than true three - quark resonances .
lattice gauge theory calculations have a vital role to play in resolving these questions . in an attempt to address these issues ,
there have been several recent calculations of the lowest - lying excited nucleon masses , emphasising in particular the extent to which the parity partners are accessible to lattice calculation .
the first calculation employed the highly - improved @xmath6 fermion action@xcite , whilst a second calculation employed domain - wall fermions@xcite .
both calculations exhibited a clear splitting between the masses of the @xmath7 and @xmath8 states , with an indication that the lowest - lying radial excitation of the nucleon has a mass considerably larger than that of the @xmath9 , casting doubt on the nature of the roper resonance . in this talk , we present a calculation of the lowest lying negative parity nucleon mass using an @xmath1-improved sheikholeslami - wohlert ( sw ) , or clover , fermion action ; preliminary results were presented in ref .
@xcite , and a comprehensive analysis appeared in ref .
@xcite . by choosing the coefficient of the improvement term appropriately ,
all @xmath1 discretisation uncertainties are removed , ensuring that the continuum limit is approached with a rate proportional to @xmath10 . for a subset of our lattices , masses of the first radial excitation of the nucleon and of the parity partner of the @xmath11
are presented . the calculation of the excited nucleon spectrum places particularly heavy demands on lattice spectroscopy .
the excited nucleon states are expected to be large ; the size of a state is expected to double with each increase in orbital angular momentum .
thus a lattice study of the excited nucleon spectrum requires large lattice volumes , with correspondingly large computational requirements .
furthermore , the states are relatively massive , requiring a fine lattice spacing , at least in the temporal direction . these requirements could be satisfied with much greater economy using the clover fermion action than using the domain - wall or overlap formulation .
thus it is important to establish that the negative parity states are indeed accessible to calculations using the clover action .
finally , by comparing the masses obtained using the clover action with a calculation , at a single quark mass , using the wilson fermion action , we also gain insight into the nature of the interaction responsible for the splitting in the parity doublet .
the layout of the remainder of this talk is as follows . in the next section
, i shall describe the construction of hadronic operators , and in particular the projection onto positive- and negative - parity states , and summarises other computational details .
section [ sec : results ] contains the results for the masses of the lowest - lying positive- and negative - parity nucleon using the sw fermion action , and for a subset of the lattices the masses obtained for the roper resonance .
the talk concludes with a discussion , a comparison with other lattice and phenomenological calculations , and prospects for future studies .
historically , there has been relatively little study of the negative - parity baryon states , though there had been numerous computations of the nucleon radial excitations .
it is therefore useful to review the rle that parity plays in the construction of baryon operators , and how this rle is manifest in the case of lattice calculations .
we illustrate the discussion through consideration of the usual nucleon interpolation operators : @xmath12 these operators have an overlap with particle states of both positive and negative parities . in order to construct operators of definite parity
, we introduce the parity - projection operator : @xmath13 and construct the correlators @xmath14 where @xmath15 is a baryon interpolating operator . on a periodic or anti - periodic lattice , the two time orderings
result in forward propagating states of positive ( negative ) parity , and backward propagating states which are anti - particles of negative ( positive ) parity states if the projection operator is chosen with positive ( negative ) sign .
thus the best delineation that may be constructed is between a forward propagating particle , and the backward propagating anti - particle of the parity partner . at large distances , where @xmath16 and @xmath17 , the correlators assume the time behaviour @xmath18 where @xmath19 and @xmath20 are the lightest positive- and negative - parity masses respectively .
for the case of the nucleon ground state , the `` diquark '' part of both @xmath21 and @xmath22 couples upper ( large ) spinor components , while that in @xmath23 involves both an upper and lower spinor component and thus vanishes in the non - relativistic limit@xcite .
thus our expectation is that the operators @xmath21 , @xmath22 should give a better overlap with the positive - parity ground state nucleon compared with operator @xmath23 , and this is confirmed in lattice simulations . for some of the lattices used in the calculation , the positive - parity nucleon mass is obtained using the non - relativistic quark operators @xcite , defined by @xmath24 corresponding to the `` large '' components in the non - relativistic limit ; this results in a factor of two reduction in computational effort at the expense of some loss of statistics .
the calculation is performed in the quenched approximation to qcd .
the standard wilson gluon action is employed , whilst the quark propagators are computed using the sheikholeslami - wohlert clover fermion action , using the non - perturbative coefficient of the clover term determined in refs .
@xcite and @xcite .
thus the hadron masses are free of all @xmath25 discretisation errors .
the propagators are computed using both local sources , and spatially extended `` smeared '' sources , using either the fuzzed - source method ( f ) or jacobi - smeared sources ( j ) as described in references @xcite and @xcite respectively .
the simulation parameters are summarized in tables [ tab : params ] .
the errors on the fitted masses are computed using a bootstrap procedure .
different numbers of configurations are used at different quark masses at some of our @xmath26 values , precluding the use of correlated fits in some of the chiral extrapolations .
thus a simple uncorrelated @xmath27 fit is performed , with the uncertainties computed from the variation in the @xmath27 .
calculations of the light hadron spectrum using these lattices were presented in references @xcite and @xcite . [ cols="^,^,^,^,^,^",options="header " , ] + lattice sizes in physical units are quoted using @xmath28 to set the scale @xcite .
the clover term removes the leading chiral - symmetry - breaking effects from hadron masses at finite @xmath29 .
since the lack of degeneracy between the positive- and negative - parity baryon states is a consequence of the spontaneous breaking of chiral symmetry , we also compare the measurements of the mass splitting at a single quark mass with those obtained using the wilson fermion action .
the masses of the lowest - lying @xmath30 and @xmath31 states are obtained from a simultaneous , four - parameter fit to the positive- and negative - parity correlators @xmath32 and @xmath33 constructed from the `` good '' nucleon operator @xmath21 of eqn .
[ eq : n1 ] . a clear signal for the mass of the lightest negative - parity state notably higher than that for the positive - parity state is observed , and is illustrated for one of our ensembles in figure [ fig : eff_mass ] .
channel ( circles ) and the @xmath31 channel ( diamonds ) at @xmath34 , with @xmath35 , using the interpolating operator @xmath21 of eqn [ eq : n1 ] . also shown as the bursts is the effective mass of the positive - parity state determined using the interpolating operator @xmath23 . ] for the chiral extrapolation of the hadron masses , we adopt the ansatz @xmath36 where we use upper - case letters to denote masses obtained in the chiral limit and @xmath37 is either @xmath30 or @xmath31 .
we include data at different volumes , but at the same @xmath26 , in the chiral extrapolations , but treat the fuzzed and jacobi - smeared data independently ; they give slightly different masses , reflecting the different contributions of the excited states in the two cases .
note that the fit to @xmath38 is formally the same as one to @xmath39 at small pion mass , but in practice provides a notably better description of the data .
it is known that there are non - analytic terms contributing to the chiral extrapolation of physical quantities , and the contribution of the non - analytic terms arising from the pion - induced baryon self energies has been explored in reference @xcite . in the quenched approximation ,
the leading , non - analytic terms can assume a different form .
from quenched chiral perturbation theory , one expects leading non - analytic terms which are linear in @xmath40 @xcite ; the coefficient of the leading term is predicted to be negative .
we therefore also attempted to fit the data at @xmath41 to the form @xmath42 fitting the @xmath30 state we found a positive value of @xmath43 , though with a very large error that would still accommodate a negative value .
thus it is unclear whether eqn .
[ eq : non_analytic ] provides a reliable form with which to extrapolate data obtained with quark masses around that of the strange to the chiral limit , emphasising once again the need for data at the smaller quark masses at which the pion cloud emerges .
the chirally extrapolated masses obtained using this procedure , with @xmath43 unconstrained , differ from those obtained using eqn .
[ eq : chiral_extrap ] by around 5% . in view of the difficulties discussed above , and because on some of our ensembles we have insufficient data points to perform a three - parameter fit , we quote our final results from fits to eqn .
[ eq : chiral_extrap ] .
an example of such a chiral extrapolation for the data at @xmath41 is shown in figure [ fig : chiral_extrap ] .
ideally , the forms eqn .
[ eq : chiral_extrap ] and [ eq : non_analytic ] should be applied to the baryons masses obtained in the infinite volume limit .
indeed , the talk by stuart wright at this workshop emphasised the importance of correctly accounting for finite - volume effects in chiral extrapolations @xcite .
our data do not admit this procedure , but an investigation of the masses of the negative - parity state obtained on the smaller and larger of our lattices at @xmath44 and @xmath45 suggests that the masses of the negative - parity state on the larger lattices are smaller than those on the smaller lattices by around 5% . .
the curves are from fits to @xmath46 using eqn .
[ eq : chiral_extrap ] . ] in order to look at the discretisation uncertainties in our data , we show in figure [ fig : continuum_extrap ] the masses in units of @xmath28 against the @xmath47 , where @xmath48 is the hadronic scale @xcite .
@xcite against @xmath10 in units of @xmath49 .
the lines are linear fits in @xmath47 to the positive- and negative - parity baryon masses . also shown
are the physical values . ] on a subset of our lattices , we have extracted the ground - state positive- and negative - parity masses using the operator @xmath23 of eqn .
[ eq : n2 ] .
the effective mass in the positive - parity channel using this operator is shown in figure [ fig : eff_mass ] ; it indicates an effective ground state mass considerably higher than that of the lightest negative - parity state .
the signal for both the positive- and negative - parity ground states degrades appreciably with decreasing pion mass , as can be seen in figure [ fig : good_bad_chiral ] where we show the extracted lowest - lying masses of both parities using each of the @xmath21 and @xmath23 interpolating operators .
a better determination of the first radial excitation would require the measurement of the cross - correlator between @xmath21 and @xmath23 , as done in reference @xcite .
however , the conclusion that the operator @xmath23 has a negligible overlap with the ground state nucleon at these values of the pseudoscalar mass seems inescapable .
( circles ) and the @xmath31 ( diamonds ) obtained using @xmath21 , and of the lowest - lying @xmath30 ( down triangle ) and @xmath31 ( up triangle ) using
@xmath23 , obtained on the @xmath50 lattices at @xmath34 . ]
the measurement of heavier excitations is inevitably noisier because of the reduced signal - to - noise ratio . on some of our lattices
we have measured the correlators for the @xmath51 @xmath11 , using an interpolating operator @xmath52 this has an overlap onto both spin-@xmath53 and spin-@xmath54 states , but these can be distinguished using a suitable spin projection , and the two parities delineated as described above .
the masses of the ground state of both positive and negative parity are shown in figure [ fig : delta ] .
further success at determining the higher spin states has been made in a recent determination of the @xmath55 masses @xcite .
( circles ) and its parity partner ( diamonds ) against the pseudoscalar mass obtained on the @xmath50 lattices at @xmath34 ]
we have seen that the use of the clover - improved fermion action enables the delineation of the positive- and negative - parity excited state masses , in accord with calculations using highly improved wilson , and domain - wall fermion actions . on a subset of our lattices
, we also computed propagators using the unimproved wilson fermion action at a single quark mass , and found a parity mass splitting consistent with that of the clover fermion action . on reflection , this is hardly surprising ; under @xmath57 spin - flavour symmetry , the low - lying positive - parity baryons can be assigned to an @xmath58 multiplet , whilst the low - lying negative - parity baryons can be assigned to an @xmath59 multiplet .
thus the mass - splitting can be likened to the @xmath60 splitting in the meson sector , which is well - known to be faithfully reproduced using the wilson fermion action .
the striking feature common to all the lattice calculations is the apparent inversion of the ordering of mass of the lowest radial excitation of the nucleon , and the mass of the parity partner of the nucleon .
all the calculations , using different gluonic and fermionic actions , suggest a splitting between the masses of the first positive - parity radial excitation and the lightest negative - parity state comparable to that between the masses of the negative - parity state and the nucleon .
in contrast , the observed lightest radial excitation of the nucleon is the roper resonance with a mass of @xmath61 , whilst the lightest negative - parity state is the heavier @xmath9 .
we can see that lattice calculations of the spectrum are already posing interesting questions about the nature of the roper resonance .
i have not discussed the issue of the nature of the @xmath5 in this talk . in the case of the positive - parity states , lattice calculations , such as that of ref .
@xcite , suggest that the quark mass behaviour of the baryon masses , in the range over which they are measured , is well described by terms corresponding to the `` centres of gravity '' of the quark masses within the hadron .
this expectation has been confirmed in ref .
@xcite , where the lightest negative - parity @xmath4 resonance has a mass considerably in excess of @xmath62 . it should be noted , however , that a recent large @xmath63 calculation has been able to accommodate a light @xmath5 @xcite .
we have seen that lattice calculations are already addressing some of the salient issues in baryon spectroscopy .
the calculation of the radial excitations and the higher spin states will present further challenges ; the need to overcome the reduced signal - to - noise ratio afflicting heavier excitations , the construction of suitable lattice operators to isolate higher excitations , and the exploitation of improved data analysis methods .
as these issues are tackled , we will be able to tackle the more ambitious projects of transition form factors and hadronic decays .
this work was supported in part by doe contract de - ac05 - 84er40150 under which the southeastern universities research association ( sura ) operates the thomas jefferson national accelerator facility , by epsrc grant gr / k41663 , and pparc grants gr / l29927 and gr / l56336 . cmm acknowledges pparc grant ppa / p / s/1998/00255 . mg acknowledges financial support from the dfg ( schwerpunkt `` elektromagnetische sonden '' ) .
propagators were computed using the t3d at edinburgh , the t3e at zib ( berlin ) and nic(jlich ) , the ape100 at nic ( zeuthen ) , and the _ calico _ cluster at jefferson laboratory . 9 f.x .
lee and d.b .
leinweber , nucl .
phys.(proc . suppl . ) 73 , 258 ( 1999 ) .
lee , nucl .
( proc . suppl . ) 94 , 251 ( 2001 ) . s. sasaki ( riken - bnl - columbia collaboration ) , nucl . phys .
( proc . suppl . ) 83 ( 2000 ) 206 ; invited talk at jefferson laboratory workshop on physics of excited nucleons , newport news , feb .
16 - 19 , 2000 , hep - ph/0004252 .
s. sasaki , t. blum and s. ohta , hep - lat/0102010 .
richards ( lhpc and ukqcd collaboration ) , nucl.phys .
( proc . suppl . ) 94 , 269 ( 2001 )
. m. gckeler _ et al . _ ( lhpc / qcdsf / ukqcd collaborations ) , hep - lat/0106022 . m. gckeler _ et al .
_ , phys .
d53 , 2317 ( 1996 ) .
m. lscher , s. sint , r. sommer , p. weisz and u. wolff , nucl .
b491 , 323 ( 1997 ) .
edwards , u.m .
heller and t.r .
klassen , phys . rev
. lett .
80 , 3448 ( 1998 ) .
p. lacock , a. mckerrel , c. michael , i.m .
stopher , and p.w .
stephenson ( ukqcd collaboration ) , phys .
d51 , 6403 ( 1995 ) .
c. best _ et al .
d56 , 2743 ( 1997 ) . d.b .
leinweber , a.w .
thomas , k. tsushima and s.v .
wright , phys .
d64 , 094592 ( 2001 ) .
r. sommer , nucl .
b411 , 839 ( 1994 ) .
m. guagnelli , r. sommer and h. wittig , nucl.phys .
b535 , 389 ( 1998 ) .
ukqcd collaboration ( k.c .
bowler _ et al .
_ ) , phys .
d62 , 054506 ( 2000 ) .
m. gckeler _ et al .
_ , phys .
b391 , 388 ( 1997 ) ; phys .
d57 , 5562 ( 1998 ) .
d. pleiter , thesis , berlin ( 2000 ) ; qcdsf collaboration , in preparation .
d. leinweber , a.w .
thomas , k. tsushima and s.v .
wright , phys .
d61 , 074502 ( 2000 ) . j.n .
labrenz and s.r .
sharpe , phys .
d54 , 4595 ( 1996 ) .
lee , d.b .
leinweber , l. zhou , j. zanotti and s. choe , hep - lat/0110164 , to appear in proceedings of lattice 2001 , berlin , aug .
w. melnitchouk _ et al _ , hep - lat/0202022 .
schat , j.l .
goity and n.n .
scoccola , hep - ph/0111082 . | we discuss the extraction of negative - parity baryon masses from lattice qcd calculations .
the mass of the lowest - lying negative - parity @xmath0 state is computed in quenched lattice qcd using an @xmath1-improved clover fermion action , and a splitting found with the nucleon mass .
the calculation is performed on two lattice volumes , and three lattice spacings enabling a study of both finite - volume and finite - lattice - spacing uncertainties .
a measurement of the first excited radial excitation of the nucleon finds a mass considerably larger than that of the negative - parity ground state , in accord with other lattice determinations but in disagreement with experiment .
results are also presented for the lightest negative - parity @xmath2 state . |
we call _ herbrand constructive _ any intermediate logic a logic stronger than intuitionistic but weaker than classical which enjoys a strong form of herbrand s theorem : for _ every _ provable formula @xmath1 , the logic proves as well an herbrand disjunction @xmath2\lor \ldots \lor a[m_{k}/\alpha]\ ] ] of course intuitionistic logic is trivially herbrand constructive , but classical logic is not : @xmath3 is arbitrary ! in between , there are several interesting logics which do have the property . yet for herbrand constructive logics there are no known natural deduction formulations with associated curry - howard correspondences , except in trivial cases .
we launch here a new series of papers to fill this void .
we begin with dummett s first - order and second - order logic @xmath4 : intuitionistic logic extended with the so - called dummett linearity axiom @xmath5 @xmath0 was introduced by dummett @xcite as an example , in the propositional case , of a many - valued logic with a countable set of truth values .
its propositional fragment is also called gdel - dummett logic , because it is based on the truth definition given in gdel s seminal paper on many - valued logics @xcite . in this case
, the logic can be formalized by corsi s sequent calculus @xcite or by the more elegant hypersequent calculus devised by avron @xcite , @xcite . surprisingly , avron s hypersequent calculus does not work for first - order @xmath0 :
only recently tiu @xcite provided a more involved version of it , which indeed corresponds to @xmath0 at the first - order .
in all this story , natural deduction is the great absent .
since it is one of the most celebrated logical deduction systems , the question is : how is that possible ?
the first issue is that @xmath0 is evidently a non - constructive system : for example , it proves the excluded middle for all negated formulas : @xmath6 ; and dummett s axiom poses even more problems . as it is well known , natural deduction was put aside by its own inventor , gentzen , precisely for the reason that he was not able to prove a meaningful normalization theorem for classical natural deduction , whilst he _ was _ for the intuitionistic case @xcite .
it indeed took a surprisingly long time to discover suitable reduction rules for classical natural deduction systems with all connectives @xcite , @xcite .
even this accomplishment , however , is still not enough : although dummett s axiom is classically provable , the known classical natural deduction systems fail to provide a refined computational interpretation of @xmath0 .
the trouble is that @xmath0 proofs are not closed under classical reductions , because during the normalization process instances of dummett s axiom are replaced by reductio ad absurdum in @xmath7-calculus @xcite and in @xcite , and by excluded middle in @xcite .
the second issue is that existential quantifiers are witnessed by multiple terms and so a parallel computational mechanism is desirable .
no curry - howard correspondence offered a suitable one until very recently @xcite .
sequent calculus solves these issues by means of _ structural rules_. classical logic is rendered by allowing more formulas on the righthand side of a sequent ; dummett s @xmath0 is rendered by allowing sequences of sequents and a communication mechanism between them .
on the contrary , natural deduction usually solves the same issues by means of new _
reduction rules_. when one wants to add some new axiom to intuitionistic natural deduction , it is enough to add it straight away or as a rule , and all the ingenuity of the construction lies in the proof transformations associated to the axiom . inspired by hypersequents , baaz , ciabattoni and fermller
@xcite did not follow the latter path and changed instead the very structure of natural deduction into an hyper version .
the resulting logical calculus is an _
hyper natural deduction
_ corresponding to gdel - dummett first - order logic ( which is not to be confused with first - order @xmath0 and can be axiomatized by adding to @xmath0 the axiom scheme @xmath8 , where @xmath9 does not occur in @xmath10 ) .
the normal form theorem , however , is only obtained by translation into the hypersequent calculus , followed by cut - elimination and backward translation : no reduction rules for hyper deductions were provided .
this last task was carried out by beckmann and preining @xcite , who formulated a _
propositional _ hyper natural deduction with a proof normalization procedure .
unfortunately , the structural rules are so complicated that the adjective `` natural '' does not fit any more .
another attempt along the `` hyper line '' has been made by hirai @xcite , with the addition of an associated lambda calculus .
one can not speak of a curry - howard correspondence , however , because subject reduction does not hold : there is no match between computational steps and proof reductions .
although hyper natural deduction is a legitimate proof system in its own right , the `` hyper approach '' is not the one we follow . for two reasons .
the first reason is that we will show that natural deduction works perfectly as it is .
there is no need to change its structure and , to render dummett s axiom , it sufficient to add the inference rule which allows to conclude unconditionally @xmath11 from two different deductions of @xmath11 : one from the hypothesis @xmath12 and one from the hypothesis @xmath13 .
we shall define simple reduction rules for proofs ending with this inference and we shall show that they are all we need to extract witnesses for existentially quantified formulas .
the second reason is that natural deduction should stay _
natural_. this is the very motivation that led to its discovery . indeed , gentzen starts his celebrated work @xcite on natural deduction and sequent calculus complaining that the proof systems known at the time were far removed from the actual mathematical reasoning .
and his main goal was to set up a formalism with the aim of `` _ _ reproducing as precisely as possible the real logical reasoning in mathematical proofs _ _ '' . to avoid betraying natural deduction s philosophical motivations
, there is no alternative but to add an inference rule that naturally mirrors the kind of reasoning corresponding to dummett s axiom , which is our approach .
one of the most attractive features of intuitionistic natural deduction is that , in a very precise sense , it does not need a truth - based semantics .
logical inferences are divided into two groups : introduction rules and elimination rules . and as gentzen @xcite himself famously suggested , introduction rules _ define _ , so to speak , the meaning of the logical constants they introduce ; elimination rules , on the other hand , are nothing but _ consequences _ of these definitions . in other words ,
introduction rules are self - justifyng , because they fix themselves the meaning of their conclusions , whereas elimination rules are sound in virtue of the meaning fixed by the introductions .
for example , the rule says that the grounds for asserting @xmath12 consist in a proof of @xmath10 from the hypothesis @xmath3 ; therefore , the elimination @xmath14 is automatically justified : if we have a proof of @xmath3 we can plug it into the proof of @xmath10 from @xmath3 , whose existence is warranted by the meaning of @xmath12 , and obtain a proof of @xmath10 .
the reverse approach works as well : we may consider elimination rules as meaning constitutive and treat introduction rules as consequences of the meaning fixed by eliminations .
in other words , meaning is determined by _ how we use _ a statement , by what we can directly _ obtain _ from the statement ; we shall adopt this _
standpoint , elaborated by dummett himself @xcite .
this idea of internal justification , as it is , can not be generalized straight away for extensions of intuitionistic logic : new inferences tend to break the harmony between introductions and eliminations .
it is at this point that brouwer s view of logic comes into play .
according to brouwer @xcite , the string of `` logical '' steps appearing in a mathematical proof is in reality a sequence of mathematical constructions . what we perceive as inference rules are instead transformations of constructions for the premises into constructions for the conclusion .
this insight finds a precise formalization by means of the curry - howard isomorphism : a proof is indeed isomorphic to an effective construction , in fact , _ it is _ , in and of itself , a construction .
since proofs are constructions , the role of semantics is just _ explaining what these constructions do_. hence , a proof - theoretic semantics of an intermediate logic is in principle always possible and is made of two ingredients : a formalization of _ proofs as programs _ and a semantical description of what these programs achieve with their calculations .
the first is obtained through the decoration of deduction trees with lambda terms , the second is the task of _ realizability_. realizability was introduced by kleene @xcite to computationally interpret intuitionistic first - order arithmetic , but it is kreisel s @xcite later version with typed terms which embodies the modern perspective on the subject . though it was initially conceived just for intuitionistic theories , realizability can be extended to intuitionistic arithmetic with markov s principle @xcite , to intuitionistic arithmetic with the simplest excluded middle @xmath15 @xcite and even all the way up to the strongest classical theories @xcite .
realizability replaces the notion of truth with the notion of _ constructive evidence_. a formula holds if it is _ realized _ by some typed program , providing some constructive information about the formula . in the following ,
we shall build a realizability interpretation for dummett s @xmath0 , inspired by krivine s realizability @xcite . by construction
, every realizer always terminates its computations and , in particular , whenever it realizes an existentially quantified formula @xmath16 , it reduces to a term of the shape @xmath17 with the property that @xmath18\lor \dots \lor a[m_{k}/\alpha]\ ] ] the circle is closed by a soundness theorem , the adequacy theorem : every formula provable in @xmath0 is realized by a closed program , which immediately implies the normalization theorem every proof reduces to a normal form and that @xmath0 is herbrand constructive .
therefore , to extract an herbrand disjunction it suffices to reduce any proof of any existentially quantified formula to a normal form , according to a very simple set of reduction rules . to find simple and terminating reduction rules for a natural deduction system is always tricky , but
once the job is done , the reductions often look so natural that they appear inevitable .
it is the effort of removing obstacles toward a good normal form what inevitably leads to these reductions , as the flow of a river leads to the sea . in the case of @xmath0 ,
the main obstacles toward witness extraction for a formula @xmath1 are configurations in which one of the hypotheses introduced by the dummett inference blocks the reduction .
for example , let us consider this proof shape : where @xmath19 has been obtained from @xmath10 by a series of elimination rules .
it is clear that no witness can be retrieved in the left branch of the proof above , because there is just a proof of @xmath3 and , magically , a `` void '' proof of @xmath10 obtained by modus ponens from @xmath3 and the arbitrary hypothesis @xmath12 .
but ca nt we just send the proof of @xmath3 to the right branch of the dummett rule and obtain a direct proof of @xmath19 , like this ? no ! in fact , the proof of @xmath3 too might depend on the hypothesis @xmath12 , so that the original proof could be and thus the previous transformation is unsound .
but the idea of sending the proof of @xmath3 to the right branch can work if the right branch is in turn moved on the left like this the reductions that we shall give generalize this transformation in order to work in every situation .
it is more convenient to express proof reductions in terms of program reductions , because for that purpose the lambda notation is superior to the proof tree notation . for this reason
, we shall define a lambda calculus isomorphic to natural deduction for @xmath0 and then define an _ head reduction strategy _ for lambda terms , inspired by krivine s strategy @xcite .
the termination of head reduction will just be a consequence of soundness of @xmath0 with respect to realizability , while the perfect match between program reductions and proof reductions will as usual be consequence of the subject reduction theorem .
the decoration of intuitionistic inferences with programs is standard and dummett s rule will be decorated in the following way the parallel operator @xmath20 is inspired by the exception operator studied in @xcite and keeps using the variable @xmath21 for communication purposes .
the variable @xmath21 has the task of sending terms from @xmath22 to @xmath23 and viceversa , as well as allowing @xmath22 to call the process @xmath23 whenever it needs it and viceversa . in section
[ section - system ] we introduce a curry - howard interpretation of intuitionistic first - order natural deduction extended with the dummett rule @xmath24 . we first describe the calculus together with its computational rules and then discuss its proof theoretical interpretation .
+ in section
[ section - systemnem ] we prove the normalization theorem and the soundness of realizability with respect to @xmath0 . + in section
[ sec : nfhd ] , we prove that @xmath0 is herbrand constructive and in particular that from any closed term having as type an existentially quantified formula , one can extract a corresponding herbrand disjunction . + in section
[ section - secondorder ] we extend the previous results to the second - order @xmath25 , achieving its first computational interpretation , for there is no known cut - elimination procedure for second - order hypersequent calculus .
in this section we describe a standard natural deduction system for intuitionistic first - order logic , with a term assignment based on the curry - howard correspondence ( e.g. see @xcite ) , and add on top of it an operator which formalizes dummett s axiom .
first , we shall describe the lambda terms and their computational behavior , proving as main result the subject reduction theorem , stating that the reduction rules preserve the type .
then , we shall analyze the logical meaning of the reductions and present them as pure proof transformations .
we start with the standard first - order language of formulas .
[ definition - languagear ] the language @xmath26 of @xmath0 is defined as follows . 1 .
the * terms * of @xmath26 are inductively defined as either variables @xmath27 or constants @xmath28 or expressions of the form @xmath29 , with @xmath30 a function constant of arity @xmath31 and @xmath32 .
2 . there is a countable set of * predicate symbols*. the _ atomic formulas _ of @xmath26 are all the expressions of the form @xmath33 such that @xmath34 is a predicate symbol of arity @xmath31 and @xmath35 are terms of @xmath26 .
we assume to have a @xmath36-ary predicate symbol @xmath37 which represents falsity .
the * formulas * of @xmath26 are built from atomic formulas of @xmath26 by the logical constants @xmath38 , with quantifiers ranging over variables @xmath39 : if @xmath40 are formulas , then @xmath41 , @xmath42 , @xmath12 , @xmath43 , @xmath44 are formulas .
the logical negation @xmath45 can be introduced , as usual , as a shorthand for the formula @xmath46 . in figure
[ fig : system ] we define a type assignment for lambda terms , called * proof terms * , which is isomorphic to natural deduction for intuitionistic logic extended with dummett s axiom . ' '' '' ' '' '' we assume that in the proof terms two distinct classes of variables appear . the first class of variables is made by the variables for the proof terms themselves : for every formula @xmath3
, we have variables @xmath47 of type @xmath3 ; these variables will be denoted as @xmath48 and whenever the type is not important simply as @xmath49 . for clarity , the variables introduced by the dummett s inference rule will be denoted with letters @xmath50 , but they are not in any syntactic category apart .
the second class of variables is made by the quantified variables of the formula language @xmath26 of @xmath0 , denoted usually as @xmath39 .
the free and bound variables of a proof term are defined as usual and for the new term @xmath51 , all of the free occurrences of @xmath21 in @xmath22 and @xmath23 are bound in @xmath51 . in the following , we assume the standard renaming rules and alpha equivalences that are used to avoid capture of variables in the reduction rules that we shall give . whenever @xmath52 and the list @xmath53 includes all the free variables of a proof term @xmath54 , we shall write @xmath55 . from the logical point of view
, the notation means that @xmath56 represents a natural deduction of @xmath3 from the hypotheses @xmath57 .
we shall write @xmath58 whenever @xmath59 , and the notation means provability of @xmath3 in intuitionistic logic with dummett s axiom .
we are now going to explain the basic reduction rules for the proof terms of @xmath0 , which are given in figure [ fig : red ] . to understand them
, we need the notions of parallel context and stack .
if we omit parentheses , any term @xmath56 can be written , not uniquely , in the form @xmath60 if we replace some @xmath61 with a `` hole '' @xmath62 $ ] to be filled , the expression above becomes a parallel context .
[ defi - parallelc ] omitting parentheses , a * parallel context * @xmath63 $ ] is an expression of the form @xmath64
\parallel_{a_{i+1}}u_{i+1}\parallel_{a_{i+2}}\ldots \parallel_{a_{n } } u_{n}\ ] ] where @xmath62 $ ] is a placeholder and @xmath65 are proof terms . for any proof term @xmath22 , @xmath66 $ ] denotes the replacement in @xmath63 $ ] of the placeholder @xmath62 $ ] with @xmath22 : @xmath67 a stack represents , from the logical perspective , a series of elimination rules ; from the lambda calculus perspective , a series of either operations to be performed or arguments to be given as input to some program . a stack is also known as a _ continuation _ , because it embodies a series of tasks that wait to be executed , and corresponds to krivine s stacks @xcite .
[ definition - stack ] a * stack * is a sequence @xmath68 such that for every @xmath69 , exactly one of the following holds : * @xmath70 , with @xmath56 proof term . *
@xmath71 , with @xmath72 . * @xmath73 , with @xmath74 . *
@xmath75 $ ] , with @xmath76 proof terms of the same type . *
@xmath77 $ ] , with @xmath23 proof term .
if no confusion with other sequences of terms arises , @xmath78 will often be written without intermediate dots , that is , as @xmath79 .
the _ empty sequence _ is denoted with @xmath80 and with @xmath81 we will denote stacks of length @xmath82 . if @xmath56 is a proof term , as usual in lambda calculus @xmath83 denotes the term @xmath84 .
we find among the reductions in figure [ fig : red ] the ordinary reductions for the intuitionistic constructs together with prawitz - style permutation rules @xcite for @xmath24 , as in @xcite .
the reduction rules for @xmath24 model the communication mechanism explained in section [ section - introduction ] . in the reduction @xmath85\parallel_{a
} v\ \mapsto\ \mathcal{c}[\,v[\lambda y^{\scriptscriptstyle b}\ , u /a^{\scriptscriptstyle b\rightarrow a}]\,]\parallel_{a } v\ ] ] we see that the term on the left is in some way stuck : the variable @xmath86 faces an argument @xmath22 of type @xmath3 ; of course , it has no idea how to use @xmath22 to produce a term of type @xmath10 ! on the contrary , the term @xmath23 knows very well how to use @xmath22 to produce something useful , because it contains the variable @xmath87 , which waits for a term of type @xmath13 .
thus , @xmath86 sends the term @xmath88 , with @xmath89 dummy , to @xmath23 , yielding the term @xmath90 $ ] .
this program is called to replace the useless @xmath91 and computation can go ahead .
we require the context @xmath63 $ ] to be parallel , because in this way types are not needed to define the reductions for @xmath24 and the calculus makes sense also in its untyped version and with curry - style typing .
we have chosen church - typing only to make clearer the intended meaning of the operations : had we omitted all the types from the terms , everything would have still worked just fine . in theorem
[ subjectred ] , we shall prove that indeed our reduction rules for @xmath24 are logically correct and preserve the type . ' '' '' ' '' ''
our goal now is to define a reduction strategy for typed terms of @xmath0 : a recipe for selecting , in any given term , the subterm to which apply one of our basic reductions .
as most typed lambda calculi are strongly normalizing and our reduction rules look fairly innocuous , one can not help but conjecture that any reduction strategy eventually terminates ; in other words , that reduction strategies are not necessary .
we do conjecture that the fragment with @xmath92 is indeed strongly normalizing . yet ,
already the proof of this weaker result appears excessively complex , to such an extent that arbitrary reduction strategies start to feel wrong , that is , to perform unnecessary computations .
we therefore leave strong normalization as an open problem and follow a more standard approach : krivine s ( weak ) head reduction strategy .
the difference is : in krivine s calculus each process has a unique head ; in our calculus each process has several heads , like the hydra monster .
this is due to the presence of the parallel operator @xmath20 . indeed ,
if we omit parenthesis , any term @xmath56 can be written , not uniquely , in the form @xmath60 the terms @xmath93 are parallel processes ; each one has its own head and may have an head redex . and as with the hydra monster , if we contract some head @xmath61 , more heads to contract might grow .
we now formally define what are the parallel processes that appear in a term and what is the head redex of a term .
[ definition - head ] * removing the parentheses , whenever a proof term @xmath56 can be written as @xmath60 each term @xmath61 , for @xmath94 , is said to be a * parallel process * of @xmath56 and is said to be an * elementary process * of @xmath56 in case it is not of the form @xmath95 . *
a * redex * is a term @xmath22 such that @xmath96 for some @xmath23 and basic reduction of figure [ fig : red ] .
* let @xmath78 be any stack .
a _ redex _
@xmath97 is said to be the * head redex * of a proof term @xmath56 in the following cases : 1 .
@xmath98 and @xmath99 ; 2 .
@xmath100 and @xmath101 ; 3 .
@xmath102 and @xmath103 ; 4 .
@xmath104\ , \sigma$ ] and @xmath105 $ ] ; 5 .
@xmath106\ , \sigma$ ] and @xmath107 $ ] ; 6 .
@xmath108 and @xmath109 ; 7 .
@xmath110 and @xmath111 .
+ we now define the head reduction of a proof term : the notion generalizes krivine s head reduction to parallel contexts .
the idea is to look for the leftmost among the head redexes of the parallel processes of a term and contract that redex .
the only subtlety is to determine exactly where the new redexes for @xmath24 start .
since the reduction for @xmath112 is completely localized either in @xmath22 or @xmath23 , it is reasonable to say that the redex starts where the subterm @xmath113 to be replaced is located .
[ definition - headred ] 1 .
the * starting symbol * of a redex @xmath114 is the symbol `` @xmath115 '' when @xmath116 for some stack @xmath117 of length @xmath82 ; it is the leftmost occurrence of the symbol `` @xmath21 '' such that @xmath118 is an elementary process of @xmath114 , when @xmath119 .
the * leftmost redex * among some redexes of a term @xmath56 is the redex whose starting symbol is the leftmost in @xmath56 among the starting symbols of those redexes .
we say that a term @xmath56 * head reduces * to @xmath120 and write @xmath121 when @xmath120 is obtained from @xmath56 by contracting the leftmost among the head redexes of the parallel processes of @xmath56 , using one of the basic reductions in figure [ fig : red ] . + for readability , parentheses are often omitted , but in order to spot the head redex of a term , one must mentally restore the parentheses that have been suppressed . in order to train our eye , we consider three examples of head reduction : @xmath122 @xmath123\ , \sigma\ , \succ\ , { { { \upiota}}}_{o}(u)\ , [ x_{0}.t_{0 } , x_{1}.t_{1}]\,\sigma\ , \succ\ , t_{0}[u / x_{0}]\,\sigma\ ] ] @xmath124 in the first case , the reduction for @xmath24 is used as third step of the head reduction , while in the third case , as first and last step .
we define the concept of normal form and normalizable term in the usual way . * a term @xmath56
is called a * head normal form * if there is no @xmath120 such that @xmath125 .
we define @xmath126 to be the set of head normal forms . * a sequence , finite or infinite , of proof terms @xmath127
is said to be a reduction of @xmath56 , if @xmath128 , and for all @xmath129 , @xmath130 .
a proof term @xmath22 of @xmath0 is ( head ) * normalizable * if there is no infinite reduction of @xmath22 .
we denote with @xmath131 the set of normalizable terms of @xmath0 .
the reductions defined in figure [ fig : red ] satisfy the important subject reduction theorem : reduction steps at the level of proof terms preserve the type , which is to say that they correspond to logically sound transformations at the level of proofs .
we first give the simple proof of the theorem , then analyze in detail its logical meaning in the next subsection .
[ subjectred ] if @xmath132 and @xmath133 , then @xmath134 . moreover , all the free variables of @xmath22 appear among those of @xmath56 .
it is enough to prove the theorem for basic reductions : if @xmath135 and @xmath136 , then @xmath134 .
the proof that the intuitionistic reductions and the permutation rules preserve the type is completely standard .
thus we are left with the @xmath24-reductions , which require straightforward considerations as well .
suppose @xmath85\parallel_{a } v\ \mapsto\ \mathcal{c}[\,v[\lambda y^{\scriptscriptstyle b}\ , u /a^{\scriptscriptstyle
b\rightarrow a}]\,]\parallel_{a } v\ ] ] since @xmath137 is a parallel context , @xmath138 and @xmath23 have both type @xmath11 .
now , @xmath22 must be of type @xmath3 , so @xmath139 is of type @xmath13 and thus @xmath140 $ ] is a correct term of type @xmath11 .
moreover , all the occurrences of @xmath87 in @xmath23 are eliminated by the substitution @xmath141 $ ] , so no new free variable is created . so far , in studying the system @xmath0 , we have given priority to the underlying lambda calculus and characterized it as a functional language endowed with parallelism and a communication mechanism .
the explanation of the reductions had little to do with logic and much with computation .
however , thanks to the subject reduction theorem , we know we could have proceeded the other way around .
namely , we could have given priority to logic and dealt only with transformation of proofs , in the style of prawitz natural deduction trees @xcite .
since it is instructive to explain directly this point of view , we are finally going to do so .
first of all , the following proof of @xmath6 is an example of natural deduction tree in @xmath0 : the standard reductions for lambda calculus still correspond to the ordinary conversions for all the logical constants of first - order logic : @xmath142 $ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$b$ } \unaryinfc{$a\rightarrow b$ } \axiomc{$\vdots$ } \noline \unaryinfc{$a$ } \binaryinfc{$b$ } \displayproof & \qquad \mbox { \textsf{converts to : } } \qquad & \axiomc{$\vdots$ } \noline \unaryinfc{$a$ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$b$ } \displayproof \\ \end{aligned}\ ] ] @xmath143 $ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$c$ } \axiomc{$[a_2]$ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$c$ } \trinaryinfc{$c$ } \displayproof & \qquad\mbox { \textsf{converts to } : } & \axiomc{$\vdots$ } \noline \unaryinfc{$a_i$ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$c$ } \displayproof\\ \end{aligned}\ ] ] @xmath144 @xmath145 $ } \unaryinfc{$\exists \alpha\ , a$ } \axiomc{$[a]$ } \noline \unaryinfc{$$ } \noline \unaryinfc{$\pi$ } \noline \unaryinfc{$$ } \noline \unaryinfc{$c$ } \binaryinfc{$c$ } \displayproof & \qquad \mbox { \textsf{converts to : } } \qquad & \axiomc{$\vdots$ } \noline \unaryinfc{$a[m/\alpha]$ } \noline \unaryinfc{$$ } \noline \unaryinfc{$\pi[m/\alpha]$ } \noline \unaryinfc{$$ } \noline \unaryinfc{$c$ } \displayproof \\ \end{aligned}\ ] ] @xmath146 $ } \displayproof & \qquad \mbox { \textsf{converts to : } } \qquad & \axiomc{$\pi[m/\alpha]$ } \noline \unaryinfc{$$ } \noline \unaryinfc{$a[m/\alpha]$ } \displayproof \\ \end{aligned}\ ] ] the permutation reductions for the terms of the form @xmath147 , are just instances of prawitz - style permutations for disjunction elimination . from the logical perspective , they are used to systematically transform , whenever possible , the logical shape of the conclusion .
this reduction is essential because the dummett inference rule does not yield much when employed to prove implications or disjunctions ; but it becomes herbrand constructive , whenever used to prove existentially quantified statements . as an example of permutation for @xmath24 , we consider the one featuring an implication as conclusion : @xmath148 $ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$f \to g$ } \axiomc{$[b\rightarrow a]$ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$f \to g$ } \rightlabel{$ { { \mathsf{d } } } $ } \binaryinfc{$f\to g$ } \axiomc{$\vdots$ } \noline \unaryinfc{$f$ } \binaryinfc{$g$ } \displayproof & \ \ \ \mbox { \textsf{converts to : } } \ \ \ & \axiomc{$[a\rightarrow b]$ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$f \to g$ } \axiomc{$\vdots$ } \noline \unaryinfc{$f$ } \binaryinfc{$g$ } \axiomc{$[b\rightarrow a]$ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$f \to g$ } \axiomc{$\vdots$ } \noline \unaryinfc{$f$ } \binaryinfc{$g$ } \binaryinfc{$g$ } \displayproof \\ \end{aligned}\ ] ] there are similar permutations for all other elimination rules , as one can see translating in natural deduction the permutations of figure [ fig : red ] . with the following notation
we denote a deduction of @xmath11 that , in order to obtain its final conclusion , combines the deductions @xmath149 of @xmath11 using _ only _ the dummett rule @xmath150 times . in other words , below the conclusions @xmath11 of the deductions @xmath149 only the dummett rule is used .
this configuration corresponds to a parallel context in our lambda calculus , as in definition [ defi - parallelc ] . with the notation
we denote a deduction of @xmath11 that , starting from @xmath10 , applies only elimination rules to obtain @xmath11 ; in particular , @xmath10 must be the main premise of the first elimination rule which concludes @xmath151 , which must be the main premise of the second elimination rule which concludes @xmath152 and so on down to @xmath11 .
this configuration corresponds to the concept of stack of definition [ definition - stack ] .
finally , we can look at the two reductions for proofs containing the dummett rule .
let us consider just the first conversion for @xmath24 , the second being perfectly symmetric : @xmath153 $ } \axiomc{$[a\rightarrow b]$ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{$a$ } \insertbetweenhyps{\hskip -3pt } \binaryinfc{$b$ } \doubleline\rightlabel{$\mathsf{el}$ } \unaryinfc{$c$ } \axiomc{$\cdots$ } \axiomc{$\mathcal{d}_{n}$ } \noline \unaryinfc{$c$ } \doubleline \rightlabel{$ { { \mathsf{d } } } $ } \insertbetweenhyps{\hskip -2pt } \quinaryinfc{$c$ } \noline \axiomc{$[b\rightarrow a]$ } \unaryinfc{$\mathcal{d}$ } \noline \unaryinfc{$c$ } \rightlabel{$ { { \mathsf{d } } } $ } \binaryinfc{$c$ } \displayproof & \mbox{\textsf{converts to : } } & \axiomc{$\mathcal{d}_{1}$ } \noline \unaryinfc{$c$ } \axiomc{$\cdots$ } \axiomc{$[a\rightarrow b]$ } \noline \unaryinfc{$\vdots$ } \noline \unaryinfc{{$a$ } } \unaryinfc{$b\rightarrow a$ } \noline
\unaryinfc{$\mathcal{d}$ } \noline \unaryinfc{$c$ } \axiomc{$\cdots$ } \axiomc{$\mathcal{d}_{n}$ } \noline \unaryinfc{$c$ } \doubleline \rightlabel{$ { { \mathsf{d } } } $ } \insertbetweenhyps{\hskip -2pt } \quinaryinfc{$c$ } \noline \axiomc{$[b\rightarrow a]$ } \unaryinfc{$\mathcal{d}$ } \noline \unaryinfc{$c$ } \rightlabel{$ { { \mathsf{d } } } $ } \binaryinfc{$c$ } \displayproof \end{aligned}\ ] ] the conversion above focuses first on the deduction @xmath154 on the left branch of the proof ; it replaces the hypothesis @xmath13 of @xmath154 with a proof of @xmath13 directly obtained from the proof of @xmath3 found on the left branch ; afterwards , it takes the deduction so generated and replaces with it the old proof of @xmath11 obtained from @xmath10 by elimination rules .
+ there is a crucial assumption about the structure of the first proof . in the left branch of the dummett rule , the hypothesis @xmath12 is used together with @xmath3 to obtain @xmath10 , which is in turn used to infer @xmath11 _ by means only of a main branch of elimination rules _ , as called by prawitz . thanks to this restriction
, the proof of @xmath3 does not end up having more open assumptions in the second proof than it has in the first proof . + but what have we gained with this reduction ?
it looks like we made no progress at all .
the hypothesis @xmath12 may be actually used more times in the second proof than in the first , because the hypothesis @xmath13 might be used several times in the deduction @xmath154 !
actually , the gain is subtle . in the left branch of the first proof
the formula @xmath10 was derived in a fictitious way : by an arbitrary hypothesis @xmath12 , bearing no relationship with @xmath11 .
since @xmath10 is _ used _ to obtain @xmath11 , we can not expect @xmath10 to provide constructive content to @xmath11 , in particular no witness if @xmath11 is an existential formula .
the conversion above gets rid of this configuration and provide a more direct proof of @xmath11 : in the new proof , if @xmath13 is employed to derive @xmath3 by modus ponens , one can discard @xmath10 and use the proof of @xmath3 coming from the first proof . the main difficulty that we face with our reduction rules for @xmath24 is _
termination_. there is hardly any decrease in complexity from before to after the reduction and the road toward a combinatorial termination proof looks barred .
we are thus forced to employ a far more abstract technique : realizability .
in this section we prove that each term of @xmath0 _ realizes _ its type and is normalizing . to this end
, we make a detour into a logically inconsistent , yet computationally sound world : the system @xmath155 , a type system which extends @xmath0 . the idea that extending a system can make easier rather than harder to prove its normalization might not seem very intuitive , but it is well tested and very successful ( see @xcite , @xcite , @xcite , @xcite ) .
@xmath155 will be our calculus of realizers .
it is indeed typical of realizability , the method we shall use , to set up a calculus with more realizers than the actual proof terms @xcite .
the idea is that a realizer is defined as a proof term that defeats every opposer and passes every termination test ; but proof terms , as opposers and testers of proof terms themselves , are not enough ; proof terms must be opposed and tested also by `` cheaters '' , terms that do satisfy the same definition of realizability , but only because they have some advantage .
these extra tests make proof terms stronger realizers than they otherwise would be .
we may imagine a realizer as a tennis player that trains himself to return fast balls thrown by a robot : if he withstands the attacks of the robot , he will perform all the more well against real weaker humans .
the system @xmath155 is not meant to be a logical system : it would be inconsistent ! the purpose of the system is not logical , but computational : to simulate the reduction rules for @xmath24 by an abort operator @xmath156 .
we define the typing rules of @xmath155 to be those of @xmath0 plus a new term formation scheme : + abort axiom : : 0.15 in with @xmath156 , @xmath157 , we shall denote some generic constant @xmath158 .
the reduction rules for the terms of @xmath155 are those for @xmath0 with the addition of a new reduction rule defined in figure [ fig : fn ] . ' '' '' ' '' '' the abort computational construct reminds krivine s @xmath159 , which removes the current continuation @xmath160 and restore a previously saved continuation @xmath161 : @xmath162 there is indeed an analogy with krivine s realizability : the terms of @xmath0 correspond to krivine s proof - like terms , whereas the terms of @xmath155 correspond to krivine s inconsistent terms that may contain @xmath159 and may realize any formula .
but in our case @xmath155 is just a _
tool for defining realizability _
, not a tool for implementing reductions , like @xmath159 in krivine case .
the role of @xmath156 will emerge later on in the proof of propositions [ proposition - simul ] and [ proposition - parallel ] .
however , by now , the intuition should be pretty clear : in the reduction @xmath163\parallel_{a } v\ \mapsto\ \mathcal{c}[\,v[\lambda y\ , u /a]\,]\parallel_{a } v\ ] ] the term @xmath164 aborts the local continuation @xmath78 .
the difficulty is that the new continuation @xmath165 $ ] , from the perspective of @xmath21 , is created out of nowhere !
therefore proving by induction that @xmath166 $ ] is a realizer would not be of great help for proving that the whole term @xmath166\parallel_{a } v$ ] is realizer . with terms of the form
@xmath167 we can instead simulate locally the global reduction above and get a stronger induction hypothesis .
the definition [ definition - stack ] of stack is of course extended to @xmath155 and the definition [ definition - head ] of head redex is extended to the terms of @xmath155 by saying that @xmath168 is the * head redex * of @xmath168 whenever @xmath22 and @xmath168 have the same type .
the reduction relation @xmath169 for the terms of @xmath155 is then defined as in definition [ definition - headred ] . in the following ,
we define @xmath170 to be the set of normalizing proof terms of @xmath155 . as usual in lambda calculus , a value represents the result of the computation : a function for arrow and universal types , a pair for product types , a boolean for sum types and a witness for existential types and in our case also the abort operator .
[ definition - value ] * a proof term is a * value * if it is of the form @xmath171 or @xmath172 or @xmath173 or @xmath174 or @xmath175 or @xmath176 or @xmath156 . *
a proof term is * neutral * if it is neither a value nor of the form @xmath112 .
we now prove a property of head normal forms that we will be crucial in the following .
it is a generalization of the well known head normal form theorem for lambda calculus and tells us that if we decompose a proof term into its elementary parallel processes , then each of them is either a value or some variable or constant applied to a list of argument .
[ proposition - hnf ] suppose @xmath56 is in head normal form and @xmath177 and that each @xmath61 is an elementary process .
then for every @xmath178 , there is some stack @xmath78 such that either @xmath179 , with @xmath180 , or @xmath181 , with the type of @xmath22 different from the type of @xmath182 , or @xmath183 or @xmath61 is a value .
by induction on @xmath56 . if @xmath56 is a value , we are done .
there are two other cases to consider .
1 . @xmath110 . by definition [ definition - head
] , the parallel processes of @xmath22 and @xmath23 are parallel processes of @xmath95 as well , so they can not have head redexes ; hence @xmath22 and @xmath23 are in head normal form . by induction hypothesis ,
@xmath22 and @xmath23 are of the desired form , thus we just have to check that if @xmath184 and for some @xmath129 , @xmath185 , with @xmath186 , then @xmath187 ( and symmetricallly for @xmath23 ) . indeed , if for some @xmath129 , @xmath188 , then @xmath189 $ ] for some parallel context @xmath63 $ ] , and therefore @xmath95 would be the leftmost head redex of itself , which is impossible since by assumption it is in head normal form
@xmath56 is neutral .
then @xmath56 can be written , for some stack @xmath78 , as @xmath190 where @xmath114 is a value or @xmath191 or @xmath192 . in the third case , we are done ; in the first and second case , @xmath193 , so @xmath194 would be the head redex of @xmath56 , unless @xmath195 , with the type of @xmath117 different from the type of @xmath196 , which is the thesis .
our main goal now is to prove the normalization theorem for @xmath0 : every proof term of @xmath0 reduces in a finite number of head reduction steps to a head normal form .
we shall employ a notion of _ classical realizability _
, a generalization of the tait - girard reducibility method @xcite that works for classical type systems .
the origins of classical realizability can be traced all the way back to parigot @xcite and krivine @xcite classical reducibility , but we present it in a fashion popularized later by krivine in his work on realizability @xcite , which is indeed a generalization of classical reducibility . thanks to the fact that one considers only head reduction , krivine - style classical realizability is slightly simpler than the notions usually employed to derive strong normalization .
given a logic , we raise a question : what kind of evidence does a proof provide other than the tiny bit `` 1 '' declaring the truth of the proven statement ?
realizability is a semantics explaining what is to be taken as _ constructive evidence for a statement _ and a technique for showing that proofs can provide such an evidence . formally , realizability is a relation between terms of @xmath155 and formulas , with terms playing the role of constructions and formulas determining what properties a construction should satisfy .
in particular , to each formula @xmath11 is associated a set of stacks @xmath197 , which represents a collection of _ valid tests _
: whenever a term passes all these tests , in the sense that it maps them into terminating programs , it is a realizer .
as prescribed by the pragmatist viewpoint , the clauses that defines realizability follow the shape of elimination rules , in order to make sure that no matter how a program is used , it always terminates .
[ definition - reducibility ] assume @xmath56 is a term of @xmath155 and @xmath11 is a formula of @xmath26 .
we define by mutual induction the relation @xmath198 ( `` @xmath56 realizes @xmath11 '' ) and a set @xmath199 of stacks of @xmath155 ( the `` valid tests for @xmath11 '' ) according to the form of @xmath11 : * @xmath198 if and only if @xmath200 and for all @xmath201 , @xmath202 * @xmath203 * @xmath204 * @xmath205 * @xmath206 { { { \ , { \mathpalette { { { \settoheight\cdotaxis{$}\vcenter{}$ } \sbox0 { \raisebox\cdotaxis { \scalebox{2.4 } { \raisebox{-1.1pt } { $ \mathsurround=0pt } .$ } } } } \dp0=0pt \sbox2{$}\bullet$ } \ifdim\ht2<\ht0 \ht0=\ht2 \fi \sbox2{$\mathsurround=0pt } .$ } \hbox to \wd2{\hss\usebox{0}\hss } } } } \,}}}\sigma \ |\ \forall t.\ ( t { \vdash}a\implies u[t / x]\,\sigma \in { \mathsf{hn}^{\star } } ) \land ( t{\vdash}b\implies
v[t / y]\ , \sigma \in { \mathsf{hn}^{\star}})\}\cup \{\epsilon\}$ ] * @xmath207||\}\cup \{\epsilon\}$ ] * @xmath208 { { { \ , { \mathpalette { { { \settoheight\cdotaxis{$}\vcenter{}$ } \sbox0 { \raisebox\cdotaxis { \scalebox{2.4 } { \raisebox{-1.1pt } { $ \mathsurround=0pt } .$ } } } } \dp0=0pt \sbox2{$}\bullet$ } \ifdim\ht2<\ht0 \ht0=\ht2 \fi \sbox2{$\mathsurround=0pt } .$ } \hbox to \wd2{\hss\usebox{0}\hss } } } } \,}}}\sigma \
|\ \forall t.\ t { \vdash}a[m/\alpha]\implies v[m/\alpha][t / x]\ , \sigma \in { \mathsf{hn}^{\star}}\}\cup \{\epsilon\}$ ] in this section we prove the basic properties of classical realizability .
they are all we need to prove the adequacy theorem [ adequacy theorem ] , which states that typable terms are realizable . the arguments for establishing the properties are in many cases standard ( see krivine @xcite ) .
we shall need extra work for dealing with terms of the form @xmath112 .
the first task is to prove that realizability is sound for all introduction and elimination rules of @xmath0 .
we start with the eliminations .
[ proposition - redelim ] 1 . if @xmath209 and @xmath210 , then @xmath211 .
if @xmath212 , then for every term @xmath213 of @xmath26 , @xmath214 $ ] .
3 . if @xmath215 , then @xmath216 and @xmath217 .
4 . if @xmath218 and for every @xmath219 , @xmath220{\vdash}c$ ] and for every @xmath221 , @xmath222{\vdash}c$ ] , then @xmath223 { \vdash}c$ ] .
if @xmath224 and for every @xmath225 and for every @xmath226 $ ] , @xmath227[w / x]{\vdash}c$ ] , then @xmath228{\vdash}c$ ] . 1 .
assume @xmath209 and @xmath210 .
let @xmath229 ; we must show @xmath230 . indeed , since @xmath231 , by definition [ definition - reducibility ] @xmath232 and since @xmath209
, we conclude @xmath230 .
2 . similar to 1 .
assume @xmath215 .
let @xmath233 ; we must show @xmath234 .
indeed , by definition [ definition - reducibility ] @xmath235 and since @xmath215 , we conclude @xmath236 .
a symmetrical reasoning shows that @xmath237 .
4 . let @xmath201 .
we must show that @xmath238\,\sigma\in{\mathsf{hn}^{\star}}$ ] . by hypothesis , for every @xmath219 , @xmath220\,\sigma \in{\mathsf{hn}^{\star}}$ ] and for every @xmath221 , @xmath222\,\sigma \in{\mathsf{hn}^{\star}}$ ] ; by definition [ definition - reducibility ] , @xmath239\,\sigma\in ||a\lor b||$ ] . since @xmath218 , we conclude @xmath223\,\sigma\in{\mathsf{hn}^{\star}}$ ]
similar to 4 .
realizability is also sound for introduction rules and the abort operator realizes any implication
. [ proposition - somecases ] 1 .
if for every @xmath231 , @xmath240{\vdash}b$ ] , then @xmath241 .
if for every term @xmath213 of @xmath26 , @xmath227{\vdash}b[m/\alpha]$ ] , then @xmath242 .
3 . if @xmath210 and @xmath243 , then @xmath244 .
4 . if @xmath245 , with @xmath246 , then @xmath247 . 5 . if @xmath248 $ ] , then @xmath249 .
if @xmath3 and @xmath10 are any two formulas , then @xmath250 . 1 .
suppose that for every @xmath231 , @xmath240{\vdash}b$ ] .
let @xmath251 .
we have to show @xmath252 . if @xmath253 , indeed @xmath254 .
suppose then @xmath255 , with @xmath256 and @xmath257 .
since by hypothesis @xmath240{\vdash}b$ ] , we have @xmath240\,\rho\in { \mathsf{hn}^{\star}}$ ] ; moreover , @xmath258\ , \rho$ ] . therefore , @xmath259 .
similar to 1 .
suppose @xmath210 and @xmath243 .
let @xmath260 .
we have to show @xmath261 . if @xmath253 , indeed @xmath262 .
suppose then @xmath263 , with @xmath246 and @xmath264 , when @xmath265 , and @xmath266 , when @xmath267 .
we have two cases . 1 .
since by hypothesis @xmath210 , we have @xmath268 ; moreover , @xmath269 . therefore , @xmath270 .
2 . @xmath267 .
since by hypothesis @xmath271 , we have @xmath272 ; moreover , @xmath273 . therefore , @xmath270 .
4 . suppose @xmath245 .
let @xmath274 .
we have to show @xmath275 . if @xmath253 , indeed @xmath276 .
suppose then @xmath277 { { { \ , { \mathpalette { { { \settoheight\cdotaxis{$}\vcenter{}$ } \sbox0 { \raisebox\cdotaxis { \scalebox{2.4 } { \raisebox{-1.1pt } { $ \mathsurround=0pt } .$ } } } } \dp0=0pt \sbox2{$}\bullet$ } \ifdim\ht2<\ht0 \ht0=\ht2 \fi \sbox2{$\mathsurround=0pt } .$ } \hbox to \wd2{\hss\usebox{0}\hss } } } } \,}}}\rho$ ] and for all @xmath278 , if @xmath279 , then @xmath280\,\rho\in { \mathsf{hn}^{\star}}$ ] and if @xmath281 , then @xmath282\ , \rho\in { \mathsf{hn}^{\star}}$ ] .
we have to show @xmath283\,\rho \in { \mathsf{hn}^{\star}}$ ] . by hypothesis @xmath284\,\rho\in { \mathsf{hn}^{\star}}$ ] ; moreover , @xmath285\,\rho { \succ}u_{i}[t / x_{i}]\ , \rho\ ] ] therefore , @xmath283 \in{\mathsf{hn}^{\star}}$ ] .
similar to 4 .
let @xmath286 .
we have to show that @xmath287 . if @xmath253 , indeed @xmath288 .
suppose then @xmath289 , with @xmath210 and @xmath257 .
we have to show that @xmath290 .
since @xmath210 and @xmath291 , we have @xmath292 .
moreover , if @xmath293 is not a redex , we are done , and if @xmath294 , the thesis follows .
it is now that the abort operator really enters the scene .
thanks to it , any reduction @xmath295 can be simulated in a purely local way .
this is possible because any such reduction affects only what is inside @xmath22 and leaves @xmath23 untouched .
then , in order to replicate the reduction is enough to substitute to @xmath21 a term @xmath296 that throws away any stack of terms it is applied to , like @xmath21 does , and then restores @xmath23 , with some substitution depending on the context .
of course , symmetrical considerations hold true for any reduction @xmath297 .
[ proposition - simul ] define @xmath298}\ ] ] @xmath299}\ ] ] with @xmath300 , @xmath89 , @xmath301 and @xmath156 occurring with the right type .
then @xmath302{\succ}^{+ } u'[{{\mathscr{a}}}/a]\ ] ] @xmath303{\succ}^{+ } v'[{{\mathscr{b}}}/a]\ ] ] we prove the first statement , the other being perfectly symmetric .
the only trouble is to formalize precisely the argument , which is otherwise intuitively obvious . to this end , we first need some simple , but tedious to prove , claims .
* _ claim 1_. every parallel process of @xmath304 $ ] is of the form @xmath305 $ ] , where @xmath56 is a parallel process of @xmath22 .
* _ claim 2_. for every parallel process @xmath305 $ ] of @xmath304 $ ] , if the starting symbol of the head redex of @xmath56 is in the @xmath31-th , from left to right , elementary process of @xmath22 , then the starting symbol of the head redex of @xmath305 $ ] is in the @xmath31-th elementary process , from left to right , of @xmath304 $ ] .
_ proof of claim 1_. by induction on @xmath22 .
let @xmath120 be a parallel process of @xmath306 $ ] .
if @xmath307 , since @xmath22 is a parallel process of itself , we are done .
suppose now @xmath308 .
then , assuming @xmath309 , we have @xmath310=(u_{1}[{{\mathscr{a}}}/a]\parallel_{b } u_{2}[{{\mathscr{a}}}/a])\ ] ] therefore , @xmath311 , @xmath312 $ ] and @xmath313 $ ] .
now , @xmath120 is a parallel process of either @xmath314 or @xmath315 ; by induction hypothesis , @xmath316 $ ] , where @xmath56 is a parallel process of @xmath317 , in the first case , and of @xmath318 in the second .
_ proof of claim 2_. by induction on @xmath22 .
let @xmath316 $ ] be a parallel process of @xmath306 $ ] , where @xmath56 is a parallel process of @xmath22 .
we have two cases . 1 . @xmath319 $ ] .
if @xmath320 and @xmath97 is its head redex , then @xmath22 is the first and unique elementary process of @xmath22 ; moreover , @xmath321)\,(\sigma[{{\mathscr{a}}}/a])$ ] , thus @xmath322 $ ] is its head redex and indeed @xmath120 is the first and unique elementary process of @xmath56 .
if @xmath323 and is a redex , then by definition [ definition - headred ] , the starting symbol of @xmath22 is the occurrence of @xmath324 such that @xmath325 is the @xmath31-th elementary process of @xmath22 and for every @xmath326 , the @xmath213-th elementary process of @xmath22 does not start with @xmath324 . since @xmath327\parallel_{b } u_{2}[{{\mathscr{a}}}/a]$ ] and @xmath328 , the starting symbol of @xmath120 is the occurrence of @xmath324 such that @xmath329)\,(\sigma [ { { \mathscr{a}}}/a])$ ] is the @xmath31-th elementary process of @xmath120 .
+ 2 . @xmath330 $ ] .
since @xmath316 $ ] and @xmath56 is a parallel process of @xmath22 with @xmath331 , there is a parallel context @xmath63 $ ] such that @xmath332 $ ] . by induction hypothesis , assuming that the starting symbol of the head redex of @xmath56 is in the @xmath31-th elementary process of @xmath56 , then the starting symbol of the head redex of @xmath120 is in the @xmath31-th elementary process of @xmath120 . since @xmath333)[t']$ ] , if @xmath213 is the number of elementary processes on the left of @xmath56 in @xmath334 $ ] , then the starting symbol of the head redex of @xmath56 is in the @xmath335-th elementary process of @xmath22 and the starting symbol of the head redex of @xmath120 is in the @xmath335-th elementary process of @xmath336 , which is the thesis .
let now us return to the main line of the proof .
suppose @xmath337 then for some parallel context @xmath137 , we have @xmath338 $ ] and @xmath339 $ ] , where @xmath340 is either obtained from @xmath341 by contracting the head redex @xmath114 of @xmath341 or @xmath342 $ ] and @xmath343 ; in the first case , it is @xmath114 that is the leftmost among the head redexes of the parallel processes of @xmath112 , whereas in the second case , it is @xmath95 . with this notation
, we have @xmath344\parallel_{a } v{\succ}\mathcal{c}[q']\parallel_{a } v\ ] ] we must show @xmath345\succ^{+ } u'[{{\mathscr{a}}}/a]\ ] ] there are several cases .
* @xmath346 and @xmath347 .
let @xmath348 $ ] .
we first need to show that @xmath349 $ ] is the leftmost among the head redexes of the parallel process of @xmath304 $ ] as well .
assume that the starting symbol of @xmath114 is in the @xmath31-th elementary processes of @xmath22 .
by claim 2 , the starting symbol of @xmath349 $ ] is in the @xmath31-th elementary process of @xmath304 $ ] .
suppose by the way of contradiction , that there is a parallel process @xmath350 of @xmath304 $ ] whose head redex has a starting symbol more on the left , that is , in the @xmath213-th elementary process of @xmath304 $ ] , with @xmath326 . by claim 1 ,
@xmath351 $ ] , where @xmath352 is a parallel process of @xmath22 . by claim 2 , the starting symbol of the head redex of @xmath352
is in the @xmath213-th elementary process of @xmath22 , which contradicts the assumption on @xmath341 and @xmath114 .
now , letting @xmath353 $ ] , @xmath316 $ ] , @xmath354 $ ] , @xmath355 $ ] , we get @xmath345=\mathcal{c}'[(\lambda x\ , s ' ) t'\ , \sigma']{\succ}\mathcal{c}'[s'[t'/x]\ , \sigma']=\mathcal{c}[s[t / x]\,\sigma][{{\mathscr{a}}}/a]=u'[{{\mathscr{a}}}/a]\ ] ] * @xmath356 and @xmath357 or @xmath358 $ ] or @xmath359 $ ] or @xmath360 and let , respectively , @xmath361 or @xmath362 $ ] or @xmath363[s / x]$ ] or @xmath364 . by exactly the same considerations of the previous case , we get @xmath345=\mathcal{c}[{{\mathscr{a}}}/a][r\,\sigma[{{\mathscr{a}}}/a]]{\succ}\mathcal{c}[{{\mathscr{a}}}/a][r'\,\sigma[{{\mathscr{a}}}/a]]=\mathcal{c}[r'\,\sigma][{{\mathscr{a}}}/a]=u'[{{\mathscr{a}}}/a]\ ] ] * @xmath365 or @xmath366\parallel_{b } s$ ] for some variable @xmath367 ( the other case is symmetric ) ; let respectively @xmath368 or @xmath369\parallel_{b } s$ ] . by exactly the same considerations of the previous case , we get @xmath345=\mathcal{c}[{{\mathscr{a}}}/a][r[{{\mathscr{a}}}/a]]{\succ}\mathcal{c}[{{\mathscr{a}}}/a][r'[{{\mathscr{a}}}/a]]=\mathcal{c}[r'][{{\mathscr{a}}}/a]=u'[{{\mathscr{a}}}/a]\ ] ] * @xmath342 $ ] and @xmath343 .
let @xmath316 $ ] , @xmath355 $ ] and @xmath354 $ ] .
we first need to show that the head redex of @xmath370={{\mathscr{a}}}\,t'\,\sigma'$ ] is the leftmost among the head redexes of the parallel processes of @xmath304 $ ] .
assume that @xmath371 is the @xmath31-th elementary process of @xmath22 , so that @xmath372 is the @xmath31-th elementary process of @xmath304 $ ] as well .
then , no parallel process of @xmath22 has an head redex whose starting symbol is in the @xmath213-th elementary process of @xmath22 , with @xmath326 . by claims 1 and 2 ,
no parallel process @xmath373 $ ] of @xmath304 $ ] , where @xmath352 is a parallel process of @xmath22 , has an head redex whose starting symbol is in the @xmath213-th elementary process of @xmath22 , with @xmath326 , otherwise the starting symbol of the head redex of @xmath352 would be in the @xmath213-th elementary process of @xmath22 as well ( claim 2 applies , since @xmath352 can not be of the form @xmath374 , given that @xmath21 is the starting symbol of the redex @xmath95 ) .
finally , we conclude @xmath345=\mathcal{c}'[{{\mathscr{a}}}\,t'\,\sigma']= \mathcal{c}'[(\lambda x\ , { { \mathcal{a}}}\,v[\lambda y\ , x / a])\ , t'\,\sigma']{\succ}\ ] ] @xmath375\,\sigma']{\succ}\mathcal{c}'[v[\lambda y\ , t ' / a]]=\mathcal{c}[v[\lambda y\ , t / a]][{{\mathscr{a}}}/a]=u'[{{\mathscr{a}}}/a]\eqno{\qed}\ ] ] we are now able to tackle the most difficult case of the adequacy theorem [ adequacy theorem ] for realizability : proving that realizability is also sound for the dummett rule .
the idea is that proposition [ proposition - simul ] allows us to use in a very strong manner an inductive hypothesis that will naturally be granted when proving the adequacy theorem .
this hypothesis is knowing that for every @xmath209 , @xmath376\in{\mathsf{hn}^{\star}}$ ] ; since one can prove that the term @xmath296 realizes @xmath12 , one can conclude with simple reasoning that the head reduction reduces @xmath22 in @xmath377 only a finite number of times and a symmetric reasoning holds for @xmath23
. had nt we the abort operator and thus the possibility of local simulation , the hypothesis that for every @xmath209 , @xmath376\in{\mathsf{hn}^{\star}}$ ] , would not be enough to conclude a great deal .
details follow .
[ proposition - parallel ] 1 . if for every @xmath209 , @xmath376\in{\mathsf{hn}^{\star}}$ ] and for every @xmath378 , @xmath379\in{\mathsf{hn}^{\star}}$ ] , then @xmath380 .
2 . if for every @xmath209 , @xmath376{\vdash}c$ ] and for every @xmath378 , @xmath379{\vdash}c$ ] , then @xmath381 . 1 . define @xmath382)\ ] ] we start by showing that @xmath383 , which establishes by means of the hypothesis that @xmath304\in{\mathsf{hn}^{\star}}$ ] .
let @xmath384 ; the case @xmath385 is trivial , so we assume @xmath386 , with @xmath231 and @xmath387
. we must show @xmath388 .
we have @xmath389 ) ) t\ , \sigma{\succ}{{\mathcal{a } } } ( v[\lambda y\ , t / a ] ) \
, \sigma{\succ}v[\lambda y\ , t / a]\ ] ] ( assuming the last reduction is possible : if not , the thesis is trivial ) . in order to obtain @xmath390\in{\mathsf{hn}^{\star}}$ ] , which is what we wanted , it is enough to show that @xmath391 . let @xmath392 ; again , the case @xmath393 is trivial , so we assume @xmath394 , with @xmath395 and @xmath396 . we must show @xmath397 .
indeed , since @xmath231 , @xmath398 + we now prove that @xmath399 by induction on the length of the reduction of @xmath304 $ ] in head normal form .
we have two cases . 1 .
assume @xmath400 so that in particular @xmath22 is in head normal form .
define @xmath401)\ ] ] since we are going again to prove the thesis by induction on the length of the reduction of @xmath402 $ ] in head normal form , we first need to show that @xmath403 , which allows us to conclude that indeed @xmath402\in{\mathsf{hn}^{\star}}$ ] .
let @xmath404 ; the case @xmath385 is trivial , so we assume @xmath386 , with @xmath405 and @xmath406 .
we have to show @xmath407 .
we have @xmath408 ) ) t\ , \sigma{\succ}{{\mathcal{a } } } ( u[\lambda y\ , t / a ] ) \
, \sigma{\succ}u[\lambda y\ , t / a]\ ] ] now , @xmath22 is in head normal form and thus by proposition [ proposition - hnf ] , @xmath409 and for each @xmath410 , either @xmath411 , with @xmath412 , or @xmath413 , with the type of @xmath278 different from the type of @xmath414 , or @xmath415 or @xmath416 or @xmath417 is a value .
therefore , @xmath418 $ ] is in head normal form , because the substitution does not create head redexes in any parallel process of @xmath22 .
+ we now prove the main thesis . by proposition [ proposition - simul ]
, @xmath402{\succ}^{+ } v'[{{\mathscr{b}}}/a]$ ] , so by induction hypothesis we conclude @xmath419 and thus @xmath399 .
2 . assume @xmath337 by proposition [ proposition - simul ] , @xmath304{\succ}^{+ } u'[{{\mathscr{a}}}/a]$ ] , so by induction hypothesis we conclude @xmath420 and thus @xmath399 .
2 . let @xmath201 .
we must show that @xmath421 . by hypothesis , for every @xmath209 , @xmath376\,\sigma \in{\mathsf{hn}^{\star}}$ ] and for every @xmath378 , @xmath379\,\sigma \in{\mathsf{hn}^{\star}}$ ] . by point 1 .
, @xmath422 . since @xmath423 we are done .
we finally prove that realizability is sound for @xmath0 : if we replace all free proof term variables of any proof term with realizers , then we get a realizer . [
adequacy theorem ] suppose that @xmath424 in the system @xmath0 , with @xmath278 having free variables among @xmath425 .
for all terms @xmath426 of @xmath26 , if there are terms @xmath427 such that @xmath428\ ] ] then @xmath429[t_1/x_1^ { { { \overline{a } } } _ { 1}}\cdots t_n / x_n^ { { { \overline{a } } } _ { n}}]{\vdash}a[{r}_1/\alpha_1\cdots { r}_k/\alpha_k]\ ] ] for any term @xmath23 and formula @xmath10 , we define @xmath430[t_1/x_1^ { { { \overline{a } } } _ { 1}}\cdots t_n / x_n^ { { { \overline{a } } } _ { n}}]\ ] ] and @xmath431\ ] ] we proceed by induction on @xmath278 . consider the last rule @xmath432 in the derivation of @xmath424 : 1 .
if @xmath433 , for some @xmath129 , then @xmath434 and @xmath435 .
so @xmath436 .
2 . if @xmath432 is the @xmath437 rule , then @xmath438 , @xmath439 so by proposition [ proposition - redelim ] , @xmath440 , for @xmath441 and @xmath442 by induction hypothesis .
3 . if @xmath432 is the @xmath443 rule , then @xmath444 , @xmath445 and @xmath446 .
so , @xmath447 , because by renaming of bound variables we can assume @xmath448 . for every @xmath449 , by induction hypothesis on @xmath22 , @xmath450{\vdash } { { \overline{c } } } $ ] . therefore
, by proposition [ proposition - somecases ] , @xmath451 .
4 . if @xmath432 is a @xmath452 rule , say left ( the other case is symmetric ) , then @xmath453 , @xmath454 and @xmath455 .
so , @xmath456 and by induction hypothesis @xmath442 .
hence , by proposition [ proposition - somecases ] we conclude @xmath457 .
5 . if @xmath432 is a @xmath458 rule , then @xmath459\ ] ] and @xmath460 with @xmath461
by induction hypothesis , we have @xmath462 ; moreover , for every @xmath449 , we have @xmath463{\vdash } { { \overline{d } } } $ ] and for every @xmath464 , we have @xmath465{\vdash } { { \overline{d } } } $ ] . by proposition [ proposition - redelim ]
, we obtain @xmath466{\vdash } { { \overline{d } } } $ ] . 6 .
the cases @xmath467 and @xmath468 are straightforward .
the cases @xmath469 and @xmath470 are similar respectively to @xmath471 and @xmath472 .
if @xmath432 is the @xmath473 rule , then @xmath474 , @xmath475 $ ] and @xmath476 .
so , @xmath477 . by inductive hypothesis @xmath478 and so @xmath479
$ ] by proposition [ proposition - redelim ] . 9 .
if @xmath432 is the @xmath480 rule , then @xmath481 , @xmath482 and @xmath483 ( with @xmath9 not occurring free in the types @xmath57 of the free variables of @xmath22 ) .
so , @xmath484 , since we may assume @xmath485 .
let @xmath213 be a term of @xmath26 ; by proposition [ proposition - somecases ] , it is enough to prove that @xmath486{\vdash } { { \overline{b } } } [ { m}/\alpha]$ ] , which amounts to showing that the induction hypothesis can be applied to @xmath22 . for this purpose ,
we observe that , since @xmath485 , for @xmath487 we have @xmath488\ ] ] 10 .
if @xmath432 is the @xmath489 rule , then @xmath490 , @xmath461 and + by induction hypothesis , for every @xmath491 , we have @xmath492{\vdash } { { \overline{d } } } $ ] and for every @xmath493 , @xmath494{\vdash } { { \overline{d } } } $ ] . by proposition
[ proposition - parallel ] , we conclude @xmath495 . 11 . if @xmath432 is the ex falso quodlibet rule , then @xmath496 , @xmath497 and @xmath498 .
now , @xmath499 and @xmath500 .
we conclude @xmath501 . as corollary of the adequacy theorem [ adequacy theorem ] ,
one obtains normalization for @xmath0 .
[ theorem - snnem ] suppose that @xmath502 is a proof term of @xmath0
. then @xmath503 .
assume @xmath504 are the free variables of @xmath56 .
we observe that @xmath505 , for @xmath506 because , given any @xmath507 , @xmath508 .
therefore , from theorem [ adequacy theorem ] , we derive that @xmath231 and since @xmath291 , we conclude @xmath509 .
in this section , we finally show that our curry - howard correspondence for @xmath0 is meaningful from the computational perspective .
we already know that every execution of every program we extract always terminate ; now we prove that in the case of _ any _ existentially quantified formula @xmath1 , every closed program of that type produces a complete finite sequence @xmath510 of possible witnesses for @xmath1 .
this means that whatever first - order model we consider , there will be an @xmath129 such that @xmath511 $ ] is true in it . in other terms , we have provided a proof that @xmath0 is herbrand constructive and a curry - howard computational interpretation of this very strong herbrand - like theorem .
such statements in first - order logic are typically drawn as consequences of the subformula property , which is in turn a corollary of full cut - elimination when sequent calculus is available .
but as in @xcite , a more primitive argument suffices here .
this is indeed providential , since not only without permutation rules for @xmath512 and @xmath513 we can have no subformula property , but surprisingly even those reductions would not suffice .
the topic of what reductions are needed is very non - trivial and left as subject of future research .
however , in a sense , herbrand constructiveness is already a _ weak _
subformula property and holds for the most interesting case of the existential quantifier , when there is actually some information to gain . for lambda calculus , instead , to enjoy the subformula property is a mere curiosity without much computational sense .
in fact , if we think that in intuitionistic logic or fragments of classical arithmetic @xcite general permutation rules are not needed to compute witnesses , it should not entirely come as a surprise that this is still the case in our framework .
if we omit parentheses , we know that every proof term in head normal form can be written as @xmath514 , where each @xmath515 is not of the form @xmath516 ; if for every @xmath129 , @xmath517 is of the form @xmath518 , then we call the whole term an _ herbrand normal form _ , because it is essentially a list of the witnesses appearing in an herbrand disjunction . formally : [ definition - hnf ] we define by induction a set of proof terms , called * herbrand normal forms * , as follows : * every proof - term @xmath519 is an herbrand normal form ; * if @xmath22 and @xmath23 are herbrand normal forms , @xmath112 is an herbrand normal form .
an herbrand normal form represents , in a straightforward way , a proof of an herbrand disjunction .
[ proposition - herbrandforms ] + suppose that @xmath520 in @xmath0 and @xmath22 is an herbrand normal form @xmath17 then for some @xmath521 @xmath522\lor \dots \lor a[m_{k}/\alpha]\ ] ] we proceed by induction on @xmath523 .
+ if @xmath524 , then @xmath525 and thus @xmath526 $ ] , which is the thesis .
+ if @xmath527 , then @xmath528 , for some @xmath529 and @xmath530 @xmath531 by induction hypothesis , @xmath532\lor \dots \lor a[m_{i-1}/\alpha]\ ] ] @xmath533\lor \dots \lor a[m_{k}/\alpha]\ ] ] hence , by repeated application of the @xmath471 inference , we get , for some @xmath534 , @xmath535\lor \dots \lor a[m_{k}/\alpha]\ ] ] @xmath536\lor \dots \lor a[m_{k}/\alpha]\ ] ] and thus @xmath537\lor \dots \lor a[m_{k}/\alpha]\ ] ] by setting @xmath538 , we obtain the thesis .
our last task is to prove that every closed realizer of any existentially quantified statement @xmath1 include an exhaustive sequence @xmath510 of possible witnesses .
[ theorem - extraction0 ] let @xmath539 be any formula .
suppose @xmath224 , @xmath56 contains neither free proof term variables nor @xmath156 , and @xmath540 .
then @xmath22 is an herbrand normal form @xmath541 and @xmath18\lor \dots \lor a[m_{k}/\alpha]\ ] ] by proposition [ proposition - hnf ] @xmath542 where for each @xmath543 , either @xmath411 , with @xmath180 or @xmath415 , with @xmath544 , or @xmath417 is a value . since @xmath417 does not contain free proof term variables other than @xmath545 , it can not be of the form @xmath411 .
moreover , @xmath546 , hence @xmath417 can not be equal to some @xmath547 , because @xmath547 must have type @xmath548 .
therefore @xmath417 is a value , according to definition [ definition - value ] , and the only possible shape compatible with its type @xmath1 is @xmath549 .
we have thus shown that @xmath22 is an herbrand normal form @xmath17 by proposition [ proposition - herbrandforms ] , for some @xmath521 @xmath550\lor \dots \lor a[m_{k}/\alpha]\ ] ] which is the thesis . as corollary
, we obtain that dummett s logic @xmath0 is herbrand constructive .
[ theorem - extraction2 ] let @xmath539 be any formula .
suppose @xmath551 then there is a proof term @xmath22 such that @xmath540 , @xmath552 and @xmath22 is an herbrand normal form @xmath541 moreover , @xmath18\lor \dots \lor a[m_{k}/\alpha]\ ] ] by the subject reduction theorem [ subjectred ] , @xmath552 . by the adequacy theorem [ adequacy theorem ] , @xmath224 and the thesis follows from theorem [ theorem - extraction0 ] .
we suggest to interpret an herbrand normal form @xmath17 in the following way .
each @xmath549 represents the result of an intuitionistic computation of a witness in a possible universe .
these witnesses have been obtained by communication coming from other intuitionistic computations in other parallel universes .
it is that process of interaction and dialogue between different possible computations that generates the herbrand normal forms .
head reduction , of course , is sequential computation .
yet , the operator @xmath20 has such a strong parallel flavour that parallel reduction strategies inevitably arise as consequence of normalization for head reduction . to see this ,
let us consider a proof term @xmath112 of @xmath0 . by the normalization theorem [ theorem - snnem ]
, the head reduction of @xmath112 reduces subterms inside the left part of the term until it is possible , afterwards it continues to reduce the right part and finally it stops .
if we consider only the first half of the reduction , we get @xmath553 for some @xmath336 in head normal form and _ not _ of the shape @xmath554 $ ] , otherwise a further reduction inside @xmath336 would be possible .
thanks to the _ perfect logical symmetry _ of the term @xmath112 , also @xmath555 is a term of the same type .
again , we can reduce @xmath556 for some @xmath557 in head normal form and _ not _ of the shape @xmath554 $ ] .
the point is that the head reductions ( [ 1 ] ) and ( 2 ) _ can be made in parallel _ and what we get , @xmath558 not only is a term of the same type of @xmath112 and with no more free variables , it also is a head normal form !
at the time of this writing , there is no known cut - free sequent calculus for second - order intuitionistic logic with dummett s axiom , which we call @xmath25 .
even if there were one , the situation would be similar to what happens in the hypersequent calculus for second - order gdel - dummett logic @xcite : there is no known cut - elimination procedure , only a semantical proof that valid statements can be proved without cuts .
this state of thing reminds the status of takeuti s conjecture @xcite , a problem which resisted the effort of the best researchers for many years in the 1950 - 60 s , and was solved constructively in 1971 by girard ( see @xcite ) .
it asked whether the now standard second - order sequent calculus was cut - free .
a cut - elimination procedure for intuitionistic second - order sequent calculus was finally obtained only through translation to natural deduction , where the powerful tait - girard reducibility settles the matter .
this shortcoming of sequent calculus is even worse in the case of hypersequent calculus , which is more complicated and no cut - elimination procedure is known at second - order . in this section ,
we consider second - order natural deduction for @xmath25 and prove the normalization of head reduction . unlike in hypersequent calculus , where second - order cut - elimination requires climbing a steep and cold combinatorial mountain , extending classical realizability to the second - oder case is a like a quiet stroll in a peaceful and sunny countryside road . indeed
, classical realizability was introduced directly in the second - order case by parigot @xcite and krivine @xcite , without even bothering with the first - order case .
we follow once again krivine s successive formulation @xcite .
the language @xmath559 of @xmath25 extends @xmath26 in the standard way , adding second order predicate variables , representing sets of individuals .
[ definition - languagear ] the language @xmath559 of @xmath25 is defined as follows . 1 .
the * terms * of @xmath559 are inductively defined as either variables @xmath27 or constants @xmath28 or expressions of the form @xmath29 , with @xmath30 a function constant of arity @xmath31 and @xmath560 .
there is a set of * predicate constant symbols * and of * predicate variables*. the * atomic formulas * of @xmath559 are all the expressions of the form @xmath33 and @xmath561 such that @xmath34 is a predicate symbol of arity @xmath31 , @xmath562 is a predicate variable and @xmath563 are terms of @xmath559 .
we assume to have a @xmath36-ary predicate symbol @xmath37 which represents falsity .
the * formulas * of @xmath559 are built from atomic formulas of @xmath559 by the logical constants @xmath38 , with quantifiers ranging over first - order variables @xmath39 and second - order variables @xmath564 : if @xmath40 are formulas , then @xmath41 , @xmath42 , @xmath12 , @xmath43 , @xmath44 , @xmath565 are formulas .
the logical negation @xmath45 can be introduced , as usual , as an abbreviation of the formula @xmath46 and the second - order existential quantification is defined as @xmath566 .
4 . as usual ,
if @xmath3 and @xmath10 are formulas of @xmath559 and @xmath562 is a predicate variable , we denote with @xmath567 $ ] the formula obtained from @xmath3 by replacing all its atomic subformulas of the form @xmath561 with @xmath568 $ ] ( without capturing free variables of @xmath10 ) .
the natural deduction for @xmath25 and @xmath569 extends respectively the natural deduction for @xmath0 and the one for @xmath155 with the following inference and reduction rules ( see girard @xcite ) : + second - order universal quantification : : + + where in the left rule @xmath562 does not occur free in the types of the free variables of @xmath3 .
reduction rule for universal quantification : : @xmath570 $ ] + the definition [ definition - stack ] of stack is of extended to @xmath569 allowing expressions @xmath571 , with @xmath10 formula , to appear in the stack , and the definition [ definition - head ] of head redex is extended to the terms of @xmath569 by saying that @xmath572 is the * head redex * of @xmath573 for every stack @xmath78 .
the reduction relation @xmath169 for the terms of @xmath569 is then defined as in definition [ definition - headred ] . in the following ,
we define @xmath170 to be the set of normalizing proof terms of @xmath569 . in order to define second - order realizability
we need the concept of _ realizability opponent _ , which is nothing but a function mapping terms of @xmath559 to arbitrary sets of stacks adapted to some fixed type .
the idea is that an arbitrary realizability opponent represents the sets of tests that an arbitrary definition of realizability requires to pass in order to declare a term to be a realizer .
[ definiton - redo ] 1 .
a stack @xmath78 of @xmath569 is said to be * adapted to a type * @xmath11 , if for all terms @xmath56 of type @xmath11 , @xmath574 is still a term of @xmath569 .
2 . a * realizability opponent * of type
@xmath575 is any function that maps each term @xmath213 of @xmath559 to a set of stacks adapted to @xmath576 $ ] .
we assume that for each realizability opponent @xmath577 of type @xmath11 there is in @xmath559 an * opponent predicate constant * @xmath578 of type @xmath575 associated to it .
realizability for @xmath569 extends realizability for @xmath155 to second - order quantification .
the idea is the usual : we would like to define @xmath579 as : for all formulas @xmath10 , @xmath580 $ ] , but we can not .
so we define @xmath579 as @xmath231 for all possible definitions of realizability which @xmath562 can be assigned to , that is , for all reducibility opponents that replace @xmath562 .
[ definition - realizability ] assume @xmath56 is a term of @xmath569 and @xmath11 is a formula of @xmath559 .
we define by mutual induction the relation @xmath198 ( `` @xmath56 is reducible of type @xmath11 '' ) and a set @xmath199 of stacks of @xmath569 according to the form of @xmath11 : * @xmath198 if and only if @xmath200 and for all @xmath201 , @xmath202 * @xmath203 if @xmath581 is atomic * @xmath582 for each realizability opponent @xmath583 * @xmath204 * @xmath205 * @xmath206 { { { \ , { \mathpalette { { { \settoheight\cdotaxis{$}\vcenter{}$ } \sbox0 { \raisebox\cdotaxis { \scalebox{2.4 } { \raisebox{-1.1pt } { $ \mathsurround=0pt } .$ } } } } \dp0=0pt \sbox2{$}\bullet$ } \ifdim\ht2<\ht0 \ht0=\ht2 \fi \sbox2{$\mathsurround=0pt } .$ } \hbox to \wd2{\hss\usebox{0}\hss } } } } \,}}}\sigma \ |\ \forall t.\ ( t { \vdash}a\implies u[t / x]\,\sigma \in { \mathsf{hn}^{\star } } ) \land ( t{\vdash}b\implies
v[t / y]\ , \sigma \in { \mathsf{hn}^{\star}})\}\cup \{\epsilon\}$ ] * @xmath207||\}\cup \{\epsilon\}$ ] * @xmath208 { { { \ , { \mathpalette { { { \settoheight\cdotaxis{$}\vcenter{}$ } \sbox0 { \raisebox\cdotaxis { \scalebox{2.4 } { \raisebox{-1.1pt } { $ \mathsurround=0pt } .$ } } } } \dp0=0pt \sbox2{$}\bullet$ } \ifdim\ht2<\ht0 \ht0=\ht2 \fi \sbox2{$\mathsurround=0pt } .$ } \hbox to \wd2{\hss\usebox{0}\hss } } } } \,}}}\sigma \
|\ \forall t.\ t { \vdash}a[m/\alpha]\implies v[m/\alpha][t / x]\ , \sigma \in { \mathsf{hn}^{\star}}\}\cup \{\epsilon\}$ ] * @xmath584|| $ for some realizability opponent $ \mathcal{b}$ of type $ \lambda \alpha b$}\}\cup \{\epsilon\}$ ] the next proposition says that in the definition of @xmath585||$ ] , we can replace @xmath586 with the realizability opponent corresponding to it , transforming in this way an intensionally defined set into an extensionally defined object .
[ proposition - comprehension ] let @xmath10 a formula of @xmath559 .
suppose @xmath583 is a realizability opponent such that @xmath587||\ ] ] then for every formula @xmath3 of @xmath559 @xmath588||=||a[\lambda \alpha b / x]||\ ] ] standard , by induction on @xmath3 ( see krivine @xcite ) . 1 .
@xmath589 , where @xmath34 is a predicate constant symbol .
then , @xmath590= \mathcal{p}(m_{1 } , \ldots , m_{n})=a[\lambda \alpha b / x]$ ] and the thesis is trivial . 2 .
@xmath591 , where @xmath592 is a predicate variable .
then , if @xmath593 , the thesis is trivial , since we have @xmath594=y(m)= a[\lambda \alpha b / x]\ ] ] so let us suppose @xmath595 .
then @xmath596||=||{\smash{\overset{\centerdot}{\mathcal{{b}}}}}(m)||=\mathcal{b}(m)=||b[m/\alpha]||=||a[\lambda \alpha b / x]||\ ] ] 3 .
the other cases are straightforward .
we extended proposition [ proposition - redelim ] by showing that realizability is also sound with respect to second - order quantification elimination .
[ proposition - redelim2 ] if @xmath579 , then for every formula @xmath10 of @xmath559 , @xmath597 $ ] .
assume @xmath579 .
let @xmath598||$ ] ; we must show @xmath599 .
let us consider a realizability opponent @xmath583 such that @xmath600||\ ] ] by proposition [ proposition - comprehension ] , @xmath601||$ ] . by definition [ definition - realizability ] , @xmath602 , and since @xmath603
, we conclude @xmath599 .
we extended proposition [ proposition - somecases ] by showing that realizability is also sound with respect to second - order quantification introduction .
[ proposition - somecases2 ] if for every realizability opponent @xmath583 of type @xmath586 , @xmath604{\vdash}a[{\smash{\overset{\centerdot}{\mathcal{b}}}}/x]$ ] , then @xmath605 .
suppose that for every formula @xmath10 of @xmath559 , @xmath604{\vdash}a[{\smash{\overset{\centerdot}{\mathcal{b}}}}/x]$ ] .
let @xmath606 .
we have to show @xmath607 . if @xmath253 , indeed @xmath608 .
suppose then @xmath609 , with @xmath610||$ ] and @xmath583 realizability opponent of type @xmath586 .
since by hypothesis @xmath604{\vdash}a[{\smash{\overset{\centerdot}{\mathcal{b}}}}/x]$ ] , we have @xmath604\,\rho\in { \mathsf{hn}^{\star}}$ ] ; moreover , @xmath611\ , \rho\ ] ] therefore , @xmath612 .
the adequacy theorem is readily extended to second - order realizability .
[ adequacy theorem2 ] suppose that @xmath424 in the system @xmath25 , with @xmath278 having free variables among @xmath425 .
let @xmath426 and @xmath613 be respectively terms of @xmath559 and realizability opponents of type @xmath614 . for every formula @xmath11 , set @xmath615 $ ] .
if there are terms @xmath427 such that @xmath616 then @xmath617[t_1/x_1^ { { { \overline{a } } } _ { 1}}\cdots t_n / x_n^ { { { \overline{a } } } _ { n}}]{\vdash } { { \overline{a } } } \ ] ] for any term @xmath23 , we define @xmath618[t_1/x_1^ { { { \overline{a } } } _ { 1}}\cdots t_n / x_n^ { { { \overline{a } } } _ { n}}]\ ] ] we proceed by induction on @xmath278 . consider the last rule @xmath432 in the derivation of @xmath424 : we just have to deal with the second - order cases , the other ones have been settled in the proof of theorem [ adequacy theorem ] . 1 .
if @xmath432 is the second - order @xmath473 rule , then @xmath619 , @xmath620 $ ] and @xmath621 .
so , @xmath622 . by inductive hypothesis @xmath623 and so @xmath624 $ ] by proposition [ proposition - redelim2 ] .
if @xmath432 is the second - order @xmath480 rule , then @xmath625 , @xmath626 and @xmath483 ( with @xmath562 not occurring free in the types @xmath57 of the free variables of @xmath22 ) .
so , @xmath627 , since we may assume @xmath628 . by proposition [ proposition - somecases2 ]
, it is enough to prove that @xmath629{\vdash } { { \overline{b } } } [ { \smash{\overset{\centerdot}{\mathcal{b}}}}/x]$ ] for every realizability opponent @xmath583 of type @xmath586 , which amounts to showing that the induction hypothesis can be applied to @xmath22 .
for this purpose , we observe that , since @xmath628 , for @xmath487 we have @xmath630\eqno{\qed}\ ] ] as consequence of the adequacy theorem [ adequacy theorem2 ] , we obtain that every typed term of @xmath25 is normalizable by head reduction .
[ theorem - sn2order ] suppose @xmath502 in @xmath25
. then @xmath631 .
we can finally prove that second - order dummett s logic @xmath25 is herbrand constructive .
[ theorem - extraction ] let @xmath539 be any formula .
suppose @xmath632 then there is a proof term @xmath22 such that @xmath540 , @xmath633 and @xmath22 is an herbrand normal form @xmath541 moreover , @xmath634\lor \dots \lor a[m_{k}/\alpha]\ ] ] as the proof of theorem [ theorem - extraction0 ] .
i would like to thank agata ciabattoni : this work arose , and greatly benefited , from conversations with her .
i would also like to thank francesco genco for interesting exchanges about the topic .
f. aschieri , _ strong normalization for ha + em1 by non - deterministic choice _ , proceedings of first workshop on control operators and their semantics 2013 ( cos 2013 ) , electronic proceedings in theoretical computer science , vol .
114 , 2013 f. aschieri , s. berardi , g. birolo , _ realizability and strong normalization for a curry - howard interpretation of ha + em1 _ , csl 2013 , leibniz international proceeding in computer science , vol .
23 , pp . 4560 , 2013 . f. aschieri , m. zorzi , _ non - determinism , non - termination and the strong normalization of system t _ , proceedings of tlca 2013 , lncs , vol
. 7941 , pp . 3147 , 2013 .
f. aschieri , m. zorzi , _ a `` game semantical '' intuitionistic realizability validating markov s principle _ , post - proceedings of types 2013 , leibniz international proceedings in informatics , vol .
26 , pp . 2444 , 2014 .
f. aschieri , m. zorzi , _ on natural deduction in classical first - order logic : curry - howard correspondence , strong normalization and herbrand s theorem _ , theoretical computer science , vol .
625 , pp . 125146 , 2016 .
m. baaz , a. ciabattoni , c. fermller , _ a natural deduction system for intuitionistic fuzzy logic _ , in : lectures on soft computing and fuzzy logic , pp . 118 , a. di nola , g. gerla eds .
, physica - verlag , 2000 .
m. parigot , _ proofs of strong normalization for second - order classical natural deduction _ ,
journal of symbolic logic , vol .
62 , n. 4 , pp . 14611479 , 1997 .
j. von plato , _
gentzen s proof of normalization for natural deduction _
, bulletin of symbolic logic , vol .
14 , n. 2 , pp . 204257 , 2008 .
d. prawitz , _ ideas and results in proof theory _ , proceedings of the second scandinavian logic symposium , pp .
. m. h. sorensen , p. urzyczyn , _ lectures on the curry - howard isomorphism _ , studies in logic and the foundations of mathematics , vol .
149 , elsevier , 2006 .
w. tait , _ normal form theorem for bar recursive functions of finite type _ , proceedings of the second scandinavian logic symposium , pp . 353267 , 1971 .
g. takeuti , _ on a generalized logical calculus _ , japanese journal of mathematics , vol .
23 , 3996 , 1953 . | dummett s logic @xmath0 is intuitionistic logic extended with dummett s axiom : for every two statements the first implies the second or the second implies the first .
we present a natural deduction and a curry - howard correspondence for first - order and second - order dummett s logic .
we add to the lambda calculus an operator which represents , from the viewpoint of programming , a mechanism for representing parallel computations and communication between them , and from the viewpoint of logic , dummett s axiom .
we prove that our typed calculus is normalizing and show that proof terms for existentially quantified formulas reduce to a list of individual terms forming an herbrand disjunction . |
high cadence observations of astronomical sources is a growing field within astrophysics , and there is a clear need for such data for the sun .
many research topics , in particular those related to the dynamic sun and the heating of its outer regions , involve the observation and modelling of wave phenomena and explosive events captured over very short timescales .
high cadence observations are also important for post - facto image reconstruction ( pfir ) techniques which require the processing of an extensive collection of images for the production of a single frame at diffraction - limited resolution . these short exposure images must be accumulated over timescales sufficiently small so that atmospheric turbulence is effectively `` frozen out '' , and the solar features remain unchanged .
since the discovery of solar oscillations in the 1960s @xcite , and their subsequent confirmation by , there has been a multitude of observational evidence presented verifying the widespread existence of oscillations in the solar atmosphere ( see the recent review by ) .
oscillations throughout the solar atmosphere have enabled scientists to probe the underlying physics that are involved with energy transfer and coronal heating , and have been suggested as candidates to explain one of the main unanswered questions in solar physics
_ why is the outer solar atmosphere hotter than its surface ?
_ however , in recent years a need to probe high - frequency oscillations has arisen .
have shown that the energy contribution of high - frequency waves to atmospheric heating may be significant .
furthermore , have indicated that quasi - periodic oscillations shorter than 100 s may be a vital mechanism in the creation of the multi - million degree solar corona .
thus , a sensitive camera system with a suitably high cadence to satisfy the nyquist parameter is paramount for the study of high - frequency oscillations .
such a system is not only capable of studying wave motion , but may also be utilized to examine highly dynamic phenomena in the lower solar atmosphere .
recently , fast - moving plasma with velocities exceeding 200 kms@xmath1 has been observed by , indicating the abundance of highly dynamic structures in the chromosphere .
additional investigations may include the search for hard x - ray non - thermal electron precipitation sites in the event of flare activity .
report short duration spikes associated with hard x - ray emission from solar flares , and in conjunction with the impact of fast - moving non - thermal electrons , leads to enhanced emission over inherently short time scales ( @xmath2 s ; ) .
this , coupled with the detection of fast fluctuations of h@xmath3 emission from a flare kernel on timescales of @xmath4 s by , demonstrates the need for high - cadence solar imaging techniques to capture these dynamic processes on their intrinsic scales .
the search for rapid , often low - amplitude intensity variations requires a highly sensitive camera system providing accurate , sustained frame rates , accompanied by good local seeing at the telescope facility .
here we report the construction of a new high - speed camera system , developed by queen s university belfast ( qub ) , and its commissioning as a common - user instrument at the dunn solar telescope ( dst ) facility run by the national solar observatory in the sacramento peak mountains , new mexico , usa .
this system , named _
rapid oscillations in the solar atmosphere _ ( rosa ) , will present users with the ability to observe the sun simultaneously in up to six wavelengths or wavebands . a detailed guide
to rosa s hardware is described in section [ instrumentation ] , while in section [ observing ] a description of available observing modes is presented . in section [ interface ] rosa s graphical user interface ( gui ) is explained , followed by guidelines for obtaining calibrated data in section [ calibration ] . finally , concluding remarks , in addition to sample images from the instrument commissioning , are given in section [ conclusions ] .
in the late 1990 s , a high - speed , two - camera solar imaging system named the _
solar eclipse corona imaging system _ ( secis ; ) began development .
this instrument was designed to rapidly capture images of the solar corona during a total eclipse and produced some very interesting results ( see _ e.g. _ , @xcite ; ) . however , in order to image the lower solar atmosphere at high time resolution , a new camera system needed to be developed .
the _ rapid dual imager _ ( rdi ; ) was designed as a followup to the highly successful secis camera system and consists of two 502 @xmath0 494 pixel@xmath5 ccds , operating at speeds of up to 20 frames per second .
even with a relatively small number of pixels and onboard storage , rdi was able to show the need for a dedicated high - cadence multi - camera system ( see _ e.g. _ ) and acted as a pre - cursor to the capabilities possessed by rosa .
the rosa system consists of six individual frame transfer ccds , each with their own dedicated server for data acquisition and storage .
all cameras are triggered via a precision control unit or `` sync '' box .
the ccd cameras are from andor technologies of belfast , with the du-885-vp model chosen comprising of a @xmath6 pixel@xmath7 area .
they are cooled up to 100 @xmath8c below ambient temperature through use of a thermoelectric peltier cooler .
maintaining low readout noise necessary for fast readout rates , each rosa camera can read out over 30 frames per second in full - chip mode and over 200 frames per second when the ccd is windowed .
the ccd characteristics specific to the rosa system are presented in table [ table1 ] and further described in section [ emccd ] .
.detailed description of rosa camera characteristics . [ cols="<,^",options="header " , ] + @xmath9 : + @xmath10 : + @xmath11 : + @xmath12 : + @xmath13 : + @xmath14 : + @xmath15 : +
rosa was successfully commissioned on the 76 cm dst during august 2008 ( fig [ rosa_bench ] ) .
it utilizes the dst s large optical benches and high - order adaptive optics to achieve multi - wavelength observations of the lower solar atmosphere . due to a wide range of optical components available
, a large number of filter combinations can be achieved .
however , details of several key optical elements are outlined here . a zeiss universal birefringent filter ( ubf ; ) may be implemented for narrowband ( @xmath160.2 ) imaging . however , due to a significant decrease in the transmission profile of the ubf at shorter wavelengths ( @xmath17% at 4000 compared to @xmath18% at 7000 ; ) , it is more desirable to allow the ubf to obtain filtergrams in the `` red '' portion of the optical spectrum .
typically , the ubf is most commonly used for chromospheric h@xmath3 core imaging at 6562.8 . however , in addition to optical imaging , the ubf may be used in conjunction with a wollaston prism to capture photospheric magnetic field information at high spatial and temporal resolution . by tuning the ubf into the wing of the magnetically sensitive fe ( 6302.5 ) absorption line , two rosa cameras may be used to capture simultaneous stokes v observations of left- and right - hand circularly polarized light , allowing the magnitude of the line - of - sight magnetic field component to be studied . if it is desirable to obtain magnetic - field information in the photosphere , simultaneous narrowband h@xmath3 core imaging can still be obtained using an additional tunable zeiss filter . for imaging in the `` blue '' ,
independent halle / oriel filters may be used which have a more efficient transmission curve for this portion of the electromagnetic spectrum .
examples include wideband continuum and g - band filters for photospheric imaging , as well as a calcium - k core filter for observations of the upper - photosphere / lower - chromosphere .
typically , observers will choose to reduce the overall field - of - view size in exchange for higher spatial resolution . in order to do this
, acquired images should have a sampling of @xmath160.07@xmath19 per pixel to match the telescope s diffraction - limited resolution in the `` blue '' portion ( _ e.g. _ @xmath20 ) of the optical spectrum to that of the ccd .
doing so places @xmath21 km of the solar surface on each pixel , providing @xmath22 km spatial resolution and an overall field - of - view size of @xmath23 km@xmath7 . maintaining the same spatial sampling across all rosa cameras will insure the resulting data has the same field - of - view .
however , users may choose to match the telescope s diffraction - limited resolution in each of their desired bandpasses to that of the ccd , thus providing optimal spatial resolution on all rosa cameras .
details of the most common filters may be found in table [ table2 ] .
exposure times listed are typical of quiet - sun , disk - centre observations filling approximately one - third of the well depth per exposure . utilizing count rates obtained during the rosa commissioning run ( incorporating the default plate - scale sampling of 0.07@xmath19 per pixel ) and the camera s quantum efficiency curve ( upper panel of figure [ qe_noise ] ) ,
photon count - rate statistics can be derived , as listed in the fifth column of table [ table2 ] .
since rosa utilizes a host of independent computers and storage devices , it is not necessary to run all six cameras during a single observing sequence .
indeed , any combination of cameras may be used depending on the observers scientific requirements .
a user can individually adjust camera operating temperatures , frame rates and exposure times through use of a simple graphical user interface ( gui ; fig [ gui ] ) .
the gui utilizes a java - script interface to create a stable platform incorporating drop - down menus in an easy - to - follow environment . through the gui
, users can also select which cameras to run , and modify their triggering sequence to maximize individual camera frame rates and insure the synchronization of specific cameras .
the gui also allows the user to select which type of observations will be acquired by selecting the desired option .
these options include solar data acquisition , dark images with a closed shutter , bias frames using a zero - second exposure time and flat field images .
each choice will result in the saved filename being modified to include the desired option , thus making post - acquisition file searching much easier .
once an observing sequence has commenced , the gui displays sample images coming from each of the cameras , refreshed approximately every second .
a detailed description of the gui and a guide to its adjustable parameters may be found in the rosa user manual , available at .
due to the high sensitivity of the rosa system , it is imperative to accompany data imaging sequences with a range of additional observations . it is these supplementary observations which allow accurately processed images to be produced via the rosa - specific data reduction pipeline . to compensate for dark current and read - out noise , sufficient dark images ( @xmath24 ) with exposure times equal to the data acquisition must be obtained and subtracted from the science images . many flat - field images ( @xmath25 ) should also be taken through an un - flat telescope mirror coupled to a random guider to allow an accurate gain table to be created for each rosa camera .
these key calibration procedures will enable science data to be corrected for camera inconsistencies as well as for variable light levels across the incident beam . at this stage
, pfir techniques such as speckle @xcite or multi - object multi - frame blind deconvolution @xcite may be implemented .
indeed , have shown how modern pfir techniques can produce reconstructed images which are photometrically accurate , allowing reliable studies of solar structures to be undertaken .
however , image reconstruction processes are extremely cpu intensive and this must be taken into consideration before attempting to reduce the data .
for example , during the commissioning run a single rosa camera obtained over 105,000 raw g - band filtergrams in only one hour of observing . with excellent seeing conditions ,
an image reconstruction of @xmath26 was deemed suitable , providing a reconstructed cadence of 0.5 s. thus , over 6500 separate reconstructions were required , with each individual process taking approximately 30 cpu minutes to complete on a modern intel xeon processor . if access to only one computer is possible ( _ e.g. _ preparing the data solely on the user s desktop pc ) , then reconstructing all 6500 images would take in excess of 135 days , with additional time being required to process the data from the remaining five rosa cameras . as telescope schedules often grant individuals in excess of seven days observing time , and each day commonly provides @xmath27 hours of good seeing conditions , it is imperative to consider how and where the acquired observations will be reduced so excessive time delays are not experienced . in order to co - align each rosa camera , additional calibration images should be acquired .
images of an air - force target will allow compensation for general rotation and image mirroring , while acquiring a burst of grid images will enable small - scale inter - camera image rotation to be evaluated and removed .
with these camera positioning artifacts removed , image destretching can be accomplished through use of the rosa data reduction pipeline . under normal circumstances , a @xmath28 grid , equating to a @xmath29 separation between spatial samples ( for diffraction - limited resolution at 4170 ) , is used to evaluate local offsets between successive images , allowing compensation for spatial distortions caused by atmospheric turbulence and/or air bubbles crossing the entrance aperture of the telescope .
the fine destretching grid implemented in this process allows compensation for small - scale seeing conditions of @xmath30 to @xmath31 in size . to help facilitate and expedite data preparation ,
qub have developed a parallel processing cluster dedicated to rosa data reduction .
this cluster consists of 25 intel xeon quad - core cpus providing 100 processing nodes on a dedicated private network .
users of the rosa instrument may opt for all data reduction to be carried out on the qub cluster , or may avail of the freely distributed rosa pipeline for use on their own reduction cluster . the accurate reconstruction , co - alignment and destretching of multiple atmospheric heights will promote the use of multi - wavelength studies , whereby the cause of dynamic phenomena can be probed and analysed , with the subsequent effect on other layers of the solar atmosphere evaluated .
in recent years , the solar community has expressed a strong desire for a high - cadence , synchronized , multi - camera system to be introduced which would allow unprecedented spatial _ and _ temporal resolutions of solar structures over a wide range of atmospheric heights . with the successful commissioning of the rosa instrument
, it is now possible to observe solar structures at high cadences in up to six independent wavelengths simultaneously .
figure [ rosa_science ] shows the capabilities of observing in at least four synchronized wavelengths ( g - band filtergrams , magnetograms from circularly - polarized light at 6302.3 , ca - k core imaging at 3933.7 and narrowband h@xmath3 core filtergrams ) .
these types of high - resolution images present an excellent opportunity to scientists , whereby magnetic structures can be investigated at multiple atmospheric heights with temporal resolutions exceeding ten frames per second . due to the rack - mounted nature of the rosa instrument , future upgrades to the system
can be readily carried out .
as technology improves and the size of components decrease , rosa hardware can be updated continually to keep it at the forefront of high - cadence imaging .
such upgrades will include the increased capacity of onboard storage , an improved connectivity for external media ( _ e.g. _ e - sata and firewire connections ) and a turbulence - reducing method of camera cooling ( _ e.g. _ via water cooling ) . with the presence of backside illuminated ccds now appearing in the marketplace , existing rosa cameras
may be replaced with such devices . in instances where low exposure times are required ,
this new breed of camera may lead to performance boosts in photon - starved regimes by offering a higher quantum efficiency when compared to traditional front - illuminated devices .
the ability to adapt rosa with continually changing demands and computer architecture has identified rosa as a bridging instrument for the advanced technology solar telescope and places a high - cadence solar imager in line to become a first - light instrument on this revolutionary facility in 2015 .
this work was supported by the u.k .
science and technology facilities council and dbj wishes to thank the science and technology facilities council for the award of a post - doctoral fellowship .
fpk is grateful to awe aldermaston for the award of a william penney fellowship .
pjc thanks the northern ireland department for employment and learning for a phd studentship .
observations were obtained at the national solar observatory , operated by the association of universities for research in astronomy , inc .
( aura ) , under cooperative agreement with the national science foundation .
finally we would like to thank the technical staff at the dst for perseverance in the face of seemingly endless camera and filter arrangements .
alamanni , n. , bertello , l. , righini , a. , cavallini , f. , ceppatelli , g. : 1990 , _ astron .
_ , * 231 * , 518 balasubramaniam , k. s. : 2002 , _ astrophys . j. _ , * 575 * , 553 banerjee , d. , erdlyi , r. , oliver , r. , oshea , e. : 2007 , _ solar phys . _ , * 246 * , 3 beckers , j. m. , dickson , l. , joyce , r. s. : 1975 , _ a fully tunable lyot - hman filter ( afcrl - tr-75 - 0090 ; bedford : afcrl ) _
beebe , h. a. , johnson , h. r. : 1969 , _ solar phys .
_ , * 10 * , 79 bonaccini , d. , righini , a. , cavallini , f. , ceppatelli , g. : 1989 , _ astron
. astrophys .
_ , * 217 * , 368 brown , j. c. : 1971 , _ solar phys . _ , * 18 * , 489 coates , c. g. , denvir , d. j. , mchale , n. g. , thornbury , k. d. , hollywood , m. a. : 2004 , _ journal of biomedical optics _ , * 9 * , 1244 denvir , d. j. , conroy , e. : 2003 , _ proc .
spie _ , * 4877 * , 55 deubner , f. -l . : 1975 , _ astron .
_ , * 44*,371 hanisch , r. j. , farris , a. , greisen , e. w. , pence , w. d. , schlesinger , b. m. , teuben , p. j. , thompson , r. w. , warnock , a. : 2001 , _ astron . astrophys .
_ , * 376 * , 359 hasan , s. s. , van ballegooijen , a. a. : 2008 , _ astrophys . j. _ , * 680 * , 1542 jess , d. b. , andi , a. , mathioudakis , m. , bloomfield , d. s. , keenan , f. p. : 2007a , _ astron .
_ , * 473 * , 943 jess , d. b. , mcateer , r. t. j. , mathioudakis , m. , keenan , f. p. , andic , a. , bloomfield , d. s. : 2007b , _ astron .
_ , * 476 * , 971 katsiyannis , a. c. , williams , d. r. , mcateer , r. t. j. , gallagher , p. t. , keenan , f. p. , murtagh , f. : 2003 , _ astron .
_ , * 406 * , 709 kiplinger , a. l. , dennis , b. r. , frost , k. j. , orwig , l. e. , emslie , a. g. : 1983 , _ astrophys
. j. lett .
_ , * 265 * , l99 leighton , r. b. , 1960 , _ iaus _ , * 12 * , 321 phillips , k. j. h. , read , p. d. , gallagher , p. t. , keenan , f. p. , rudawy , p. , rompolt , b. , berlicki , a. , buczylko , a. , diego , f. , barnsley , r. , smartt , r. n. , pasachoff , j. m. , babcock , b. a. : 2000 , _ solar phys .
_ , * 193 * , 259 porter , l. j. , klimchuk , j. a. , sturrock , p. a. : 1994 , _ astrophys . j. _ , * 435 * , 502 rimmele , t. r. : 2004 , _ astrophys . j. _ , * 604 * , 906 uitenbroek , h. , tritschler , a. : 2006 , _ astrophys . j. _ , * 639 * , 525 van noort , m. , rouppe van der voort , l. , lfdahl , m. g. : 2005 , _ solar phys . _ ,
* 228 * , 191 van noort , m. j. , rouppe van der voort , l. h. m. : 2006 , _ astrophys
. j. lett .
_ , * 648 * , l67 vernazza , j. e. , avrett , e. h. , loeser , r. : 1981 , _ astrophys . j. supp . _ , * 45 * , 635 wang , h. , qiu , j. , denker , c. , spirock , t. , chen , h. , goode , p. r. : 2000 , _ astrophys . j. _ , * 542 * , 1080 weigelt , g. , wirnitzer , b. : 1983 , _
optics letters vol . 8
_ , * 7 * , 389 williams , d. r. , phillips , k. j. h. , rudawy , p. , mathioudakis , m. , gallagher , p. t. , oshea , e. , keenan , f. p. , read , p. , rompolt , b. : 2001 , _ mon . not .
_ , * 326 * , 428 williams , d. r. , mathioudakis , m. , gallagher , p. t. , phillips , k. j. h. , mcateer , r. t. j. , keenan , f. p. , rudawy , p. , katsiyannis , a. c. : 2002 , _ mon . not .
_ , * 336 * , 747 wger , f. , von der lhe , o. , reardon , k. : 2008 , _ astron .
_ , * 488 * , 375 | rapid oscillations in the solar atmosphere ( rosa ) is a synchronized , six camera high cadence solar imaging instrument developed by queen s university belfast .
the system is available on the dunn solar telescope at the national solar observatory in sunspot , new mexico , usa as a common - user instrument .
consisting of six 1k @xmath0 1k peltier - cooled frame - transfer ccd cameras with very low noise ( 0.02 15 e s@xmath1 pixel@xmath1 ) , each rosa camera is capable of full - chip readout speeds in excess of 30 hz , or 200 hz when the ccd is windowed . combining multiple cameras and fast readout rates
, rosa will accumulate approximately 12 tb of data per 8 hours observing .
following successful commissioning during august 2008 , rosa will allow multi - wavelength studies of the solar atmosphere at high temporal resolution . |
in the variational description of surfaces the following functionals are of primary importance : * the area @xmath0 , where @xmath1 is the area element , is preserved by isometries . * the total gaussian curvature @xmath2 , where @xmath3 is the gaussian curvature , is topological invariant . * the total mean curvature @xmath4 , where @xmath5 is the mean curvature , depends on external geometry of the surface . *
the willmore energy @xmath6 is invariant with respect to mbius transformations .
geometric discretizations of the first three functionals for simplicial surfaces are well known . for the area functional it is obvious .
the local gaussian curvature at a vertex @xmath7 is defined as the angle defect @xmath8 where @xmath9 are the angles of all triangles ( see fig .
[ f.alpha-beta ] ) at vertex @xmath7 .
the total gaussian curvature is a sum over all vertices @xmath10 .
the local mean curvature at an edge @xmath11 is defined as @xmath12 where @xmath13 is the length of the edge and @xmath14 is the angle between the normals to the adjacent faces at @xmath11 ( see fig . [ f.bending ] ) .
the total mean curvature is the sum over all edges @xmath15 .
these discrete functionals possess the geometric symmetries of the smooth functionals mentioned above . until recently a geometric discretization of the willmore functional was missing . in this paper
we introduce a mbius invariant energy for simplicial surfaces and show that it should be treated as a discrete willmore energy .
let @xmath16 be a simplicial surface in 3-dimensional euclidean space with the set of vertices @xmath17 , edges @xmath18 and ( triangular ) faces @xmath19 .
we define a conformal energy for simplicial surfaces using circumcircles of their faces . each ( internal ) edge @xmath20 is incident to two triangles .
consistent orientation of the triangles naturally induces an orientation of the corresponding circumcircles .
let @xmath21 be the external intersection angle of the circumcircles of the triangles sharing @xmath11 , which is the angle between the tangent vectors of the oriented circumcircles .
now we are ready to define a new energy functional for simplicial surfaces .
local conformal ( discrete willmore ) energy at a vertex @xmath7 is given by the sum over all incident edges @xmath22 the conformal ( discrete willmore ) energy of the simplicial surface is the sum over all vertices @xmath23 where @xmath24 is the number of vertices of @xmath16 .
fig.[f.denergy ] presents two neighboring circles with their external intersection angle @xmath25 as well as a view `` from the top '' at a vertex @xmath7 showing all @xmath26 circumcircles passing through @xmath7 with the corresponding intersection angles @xmath27 . for simplicity we will consider only simplicial surfaces without boundary .
note that the energy we defined is obviously invariant with respect to mbius transformations .
this invariance is an important property of the classical willmore energy defined for smooth surfaces ( see below ) .
note also that the changing of the orientation of both circles preserves the angle @xmath21 .
this shows that the energy @xmath28 is well defined for non - oriented simplicial surfaces as well .
the star @xmath29 of the vertex @xmath7 is the subcomplex of @xmath16 comprised by the triangles incident with @xmath7 .
the vertices of @xmath29 are @xmath7 and all its neighbors .
we call @xmath29 convex if for any its face @xmath30 the star @xmath29 lies to one side of the plane of @xmath19 and strictly convex if the intersection of @xmath29 with the plane of @xmath31 is @xmath31 itself .
[ p.non-negative ] the conformal energy is non - negative @xmath32 and vanishes if and only if the star @xmath29 is convex and all its vertices lie on a common sphere . the proof of this proposition is based on the following elementary lemma .
[ l.alpha-beta ] let @xmath33 be a ( not necessarily planar ) @xmath26-gon with external angles @xmath25 .
choose a point @xmath34 and connect it to all vertices of @xmath33 .
let @xmath9 be the angles ( see fig.[f.alpha - beta ] ) of the triangles at the tip @xmath34 of the obtained pyramid .
then @xmath35 and the equality holds if and only if @xmath33 is planar and convex of a convex polygon may vanish .
the corresponding cite - triangles of the pyramid lie in one plane . ] and the vertex @xmath34 lies inside @xmath33 . _
proof_. let us denote by @xmath36 and @xmath37 the angles of the cite - triangles at the vertices of @xmath33 ( see fig.[f.alpha - beta ] ) .
the claim of lemma [ l.alpha-beta ] follows from adding over all @xmath38 two obvious ( in)equalities @xmath39 all inequalities become equalities only in the case when @xmath33 is planar , convex and contains @xmath34 .
note that some of the external angles @xmath25 may vanish . as a corollary we obtain a polygonal version of fenchel s theorem @xcite .
[ c.non-negative ] @xmath40 _ proof .
_ for a given @xmath33 choose the point @xmath34 varying on a straight line encircled by @xmath33
. there always exist points @xmath34 so that the star at @xmath34 is not strictly convex , and thus @xmath41 .
_ proof of proposition [ p.non - negative]_. the claim of proposition [ p.non-negative ] is invariant with respect to mbius transformations . applying a mbius transformation @xmath42 which maps the vertex @xmath7 to infinity , @xmath43
, one observes that all the circles passing through @xmath7 become straight lines and we arrive at the geometry shown in fig.[f.alpha - beta ] with @xmath44 .
now the claim follows immediately from corollary [ c.non-negative ] .
let @xmath16 be a simplicial surface .
then @xmath45 and the equality holds if and only if @xmath16 is a ( part of a ) convex polyhedron inscribed in a sphere .
_ only the second statement needs to be proven . due to proposition [ p.non-negative ]
the equality @xmath46 implies that all vertices and edges of @xmath16 are convex ( but not necessarily strictly convex ) . deleting the edges which separate triangles lying in one plane one obtains a polyhedral surface @xmath47 with circular faces and all strictly convex vertices and edges .
proposition [ p.non-negative ] implies that for every vertex @xmath7 there exists a sphere @xmath48 with all vertices of the star @xmath29 lying on it . for any edge @xmath49 of @xmath47 two neighboring spheres @xmath50 and @xmath51 share two different circles of their common faces .
this implies @xmath52 and finally the coincidence of all the spheres @xmath48 .
discrete conformal energy @xmath53 defined above is a discrete analogue of the willmore energy for smooth surfaces @xcite @xmath54 here @xmath1 is the area element , @xmath55 the principal curvatures , @xmath56 the mean curvature , @xmath57 the gaussian curvature of the surface . here
we prefer a definition for @xmath58 with a mbius invariant integrand .
it differs from the one in the introduction by a topological invariant .
let us mention two important properties of the willmore energy : * @xmath59 and @xmath60 if and only if @xmath61 is the round sphere .
* @xmath62 ( and the integrand @xmath63 ) is mbius invariant @xcite whereas the first claim almost immediately follows from the definition , the second one is a non - trivial property .
we have shown that the same properties hold for the discrete energy @xmath53 ; in the discrete case the mbius invariance is built into the definition and the non - negativity of the energy is non - trivial . in the same way one can define conformal ( willmore ) energy for simplicial surfaces in euclidean spaces of higher dimension and space forms .
the discrete conformal energy is well defined for polyhedral surfaces with circular faces ( not necessarily simplicial ) .
consider two triangles with a common edge .
let @xmath64 be their other edges oriented as in fig.[f.formula ] . identifying vectors in @xmath65 with imaginary quaternions @xmath66 one obtaines for the quaternionic product @xmath67 where @xmath68 and @xmath69 are the scalar and vector products in @xmath65 .
the external angle @xmath70 $ ] between the circumcircles of the triangles in fig.[f.formula ] is given by one of the equivalent formulas : @xmath71 here @xmath72 is the cross - ratio of the quadrilateral . _ proof .
_ since @xmath73 , @xmath74 and @xmath75 are mbius invariant it is enough to prove the first formula for the planar case @xmath76 , mapping all four vertices to a plane by a mbius transformation . in this case
@xmath77 becomes the classical complex cross - ratio .
considering the arguments @xmath76 one easily arrives at @xmath78 .
the second representation follows from the identity @xmath79 for imaginary quaternions . finally applying ( [ product ] ) we obtain @xmath80
similarly to the smooth willmore functional @xmath58 , minimizing the discrete conformal energy @xmath53 makes the surface as round as possible .
let us denote by @xmath81 the combinatorial data of @xmath16 .
the simplicial surface @xmath16 is called a geometric realization of the abstract simplicial surface @xmath81 .
critical points of @xmath28 are called simplicial willmore surfaces .
the conformal ( willmore ) energy of an abstract simplicial surface is the infimum over all geometric realizations @xmath82 [ cols="^,^,^,^,^ " , ] kevin bauer implemented the proposed conformal functional with the brakke s evolver @xcite and did some minimization numerical experiments .
examples of those are presented in fig.[f.minimizing ] . in the first and the second lines the initial configurations and the corresponding willmore surfaces minimizing the conformal energy respectively are shown .
let us call the gradient flow of the energy @xmath53 the discrete willmore flow . by this flow
the energy of the first simplicial sphere decreases to zero and the surface evolves into a convex polyhedron with all the vertices lying on a sphere .
the abstract simplicial surface of the central example is different and we obtain a simplicial willmore sphere with positive conformal energy .
the third example is a simplicial projective plane .
the initial configuration is made from squares divided into triangles ( cf .
we see that the minimum is close to the smooth boy surface known to minimize @xcite the willmore energy for projective planes . the minimization of the conformal energy for simplicial spheres is related to a classical result of steinitz @xcite who has shown that there exist abstract simplicial 3-polytopes without geometric realizations all vertices of which belong to a sphere .
we call these combinatorial types non - inscribable
. the non - inscribable examples of steinitz are constructed as follows @xcite .
let @xmath81 be an abstract simplicial sphere with vertices colored in black and white .
denote the sets of white and black vertices by @xmath83 and @xmath84 respectively , @xmath85 .
assume that the number of black vertices does not exceed the number of white vertices , @xmath86 , and there are no edges connecting two white vertices and there are edges connecting black vertices .
it is easy to see that @xmath81 with these properties can not be inscribed in a sphere .
indeed , assume that we have constructed such inscribed convex polyhedron .
then the equality of the intersection angles at both ends of an edge ( see left fig.[f.denergy ] ) implies @xmath87 and the equalities hold only if all edges connect vertices of different color . the obtained contradiction to the assumed inequality implies the claim . to construct abstract polyhedra with @xmath86 and with some edges connecting black points ,
take a polyhedron @xmath88 whose number of vertices does not exceed the number of faces @xmath89 .
color all the vertices in black , add white vertices at the faces and connect them to all black vertices of a face .
we obtain a polyhedron with black ( original ) edges and @xmath90 .
the example with minimal possible number of vertices @xmath91 is shown in fig.[f.11 ] .
the starting polyhedron @xmath88 here are two tetrahedra identified along a common face : @xmath92 .
[ f.11],scaledwidth=35.0% ] hodgson , rivin and smith @xcite found a characterization of inscribable combinatorial types , based on a transfer to the klein model of hyperbolic 3-space .
it is not clear whether there exist non - inscribable examples of non - steinitz type .
numerical experiments lead us to the following the conformal energy of simplicial willmore spheres is quantized @xmath93 note that this claim belongs to differential geometry of discrete surfaces .
it would be interesting to find a ( combinatorial ) meaning of the integer @xmath94 .
compare also with the famous classification of smooth willmore spheres by bryant @xcite , who has shown that the energy of willmore spheres is quantized @xmath95 .
the discrete willmore energy is defined for the ambient spaces ( @xmath96 or @xmath97 ) of any dimension .
this leads to combinatorial willmore energies @xmath98 where the infimum is taken over all realizations in the @xmath99dimensional sphere .
obviously these numbers build a non - increasing sequence @xmath100 which becomes constant for sufficiently large @xmath26 .
complete understanding of non - inscribable simplicial spheres is an interesting mathematical problem .
however the phenomenon of existence of such spheres might be seen as a problem in using of the conformal functional for applications in computer graphics , such as fairing of surfaces .
fortunately the problem disappears just after one refinement step : all simplicial spheres become inscribable .
let @xmath81 be an abstract simplicial sphere .
define its refinement @xmath101 as follows : split every edge of @xmath81 in two by putting additional vertices and connect these new vertices sharing a face of @xmath81 by additional edges .
[ p.refine ] the refined simplicial sphere @xmath101 is inscribable , and thus @xmath102 .
_ koebe s theorem ( see , for example , @xcite , @xcite ) claims that every abstract simplicial sphere @xmath103 can be realized as a convex polyhedron @xmath16 all edges of which touch a common sphere @xmath104 .
starting with this realization @xmath16 it is easy to construct a geometric realization @xmath105 of the refinement @xmath101 inscribed in @xmath104 .
indeed choose the touching points of the edges of @xmath16 with @xmath104 as additional vertices of @xmath105 and project the original vertices of @xmath16 ( which lie outside of the sphere @xmath104 ) to @xmath104 .
one obtaines a convex simplicial polyhedron @xmath105 inscribed in @xmath104 .
another interesting variational problem with the conformal energy is the optimization of triangulations of a given simplicial surface .
here one fixes the vertices and chooses an equivalent triangulation ( abstract simplicial surface @xmath81 ) minimizing the conformal functional .
the minimum @xmath106 yields `` an optimal '' triangulation for a given vertex data . in the case of
@xmath104 this optimal triangulation is classical .
let @xmath16 be a simplicial surface with all vertices @xmath17 on a two dimensional sphere @xmath104 .
then @xmath46 if and only if it is the delaunay triangulation on the sphere , i.e. @xmath16 is the boundary of the convex hull of @xmath17 . in differential geometric applications like numerical minimizing the willmore energy of smooth surfaces ( cf .
@xcite ) it is not natural to preserve the triangulation by minimizing the energy , and one should also change the combinatorial type decreasing the energy .
discrete conformal energy @xmath53 is not only a discrete analogue of the willmore energy .
one can show that it approximates the smooth willmore energy although the smooth limit is very sensitive to the refinement method and should be chosen in a very special way .
a computation which will be published elsewhere shows that if one chooses the vertices of a curvature line net of a smooth surface @xmath61 for the vertices of @xmath16 and triangularizes it then @xmath28 converges to @xmath62 by natural refinement . on the other hand the infinitesimal equilateral triangular lattice gives in the limit 3/2 times larger energy .
possibly the minimization of the discrete willmore energy with the vertices on the smooth surface could be used for computation of the curvature line net .
we are going to investigate this interesting and complicated phenomenon .
an accurate model for bending of discrete surfaces is important for modelling in virtual reality .
let @xmath107 be a thin shell and @xmath108 its deformation .
the bending energy of smooth thin shells is given by the integral @xcite @xmath109 where @xmath110 and @xmath5 are the mean curvatures of the original and deformed surface respectively .
for @xmath111 it reduces to the willmore energy . to derive the bending energy for simplicial surfaces
let us consider the limit of fine triangulation , i.e. of small angles between the normals of neighboring triangles .
consider an isometric deformation of two adjacent triangles .
let @xmath14 be the complement of the dihedral angle of the edge @xmath11 , or , equivalently , the angle between the normals of these triangles ( see fig.[f.bending ] ) and @xmath112 the external intersection angle between the circumcircles of the triangles ( see fig.[f.denergy ] ) as a function of @xmath14 . [ p.bending ]
assume that the circumcenters of the circumcircles of two adjacent triangles do not coincide .
then in the limit of small angles @xmath113 the angle @xmath75 between the circles behaves as follows : @xmath114 here @xmath13 is the length of the edge and @xmath115 is the distance between the centers of the circles . in @xcite
similar representations for the bending energy of simplicial surfaces were found empirically .
they were demonstrated to give convincing simulations and good comparison with real processes . in @xcite
the distance between the barycenters is used for @xmath117 in the energy expression but possible numerical advantages in using circumcenters are indicated .
using the willmore energy and willmore flow is a hot topic in computer graphics .
applications include fairing of surfaces and surface restoration .
we hope that our conformal energy will be useful for these applications and plan to work on them .
* acknowledgements*. i would like to thank ulrich pinkall for the discussion in which the idea of the discrete willmore functional was born .
i am also grateful to gnter ziegler , peter schrder , boris springborn , yuri suris and ekkerhard tjaden for useful discussions and to kevin bauer for making numerical experiments with the conformal energy . | a new functional for simplicial surfaces is suggested . it is invariant with respect to mbius transformations and is a discrete analogue of the willmore functional .
minima of this functional are investigated . as an application a bending energy for discrete thin - shells
is derived .
institut fr mathematik , technische universitt berlin , + strasse des 17 .
juni 136 , 10623 berlin , germany + e-mail:[email protected] [ [ keywords ] ] keywords : + + + + + + + + + conformal energy , willmore functional , simplicial surfaces , discrete differential geometry |
qcd sum rules on the light cone @xcite have been extensively applied to various exclusive form factors , in particular , to weak heavy - to - light transition form factors ( see , e.g. , @xcite ) . within this method ,
the relevant correlators are obtained as power expansions in @xmath6 near the light cone ( lc ) @xmath0 in terms of the distribution amplitudes of increasing twist . in practice , however , only few lowest - twist distribution amplitudes are known with a reasonable accuracy .
thus one has to rely on calculations which take into account only lowest powers of @xmath6 , at most up to terms linear in @xmath6 , without a systematic study of the effects related to higher powers of @xmath6 ( @xcite and references therein ) .
however , the off - lc @xmath7 contributions to the correlator are not parametrically suppressed compared to the contribution evaluated at @xmath0 : for instance , braun and halperin @xcite studied the pion elastic form factor with light - cone sum rules and found that contributions to the correlator from terms corresponding to higher powers of @xmath6 in the pion bethe
salpeter ( bs ) amplitude @xmath8 have in general the same @xmath9 behaviour . as we show here
, a similar situation occurs for the heavy - to - light form factor : the contributions to the correlator coming from the lc and the off - lc terms in the bs amplitude of the light meson have the same dependence in @xmath2 , @xmath10 being the heavy - quark mass .
therefore , relying on calculations performed at @xmath0 does not seem to be safe and the effects of all powers of @xmath6 should be summed and taken into account in the sum rule . in this paper , we study the correlator relevant for the extraction of the heavy - to - light form factors with sum rules and propose a method which allows one to obtain and take into account the full @xmath6-dependence of the bs amplitude of the light meson . for the sake of argument
, we present the analysis for the model with scalar constituents , the heavy scalar @xmath11 of the mass @xmath10 and the light scalar @xmath12 of the mass @xmath13 ( which we name `` quarks '' throughout the paper ) .
we study the weak transition of a heavy spinless bound state @xmath14 to the light spinless bound state @xmath15 ( which we refer to as mesons ) induced by the weak heavy - to - light @xmath16 quark transition .
the analysis of this model is technically simpler but at the same time allows one to study some essential features of the corresponding qcd case .
the paper is organized as follows : in sec .
[ sect_bs ] , we consider the bs amplitude of the light meson @xmath17 and its expansion in powers of @xmath6 . in sec .
[ sect_ff ] , we study the correlator @xmath18 , relevant for the extraction of the @xmath19 form factor within the method of sum rules . first , we show that the use of the nakanishi representation for the bs amplitude of the light meson amounts to a summation of all the off - lc effects .
we refer to the correlator obtained by this procedure as the full correlator .
second , we make use of the expansion of the bs amplitude near the lc and obtain a different form of the correlator corresponding to this lc expansion .
we show that for the quark quark interaction dominated by one - boson exchange at small distances ( similar to qcd , where the interaction is dominated by one - gluon exchange ) any term corresponding to @xmath20 in the bs amplitude gives a contribution to the correlator which behaves as @xmath21 , independent of @xmath22 .
keeping the term @xmath0 only , gives the lc correlator .
we then discuss the procedure of making the borelization of the correlator and obtaining the sum rule for the full and the lc correlators .
section [ numerical ] presents the numerical analysis of the full and the lc correlators for charm and beauty decays .
it is shown that at @xmath23 the lc correlator systematically overestimates the full correlator , the difference being numerically @xmath5% , independent of the mass of the heavy quark .
section [ conclusions ] summarizes our results .
the bs amplitude is defined according to @xmath24 as a function of the variable @xmath25 , the amplitude may be represented by the fourier integral @xmath26 where the @xmath27-integration runs from @xmath28 to @xmath29 as follows from the analytic properties of feynman diagrams .
nakanishi proposed to parametrize the kernel @xmath30 as @xcite @xmath31 ^ 3 } , \end{aligned}\ ] ] where the function @xmath32 has no singularities in the integration regions in @xmath33 and @xmath27 .
the bs equation for the function @xmath34 leads to an equation for the function @xmath32 .
the @xmath35-integral in ( [ nak2 ] ) is the second derivative w.r.t .
@xmath36 of the feynman propagator of a scalar particle of mass @xmath37 [ with @xmath38 in coordinate space .
making use of the explicit expression for this propagator @xcite , we obtain the following expansion near the light cone @xmath0 : @xmath39 + o(x^4 ) . \end{aligned}\ ] ] interestingly , this is not a pure power series but it contains also terms involving @xmath40 . inserting ( [ prop ] ) into ( [ nak2 ] )
leads to the light - cone expansion of @xmath30 : @xmath41 + x^4 \left[\,g_2(\xi)+\log(-x^2 m^2)h_2(\xi)\right ] + \cdots . \end{aligned}\ ] ] the functions @xmath42 and @xmath43 here may be expressed in terms of @xmath32 .
for instance , @xmath44 , \nonumber\\ h_1(\xi)&=&-\frac{1}{128\pi^2}\int\limits_0^\infty dz\ , g(z,\xi).\end{aligned}\ ] ] the presence of the logarithmic terms leads to complications : let us go back to eq .
( [ nak2 ] ) and expand the exponential @xmath45 .
only the first term corresponding to @xmath0 ( the light - cone ) is finite , whereas all higher terms lead to divergent @xmath35-integrals ; this is a consequence of the presence of the terms @xmath40 . after performing the wick rotation in the @xmath35-space
, we may cut the @xmath46-integration at @xmath47 . in this way
, we obtain a regularized pure _ power _ series in @xmath6 for @xmath30 and thus for @xmath48 .
the introduction of the cutoff @xmath49 in the momentum - space integrals is equivalent to the introduction of a regulator @xmath50 in coordinate space : @xmath51 the initial expansion ( [ logseries ] ) is reproduced by setting @xmath52 . for a nonzero @xmath50 ,
we can represent @xmath53 as a power series in @xmath6 and insert this expansion in ( [ logseries ] ) , obtaining @xmath54 with the first three distribution amplitudes @xmath55 given by @xmath56 as a result , instead of the series expansion for the bs wave function in terms of powers and logarithms of @xmath6 with finite coefficients , we obtained a pure power series , in which only the first term , corresponding to @xmath0 , is finite , whereas all higher distribution amplitudes depend on the regulator parameter @xmath50 and become infinite in the limit @xmath52 ( @xmath57 ) .
the regularised lc expansion of the bs amplitude reads @xmath58 notice that in our case the full bs amplitude does not depend on the regulator parameter @xmath50 . however , any truncation of the series at some @xmath22 leads to an explicit dependence of @xmath59 on @xmath50 in such a way , that it diverges as @xmath52 .
the bs amplitude in momentum space reads @xmath60
^ 3}. \end{aligned}\ ] ] the kernel @xmath32 determines now the bound - state properties . for instance , the decay constant of a bound state @xmath61 is given by the relation @xmath62 performing the @xmath35-integration in eq .
( [ nak2 ] ) for @xmath63 leads to @xmath64 the light - cone wave function is related to the kernel @xmath32 as follows @xcite : @xmath65 ^ 2}.\end{aligned}\ ] ] respectively , the distribution amplitudes @xmath55 may be calculated as @xmath66-moments of @xmath67 . for the function ( [ 2.14 ] ) ,
only the zero @xmath66-moment leading to @xmath68 is finite , all higher moments require a cut - off in @xmath66 precisely as in ( [ 2.9 ] ) . for the case of an interaction between the constituents via exchange of a massless boson , the solution to the bs equation in the ladder approximation takes a simple form @xcite : @xmath69 the function @xmath70 here is a smooth function which takes finite nonzero values at the end - points .
only the end - point behavior is crucial for the heavy - to - light correlator , therefore the precise form of the function @xmath70 and its normalization are not essential for our analysis ; of importance for us is only the fact that @xmath71 .
hereafter we just set @xmath72 and do not care about the overall factor @xmath73 .
this factor turns out to be the same in the full and the light - cone correlators which we calculate and compare in the next section . closing this section ,
let us emphasize that the @xmath4-functional @xmath33-dependence of the kernel @xmath32 leads to a nontrivial @xmath66-dependence of the light - cone wave function @xmath74 ^ 2}.\end{aligned}\ ] ] the @xmath75 tail of the bound - state lc wave function is a consequence of the massless - boson exchange ( and is similar in this respect to the power - like behavior of the pion light - cone wave function in qcd ) .
in this section , we consider the correlator @xmath76 and its borel transform in the variable @xmath77 relevant for the extraction of the @xmath19 form factor . since @xmath78 is fixed , this correlator depends on the two variables @xmath77 and @xmath79 .
making use of the hadronic degrees of freedom , by inserting the complete system of hadron states into the correlator , we obtain the so - called phenomenological representation for @xmath80 , which we denote by @xmath81 : @xmath82 where @xmath83 is the threshold of the hadronic continuum , containing the @xmath84-meson plus light mesons , i.e. , @xmath85 , @xmath86 , etc . states .
respectively , @xmath87 . the decay constant @xmath88 of the @xmath84-meson is defined as @xmath89 and the form factor is given by @xmath90 eq .
( [ phen ] ) gives the single dispersion representation for @xmath80 in terms of the hadron degrees of freedom .
one can calculate the correlator @xmath80 also by making use of the underlying field - theoretic degrees of freedom and obtain in this way a different theoretical expression for @xmath80 , which we call @xmath91 .
this expression may also be written as a single dispersion representation in @xmath77 : @xmath92 quark hadron duality @xcite postulates that @xmath93 and @xmath94 are equal to each other after a proper smearing is applied to both of them .
the smearing may be implemented , e.g. , by applying the borel transform @xcite . calculating the borel image of @xmath93 and @xmath94 with the parameter @xmath95 , such that @xmath96
, we obtain the sum rule for @xmath80 : @xmath97 let us ignore for a moment the dependence of @xmath98 on @xmath79 and introduce the continuum subtraction point @xmath99 ( an effective continuum threshold , different from the physical continuum threshold @xmath83 ) according to @xmath100 for sufficiently large @xmath101 , one expects @xmath102 ( up to oscillating terms @xcite ) .
the region near the continuum threshold @xmath103 is clearly below the duality region : recall that @xmath104 , while @xmath105 .
thus , in the vicinity of @xmath83 , @xmath106 . as a result ,
the continuum subtraction @xmath99 as defined by ( [ s0 ] ) depends on the borel parameter ( and on @xmath79 ) and lies above the physical threshold : @xmath107 .
then the sum rule ( [ sr_1 ] ) takes the form @xmath108 with the correlator to which the @xmath101-cut is applied ( hereafter referred to as the cut correlator ) @xmath109 if one had known the @xmath110- and @xmath79-dependent solution @xmath111 to ( [ s0 ] ) , which encodes the full nonperturbative dynamics , the expression ( [ srff ] ) would have been exact . taking @xmath99 constant
makes it approximate .
one may then impose the stability criterium which is expected to guarantee the extraction of the true form factor in several different ways ( see , e.g. , @xcite ) .
we shall now concentrate on different possibilities to calculate @xmath91 and compare the corresponding results .
feynman diagrams for the correlator @xmath112 .
the normal and bold lines denote the heavy @xmath11 and the light @xmath12 particles , respectively , the wavy line corresponds to the scalar particle which mediates the @xmath11@xmath12 interaction ( a scalar analog of the gluon in qcd ) ; @xmath113 is the state - vector of the @xmath114 bound - state with the momentum @xmath115 .
the contribution of the first diagram is expressable through the bs wave function of the light meson and the full @xmath11-propagator . in the region of not very large @xmath116 , @xmath117 , the second and the third terms are suppressed by higher powers of the heavy - quark mass ( @xmath2 and @xmath21 , respectively ) compared to the first term and will not be considered.,width=453 ] we now calculate this correlator using the underlying field - theoretic degrees of freedom , and denote the result as @xmath93 .
the corresponding feynman diagrams are shown in fig .
[ fig:1 ] . in the region of not too large timelike momentum transfer @xmath116 ,
@xmath117 , the first diagram gives the main contribution whereas the contributions of the other diagrams are suppressed by higher powers of the heavy - particle mass @xmath10 .
neglecting the contributions of the power - suppressed diagrams leads to @xmath118 where @xmath119 is the full propagator of the heavy quark . approximating the full propagator with the free propagator gives @xmath120 we can proceed further in two different ways : a. first , the correlator ( [ 3.5 ] ) may be expressed in terms of the bs amplitude in momentum space @xmath121 making use of ( [ bsmom ] ) gives @xmath122 ^ 3 [ m_q^2-(p - k)^2-i0]}. \end{aligned}\ ] ] introducing the feynman parameter @xmath123 and performing the @xmath124-integration , we find @xmath125 ^ 2},\nonumber\\ & & \mu^2\equiv z+m^2-\xi(1-\xi)m^2,\qquad
p^2=p^2(1-\xi)-\xi(1-\xi)m^2 + q^2\xi.\end{aligned}\ ] ] from this expression , it is straightforward to obtain the single dispersion representation for @xmath93 in the form @xmath126 with @xmath127 the spectral density @xmath128 vanishes at the lower limit of the @xmath101-integration at @xmath129 .
notice that the quantity @xmath13 in the lower limit of the @xmath101-integration is the mass of the light spectator . for more details about the analytic properties of the three - point functions , we refer to @xcite . performing an integration by parts in ( [ disp_th_2 ] )
, we obtain the spectral representation in the `` standard '' form ( [ disp_th ] ) with @xmath130 the borel images of the two representations ( [ disp_th_2 ] ) and ( [ disp_th ] ) read , respectively , @xmath131 and @xmath132 when the @xmath101-integration extends to infinity , the equality of both formulas is demonstrated by integration by parts .
we now want to obtain the cut borel image of eq .
( [ 3.9 ] ) .
recall , that the cut at @xmath99 is applied to the borel transform of the spectral representation for the correlator in the form ( [ 3.5 ] ) . then
, the relation @xmath133 leads to the appearance of an additional surface term in the cut spectral representation of the form eq .
( [ 3.17 ] ) : @xmath134 moreover , if one works with the borel image in the form ( [ 3.20 ] ) , the surface term gives the main contribution to the correlator for large values of @xmath110 .
the sum rule for the form factor can now be written in two equivalent ways : @xmath135 b. another possibility to calculate @xmath91 is to substitute the light - cone expansion of @xmath48 , eq .
( [ lcexpansion ] ) , into eq .
( [ 3.5 ] ) : @xmath136 the functions @xmath137 are related to @xmath32 via eqs .
( [ 2.6 ] ) and ( [ 2.9 ] ) .
performing the @xmath138 and @xmath124 integrations gives @xmath139 ^ 3}+\cdots.\end{aligned}\ ] ] here @xmath140 .
this series ( [ 3.23 ] ) may be written as the series of spectral representations @xmath141 with @xmath142 calculable via @xmath55 .
for instance , for @xmath143 one finds @xmath144 the corresponding uncut borel image reads @xmath145 or , equivalently , @xmath146 \exp({-s_\xi/{2\mu_b^2 } } ) , \end{aligned}\ ] ] with @xmath147 in ( [ 3.24 ] ) and ( [ 3.24a ] ) the dots stand for terms corresponding to higher @xmath22 .
clearly , their contributions are suppressed with powers of the borel parameter @xmath110 and vanish in the limit @xmath148 .
usually , the borel parameter has the following dependence on the heavy quark mass : @xmath149 @xcite , where @xmath150 stays finite in the limit @xmath151 .
then , the terms corresponding to higher @xmath22 are not suppressed at @xmath151 .
now , let us consider the cut borel image of eq .
( [ 3.23a ] ) . as explained above
, the cut is applied to the dispersion representation in the form ( [ 3.5 ] ) . therefore ,
if we want to work with the spectral densities @xmath152 , we should take into account the surface terms , similar to eq .
( [ 3.20 ] ) .
the cut borel image takes the form @xmath153\frac{\exp(-s_0/2\mu_b^2)}{2 } \\
\nonumber & & + \mbox{contributions of terms corresponding to higher $ n$}.\end{aligned}\ ] ] making use of the distribution amplitudes @xmath55 , we rewrite the correlator ( [ cutlc ] ) as @xmath154 \exp({-s_\xi/{2\mu_b^2}})\theta(s_\xi < s_0)\nonumber\\ & -&4\exp(-s_0/{2\mu_b^2 } ) \left [ \frac{\phi_1(\xi_0)}{m_q^2 } + \frac{\phi'_1(\xi_0)}{4\mu_b^2(1-\xi_0)}+\frac{\phi'_1(\xi_0)}{s_0 } \right]+\cdots , \end{aligned}\ ] ] where @xmath155 and @xmath156 stand for contributions of the terms corresponding to @xmath157 .
the light - cone correlator corresponds to the first term in the expansions ( [ cutlc ] ) and ( [ cutlc1 ] ) , calculated with the cut parameter @xmath158 specific for the lc approximation ( see the next section for details ) : @xmath159 keeping only this term leads to the following sum rule : @xmath160 let us study the conditions under which the contributions of higher @xmath22 are parametrically suppressed compared to the contribution of the light - cone @xmath161 . \1 . the heavy - meson mass is related to the heavy quark mass by @xmath162 , @xmath163 .
one can then check that the dominant contribution to the integral comes from the end - point region @xmath164 .
therefore , in order to have contributions of higher @xmath22 suppressed by powers of @xmath2 compared to the @xmath161 contribution , one needs , e.g. , the following end - point behavior : @xmath165 such a behavior can be obtained , e.g. , for the light - cone wave function @xmath166 @xmath167 being the size parameter of the light meson . in this specific case the functions @xmath168 have the necessary behavior ( [ nb ] ) in the end - point region . however , for realistic wave functions obtained as solutions to the bs equation , the distribution amplitudes @xmath168 for different @xmath22 have a similar behavior in the end - point region and therefore the lc expansion for the form factor has no small parameter . in this case , the term @xmath161 is not parametrically enhanced compared to terms of higher @xmath22 , and , in order to obtain the form factor , terms for all @xmath22 should be summed .
in other words , the transverse motion of the light quark is essential , the longitudinal light - cone distribution amplitude @xmath169 is not sufficient , and the knowlegde of the full wave function @xmath170 is necessary to obtain the @xmath171 form factor .
one might expect to have a power suppression of the off - lc terms corresponding to @xmath172 in the limit @xmath148 , within the `` local - duality '' sum rule @xcite , taking into account that in standard svz sum rules the choice @xmath148 suppresses condensate contributions . within the context of light - cone sum rules ,
the limit @xmath148 is , however , tricky : in the uncut light - cone correlator the limit @xmath148 indeed suppresses the off - lc effects ( @xmath1 terms ) .
however , after applying the cut , this property is lost : obviously , the surface terms in the cut correlator ( [ cutlc ] ) corresponding to @xmath172 remain finite in the limit @xmath148 and can give a sizeable contribution .
there is also another subtlety : a specific feature of the local - duality sum rule is that the observables are determined to great extent by the value of the continuum subtraction point @xmath99 .
therefore one can not apply the standard sum - rule stability criteria to extract the physical value of an observable . nevertheless , _ assuming _ the duality interval to be process - independent and fixing @xmath99 from one observable allows one to predict other observables .
however , as we shall see in the next section , when the standard criterion to fix the continuum subtraction @xmath99 is applied , its value for the full and for the lc correlators turn out to be different from each other . finally , we conclude that for the realistic case of the interaction dominated by one - boson exchange at short distances , the lc contribution does not dominate the cut correlator parametrically , i.e. , the off - lc terms are not suppressed by any large parameter compared to the lc term . to understand
how well the lc contribution numerically compares with the full result , we calculate the correlator @xmath173 , eq .
( [ gammaborelfullcut ] ) , and the lc correlator , eq .
( [ gammaborellc ] ) , making use of the bs wave function obtained as solution to the bs equation with a realistic potential dominated by a one - boson exchange at small separations .
in this section we address the case @xmath143 , and therefore do not write explicitly the argument @xmath79 . for the bs kernel @xmath174 it is straightforward to calculate the spectral densities @xmath175 and @xmath176 .
we shall analyse the correlators for the light meson in the final state .
we may therefore set @xmath177 and find @xmath178\theta(s-(m_q+m)^2),\nonumber\\ \delta_{\rm lc}(s)&=&\frac{m_q^2(s - m_q^2)}{s^3}\theta(s - m_q^2 ) .
\end{aligned}\ ] ] here @xmath179 and @xmath180 is just @xmath181 of eq .
( [ 3.23a ] ) . clearly , in the limit @xmath182 , both quantites coincide .
notice the relation @xmath183 which can be checked explicitly .
let us now discuss the values of the parameters to be used in numerical estimates . for the heavy - quark mass
the situation is obvious : to study the beauty - meson decay , we choose the value @xmath184 gev used in light - cone sum - rules @xcite . the same value is employed in quark - model calculations @xcite . to describe decay of the charm meson
, we set @xmath185 gev according to sum rules @xcite .
the relevant choice of the light - quark mass parameter @xmath13 requires clarification .
our interest is to understand with the help of the simplified model under consideration the corresponding qcd calculation .
we therefore choose the numerical parameters relevant for qcd .
since the light - quark mass parameter @xmath13 appears in the framework of the bs equation , it should be understood as the effective quark mass which takes into account nonperturbative effects related to spontaneous chiral symmetry breaking in the soft region , i.e. , the constituent quark mass . in @xcite
, the constituent quark mass was calculated through the quark condensate in qcd with the result @xmath186 mev at the chiral - symmetry breaking scale @xmath29 gev .
the scale - dependence of the constituent mass of the light quark above 1 gev was also reported in @xcite . in sum - rule analyses @xcite it was found , that the relevant infrared factorization scale for a heavy - meson decay is @xmath187 .
we therefore make use of the constituent quark mass evaluated at this scale . employing the results from @xcite , we find the relevant value of the constituent mass of the light quark @xmath188 mev for beauty - meson decay , and @xmath189 mev for charm - meson decay .
these values of the quark masses will be used in numerical estimates below .
[ cols= " < , > " , ] we introduce the parameters @xmath4 related to the continuum subtraction @xmath99 by @xmath190 to fix @xmath4 , we follow the standard procedure @xcite : namely , we require that the quantity @xmath191 reproduces the heavy - meson mass , i.e. , @xmath192 both for the lc and the full spectral densities , where @xmath193 for the lc correlator and @xmath194 for the full correlator .
the quantity @xmath4 as defined by ( [ delta2 ] ) depends on @xmath79 and @xmath110 : eq .
( [ delta2 ] ) is just the definition of the implicit function @xmath195 .
such a procedure of fixing @xmath4 for the light - cone correlator was employed e.g. in @xcite .
recall , however , that there is no unique way to fix @xmath4 : one may require instead that @xmath196 for @xmath197 .
moreover , since the spectral densities and the thresholds in eq .
( [ delta ] ) are different for @xmath198 and @xmath199 , also the numerical values of @xmath198 and @xmath199 obtained from eq .
( [ delta2 ] ) are different .
we discuss here only the case @xmath143 .
taking into account the lack of a unique way to introduce @xmath4 , we shall not consider the @xmath110-dependent @xmath198 and @xmath199 .
we rather determine the constant values @xmath198 and @xmath199 such that the relation ( [ delta2 ] ) is satisfied only for one specific value of @xmath150 .
[ fig:3 ] presents the numerical results for beauty - meson decay : @xmath200 gev@xmath201 gev and @xmath188 mev .
we plot the borel curves for the full and for the lc correlators for two different values of @xmath4 : the left column shows the results for @xmath202 gev and @xmath203 gev . in this case , the relation ( [ delta2 ] ) is fulfilled at the relatively low value @xmath204 gev : namely , @xmath205 gev for @xmath143 and @xmath204 gev .
the right column present the results for @xmath206 gev and @xmath207 gev . in this case , @xmath205 gev for @xmath143 and @xmath208 gev .
the first row shows @xmath209 calculated with the lc and the full correlators vs @xmath150 .
the second row presents the quantity @xmath210 for the lc and the full correlators .
finally , the third row gives the ratio of the full to the lc correlators .
[ fig:4 ] gives the results for charm - meson decay : @xmath211 gev , @xmath185 gev and @xmath189 mev ( left column ) . to illustrate the influence of the light - quark mass on the off - lc effects , we present also the results for @xmath212 mev ( right column ) .
the continuum subtraction parameter @xmath4 is fixed from the relation @xmath213 at @xmath214 gev .
to show the origin of the difference between the cut full and light - cone correlators , we consider the limit @xmath215 and @xmath148 , with @xmath216 . in this case
explicit expressions for the correlators may be obtained : @xmath217+o(m^4/\delta^2_{\rm th } ) + o(\delta^3_{\rm th}/m_q).\end{aligned}\ ] ] for @xmath216 ( @xmath218 ) , the uncut and the cut correlators ( both full and lc ) behave quite differently for large @xmath10 : @xmath219 thus , the cut correlator picks up only a small fraction of the full correlator from the region not far from the threshold .
( for @xmath220 , both the cut and the uncut correlators have a similar behavior @xmath221 . ) now , fixing @xmath198 and @xmath199 according to the standard procedure ( [ delta2 ] ) , we express these quantities via the binding energy of the heavy meson @xmath222 defined according to @xmath223 : turns out to be different from @xmath198 .
this has the following origin : if we include @xmath224 terms in the lc expansion of the cut correlator ( [ cutlc ] ) , cut them at @xmath225 , and determine the latter from eqs .
( [ delta ] ) and ( [ delta2 ] ) , then @xmath226 . since we have included only one @xmath227 term , we obtain @xmath228 . ]
@xmath229+\cdots , \ ] ] leading to @xmath230 and thus @xmath231 in the expressions above the dots denote terms containing higher powers of @xmath232 .
this example illustrates that the off - lc effects may play an essential role in the cut correlators , as their contribution is not suppressed by any large parameter : the quantites @xmath13 and @xmath222 have the same order of magnitude .
let us emphasize that we compare the full and the light - cone correlators evaluated at different values of the cut parameters @xmath199 and @xmath198 . from our point of view
this very comparison is relevant if one wants to understand the error due to taking into account only the light - cone @xmath233 contribution to the correlator and neglecting terms containing higher powers of @xmath6 .
one could also compare the correlators for the same value of @xmath4 .
the difference between the full and the lc correlators is only slightly reduced in this case , the ratio still remaining @xmath234 .
this can be seen by setting @xmath235 in ( [ 4.8 ] ) .
the following lessons may be drawn from the results presented in this section : * the off - lc effects play an essential role in the cut correlator , as they are not suppressed by any large parameter . numerically , the difference between the full and the lc correlators , evaluated at the same value of the borel parameter , is 10@xmath23620% .
this difference is due to the off - lc effects . * the functions @xmath237 and
@xmath238 have the same shape , but @xmath237 lies well above @xmath238 , if the standard procedure of fixing @xmath199 and @xmath198 ( [ delta2 ] ) is used .
the values of both ( th and lc ) correlators obtained with @xmath198 and @xmath199 tuned at @xmath204 gev ( left column in fig .
[ fig:3 ] ) are greater than the values of the respective correlators obtained with @xmath198 and @xmath199 tuned at @xmath208 gev ( right column in fig .
[ fig:3 ] ) .
the local stability is better when one fixes @xmath4 from ( [ delta2 ] ) at a larger value of @xmath150 ( right column in fig .
[ fig:3 ] ) . in this case , both borel curves for @xmath239 and @xmath240 show a good stability in @xmath150 . nevertheless , still @xmath239 is much larger than @xmath240 !
this illustrates that the borel stability _ per se _ does not guarantee the extraction of the correct physical value . *
the difference between @xmath239 and @xmath240 increases with increasing mass of the light quark .
therefore , this difference is expected to be greater for the heavy mesons @xmath241 and @xmath242 , containing the strange @xmath101-quark , than for @xmath243 and @xmath244 .
in this paper , we studied the correlator @xmath245 which is one of the basic objects for extracting the heavy - to - light form factor within the method of sum rules .
we have shown that to leading @xmath2-accuracy this correlator may be calculated through the bs amplitude of the light meson @xmath246 expanding the bs amplitude near the light cone @xmath0 generates the light - cone expansion of the correlator . making use of the nakanishi representation for the bs amplitude , we obtained dispersion representations for the full and the lc correlators in terms of the kernel @xmath32 of the nakanishi representation .
we studied the full and the light - cone correlators and their borel transforms depending on the properties of the kernel @xmath32 .
we then made use of the known solution for @xmath32 in a model with light scalar particles interacting by an exchange of a massless boson .
this relatively simple model provides a good laboratory for studying qcd since the corresponding bound - state wave functions have properties similar to the properties of hadron wave functions in qcd .
we calculated the full and the light - cone correlators and their borel transforms in the variable @xmath77 , and studied these correlators for various prescriptions to fix the heavy - hadron continuum subtraction and in various regions of the parameters relevant for extracting the heavy - to - light form factors .
this work thus represents the first systematic study of the off - light - cone effects in qcd sum rules for heavy - to - light form factors .
* we have seen that after performing the borel transform the light - cone correlator provides numerically the bulk of the full correlator , although parametrically the off - lc effects are not suppressed compared to the lc contribution .
this observation holds for various prescriptions of fixing the heavy - hadron continuum subtraction point ( i.e. , a cut applied to the correlator for isolating the contribution of the heavy hadron of interest in the initial state ) and a wide range of masses of particles involved in the decay process .
* we demonstrated that , nevertheless , the difference between the cut full and the cut light - cone correlators always remains nonvanishing .
for example , fixing the continuum subtraction points the by standard criteria , we have found the following relation for the cut borel transforms of the full and the lc correlators for @xmath151 , @xmath148 , and @xmath143 : @xmath247 the correction being always negative . here
@xmath13 is the effective constituent mass of the light quark , which emerges from the bs equation , @xmath248 , and @xmath222 is the binding energy of the heavy meson @xmath162 .
taking into account that the constituent quark mass remains finite in the chiral limit , we come to the following important conclusion : _ in heavy - to - light decays , there exists no rigorous theoretical limit in which the cut lc correlator coincides with the cut full correlator_. + thus , the off - light - cone effects in sum rules for heavy - to - light correlators are not negligible and should be taken into account .
* we note that the borel curves for the full and the lc correlators have similar shapes .
however the light - cone correlator systematically _ overestimates _ the full correlator , the difference at small @xmath79 being @xmath5% in a wide range of the heavy - quark mass relevant for charm and beauty decays .
we want to point out that the similarity of the borel curves for the full and the lc correlators implies that the systematic difference between the correlators can not be diminished by a relevant choice of the criterion for extracting the heavy - to - light form factor . + the observed effect might suggest a systematic uncertainty in the results for form factors obtained within light - cone sum rules . as follows from the relation ( [ result ] ) , this uncertainty is expected to be larger for decays of heavy mesons containing the strange quark , @xmath241 and @xmath242 , than for the @xmath243 and @xmath244 mesons .
this issue deserves further investigation .
finally , we point out the following : although the model , discussed here , in many aspects differs from qcd , this model mimics correctly those features which are essential for the effects discussed
. therefore , many of the results obtained in this paper are valid also for qcd .
in particular , the expression ( [ result ] ) suggests the following relationship between the light - cone and the full correlators in qcd for large values of @xmath10 and @xmath110 : @xmath249 in numerical estimates , we used the parameters relevant for @xmath243 and @xmath244 decays .
we therefore believe that also the numerical estimates for higher - twist effects obtained in this work provide a realistic estimate for higher - twist effects in qcd . _
_ we are grateful to vladimir braun , pietro colangelo , and alexander khodjamirian for valuable comments on the preliminary version of the paper .
we thank vittorio lubicz and matthias neubert for interesting and stimulating discussions .
d. m. gratefully acknowledges financial support from infn , university `` roma tre '' , and the austrian science fund ( fwf ) under project p17692 .
30 i. i. balitsky , v. m. braun , and a. v. kolesnichenko , nucl .
* b312 * , 509 ( 1989 ) ; v. m. braun and i. filyanov , z. phys .
* c44 * , 157 ( 1989 ) ; v. i. chernyak and i. r. zhitnitsky , nucl .
* b345 * , 137 ( 1990 ) .
colangelo and a. khodjamirian , _ qcd sum rules : a modern perspective _ , in _ at the frontier of particle physics _ , edited by m. shifman ( world scientific , singapore , 2001 ) , vol . 3 , p. 1495
[ hep - ph/0010175 ] .
p. ball and v. m. braun , phys .
* d58 * , 094016 ( 1998 ) .
p. ball and r. zwicky , phys .
rev . * d71 * , 014015 ( 2005 )
. v. m. braun and i. halperin , phys .
b328 * , 457 ( 1994 ) .
n. nakanishi , phys . rev . * 130 * , 1230 ( 1963 ) . n. n. bogolyubov and d. v. shirkov , _ quantum fields _ , nauka , moscow ( 1980 ) . v. a. karmanov and j. carbonell , eur .
phys . j. * a27*,1 ( 2006 ) .
m. shifman , a. vainshtein , and v. zakharov , nucl . phys . *
b147 * , 1 ( 1979 ) .
m. shifman , prog .
. suppl . * 131 * , 1 ( 1998 ) [ hep - ph/9802214 ] .
m. jamin and b. lange , phys . rev . * d65 * , 056005 ( 2002 ) .
t. kleinschmidt , hep - ph/0409039 ( unpublished ) .
d. melikhov , phys . rev . *
d53 * , 2460 ( 1996 ) ; phys . rev . *
d56 * , 7089 ( 1997 ) ; eur .
direct * c4 * , 2 ( 2002 ) [ hep - ph/0110087 ] ; w. lucha , d. melikhov , and s. simula , phys .
* d75 * , 016001 ( 2007 ) . v. a. nesterenko and a. v. radyushkin , phys . lett . *
115b * , 410 ( 1982 ) .
w. lucha and d. melikhov , phys . rev . * d73 * , 054009 ( 2006 ) .
d. melikhov and b. stech , phys . rev . * d62 * , 014006 ( 2000 ) .
p. ball , phys .
b641 * , 50 ( 2006 ) .
d. melikhov and s. simula , eur .
journal * c37 * , 437 ( 2004 ) .
w. lucha , d. melikhov , and s. simula , phys . rev . *
d74 * , 054004 ( 2006 ) . | we report the first systematic analysis of the off - light - cone effects in sum rules for heavy - to - light form factors .
these effects are investigated in a model based on scalar constituents , which allows a technically rather simple analysis but has the essential features of the analogous qcd calculation .
the correlator relevant for the extraction of the heavy - to - light form factor is calculated in two different ways : first , by adopting the full bethe
salpeter amplitude of the light meson and , second , by performing the expansion of this amplitude near the light cone @xmath0 .
we demonstrate that the contributions to the correlator from the light - cone term @xmath0 and the off - light - cone terms @xmath1 have the same order in the @xmath2 expansion .
the light - cone correlator , corresponding to @xmath0 , is shown to systematically overestimate the full correlator , the difference being @xmath3 , with @xmath4 the continuum subtraction parameter of order 1 gev .
numerically , this difference is found to be @xmath5% . |
the x ray source 16 has several unusual characteristics .
though , like all soft x - ray transients , it is a low - mass binary system , its 1.7@xmath1 secondary orbiting a 5.5@xmath2 black hole ( shahbaz et al .
1999 ) is considerably more massive than is typical in such systems ( see e.g. a review by tanaka & shibazaki 1996 ) .
its mass and spectral type ( f3-f4 ) imply that the donor star is near ( or just beyond ) the end of its main - sequence life - time .
in fact kolb et al . ( 1997 ) and kolb ( 1998 ) assert that it appears to be crossing the hertzsprung gap , i.e. expanding towards the giant branch , which would imply a very large mass transfer rate , @xmath3 g s@xmath4 ( king & kolb 1999 )
. even if the secondary star in 16 is still on the main sequence , as argued by regs , tout & wickramasinghe ( 1998 ) , the inferred mass transfer rate , @xmath5 g s@xmath4 , is orders of magnitude greater than @xmath6 estimates in other x - ray transients ( see e.g. chen , shrader & livio 1997 ) .
moreover , the values quoted above are close to the critical rate , above which the irradiated accretion disc in 16 would be stable with respect to the dwarf - nova type instability ( van paradijs 1996 ; see dubus et al .
1999b for a most recent discussion of the critical rate for this instability ) , that is generally thought to be responsible for the transient behaviour of low - mass x - ray binary systems .
the outbursts themselves are also rather atypical . like other black hole ( low mass )
x - ray transient systems ( bhxts ) , 16 has been quiescent for more than 30 years before entering the active phase in 1994 .
however , contrary to the behaviour of ` conventional ' bhxts , the first outburst was followed in 1995 by two others which displayed hard x - ray spectra . finally , after the last burst of emission in july / august 1995 the system settled into x
ray quiescence , with luminosity of @xmath7 erg s@xmath4 .
this quiescent state ended around april 25 , 1996 when a new , soft x
ray outburst began , which is the main subject of this paper .
orosz et al .
( 1997 ) obtained bvri photometry close to the onset of the outburst and found the rise in x - ray flux delayed by 6 days with respect to the optical increase .
consequently , hameury et al . (
1997 ) showed that the rise to outburst , and in particular the x - ray ` delay ' , are very well described by the dwarf - nova type disc instability model ( dim ) if the accretion disc is truncated at @xmath8 @xmath9 ( @xmath10 is the schwarzschild radius ) .
however , the success of this model is put into perspective by the subsequent behaviour of the system during the outburst , when , after the first local maximum , the soft x - ray light curve rose to a higher luminosity ` flaring ' plateau .
the present theoretical understanding tells us that this type of light - curve can not be produced based on the standard dim ( hameury et al .
1998 ) .
the behaviour of the optical luminosity during the 1996 outburst of 16 also seems to defy the generally accepted idea ( based on a solid observational background ) that in low mass x - ray binaries ( lmxbs ) the optical emission from the accretion disc is due to reprocessing of x - rays in the outer disc .
the fact that in 16 the optical flux decreases while the x - ray flux increases and then fluctuates around an approximately constant value is difficult to reconcile with the x - ray reprocessing model ( hynes et al .
finally , the list of unusual properties of 16 should be completed by the presence of a superluminal ` jet ' seen during the 1994 outburst ( hjellming & rupen 1995 ) . in this paper
we argue that the puzzling behaviour of this system can be readily reconciled with the dim , if we take into account the fact that the mass transfer rate in 16 is likely to be rather close to the stability limit . in this case ,
irradiation of the secondary during an outburst may be enough to increase the mass transfer rate by a factor of a few and thereby push the system into the stable regime .
this scenario and its expected effect on the observed x - ray light curve are discussed in detail in [ xray ] .
furthermore , in [ opt ] we speculate that the reason the optical flux does not rise with increasing x - ray luminosity is a reduction in the amplitude of the outer disc warping .
using a flared planar disc ` approximation ' to describe a warped disc , we show that as a result , the intercepted x - ray flux and therefore the observed optical emission is reduced .
we further suggest in [ end ] that the flattening of the outer disc and a consequent decrease in x - ray irradiation may also be responsible for the ultimate end of the stable accretion phase , by raising the stability criterion .
we conclude with a final discussion and a summary in [ disc ] .
following shahbaz et al . ( 1999 ) , in our estimates below we assume that 16 contains a 7@xmath11 black hole and that the mass ratio between the primary mass @xmath12 and the secondary mass @xmath13 is @xmath14 .
we also adopt a distance to the system of 3.2@xmath15 ( hjellming & rupen 1995 ) .
figure [ ltcrv](a ) shows the progress of the 1996 x - ray outburst of 16 .
the asm ( 2 - 12 kev ) ` soft ' x - ray flux began to rise 6 days after the start of the optical outburst and 15 days later attained a maximum corresponding to @xmath16 ( hereafter we use spectral fits by sobczak et al .
1999 to convert asm fluxes to bolometric luminosity values ) , where @xmath17 is the eddington luminosity .
after @xmath18 days of a roughly exponential decline which followed the first maximum , the soft x - ray flux fell by about @xmath19 .
it then began to rise again , and after about two months from the onset of the outburst , reached a strongly flaring plateau with _ total _
luminosity varying around @xmath20 .
this flaring state , which sobczak et al .
( 1999 ) identify with the very high spectral state , continued roughly until day 200 of the x - ray outburst . over the next @xmath21 days
the luminosity fell to less than a third of its peak value .
note that this local minimum was followed by a @xmath22 day long reflare , which we do not try to address in this paper ( but see [ end ] ) .
our goal here is to apply the ( modified ) dim to the main 1996 outburst of 16 since , as discussed above , this model was successfully used to describe the beginning of this event .
various type of ` reflares ' ( observed also in dwarf novae ) do not have a clear explanation in the context of the dim ( see e.g. hameury , lasota & warner 2000 ) .
one can speculate that the long rise - time of the 16 s reflare could be due to an inside out outburst from a non truncated disc , while its flat - topped lightcurve suggests a quasi steady accretion phase .
however , in view of incomplete optical coverage and uncertainties in models discussed below , we feel that an attempt at serious modeling of the reflare would be pure guesswork .
it is very suggestive that the e - folding rise time of @xmath23 day and the e - folding decay time of @xmath24 days , observed in the first 27 days of the x - ray outburst of 16 , are characteristic of the so - called fred - type light curves seen in many x - ray transients ( see e.g. chen , shrader & livio 1997 ) . the rise and decay ( see figs .
[ ltcrv][b ] and [ ltcrv][c ] ) of the optical and uv luminosity also follow the pattern of a fred outburst with an e - folding decline time close to @xmath25 days , considerably longer than that for x - rays .
this behaviour naturally follows from the dim in which the outer regions of the accretion disc are irradiated by the x - rays from the inner region ( king & ritter 1998 ; dubus , lasota & hameury 1999a ) .
one could conclude , therefore , that 16 was producing a ` normal ' x - ray transient outburst when , on day 27 a new event occurred which stopped the decline of the x - ray flux . in what follows
, we will argue that the change in the outburst pattern was due to an increase in the mass transfer rate , which moved the system parameters into the range corresponding to _ stable _
accretion for an x - ray irradiated disc .
the minimum mass transfer rate necessary for a stable , steady - state accretion in an x - ray irradiated disc may be written as ( dubus et al .
1999b ) @xmath26 here the phenomenological ` parameter ' ( in reality a function of radius and time , see [ opt ] ) @xmath27 provides a simple description of the disc irradiation properties through : @xmath28 where @xmath29 is the irradiation temperature . @xmath30 is the value that represents best the average irradiation flux intercepted by the the outer disc regions in lmxbs ( see dubus et al .
1999b is deduced from applications of this formula to observations . ] ) .
we assumed that in outburst the disc outer radius @xmath31 expands to @xmath32 of the primary roche lobe radius ( smak 1999b ) , so that @xmath33 let us note , however , that if the outward propagating heat front ( hameury et al . 1998 ) did not reach the outermost disc regions , the disc expansion would be negligible , and its outer edge would extend only to @xmath34 of the roche lobe radius .
the critical accretion rate given by eq .
( [ mcrirr ] ) is very uncertain , as discussed in dubus et al . (
1999a ) , and should be considered an order of magnitude estimate only . for the accretion in an irradiated disc to be stable the mass - transfer rate from the secondary has , therefore , to be larger than @xmath35 . for 16
this corresponds to @xmath36 ( @xmath37 ) . since @xmath38 during the plateau phase of the outburst ,
the observations are entirely consistent with a _
stable _ disc accreting at @xmath39 . for our arguments to work , however
, the mass transfer rate from the secondary must be smaller than critical value of @xmath40 quoted above , since otherwise 16 would not be a transient , at least according to the dim .
so what is known about @xmath41 in this system ?
estimates of the mass transfer rate in 16 quoted in the literature differ by three orders of magnitude . at the high end , kolb et al .
( 1997 ) and king & kolb ( 1999 ) argue that the secondary star in 16 is crossing the hertzsprung gap between the main sequence and the red giant branch , and is therefore undergoing a rapid envelope expansion ( on a time scale @xmath42 years ) .
they conclude that this implies a mass transfer rate of order @xmath43 . obviously , taken at face value
, this result suggests that 16 can not be a transient system .
regs et al .
( 1998 ) show , however , that the type f3-f4 donor can still lie on the main sequence if convective overshooting at the core
envelope boundary is taken into account in stellar modeling .
the presence of such ( or equivalent ) mixing is expected for main sequence stars in the relevant mass range ( see regs et al .
1998 and references therein ) . based on this argument , regs et al .
( 1998 ) derive a considerably lower value for the mass transfer rate in 16 , estimating that @xmath44 .
if this estimate is correct , the quiescent accretion disc in 16 is still unstable , though with the mass transfer rate rather close to the critical one . in this case ,
an increase in @xmath41 by a factor of a few during the outburst would stabilize the accretion , and produce the ` plateau ' in the light curve , just as observed during the 1996 outburst .
interestingly , hameury et al . (
1997 ) showed that the spectrum of 16 in quiescence as well as its rise to outburst in 1996 can be very well modeled using the standard dim operating in a truncated disc , assuming that the mass transfer rate is @xmath45 , within the range considered plausible by regs et al .
( 1998 ) .
these relatively high values of @xmath41 are somewhat difficult to reconcile with the estimate of van paradijs ( 1996 ) , who used the standard method of deducing the mass transfer rates in x - ray transients by dividing the mass accreted during the outburst by the recurrence time . by estimating the total emission observed during the 19941995 period of activity and assuming a recurrence time of @xmath46 years van paradijs ( 1996 ) obtained @xmath47 ( see also menou et al .
1999 ) . however , the validity of this method for the recent outburst cycle of 16 is questionable , since the observed outbursts clearly do not occur with even rough regularity . between august 1995 and
april 1996 16 was in true quiescence so the recurrence time for the 1996 outburst was only 9 months .
the relevant mass transfer rate may then be estimated by dividing the mass accreted during the 1996 outburst ( which was roughly @xmath48 , assuming the standard @xmath49 radiative efficiency ) by this recurrence time .
the result , @xmath50 , is in reasonable agreement with convective overshooting models of regs et al .
this , of course , does not resolve the problem of the pre1994 phase , namely that no outbursts were observed during the preceding 30 years .
overall we conclude that there is significant evidence pointing to the mass transfer rate in 16 being rather _ close _ to the stability limit of an irradiated disc , at least during the period directly preceding the 1996 outburst .
furthermore , we can confidently state that the transient nature of this system as well as the shape of the x - ray light curve seen in 1996 can be attributed to the dwarf nova type disc instability only if @xmath41 in quiescence was below , but not too far from the critical mass transfer rate .
the former condition is necessary to have outbursts at all , and the latter is desirable so that a reasonable increase in @xmath41 during the first part of the outburst is sufficient to stabilize the disc .
in addition , the same value of @xmath41 is required to explain all the rise to outburst properties .
an obvious reason for an increase in @xmath41 is x - ray irradiation of the secondary star . according to phillips , shahbaz & podsiadlowski ( 1999 )
, x - ray heating strongly affects the vertical structure of the irradiated outer layers of the secondary . assuming an x - ray luminosity of @xmath51 erg s@xmath4
, they obtain for 16 a ( maximum ) ratio of the irradiating to the intrinsic stellar flux @xmath52 .
they were considering batse ( 20 -200 kev ) x - ray fluxes observed during the march 1995 event , whereas during a typical outburst most of the energy is emitted in softer ( 2 - 20 kev ) x - rays ( sobczak et al . 1999 ) . at the first maximum in the 1996 outburst , the total x - ray luminosity was approximately 8 times larger than the value used by phillips et al .
( 1999 ) , so the flux ratio would instead be @xmath53 and the expected effect considerably larger . thus , an increase of the mass - transfer rate due to irradiation should be expected .
in addition , in a recent seminal article , smak ( 1999a ) showed that mass transfer enhancements play a fundamental role in dwarf nova outbursts ( see [ disc ] ) and therefore they should also be expected in x ray transients .
as pointed out by phillips et al .
( 1999 ) x - ray irradiation does not directly affect the vicinity of the @xmath54 point but heats up matter located higher in altitude .
this will induce a delay , because the heated matter has to move to the roche nozzle before falling onto the accreting object .
such a delay was observed in dwarf novae ; when the heating of the secondary by uv and optical emission is clearly observed , brightening of the hot spot due to the increased mass transfer appears a few days after the outburst maximum ( smak 1995 ) .
if the increased mass - transfer rate is due to irradiation , this delay must be shorter than the disc viscous time , in order to keep @xmath6 approximately constant . in the opposite case
, the light - curve would instead have an exponential shape ( hameury , king & lasota 1988 ; augusteijn , kuulkers & shaham 1993 ; hameury , lasota & hur 1997 ) .
the disc viscous time is @xmath55 using the expression for midplane temperature given for the ` hot ' shakura & sunyaev ( 1973 ) solution . at 16 s outer disc radius this time is longer than the total duration of the 1996 activity ( @xmath56 days ) , at least for values of @xmath57 - 0.2 assumed in the dim . in our picture , irradiation of the secondary at the time of the first maximum in the x - ray light curve , would increase the mass transfer rate from the companion . to be consistent with observations , this increase should affect the accretion rate onto the black hole after about two weeks , a considerably shorter time period than @xmath58 above . in this case , however , the relevant time scale is not that given by eq .
( [ tvis ] ) , but instead the time it takes for a surface density front , created by a sudden increase in mass transfer , to diffuse towards the inner disc regions . in the presence of a warp , matter transfered from the companion
is fed to the disc not at the outer edge but at radius @xmath59 ( see e.g. wijers & pringle 1999 ) , where @xmath60 is the so called circularization radius corresponding to the specific angular momentum of the mass transferred through the @xmath54 lagrangian point .
the time in which this surface density excess will reach the black hole is then given by ( hameury et al .
1997 ) @xmath61 where @xmath62 is the width of the surface density contrast .
a density excess with @xmath63 ( see e.g. frank et al .
1987 ) would diffuse to the central black hole in @xmath64 days in agreement with our scenario . as we pointed out above , this number should be considerably longer than the expected delay between irradiation of the secondary and the increase in mass transfer , as required to produce a flat light - curve .
although it is very difficult to estimate how long it takes for the heated matter to reach the @xmath54 point ( see hameury et al .
1993 , for discussion of a related problem ) , a time of @xmath65 few days is reasonable in the sense that it would imply subsonic speeds .
our explanation of the ` flat top ' x - ray light - curve of 16 is very similar to the one proposed by king & cannizzo ( 1998 ) for light - curves of z cam - type dwarf - nova systems ( see their fig .
2 ) . these authors do not invoke irradiation to explain the increased mass - transfer rate , however , but attribute mass transfer variations to a starspot .
interestingly , the king & cannizzo ( 1998 ) model is more successful in our case than it is in explaining z cam ` standstills ' . in z cam
the luminosity is observed to stick at a constant level halfway down from maximum , whereas in light curves produced by king & cannizzo the the ` standstill ' is at a luminosity _ higher _ than the outburst maximum , as in 16 .
king & cannizzo expect this to be the result of the high mass - transfer enhancement factor ( 6 ) used in their calculation . in reality , however , the mass - transfer rate corresponding to stable disc accretion is , by construction , _ always _ higher than the maximum outburst mass accretion rate .
the reason ( see e.g. hameury et al . 1998 ) is that at outburst maximum , before the cooling wave begins to propagate , the accretion rate in the disc is almost constant , i.e. at the outer disc edge ( or at the hot disc outer edge ) it is close to the accretion rate corresponding to the critical surface density ( @xmath66 marking the end of the ` hot ' branch of the s - curve representing disc equilibria )
. a stable mass - transfer rate must be larger than this value .
there is now a general consensus that optical emission in persistent and transient ( in outburst ) low - mass x - ray binaries is due to reprocessing of x - rays by the outer disc ( see e.g. van paradijs & mcclintock 1995 ) .
this is also the case for 16 , since a simple estimate of the expected optical flux from a non - irradiated disc which reproduces the observed x - ray emission falls short of the observed optical flux by roughly an order of magnitude ( see fig .
[ spectra ] ) .
moreover , as we pointed out in [ xray ] , the optical light - curve is entirely consistent with the decline from maximum during a typical fred - type outburst in an irradiated disc .
longer uv and optical decline time scales ( @xmath25 days , as compared to @xmath24 days for the x - ray flux ) are simply due to the fact that as the irradiating flux declines , the outer disc edge becomes cooler and the peak of the emission moves into the optical band , thus compensating for the decrease in the total emission from the outer disc .
however , this simple picture clearly can not explain the behaviour of the optical flux at times later than @xmath67 days after the onset of x - ray outburst . at this time ,
x - ray flux begins to increase , while the optical and uv emission continues to decline .
here we argue how the scenario for the x - ray emission of 16 described in [ xray ] can reconcile these observations with the x - ray reprocessing origin of the optical emission . as shown by dubus et al .
( 1999b ) the outer regions of a planar accretion disc can not intercept the x - rays emitted by a point source located at the midplane .
therefore , in order for the outer disc to be irradiated , it must be warped ( the other possible way for the outer disc to see the x - rays - an extended irradiating source - fails to explain why the outer disc is _ effectively _ geometrically thick while the vertical equilibrium implies very thin discs ) .
the origin and propagation of warps in accretion discs is still an open question ( see e.g. pringle 1999 ) .
here we consider two possible regimes : one with low viscosity , when the warp propagation relies on sound waves ; and another with high viscosity , when the warp evolution is driven by diffusion .
as we shall see , in both cases the warp amplitude in 16 is likely to decay during an outburst . if the viscosity is low , i.e. if @xmath68 , where @xmath69 is the half thickness of the disc , the warp can propagate as a non dispersive wave at approximately the speed of sound ( papaloizou & lin 1995 )
. this regime can be relevant if the outer disc regions are not affected by the propagation of the heat front during the outburst . in the standard dim the heat
front passage changes the viscosity parameter @xmath70 from a low , cold ( @xmath71 ) to a high , hot ( @xmath72 ) value .
therefore , if the heat front does not reach the outer disc it is conceivable ( since we do nt know the physical mechanism supposedly responsible for the change of @xmath70 ) that in the outer regions we would still have @xmath73 , while @xmath74 ( see e.g. dubus et al .
in such a case the increased stream of mass transfered from the secondary deposits matter moving in the orbital plane at @xmath59 . since various parts of the disc
communicate efficiently through sound waves , this new component will exert a torque on the outer disc reducing the warp on a time scale of a few forced precession periods , where the precession period is @xmath75 days ( larwood 1998 )
. on the other hand , the whole disc could be in a high @xmath70 state during an outburst .
since in a standard accretion disc @xmath76 in the outer regions , @xmath77 and the warp propagation is driven by viscous processes .
the relevant viscosity is the one corresponding to the vertical shear .
the ratio of this kinematic viscosity coefficient to the standard ( radial ) one is approximately @xmath78 , for @xmath79 ( papaloizou & pringle 1983 ) , and therefore the warp damping time is @xmath80 . at the outer disc edge of 16
, @xmath81 is then about @xmath82 days for @xmath83 , in very good agreement with our scenario .
note that in this regime the increase of mass transfer would not affect the warp .
pringle ( 1996 ) found that warp can be radiation driven . in such a case the warp s viscous decay could be prevented by irradiation .
for this to happen , the growth rate of the radiative instability must be shorter than the viscous damping time @xmath81
. this condition can be written as ( e.g. wijers & pringle 1999 ) : @xmath84 where @xmath85 is the accretion efficiency and @xmath86 is the critical ratio of the radiative growth to the viscous damping times .
one can see that for 16 the inequality above can be satisfied only for @xmath70 values higher than the ones usually assumed in the dim . unless such values are assumed a warp will be viscously damped on a time
scale estimated above .
whatever the mechanism of warp decay , it would result in the reduction of the irradiating flux intercepted by the disc , and a consequent decrease in the observed optical flux from the system . to illustrate this argument we calculated a series of optical spectra from uniformly accreting thin discs with varying degree of irradiation .
the value of the mass accretion rate was chosen to reproduce the black body component of x - ray emission ( sobczak et al .
1999 ) . in our simple treatment here
we specify neither the origin nor the structure of the warp . however , since we need some description of the photosphere of the disc above the orbital plane as a function of radius , we use the prescription , @xmath87 chosen by analogy with formulae used in the literature , which ( despite being based on an incorrect assumption about irradiated discs ) seem to give a correct empirical description of the reprocessed x ray flux ( see dubus et al .
all other properties of the warped disc , as far as irradiation is concerned , are described by @xmath88 , defined in eq .
( [ tirr2 ] ) .
we use the prescription for irradiation of the outer disc by the inner disc edge ( e.g. see shakura & sunyaev 1973 ; king & ritter 1998 ) , which combined with eq . ( [ z ] ) above gives @xmath89 where @xmath90 . note that for the disc
disc irradiation geometry , the strength of irradiation is quadratic in @xmath91 , since it depends on the projections of both the emitting and irradiated annuli .
of course the exact dependence of @xmath92 on the disc radius is important in determining the shape of the optical spectrum .
however , there are many other highly uncertain quantities in the calculation ( e.g. radial profile of the albedo in the outer disc , angular distribution of the irradiating flux , details of radiative transfer in the atmospheres of irradiated discs ) which all contribute significantly to the appearance of the disc in the optical band .
in addition , we are using a flared planar disc ` approximation ' to describe a warped disc .
since all these uncertainties are hidden inside our parameter @xmath93 , the exact choice of @xmath94 is not very important .
as far as we are concerned , eqs .
( [ z ] ) and ( [ c ] ) simply describe the distribution of the irradiating flux with radius and _ do not _
result from some assumed vertical disc structure .
note especially that we _ do not _ assume that the irradiated disc is isothermal or adopt a particular value for the disc aspect ratio at the outer edge .
the resulting spectra computed for different values of @xmath95 are shown in figure [ spectra ] .
for comparison we also plotted the dereddened spectra of 16 observed ( in order of decreasing flux ) on may 14 , june 8 , june 20 , june 30 , and july 22 , 1996 with _ hst_/fos red prism .
( note that these spectra correspond to the optical and uv data points shown on figure [ ltcrv ] . )
since the contribution from the secondary is quite significant , especially at later times , we have subtracted from the data an estimate of the quiescent optical spectrum of 16 , adjusted according to orbital phase ( see hynes et al .
1998 and hynes 1999 for a fuller description of data processing ) .
the broad hump centered at @xmath96 which remains after subtraction in the last two spectra has a similar spectrum to the secondary , and we suspect is probably due to a residual contribution from the companion .
this suggests brightening of the secondary in outburst due to x - ray heating , supporting our irradiation scenario . though the model spectra are not intended as a formal fit to the data because of many simplifications in our calculation , the agreement between the five sets of spectra is fairly good .
figure [ spectra ] shows qualitatively that by decreasing the degree of warping ( as described by @xmath93 ) , and therefore irradiation , of the outer disc , we can mimic the observed evolution of 16 in the optical band .
note how with decreasing @xmath95 the model optical spectra become softer , just as observed , even allowing for some residual contribution from the secondary in the data .
one should keep in mind that all model spectra shown in figure [ spectra ] were computed for a planar disc with @xmath97 and a fixed outer radius , given by eq .
( [ rdisc ] ) .
this means varying @xmath98 in our model corresponds simply to different values of the photospheric height at the outer radius .
a more realistic description of the effects caused by enhanced mass transfer should , of course , include both changes in the disc profile ( different functional form for @xmath94 ) as well as possible changes in the disc outer radius .
we feel that as long as we use a flared planar disc to approximate the effects of the warp , the exact disc shape is beyond the scope of this paper .
similarly , it is difficult to calculate the change in the outer radius of the disc from first principles .
however , observations and numerical simulations of dwarf nova discs ( e.g. smak 1984 ; 1999b ) show that at the onset of the outburst @xmath31 expands by about @xmath19 ( from the canonical @xmath99 to nearly @xmath32 of the roche radius ) for a constant mass transfer rate from the companion . by comparison ,
simulations with enhanced mass transfer ( say by a factor of 30 ) simply decrease this value to @xmath100 , so the difference is too small to be constrained by the data .
because of many uncertainties in the physics of irradiation , it is very difficult to infer the time scale of the warp decay from the optical data . in both regimes considered in the previous section , our estimates of the decay time ( a few forced precession times in the low viscosity case [ papaloizou & terquem 1995 ; larwood 1998 ] or the viscous damping time in a high viscosity case ) are of the same order as the duration of the plateau phase of the outburst .
it is therefore very tempting to speculate that the decay of the warp is not only responsible for the decline in the observed optical emission , but can also , by reducing heating through irradiation , terminate the outburst .
even if the mass transfer rate from the secondary remains high , a decrease in the irradiating flux can cause an increase in the minimum rate required for stable accretion ( see eq .
[ mcrirr ] ) which is sufficient to bring the disc back into the unstable regime , and therefore , partially shut off the accretion in the inner disc where x - rays are produced .
however , since the disc will most likely remain close to the stability limit , a very short period of mass accumulation at the outer edge is required to trigger the next outburst , just as was observed for 16 in 1997 ( e.g. sobczak et al .
we have shown that x - ray and optical observations of 16 during its 1996 outburst can be reconciled with the standard disc instability scenario under the assumptions that the outer accretion disc is warped and that the primary ` standard ' outburst triggers a burst of mass transfer from the secondary . in our picture , the enhancement in the mass transfer rate was sufficient to create a period of stable accretion , which was observed as a @xmath101 month plateau in the x - ray light curve .
we have argued that this is entirely plausible , since available evidence is not inconsistent with the mass transfer rate in 16 being close to the stability limit of an irradiated disc .
thus a rather small enhancement in @xmath41 is required to stabilize it .
in addition , a burst of mass from the secondary is likely to decrease the warp amplitude in the outer disc .
this effect explains why the optical light curve showed a decline as the x - ray flux was increasing , and can also account for the spectral evolution of the optical flux .
warped disc physics is rather complicated and calculating the thermal viscous instability in a warped disc ( not to mention the effects of the enhanced mass transfer ) is out of the question , at present . because of this , the arguments presented in this paper are based on the calculations for planar discs and are necessarily of a qualitative nature .
a lot more work is necessary before we can venture to make detailed quantitative comparisons between model predictions and what is observed .
it is also important to emphasize that although the overall ratio of x - ray to optical flux ( determined by the value of @xmath88 at the outer disc edge ) is well constrained by observations , the detailed optical spectra of irradiated discs depend on many unknown parameters .
while the angular distribution of irradiating flux may in some cases be deduced from detailed modeling of x - ray spectra , the exact shape of the disc as well as the x - ray albedo can only be determined from future theoretical calculations . despite these uncertainties
, we believe that the picture presented here is realistic and is consistent with all the available data on the 1996 outburst of 16 .
one should perhaps view our result as a test of the hypothesis that this outburst was due to a dwarf nova type disc instability .
it is important to note that in our arguments we rely on only two major assumptions , which are both supported by extensive circumstantial evidence .
disc warping emerges as a natural conclusion from studies of x - ray irradiated outer discs , and a mass transfer increase during the outburst seems equally plausible when we consider irradiation of the secondary .
similar modifications are also required in the application of the dim to standard ( u gem type ) dwarf novae . in these systems
there is clear observational evidence for irradiation
induced enhancement in the rate of mass transfer from the secondary ( e.g. smak 1995 ) .
smak ( 1999a ) studied the two types of ` inside out ' outbursts which are observed in dwarf novae : the ` narrow ' and ` wide ' ones .
he showed , that their properties ( various widths but similar amplitudes ) can be explained if during the outburst the mass transfer rate is increased .
narrow outbursts correspond to moderate ( factor 2 ) enhancements whereas wide outbursts correspond to major enhancements which temporarily put the accretion disc into the stable regime .
it is , therefore , reasonable to expect that the variety of light curve types observed in x ray transient systems may be due , at least in part , to mass transfer variations .
16 is an exceptional system but is not the only one showing a ` flat top ' outburst light curve ( see e.g. chen et al .
one may hope , therefore , that such systems could be observed in a not too distant future .
multi - wavelength observations of such a transient event , in particular the correlation between x - ray and optical fluxes , could provide a useful test of our scenario .
at least one feature of the observed behaviour of 16 still remains a puzzle .
if the mass transfer rate in quiescence is always close to the stability limit , it is unclear how this system could have sustained a @xmath46 years quiescence period . at such a high rate of accretion
, the mass accumulation period at the outer edge of the disc would be very short ( on the order of a few months to a year ) , with the successive outbursts following each other after a correspondingly short time .
in fact , the behaviour observed in this system since 1994 should be the norm , rather than the exception .
one possible explanation can be that during some periods accretion in 16 is non conservative , e.g. king & kolb ( 1999 ) propose that a lot of matter can be lost into a jet . however , more theoretical work as well as observations are required before we can assemble a comprehensive picture of this fascinating system .
we thank caroline terquem for enlightening discussions and advice and phil charles for carefully reading the manuscript .
we thank the third referee of this article ( first for a&a ) for a useful , critical and professional report .
we are grateful to greg sobczak for the x - ray light - curve data .
this research was supported in part by the national science foundation grant no .
phy94 - 07194 , and by nasa through chandra postdoctoral fellowship grant # pf8 - 10002 awarded by the chandra x - ray center , which is operated by the smithsonian astrophysical observatory for nasa under contract nas8 - 39073 .
_ hst _ data was supported by nasa through grant number go-6017 - 01 - 94a from the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
we thank carole haswell for providing these data . | we show that the 1996 outburst of the x - ray binary transient system 16 can be explained by the standard dwarf - nova type disc instability , followed by an episode of enhanced mass transfer from the secondary if the mass transfer rate in 16 is within a factor @xmath0 of the stability limit .
we argue that irradiation of the secondary during the onset of the outburst driven by the thermal instability in the outer disc can increase the mass transfer rate above the minimum value required for stable accretion .
this will then produce the period of near - constant x - ray emission seen in this system .
this scenario can also explain the observed anti - correlation between the optical and x - ray fluxes .
it is generally accepted that optical emission in low - mass x - ray binaries is produced by irradiation of the outer disc by x - rays .
there is also strong circumstantial evidence that in order for the outer disc to see the irradiating flux , it must be warped .
depending on the warp propagation mechanism , either a burst of mass from the secondary or viscous decay are likely to decrease the degree of warping , thereby causing the decrease in the observed optical flux while the x - ray flux remains constant or even increases , exactly as observed in 16 . finally , the decrease of the disc warping and , therefore , irradiation will cause the disc to become unstable once again , terminating the outburst .
# 1a&a , # 1 # 1acta astr . , # 1 # 1a&as , # 1 # 1ara&a , # 1 # 1aj , # 1 # 1apj , # 1 # 1apjs , # 1 # 1mnras , # 1 # 1nature , # 1 # 1pasj , # 1 # 1pasp , # 1 # 1phys .
lett .
a. , # 1 # 1prog .
theor .
phys . , # 1 # 1science , # 1 # 1solar phys . ,
# 1 # 1space sci .
rev .
, # 1 16gro j165540 |
the possible discovery of an accelerating universe from observations of type ia supernovae ( perlmutter 1998 , riess 1999 ) has led to a resurgence of interest in the possibility that the universe is dominated by a cosmological constant ( for a recent review see turner 1999 ) .
a number of authors have shown how observations of distant type ia supernovae ( sn ) can be combined with observations of cmb anistropies to constrain the cosmological constant and matter density of the universe ( white 1998 , tegmark 1998 , lineweaver 1998 , garnavich , 1998 , efstathiou and bond 1999 , tegmark 1999 , efstathiou 1999 ) .
for example , efstathiou ( 1999 , hereafter e99 ) combine the large sn sample of the supernova cosmology project ( perlmutter 1998 , hereafter p98 ; we will refer to these supernovae as the scp sample ) with a compilation of cmb anisotropy measurements and find @xmath13 and @xmath14 ( @xmath15 confidence errors ) for the cosmic densities contributed by matter and a cosmological constant respectively .
these results are consistent with a number of other measurements , including dynamical measurements of @xmath0 , the large - scale clustering of galaxies and the abundances of rich clusters of galaxies ( turner 1999 , bridle 1999 , wang 1999 ) . 3.0 truein the observational evidence for an accelerating universe has stimulated interest in more general models containing a component with an arbitrary equation of state , @xmath16 with @xmath17 .
examples include a dynamically evolving scalar field ( see _ e.g. _ ratra and peebles 1988 and caldwell , dave and steinhardt 1998 , who have dubbed such a component ` quintessence ' ; we will refer to this as a ` q ' component hereafter ) and a frustrated network of topological defect ( spergel and pen 1997 , bucher and spergel 1999 ) .
in particular , steinhardt , wang and zlatev , 1998 , have pointed out that for a wide class of potentials , the evolution of a q - like scalar field follows ` tracking solutions ' , in which the late time evolution is almost independent of initial conditions .
the purpose of this paper is three - fold .
firstly , to illustrate how the constraints on @xmath0 and @xmath1 can be improved by extending the redshift range of the supernovae samples . at low redshifts ,
the magnitude - redshift relation is degenerate for models with the same value of the decellaration parameter @xmath18 ( @xmath19 ) .
this degeneracy can be broken by observing supernovae at redshifts @xmath20 ( see , for example , goobar and perlmutter , 1995 ) .
thus , by extending the redshift range of the current supernovae samples it should be possible to set tighter limits on @xmath0 and @xmath1 independently .
this is important because there are significant worries that the sn data may be affected by grey extinction , evolution , or some other systematic effect .
the consistency of sn constraints on @xmath0 and @xmath1 with those derived from the cmb anisotropy measurements would provide an important consistency check of systematic errors in the sn data and the interpretation of the cmb data .
secondly , we estimate the accuracy with which a more general q - like equation of state can be constrained by high redshift sn and cmb data .
thirdly , we use the current sn and cmb anisotropy data to constrain q - like models in a spatially flat universe and in a universe with arbitrary spatial curvature .
the predicted peak magnitude - redshift relation is given by @xmath22 where @xmath23 is related to the peak absolute magnitude by @xmath24 . and
@xmath25 is the hubble constant - free luminosity distance . to compute the luminosity distance , we ignore gravitational lensing and use the standard expression for a universe with uniform density
( see _ e.g. _ peebles 1993 ) , @xmath26,\ ] ] @xmath27^{1/2 } } \quad ( 2)\ ] ] where @xmath28 and @xmath29 if @xmath30 and @xmath31 for @xmath32 .
[ tab1 ] * table 1 : fisher matrix errors , @xmath0 and @xmath1 . * & scp & & + @xmath33 & & @xmath34 & 1.5 & 1.0 & 1.5 + @xmath35 & @xmath36 & @xmath37 & @xmath38 & @xmath39 & @xmath40 + @xmath41 & @xmath42 & @xmath43 & @xmath44 & @xmath45 & @xmath46 + @xmath47 & @xmath48 & @xmath49 & @xmath50 & @xmath51 & @xmath51 + + & scp & & + @xmath33 & & @xmath34 & 1.5 & 1.0 & 1.5 + @xmath35 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & @xmath56 + @xmath41 & @xmath57 & @xmath58 & @xmath59 & @xmath60 & @xmath61 + @xmath47 & @xmath62 & @xmath60 & @xmath55 & @xmath63 & @xmath64 + we assume that we observe @xmath65 supernovae , with peak magnitude @xmath66 , ( corrected for k - term , decline rate - luminosity relation , reddening _ etc _ ) , magnitude error @xmath67 and redshift @xmath68 , from which we want to determine a set of parameters @xmath69 by maximising the likelihood function , @xmath70 in this section we assume that the parameters @xmath69 are @xmath0 , @xmath1 and @xmath23 ( defined in equation 1 ) .
an estimate of the covariance matrix , @xmath71 , for these parameters for a given sn data set is given by the inverse of the fisher matrix @xmath72 ( kendall and stewart 1979 ) .
the marginalized error on each parameter ( given by @xmath73 ) is listed in table 1 for several assumed supernova datasets .
the column labelled scp gives the fisher matrix errors on @xmath0 , @xmath1 and @xmath74 derived for sample c ( @xmath75 supernovae ) of p98 , _ i.e. _ assuming the magnitude errors , intrinsic magnitude scatter and redshift distribution of the real sample .
the next two columns give the expected errors for the scp sample supplemented by 20 supernovae with a peak magnitude error of @xmath76 magnitudes and a gaussian redshift distribution of dispersion @xmath77 and mean redshift @xmath78 and @xmath79 .
the upper redshift limit is close to the maximum for feasible spectroscopic measurements with @xmath80 metre - class telescopes ( see goobar and perlmutter 1995 ) .
as these authors comment , ground based spectroscopy at optical wavelengths becomes prohibitively expensive for supernovae at higher redshifts because of the strong k - correction .
the last two columns give the errors for a sample twice as large as the scp sample supplemented by 40 supernovae with mean redshift of @xmath34 and @xmath79 .
we adopt a background cosmology with @xmath81 and @xmath82 as indicated by the joint likelihood analysis of the scp sample and cmb anisotropies described in e99 . from table 1
we see that the fisher matrix analysis of the scp sample gives relatively large errors on @xmath1 and @xmath0 , in agreement with the likelihood analysis presented by p98 . however , by adding @xmath83 sn at @xmath21 , the errors on @xmath1 and in particularly @xmath0 are reduced significantly .
the last column shows that an enhanced scp sample together with @xmath84 sn at @xmath85 ( a formidable , but feasible observing programme ) can provide a tight constraint on @xmath0 .
the parameters @xmath1 and @xmath0 are , of course , highly correlated .
this is illustrated in figure 1 which shows @xmath86 , @xmath87 and @xmath88 error ellipses in the @xmath1-@xmath0 plane after marginalizing over @xmath89 assuming a uniform prior distribution .
( the components of the new fisher matrix after marginalization are given by @xmath90 , @xmath91 , @xmath92 . )
the points in the figure show the results of monte - carlo calculations , where we have simulated the observational samples and determined the parameters @xmath93 by maximising the likelihood function ( equation 2 ) . by diagonalizing the matrix @xmath94 we can find the orthogonal linear combinations @xmath95 and @xmath96 defining the major and minor axes of the likelihood contours shown in figure 1 .
the distributions in these orthogonal directions are shown in figure 2 and compared with the distributions determined from the monte - carlo simulations .
the monte - carlo distributions are very close to gaussians and show that the fisher matrix gives an extremely accurate description of the errors in the @xmath0@xmath1 plane .
5.4 truein although the errors in @xmath0 and @xmath1 are significantly reduced by the addition of high redshift supernovae over those of the scp sample , they are still quite large in the parallel direction @xmath97 .
this means that it is difficult to set tight limits on @xmath1 from sn measurements alone .
the constraints on the spatial curvature @xmath98 are even weaker . for example , for the larger sample shown in figures 1 and 2 , the @xmath99 error on @xmath98
is @xmath100 .
this can be reduced by extending the range to even higher redshifts ( see section 2.2 ) or by combining the sn data with cosmic microwave background anisotropies , as has been done by several authors ( white 1998 , lineweaver 1998 , garnavich 1998 , tegmark 1999 , e99 ) .
cmb anisotropy measurements , especially with future satellites such as map and planck , are capable of setting tight constraints on the locations of the acoustic peaks in the cmb power spectrum .
following e99 , we define an acoustic peak location parameter @xmath101 to be the ratio of the peak position in a model with arbitrary cosmology compared to that in a spatially flat model with zero cosmological constant .
( this parameter depends weakly on the matter content of the universe and on the spectral index of the fluctuations , but we ignore these small dependences in what follows ) .
cmb measurements are therefore capable of fixing @xmath102 , defining a degeneracy direction in the @xmath1@xmath0 plane given by @xmath103 ( see efstathiou and bond 1998 ) .
the results in the lower panel of table 1 show the fisher matrix analysis of the sn samples including the constraint imposed by equation ( 5 ) .
as is well known , the combination of sn and cmb measurements can break the degeneracy between @xmath1 and @xmath0 and it should be possible to determine these parameters with an error of less than @xmath104 with an enlarged supernova sample assuming , of course , that systematic errors are unimportant .
although the errors on @xmath1 from sn measurements alone converge relatively slowly as the redshift range is increased , consistency of the cosmological parameter estimates provides a strong test of systematic errors in the sn data . if we believe that systematic errors are unimportant , and that our interpretation of the cmb anisotropies ( in terms of adiabatic cdm - like models ) are correct , then current data already constrain @xmath0 and @xmath1 to high precision ( see fig 5 of e99 ) .
consistency requires that the likelihood contours for a high redshift supernova sample converge to the same answer .
3.0 truein observations of very distant supernovae at @xmath105 may be possible with a next generation space telescope ( _ e.g. _ miralda - escude and rees 1997 , madau 1998 , livio 1999 ) .
we will not analyse the feasibility of such observations here .
rather , we note from figures 1 and 2 that the major axis of the error ellipses in the @xmath1 @xmath0 tilt and become more vertical as the redshift range of the sn sample is increased .
this is because the magnitude redshift relations for models with very different values values of @xmath1 and the same @xmath0 converge at higher redshifts .
the convergence redshift depends on @xmath0 and lies between @xmath106@xmath107 for @xmath0 in the range @xmath108@xmath86 ( see figure 1 of melnick , terlevich and terlevich , 1999 ) .
this is illustrated by figure 3 , which shows the @xmath86 , @xmath87 and @xmath109 likelihood contours determined from the fisher matrix for a sample consisting of twice the scp sample , @xmath110 sn with @xmath111 , @xmath112 , and @xmath84 sn with @xmath113 , @xmath114 . as expected , these contours are almost vertical in the @xmath0@xmath1 plane . a sample of supernovae ( or some other distance indicator such as hii galaxies , melnick 1999 ) at redshifts @xmath115 can therefore produce a tight constraint on @xmath0 independently of the value of @xmath1 .
in this section , we analyse the constraints that sn can place on an arbitrary equation of state . we first consider a constant equation of state .
models of this type ( see bucher and spergel ) include a frustrated network of cosmic strings ( @xmath116 ) and a frustrated network of domain walls ( @xmath117 ) .
a constant equation of state is also a good approximation to a q component obeying tracker solutions .
tracker solutions are discussed in section 3.2 .
constraints on generalised forms of dark matter with anisotropic stress are discussed by hu ( 1999 ) and will not be considered here . 3.0 truein
if we include a q - like component with equation of state @xmath16 , the expression for the term @xmath118 in the luminosity distance ( equation 2 ) is modified to @xmath119^{1/2 } } .
\end{aligned}\ ] ] the addition of the parameter @xmath8 means that it is not possible to constrain all of the parameters @xmath0 , @xmath120 , @xmath8 to high accuracy from the supernova data alone ( see section 4.2 ) .
thus , garnavich ( 1998 ) analyse the high - z supernovae search ( hzs ) sample ( riess 1999 ) assuming a spatially flat universe and find that @xmath121 at @xmath122 confidence . a similar analysis of the scp sample by perlmutter , turner and white ( 1999 ) yields @xmath123 .
table 2 lists the results of a fisher matrix analysis for a q - like component with a constant @xmath8 .
here we have applied the constraints @xmath124 and @xmath125 .
the upper table gives results for the supernovae magnitude - redshift relation alone assuming a spatially flat universe with @xmath126 and @xmath127 .
the constraints on @xmath8 from a sample such as the scp data are quite poor and improve relatively slowly as the sample is extended to higher redshift because of a strong degeneracy between @xmath8 and @xmath0 in the magnitude - redshift relation .
this is illustrated in figure 4 , which shows the analogue of figure 2 for q - like models . as the supernovae sample is extended to higher redshift , the likelihood contours narrow but @xmath8 and @xmath0 remain strongly degenerate .
[ tab2 ] * table 2 : fisher matrix errors , @xmath0 , @xmath120 and @xmath8 . * & scp & & + @xmath33 & & @xmath34 & 1.5 & 1.0 & 1.5 + @xmath35 & @xmath128 & @xmath129 & @xmath130 & @xmath130 & @xmath52 + @xmath131 & @xmath132 & @xmath133 & @xmath134 & @xmath135 & @xmath136 + @xmath47 & @xmath59 & @xmath59 & @xmath59 & @xmath137 & @xmath61 + + & scp & & + @xmath33 & & @xmath34 & 1.5 & 1.0 & 1.5 + @xmath35 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 + @xmath131 & @xmath143 & @xmath144 & @xmath38 & @xmath145 & @xmath40 + @xmath47 & @xmath146 & @xmath62 & @xmath147 & @xmath148 & @xmath148 + + & scp & & + @xmath33 & & @xmath34 & 1.5 & 1.0 & 1.5 + @xmath35 & @xmath128 & @xmath129 & @xmath149 & @xmath150 & @xmath151 + @xmath152 & @xmath143 & @xmath153 & @xmath154 & @xmath155 & @xmath146 + @xmath131 & @xmath156 & @xmath156 & @xmath157 & @xmath136 & @xmath158 + @xmath47 & @xmath59 & @xmath59 & @xmath59 & @xmath61 & @xmath61 + the situation is dramatically improved by the addition of constraints from cmb anisotropies .
the addition of a q - like component affects the location of the doppler peaks ( see caldwell 1998 , white 1998 ) and , in analogy with equation ( 5 ) , an accurate determination of the cmb power spectrum imposes the constraint @xmath159 the second panel of table 2 shows the constraints derived on an arbitrary equation of state by combining supernovae data with the cmb constraint of equation ( 7 ) . for spatially flat models ,
the combination of sn and cmb anisotropies constrains @xmath8 to an accuracy of better than @xmath160 , sufficient to set tight constraints on the physical parameters of q - like models ( for example , whether one requires contrived potentials , see section 4 ) .
however , the constraints on @xmath8 improve relatively slowly as the sn sample is extended to higher redshift .
similar conclusions apply if the assumption of a spatially flat universe is relaxed ( see the lower panel of table 2 ) . in that case , the parameters @xmath0 and @xmath120 can be determined to high precision , but the constraints on @xmath8 improve slowly as the sn sample is increased .
this implies that it is worth analysing the constraints on q - like models with arbitrary spatial curvature using current sn and cmb data ( see section 4.2 ) .
5.4 truein 5.4 truein in the previous section we have investigated the simplified case of a constant @xmath8 . if , in fact , the q - like component arises from a slowly rolling scalar field evolving in a potential @xmath161 , the equation of state of the @xmath162 component will vary as a function of time .
the equations of motion of the @xmath162 field can be written in the following compact form ( steinhardt , wang and zlatev , 1998 ) @xmath163 where primes denote derivatives with respect to @xmath162 , @xmath164 , @xmath165 and @xmath166 is the scale factor of the cosmological model . for a wide class of potentials , and almost independently of the initial conditions , the evolution of @xmath162 locks on to a tracking solution in which @xmath162 and @xmath8 vary slowly ( see zlatev , wang and steinhardt 1998 , steinhardt 1998 ) .
examples of the evolution of @xmath8 and @xmath120 at late times are shown in figure 5 for three forms of the potential @xmath161 . in each case , the evolution of @xmath8 at @xmath167 is well approximated by @xmath168 where @xmath169 is a small number determined from the value of @xmath170 at the present time .
figure 7 shows the relations between @xmath120 , @xmath8 and @xmath169 at the present time derived from the solutions to equation 4 for the three potentials considered in figure 5 .
the minimum value of @xmath169 is about @xmath171 , reflecting the fact that @xmath162 is evolving relatively slowly even at late times . with the approximation of equation ( 9 )
, the energy density of the @xmath162 component evolves according to @xmath172 ^ 2 \right).\ ] ] note also that with the approximation of equation ( 9 ) , the tracker equation ( 8) becomes an algebraic equation relating @xmath173 to @xmath8 , @xmath120 and @xmath169 ( @xmath174 in the matter dominated era ) . a small value of @xmath175 to @xmath176
can not be determined accurately from sn and cmb observations because it is highly degenerate with @xmath8 and @xmath0 . as we will show in the next section , the introduction of the parameter @xmath169 provides a convenient way of testing the sensitivity of constraints on q - like models to the time evolution of @xmath8 . 3.0 truein we note that huterer and turner ( 1998 ) have recently proposed a prescription for reconstructing the potential of a @xmath162-like component directly from the magnitude - redshift relation of type ia supernovae .
this approach may produce interesting constraints if the field @xmath162 is rapidly evolving at late times . for tracker solutions , however , the equation of state changes so slowly that it would be difficult to distinguish the true potential from a perfectly flat one .
in this section , we use current sn and cmb data to constrain the equation of state of the universe .
the analysis closely follows that presented in e99 .
we use the sample of @xmath75 type ia sn of fit c of p98 and adopt the likelihood analysis described by e99 ( including a parametric fit to the luminosity - decline rate correlation ) , modifying the expression for luminosity distance to incorporate the parameters of the q - like model .
the cmb data that we use are plotted in figure 1 of e99 .
we perform a likelihood analysis for these data assuming scalar adiabatic perturbations , varying the amplitude of the fluctuation spectrum , the scalar spectral index , the physical densities of the cdm and baryons @xmath177 , @xmath178 , is the hubble constant in units of @xmath179 . ] and the doppler peak location parameter @xmath102 .
modifications to the cmb power spectrum arising from spatial fluctuations in the q component are ignored as these are negligible in the slowly evolving @xmath162 models considered here ( see caldwell 1998 , huey 1998 ) .
we integrate over the cmb likelihood assuming uniform prior distributions of the parameters to compute a marginalized likelihood for @xmath102 as described in e99 .
the likelihood functions for the parameters @xmath8 , @xmath0 and @xmath120 presented below are constructed from the expression for the angular diameter distance to the last scattering surface and the probability distribution of @xmath102 .
figure 7 shows the constraints on @xmath8 and @xmath0 for spatially flat universes .
the different line types show the constraints for three different values of the parameter @xmath169 characterising the evolution of @xmath8 , @xmath180 ( solid lines ) , @xmath181 ( dotted lines ) and @xmath182 ( dashed lines ) .
as described in the previous section , these values span the range found for tracker solutions for a variety of potentials .
these rates of evolution are so low that they have very little effect on the likelihood contours .
the constraints plotted in figure 7 are in very good agreement with those derived by garnavich ( 1998 ) from an analysis of the hzs sample , and with the analysis of the scp sample ( perlmutter , turner and white 1999 ) and of the combined hzs and scp samples ( wang 1999 ) .
the fact that the constraints are weakly dependent on the size of the sn sample is a consequence of the strong degeneracy between @xmath8 and @xmath0 discussed in section 3.2 .
5.5 truein figure 7b shows the results of combining the sn likelihoods with those determined from the cmb .
the likelihood peaks at @xmath127 , @xmath183 .
qualitatively , these results are similar to those of perlmutter ( 1999 ) ; the favoured cosmology has an equation of state @xmath127 and @xmath8 is contrained to be less than @xmath184 at the @xmath9 level .
however , in detail , the constraints in figure 7b are somewhat less stringent than those of perlmutter _
et al . _ , , allowing a broader range in @xmath0 ( @xmath185 at the @xmath9 level ) .
this is because perlmutter include constraints on the power spectrum of galaxy clustering based on the data compiled by peacock and dodds ( 1994 ) . in our view
this is dangerous because it requires a specific assumption concerning the distribution of galaxies relative to the mass .
qualitatively , for nearly scale - invariant adiabatic models , galaxy clustering imposes a constraint on the parameter combination @xmath186 of @xmath187 , if galaxies are assumed to trace the mass fluctuations on large scales ( efstathiou , bond and white 1992 , maddox , efstathiou and sutherland 1996 ) . combined with measurements of the hubble constant ( for which perlmutter adopt @xmath188 ) , galaxy clustering leads to a constraint of @xmath189 , partly breaking the degeneracy between @xmath8 and @xmath0 .
the combined sn and cmb analysis in figure 7b provides constraints which are nearly as tight , but are much less model dependent .
the constraints of figure 7b place strong limits on q - like models . for tracking solutions
, the constraint @xmath190 @xmath184 excludes steep potentials ( _ e.g. _ @xmath191 with @xmath192 ) and the data clearly favour a standard cosmological term ( @xmath193 ) .
these limits on @xmath8 are very close to the lower limit ( @xmath194 ) allowed for ` physically well motivated ' tracker solutions ( steinhardt , wang and zlatev , 1998 , _ i.e. _ smooth potentials with simple functional forms ) . with a slight improvement of the observations one
may be forced to fine - tune the shape of the potential to construct a viable quintessence model .
the constraints of figure 7b are somewhat stronger than those of wang ( 1999 ) , who perform a ` concordance analysis ' of q - like models using a number of observational constraints including those from type ia supernovae and cmb anisotropies .
these authors conclude limits of @xmath195 .
the difference is caused by the different methods of statistical analysis .
the concordance analysis of wang leads , by construction , to more conservative limits than the maximum likelihood analysis and is more robust to systematic errors in any particular data set . however , provided systematic errors are negligible in the cmb and sn datasets , then the constraints of figure 7b derived by combining likelihoods should be realistic .
these small differences in the upper limits on @xmath8 are important because they can place significant restrictions on the physics . as stressed in the previous paragraph , the upper limit of @xmath196 places strong constraints on tracker models with simple potentials .
figure 8 shows the results of a likelihood analysis of the sn and cmb data , but now allowing arbitrary spatial curvature .
we show two projections of the likelihood distributions , marginalizing over @xmath120 in figures 8(a ) and 8(c ) and over @xmath8 in figures 8(b ) and 8(c ) . the constraints ,
although weaker than those presented in figure 7 , are interesting nevertheless .
the combined sn and cmb likelihoods give a @xmath197 upper limit on @xmath8 of @xmath198 ( figure 8d ) and a maximum likelihood solution of @xmath199 , @xmath200 irrespective of the value of @xmath8 .
evidently , the sn and cmb data constrain us to a nearly spatially flat universe dominated either by a cosmological constant , or a q - like component with an equation of state @xmath198 .
observations of distant type ia supernovae have provided important evidence that the universe may be dominated by a cosmological constant ( p98 , riess 1999 ) . however , the constraints in the @xmath1@xmath0 plane from current data are degenerate along a line defined by @xmath201 ( figure 1 ) .
this degeneracy can be reduced significantly by extending the redshift range of the supernovae sample .
for example , with @xmath83 additional supernovae at redshift @xmath202 the errors in @xmath0 and @xmath1 could be reduced to @xmath203 and @xmath204 .
a sample of supernovae at @xmath105 could provide an accurate estimate of @xmath0 that is independent of the value of @xmath1 .
the combination of supernovae and cmb anisotropy measurements can break the degeneracy between @xmath1 and @xmath0 if the initial fluctuations are assumed to be adiabatic and characterised by a smooth fluctuation spectrum .
this method applied to recent supernovae and cmb data suggests a nearly spatially flat universe dominated by a cosmological term with @xmath205 ( lineweaver 1998 , garnavich 1998 , tegmark 1999 , e99 ) .
the only plausible way of avoiding this conclusion is to appeal to some systematic effect in the supernovae data , for example , grey dust or an evolutionary effect in the supernovae data such as a metallicity dependence ( see _ e.g. _
p98 for a discussion ) .
the degeneracy breaking afforded by extending the supernovae data to higher redshift would provide an important consistency check of such systematic effects and also on the interpretation of the cmb anisotropy data the constraints on quintessence - like models with an equation of state @xmath7 improve relatively slowly as the supernovae data are extended to higher redshift . the most promising way of constraining @xmath8 seems to be to combine supernovae and cmb measurements .
we have carried out a joint likelihood analysis of cmb anisotropy observations and the scp supernovae data .
for a spatially flat universe we derive a @xmath197 upper limit of @xmath206 . for universes of arbitrary spatial curvature ,
the @xmath9 upper limit is @xmath207 .
the combined sn and cmb likelihood peaks at @xmath199 and @xmath208 irrespective of the value of @xmath8 , suggesting that the universe is almost spatially flat .
the @xmath9 upper limit of @xmath206 for spatially flat universes is close to the minimum value of @xmath209 allowed for simple quintessence - models .
this suggests that some fine tuning of the potential may be required to construct a viable quintessence model .
* acknowledgements .
* i thank richard ellis , paul steinhardt and roberto terlevich for useful discussions and pparc for the award of a senior fellowship .
i also thank sarah bridle , anthony lasenby , mike hobson and graca rocha for allowing me to use their compilation of cmb anisotropy data .
livio m , 1999 , to be published in the proceedings of type ia supernovae : theory and cosmology ( held at the university of chicago , 29 - 31 october 1998 ) , edited by jens niemeyer and james truran ( cambridge univ .
press , cambridge , uk ) .
astro - ph/9903264 .
turner m.s . , 1999 , to be published in the proceedings of type ia supernovae : theory and cosmology ( held at the university of chicago , 29 - 31 october 1998 ) , edited by jens niemeyer and james truran ( cambridge univ .
press , cambridge , uk ) .
astro - ph/9904049 . | the magnitude - redshift relation for type ia supernovae is beginning to provide strong constraints on the cosmic densities contributed by matter , @xmath0 , and a cosmological constant , @xmath1 , though the results are highly degenerate in the @xmath0@xmath1 plane . here
we estimate the constraints that can be placed on a cosmological constant or quintessence - like component by extending supernovae samples to high redshift .
such measurements , when combined with constraints from anisotropies in the cosmic microwave background , could provide an important consistency check of systematic errors in the supernovae data .
a large campaign of high - z supernovae observations with 10 metre class telescopes could constraint @xmath0 to a ( @xmath2 ) accuracy of @xmath3 and @xmath1 to @xmath4 . a sample of supernovae at redshift @xmath5 , as might be achievable with a next generation space telescope ,
could constrain @xmath0 to an accuracy of about @xmath6 independently of the value of @xmath1 . the constraints on a more general equation of state ,
@xmath7 , converge slowly as the redshift of the supernovae data is increased .
the most promising way of setting accurate constraints on @xmath8 is by combining high - z supernovae and cmb measurements . with feasible measurements it should be possible to constrain @xmath8 to a precision of about @xmath3 , if the universe is assumed to be spatially flat .
we use the recent supernovae sample of perlmutter and observations of the cmb anisotropies to constraint the equation of state in quintessence - like models via a likelihood analysis .
the @xmath9 upper limits are @xmath10 if the universe is assumed to be spatially flat , and @xmath11 for universes of arbitrary spatial curvature .
the upper limit derived for a spatially flat universe is close to the lower limit ( @xmath12 ) allowed for simple potentials , implying that additional fine tuning may be required to construct a viable quintessence model
. 0.2 truein = -0.6 truein |
negative refraction is a very interesting phenomenon which nowadays attracts a lot of attention .
intuitively , negative refraction can be imagined as the phenomenon at the interface between two media when the refracted and incident beams of light remain on the same side of the interface normal .
most frequently , the work by veselago @xcite is cited as the one which represents the early efforts on the negative refraction studies . for
a brief history of related studies performed before @xcite , the reader may refer , for example , to @xcite .
it is quite often believed that negative refraction appears at the interface of artificial media , such as metamaterials @xcite or photonic crystals @xcite .
however , negative refraction has also been observed at interfaces of conventional media , for example , anisotropic materials @xcite , or metallic wedges @xcite . the work by veselago @xcite considers the most straightforward way to obtain negative refraction .
according to veselago , if a medium has both negative permittivity , @xmath0 , and negative permeability , @xmath1 , the resulting refractive index is also negative , @xmath2 , thus providing negative refraction at the interface with a medium whose index @xmath3 is positive .
this idea is utilized in metamaterials , where one uses the overlapping of permittivity and permeability resonances to achieve simultaneously negative @xmath4 and @xmath5 @xcite .
the obvious disadvantage of such an approach is that , as with any resonant media , metamaterials are highly absorptive .
therefore , theoretical studies of negative refraction in metamaterials should consider the propagation of light in lossy media . from many standard textbooks ( for example , @xcite ) we know that , unlike lossless media , the waves transmitted into lossy media are , in general , inhomogeneous ( or nonuniform ) , that is , their planes of constant phase and constant amplitude are not parallel .
thus as a first step to understand negative refraction in metamaterials we need to consider the formation of inhomogeneous waves at the interface of lossy media , characterized by complex permittivity and permeability .
inhomogeneous waves in conventional conducting media ( with complex @xmath4 and @xmath6 ) have been thoroughly studied and discussed in many classical textbooks .
some generalizations of the classical theory were made in @xcite .
however , the classical theory of inhomogeneous waves does not consider negative refraction .
the authors of @xcite made an attempt to apply the formalism of inhomogeneous waves in order to explain the results of numerical simulations of negative refraction in a metamaterial prism and slab .
unfortunately , they used the formulas of the classical theory , where they formally substituted the complex refractive index of the metamaterial by a negative value @xmath3 , defined as @xmath7 .
however , in an inhomogeneous wave propagating in a lossy medium , the actual refractive index and attenuation coefficient depend on the incident angle and , therefore , the value @xmath3 loses its meaning of a complex refractive index @xcite .
this means that the criterion of negative refraction proposed by veselago @xcite and based on negative @xmath3 should be reconsidered . in order to extend the classical theory of inhomogeneous waves , in this work we study negative refraction of inhomogeneous waves at the interface of lossy isotropic media . unlike the previous authors @xcite we use a parametrization of inhomogeneous waves which allows us to find explicitly all the necessary parameters of transmitted waves , up to their signs . in @xcite , to obtain these parameters , one needs to find a unique solution of a system of four nonlinear equations with multivalued inverse trigonometric and hyperbolic functions . in @xcite ,
the parameters of transmitted waves are given only in a quadratic form .
additionally , unlike @xcite , we consider a more general problem , where both media at the interface are , in addition to a complex permittivity @xmath4 , characterized by a complex permeability @xmath5 .
after the formal derivation of all the necessary equations , we show that the inhomogeneity of transmitted waves affects negative refraction .
we demonstrate that the recently reported criterion of negative refraction that requires negative real parts of @xmath4 and @xmath5 @xcite is applicable only in the case of a homogeneous incident wave at the interface between a lossless and lossy media . in a more general situation ,
when the incident wave is inhomogeneous , or both media are lossy , the criterion of negative refraction becomes dependent on an incident angle .
in particular , we show that negative refraction can be realized in conventional lossy materials if their interfaces are properly oriented .
let us consider a plane monochromatic wave propagating in a lossy isotropic medium with complex permittivity @xmath8 and permeability @xmath9 .
the electric , @xmath10 , and magnetic , @xmath11 , field vectors of such a wave are given by @xmath12 where @xmath13 and @xmath14 are complex amplitude vectors , @xmath15 and @xmath16 are the wavevector and a position vector , respectively , with @xmath17 and @xmath18 being the frequency and time . in a lossy medium the wavevector @xmath15 is complex , that is , @xmath19 , where @xmath20 and @xmath21 are real phase and attenuation vectors , respectively . if the vectors @xmath20 and @xmath21 are not parallel , the planes of constant phase and constant amplitude ( defined as @xmath22const and @xmath23const , respectively ) are not parallel either , and the corresponding wave is inhomogeneous .
we can formally define the refractive index @xmath24 and attenuation coefficient @xmath25 of an inhomogeneous wave through the lengths of @xmath20 and @xmath21 as @xmath26 and @xmath27 , where @xmath28 is the speed of light in vacuum .
such definition retains the physical meaning of the refractive index and attenuation coefficient ; namely , @xmath24 equals to the ratio of @xmath28 to the phase velocity and @xmath25 determines the distance at which the amplitude is reduced to @xmath29 times .
note that both @xmath24 and @xmath25 are real and positive by definition , since we determine them through the length of a real vector . by separately equating the real and imaginary parts in the dispersion equation @xmath30 , where the complex value @xmath31 is defined as @xmath7
, we find @xcite @xmath32 in which @xmath33 is the angle between @xmath20 and @xmath21 .
note that inhomogeneous waves with @xmath34 can exist in lossless media ( @xmath35 ) if @xmath36 in ; that is , if @xmath37 . for example , such a kind of inhomogeneous wave describes total internal reflection , where a wave at the boundary of two lossless media propagates along the boundary and attenuates in the direction perpendicular to the interface . additionally , and indicate that @xmath38 and @xmath39 only if @xmath40 ; that is , if @xmath41 ( homogeneous damped waves ) .
thus the complex quantity @xmath3 loses its traditional meaning of refractive index for inhomogeneous waves ( @xmath42 ) .
therefore , a natural question occurs : how should we reinterpret the criterion of negative refraction that is commonly believed to be @xmath43?.
to answer this question , we examine the formation of inhomogeneous waves at the interface of two lossy isotropic media : the first medium is characterized by a complex permittivity @xmath44 and permeability @xmath45 , and the second medium by @xmath46 and @xmath47 .
the incident wave comes from the first medium , and a unit interface normal @xmath48 points to the second medium ( see figure [ fig : sketch ] ) .
orientation of the phase and attenuation vectors at the interface ( white plane ) of two lossy isotropic media .
the blue and red planes are the incidence planes for the phase and attenuation vectors , respectively . ]
the wavevectors @xmath49 , @xmath50 , and @xmath51 of the incident , reflected and transmitted waves can be written as the sum of two vectors that are parallel and perpendicular to the interface @xcite : @xmath52 the parallel components @xmath53 are continuous across the interface : @xmath54 = [ { \boldsymbol{\hat{q}}}\times[{\boldsymbol{k}}_{\rm r}\times{\boldsymbol{\hat{q } } } ] ] = [ { \boldsymbol{\hat{q}}}\times[{\boldsymbol{k}}_{\rm t}\times{\boldsymbol{\hat{q}}}]].\ ] ] the normal components have magnitudes @xmath55 , @xmath56 and @xmath57 , where @xmath58 for the most general case , all three wavevectors @xmath59 ( with @xmath60 , @xmath61 , and @xmath62 ) are complex , that is , @xmath63 . therefore , @xmath64 and @xmath65 , where @xmath66 , \qquad & { \boldsymbol{p } } '' = [ { \boldsymbol{\hat{q}}}\times[{\boldsymbol{k}}_{\rm i}''\times{\boldsymbol{\hat{q } } } ] ] , \label{eq : p'p '' } \\ & q_\alpha ' = ( { \boldsymbol{k}}_\alpha'\cdot{\boldsymbol{\hat{q } } } ) , & q_\alpha '' = ( { \boldsymbol{k}}_\alpha''\cdot{\boldsymbol{\hat{q}}}).\end{aligned}\ ] ] by separately equating the real and imaginary parts in , we obtain two sets of equations for the phase vectors @xmath67 and attenuation vectors
@xmath68 : @xmath69 according to the phase vectors @xmath67 and attenuation vectors @xmath68 , in general , lie in two different planes ( see figure [ fig : sketch ] ) .
in other words , at the plane interface of two lossy isotropic media , there are , in general , two incidence planes : the incidence plane for the phase vectors ( spanned by @xmath70 and @xmath48 with the normal @xmath71 $ ] ) and attenuation vectors ( spanned by @xmath72 and @xmath48 with the normal @xmath73 $ ] ) @xcite .
two incidence planes result in two sets of equalities for the snell s law , for the phase and attenuation vectors , respectively .
we find them by equating the magnitudes of the real and imaginary parts in and using the definitions of @xmath24 and @xmath25 : @xmath74 where @xmath75 and @xmath76 are the angles between the unit normal @xmath48 and the associated phase vectors @xmath67 and attenuation vectors @xmath68 , respectively .
before we focus on the transmitted wave , we make a simple remark on the reflected wave .
the reflected wave has the same inhomogeneity as that of the incident wave , so that @xmath77 and @xmath78 ( the reflection angles are @xmath79 and @xmath80 ) .
this implies that if the first medium is lossless and the incident wave is homogeneous , then the reflected wave is also homogeneous , even if the second medium is lossy @xcite .
now we examine the transmitted wave whose phase and attenuation vectors are given by . in
we need to determine the projections @xmath81 and @xmath82 , which are the real and imaginary parts of the complex projection @xmath83 .
note that we can not uniquely determine @xmath84 by , since @xmath84 is in the quadratic form .
standard textbooks choose a positive sign for @xmath84 , or equivalently @xmath85 and @xmath86 , because the phase vector and , consequently the phase velocity , are outgoing from the interface only for this choice of sign .
we now know , however , that the phase velocity can be incoming towards the interface if negative refraction takes place . thus a careful study is necessary to unambiguously determine the sign of @xmath84 . for this purpose
we introduce a complex dimensionless parameter @xmath87 defined as @xmath88 for its real and imaginary parts , we find @xmath89 and @xmath90 .
therefore , taking into account that @xmath81 and @xmath82 are real by definition , we obtain @xmath91 , and @xmath92 , where @xmath93 . to find the signs of @xmath81 and @xmath82 we express them as @xmath94 where @xmath95 and @xmath96 are their signs .
then , since @xmath97^{1/2 } = s's''\sqrt{\xi''^2}=s's''|\xi''|$ ] , we find that @xmath98 thus the signs @xmath99 and @xmath100 are the same if @xmath101 , and opposite if @xmath102 . specific values of @xmath99 and @xmath100 can be selected only by applying an additional constraint : that the energy flux in the second medium must be directed away from the interface .
mathematically we write this constraint as @xmath103 , meaning that the projection of the time - averaged poynting vector @xmath104 of the transmitted wave on the interface normal must be non - negative . to calculate @xmath81 and @xmath82 by , we need to express @xmath105 and @xmath106 through some known values such as permittivities , permeabilities , incident angles , etc . to do so we substitute into and then separate the real and imaginary parts .
as a result we obtain @xmath107 using , we express the terms with @xmath108 and @xmath109 in and through the parameters of the incident wave : @xmath110 where the angle @xmath111 between @xmath70 and @xmath72 can be found from for the incident wave
. then we obtain @xmath112 having clarified the sign issue of the complex projection @xmath84 , we now turn to the refractive index @xmath113 and attenuation coefficient @xmath114 . from
and we have @xmath115 , which is equal to @xmath116 . therefore , we obtain @xmath117^{1/2}$ ] where we have chosen the `` + '' sign since @xmath118 is positive by definition .
similarly , we obtain @xmath119^{1/2}$ ] . with the help of , we find that @xmath113 and @xmath114 can be written as @xmath120 equations and clearly show that the refractive index @xmath113 and attenuation coefficient @xmath114 depend not only on the material properties of the second medium , but also on the incident angles @xmath121 and @xmath122 .
to summarize the results of this section , we describe how one can find the parameters of the transmitted wave if the corresponding parameters of the incident wave are known :
( i ) calculate @xmath105 and @xmath106 from and ( @xmath111 can be found from ) ; ( ii ) find the refractive index @xmath113 and attenuation coefficient @xmath114 from and ; ( iii ) with the obtained @xmath113 and @xmath114 calculate the transmission angles @xmath123 and @xmath124 using and ; ( iv ) find the signs of @xmath81 and @xmath82 in , utilizing with the additional constraint @xmath103 , where the time - averaged poynting vector of the transmitted wave , @xmath104 , should be calculated separately . before concluding this section , we wish to say a few words about the limitations of the model .
it is known that , at optical frequencies , the permeability @xmath125 loses its usual physical meaning @xcite . to overcome this problem the authors of @xcite proposed describing the linear response of a negatively refractive medium by a spatially dispersive permittivity tensor @xmath126 .
however , they showed that , as long as the spatial dispersion is restricted to terms @xmath127 , the formal description of a medium by @xmath128 and @xmath125 is adequate , if an effective permeability is introduced , even if this permeability loses its original physical meaning .
thus our model implies a weak spatial dispersion and the effective permeabilities @xmath129 and @xmath130 that do not have a direct physical meaning .
in addition , we note that the spatial dispersion is formally present in our result , because the obtained refractive index and attenuation coefficient depend on an incident angle ( see and ) , and consequently , on a wavevector of the incident wave .
as a first example , we consider a single interface where the first medium is a lossless dielectric ( @xmath131 ) and the second medium is lossy and isotropic ( see figure [ fig : interface ] ) .
we further assume that the incident wave is a homogeneous plane wave , which is characterized by the real wavevector @xmath132 ( @xmath133 and @xmath134 ) , refractive index @xmath135 , and incident angle @xmath136 . according to and
, the condition @xmath133 means that @xmath137 and @xmath138 . as a consequence the transmission angle @xmath139 .
thus the attenuation vector @xmath140 of the transmitted wave is normal to the interface for any incident angle . in other words , equiamplitude planes of the transmitted wave
are always parallel to the interface @xcite .
orientations of the phase and attenuation vectors at the interface between a lossless dielectric and a lossy isotropic medium .
( a ) and ( b ) are cases of positive ( @xmath101 ) and negative ( @xmath102 ) refraction . note
that @xmath133 is assumed for both cases . ] for the case of a single interface and can be simplified as @xmath141 with these @xmath105 and @xmath106 we can calculate the refractive index @xmath118 and attenuation coefficient @xmath142 of the transmitted wave using and , and then the transmission angle @xmath123 using . to find the orientation of @xmath143 and @xmath140 , we need to determine the signs of their projections @xmath81 and @xmath82 in . in [ sec : kp ] we obtain , which says that , in any lossy medium , the projection of an attenuation vector onto a time - averaged poynting vector is always positive . in our case
reads as @xmath144 , where @xmath104 is the time - averaged poynting vector of the transmitted wave . by substituting @xmath140 from into and taking into account that @xmath137 , we find @xmath145 . since the condition for the energy flux to be directed away from the interface is @xmath103 , we eventually find that @xmath82 must be positive
that is , the attenuation vector @xmath140 of the transmitted wave is outgoing from the interface .
the condition @xmath146 means that we should choose @xmath147 in . as a result
, we obtain that the sign of the projection @xmath81 is given by @xmath148 .
thus , if @xmath102 we have @xmath149 and the phase vector is incoming towards the interface , and hence negative refraction takes place .
therefore , the criterion of negative refraction is @xmath102 , which , according to , is equivalent to @xmath150 thus , starting from the most general case of light wave transmission through two lossy isotropic media , we have rediscovered the well - known criterion for negative refraction , which implies negative @xmath151 and @xmath152 @xcite .
we note that the angle @xmath153 between @xmath143 and @xmath140 is equal to @xmath123 , since @xmath139 . therefore ,
if the condition of negative refraction is satisfied and @xmath154 , we have @xmath155 . in other words ,
negative refraction is accompanied by the formation of inhomogeneous waves with an obtuse angle between equiamplitude and equiphase planes ( see figure [ fig : interface](b ) ) . using this argument
, we can now clarify the relation between negative refraction and the condition of @xmath43 : at an angle @xmath156 the cosine in becomes negative , which results in a negative @xmath157 on the right - hand side of the equation ( @xmath158 is positive in lossy media ) .
in particular , in the case of negative refraction at normal incidence @xmath159 and @xmath160 in . as a result , we find @xmath161 and @xmath162 . in this case
the choice between a negative cosine and negative @xmath157 is a matter of preference ; hence our theory does not violate the commonly used criterion for negative refraction ; namely , that @xmath157 is negative .
but our theory predicts more than that : it says that , if an incidence angle is different from zero , the use of @xmath157 and @xmath158 instead of @xmath24 and @xmath25 can result in an error . to be more quantitative
we now assume that the first medium is vacuum and the second medium is the negative - index metamaterial reported as structure 3 in @xcite . to retrieve the permittivity and permeability of this metamaterial we digitized the relevant data from @xcite and then approximated it by the drude model using parameter fitting @xcite .
according to the retrieved parameters , the value of @xmath106 is negative in the spectral region between 710 and 800 nm , irrespective of the incidence angle .
figures [ fig : mtheta](a)(c ) show the change of @xmath113 , @xmath114 and @xmath123 , calculated by , and , as functions of the wavelength and the incidence angle @xmath163 . at normal incidence , as was mentioned before , the spectral dependencies of @xmath113 and @xmath114 ( black curves in figure [ fig : mtheta ] ) reproduce the absolute values of the corresponding experimental data from @xcite . for example , the spectral dependence of @xmath113 at normal incidence is the modulus of a well - known bell - gap shape which one usually observes in experiments ( see figure 3 in @xcite ) .
we see that both @xmath113 and @xmath114 increase as @xmath163 increases . from figure
[ fig : mtheta](b ) we find that the absorption is minimum at @xmath164 .
as @xmath163 increases from zero , the transmission angle @xmath123 very quickly approaches almost @xmath165 , which means that @xmath143 becomes nearly parallel to the interface ( see figure [ fig : mtheta](c ) ) .
unfortunately , at incidence angles different from zero , we can not directly compare the above example with the experimental data , since the metamaterial in @xcite is anisotropic .
instead , we consider @xmath113 , @xmath114 and @xmath123 in figure [ fig : mtheta ] as characteristic parameters of some hypothetical isotropic metamaterial whose permittivity and permeability are represented by the drude model .
since the majority of existing metamaterials are highly anisotropic , the assumption of such isotropic metamaterials in this work may raise questions .
however , the first isotropic metamaterials have been fabricated very recently @xcite . in section [ sec :
laws ] we have shown that the sign of the @xmath106 parameter determines the directions of the phase and attenuation vectors in the transmitted wave ( see ) . in particular , for a single interface , we have shown that , if @xmath102 , then the phase vector of the transmitted wave points towards the interface .
that is , @xmath102 is the criterion of negative refraction . according to , in general
, @xmath106 consists of two terms .
the first term is associated with the material properties of a medium below the interface , while the second term depends on the incident angles and thus is associated with the geometry of the problem .
hereinafter we will refer to the first and second terms of @xmath106 in as the _ material _ and _ geometric _ terms , respectively . we have already seen that for a single interface the geometric term is zero .
since the sign of @xmath106 affects the criterion of negative refraction , we expect that this criterion will change if we manage to find a case where the geometric term is nonzero .
the attenuation vector of any incident wave coming from a lossless medium is zero and therefore has a zero parallel component .
since the parallel components of attenuation vectors must be continuous across the interface ( see ) , the attenuation vector of the transmitted wave must be perpendicular to the interface in order to also have a zero parallel component .
this argument holds for any number of subsequent interfaces which are parallel to the first one .
therefore , to obtain a nonzero geometric term , we need to consider a case where an incident wave has a nonzero attenuation vector that is tilted relative to the interface .
this is most easily realized by a lossy prism ( see figure [ fig : prism ] ) where the attenuation vector of the wave transmitted through the first prism interface forms the angle @xmath166 equal to the prism angle with the second prism interface . as a result ,
the parallel components of the attenuation vectors at the second prism interface become nonzero , and accordingly the corresponding geometric term becomes nonzero . if , in addition , we want to obtain negative refraction at the exit of the prism , the medium below the second prism interface must be lossy , since in lossless media the poynting and phase vectors are co - directional and the poynting vector is necessarily directed away from the interface .
the metallic prism lying on the semi - infinite surface of another metal and surrounded by vacuum .
the incoming wave is incident normally on the first prism interface .
the phase and attenuation vectors are marked in blue and red , respectively . at this stage
the orientations of the phase and attenuation vectors in the underlying metal are unknown and indicated only schematically . ] after these considerations , we now investigate the following problem .
a metallic prism lies on a semi - infinite surface of another metal ( see figure [ fig : prism ] ) and surrounded by vacuum .
the prism and underlying metal are characterized by the complex permittivities @xmath46 and @xmath167 , respectively , and permeabilities @xmath168 .
a homogeneous incident wave comes normally from vacuum to the first prism interface .
this wave is characterized by a real wavevector @xmath169 , zero attenuation vector @xmath170 , refractive index @xmath171 , attenuation vector @xmath172 and incident angle @xmath173 .
using the procedure described at the end of section [ sec : laws ] , we find that the wave transmitted through the first prism interface is characterized by the complex wavevector @xmath174 , with the phase vector @xmath175 and attenuation vector @xmath176 perpendicular to the interface ; refractive index @xmath177 and attenuation coefficient @xmath178 , where @xmath179 and @xmath180 are given by @xmath181 ; and transmission angles @xmath182 .
since the @xmath183 parameter for the first interface has the imaginary part @xmath184 ( see ) , refraction at this interface is always positive .
if we are interested only in the principal ray far away from the prism s edge ( such that we can neglect the effects of multiple reflections inside the prism and diffraction on the prism edge ) , then at the second prism interface we can formulate the following problem ( see figure [ fig : pdirections ] ) : the incident wave is characterized by the complex wavevector @xmath185 with a zero angle @xmath111 between @xmath70 and @xmath72 ; refractive index @xmath186 and attenuation coefficient @xmath187 , where @xmath157 and @xmath158 are given by @xmath188 ; and incidence angles @xmath189 , where @xmath166 is the prism angle .
we seek to find the orientation of the phase vector @xmath143 and attenuation vector @xmath140 of the transmitted wave relative to the interface .
according to and , the parameter @xmath87 for the considered interface is given by @xmath190 and @xmath191 , or taking into account and , by @xmath192 where @xmath106 has the geometric term @xmath193 , which is nonzero and negative for any nonzero prism angle @xmath166 . as we have shown in section [ sec : laws ] , the directions of @xmath143 and @xmath140 relative to the interface are specified by the sign of @xmath106 ( see ) .
for example , when @xmath106 is such that the sign @xmath99 of the projection @xmath194 is negative , then @xmath143 is directed towards the interface , and hence negative refraction takes place . according to
, @xmath106 is positive if @xmath195 and negative otherwise . using and
, we categorize all possible orientations of @xmath143 and @xmath140 into the following cases ( see figure [ fig : pdirections ] ) : * @xmath196 and @xmath197 . both @xmath143 and @xmath140
are directed away from the interface .
refraction is positive . *
@xmath196 and @xmath198 . both @xmath143 and @xmath140
are directed towards the interface .
refraction is negative . *
@xmath199 and @xmath200 while @xmath201 .
vector @xmath143 is directed away from the interface , while @xmath140 points towards the interface .
refraction is positive . *
@xmath199 and @xmath202 while @xmath147 .
vector @xmath143 is directed towards the interface , while @xmath140 points away from the interface .
refraction is negative .
it might seem strange that the attenuation vector in cases ( b ) and ( c ) points towards the interface .
naively , one can come to the conclusion that in these cases the amplitude of the transmitted wave grows exponentially as the wave moves away from the prism , which violates the energy conservation law .
however , the attenuation vector pointing towards the interface merely reflects the fact that the amplitude of the transmitted wave decreases in the direction from the sharp to the blunt end of the prism , since the thicker part of the prism absorbs more . to resolve the confusion
, we note that there is always a fraction of the incident wave that propagates without going through the prism ( to the left of the prism edge in figure [ fig : prism ] ) .
therefore , the growth of the transmitted wave amplitude is always limited by the amplitude of the incident wave . thus cases ( b ) and
( c ) do not violate the energy conservation law .
schematic representation of the interface problem formulated for the second interface of the prism shown in figure [ fig : prism ] .
depending on the problem parameters , the orientation of the phase @xmath143 and attenuation @xmath140 vectors of the transmitted wave can be divided into four possible cases ( a)(d ) . ] which of ( a)(d ) is actually the case depends on the additional condition that require the time - averaged poynting vector @xmath104 of the transmitted wave to be directed away from the interface . since at the considered interface
the parallel component of @xmath140 is nonzero , we can no longer apply the inequality to uniquely determine the sign of @xmath82 . instead , we need to find an expression for @xmath104 and directly consider which of the cases ( a)(d ) does not violate the inequality @xmath103 . to simplify the task , we assume that the incident wave is p - polarized ; that is , in figure [ fig : prism ] the electric field vector of the wave incident on the first prism interface lies in the plane spanned by vectors @xmath203 and @xmath204 ( plane of the figure ) .
we additionally assume that the frequency of the incident wave is less than the plasma frequency of the underlying metal ; that is , @xmath205 . in [ sec : poynting ] we show ( see ) that for p - polarized incident wave the condition @xmath103 is equivalent to @xmath206 .
taking into account that @xmath207 and @xmath208 while @xmath205 and @xmath209 , we rewrite the latter inequality as @xmath210 equation is a necessary and sufficient condition for the energy flux in the underlying metal to be directed away from the interface .
thus we need to find under which conditions the cases ( a)(d ) do not violate the inequality of . consider each of the cases separately : * if @xmath196 and @xmath197 , then gives us @xmath211 or @xmath212 * if @xmath196 and @xmath198 , then gives us @xmath213 or @xmath214 * if @xmath199 and @xmath200 while @xmath201 , then gives us @xmath215 , meaning this case is never realized . * if @xmath199 and @xmath202 while @xmath147 , then gives us @xmath216 , that is , this case is always realized .
thus we have found that , if @xmath196 , then cases ( a ) and ( b ) can be realized , if , respectively , conditions of and are fulfilled . if @xmath199 , then only case ( d ) is realized . in the above
we have pointed out that cases ( b ) and ( d ) correspond to negative refraction at the second prism interface . in case ( b )
negative refraction is realized if both conditions of and are fulfilled . since usually @xmath217 , the condition of
can be realized only if @xmath105 in its right - hand side is negative .
according to , @xmath105 is negative if @xmath218 . combining the latter inequality with , we find that the necessary but not sufficient condition of negative refraction for case ( b ) is @xmath219 . in case ( d ) the criterion of negative refraction is much simpler ,
namely , the prism angle @xmath166 must satisfy the condition @xmath220 according to , to realize negative refraction in case ( d ) , the losses in the prism must be higher than that in the underlying metal , that is , @xmath221 . to give a quantitative example
, we consider an aluminum prism lying on a silver substrate .
the permittivities of both metals we approximate by the drude model @xmath222 , where for aluminum @xmath223 s@xmath224 , @xmath225 s@xmath224 and for silver @xmath226 s@xmath224 , @xmath227 s@xmath224 @xcite . figure [ fig : psid ] shows the angle @xmath228 as a function of the wavelength @xmath229 , where @xmath230 and @xmath231 are the imaginary parts of the permittivities for silver and aluminum , respectively . according to , negative refraction is realized for any prism angles @xmath166 that lie above the curve in figure [ fig : psid ] . as we see , in order to observe negative refraction in our example , the prism angles must exceed @xmath232@xmath233 .
angle @xmath228 as a function of the wavelength @xmath229 , where @xmath230 and @xmath231 are the imaginary parts of the permittivities for silver and aluminum , respectively . ] to conclude this section , we have shown that negative refraction can be realized in conventional lossy media , if their interfaces are properly oriented .
please note that what we report is negative refraction of a wavevector , that is , of a phase velocity .
this effect should be distinguished from negative refraction of energy flux , that is , of poynting vector . in anisotropic and lossy
media the direction of poynting vector does not necessarily coincide with the direction of the wavevector @xcite . therefore , negative refraction of energy flux does not automatically imply negative refraction of phase velocity .
negative refraction of energy flux in conventional media has already been reported for anisotropic materials @xcite and metallic wedges @xcite .
it was also predicted for bulk metals in case of specially polarized incident light @xcite . however , as far as we know , negative refraction of phase velocity in conventional media has never been reported before .
in conclusion we have studied the formation of inhomogeneous waves at the interface between two lossy isotropic media .
we have found explicit expressions ( up to the sign ) for the propagation constants , refractive index and attenuation coefficient of transmitted inhomogeneous waves .
we have shown that the inhomogeneity of transmitted waves can affect negative refraction .
in particular , we have shown that , if a homogeneous wave is incident on the interface between a lossless and lossy isotropic media , negative refraction is accompanied by the formation of an inhomogeneous wave with an obtuse angle between the equiphase and equiamplitude planes . in this case , the criterion of negative refraction is equivalent to the well - known one that requires negative real parts of permittivity and permeability .
however , in a more general case , when the incident wave is inhomogeneous or both media are lossy , the criterion of negative refraction becomes dependent on an incident angle . in particular , we have shown that negative refraction can be realized in the system of a metallic prism lying on the surface of another metal .
thus , negative refraction can be realized even in conventional lossy materials if their interfaces are properly oriented .
this work was supported by a grant - in - aid for scientific research from the ministry of education and science of japan .
part of the work by v yu fedorov was also supported by the japan society for the promotion of science ( jsps ) . [
[ sec : k''p ] ] in this appendix we show that , in any lossy medium , the projection of an attenuation vector @xmath21 onto a time - averaged poynting vector @xmath234 is always positive .
the maxwell equations for the electric @xmath10 and magnetic @xmath11 field vectors of a monochromatic plane wave in a medium with complex permittivity @xmath4 and permeability @xmath5 can be written as @xmath235 taking the scalar product of with @xmath236 and with @xmath237 , and summing the results , we obtain @xmath238 + [ { \boldsymbol{k}}\times{\boldsymbol{e}}]\cdot{\boldsymbol{h}}^ * = \varepsilon_0\varepsilon\omega|{\boldsymbol{e}}|^2 + \mu_0\mu\omega|{\boldsymbol{h}}|^2\ ] ] or @xmath239 + [ { \boldsymbol{e}}^*\times{\boldsymbol{h } } ] \right ) = \varepsilon_0\varepsilon\omega|{\boldsymbol{e}}|^2 + \mu_0\mu\omega|{\boldsymbol{h}}|^2.\ ] ] since @xmath240 + [ { \boldsymbol{e}}^*\times{\boldsymbol{h } } ] = 2\,\mathrm{re}\{[{\boldsymbol{e}}\times{\boldsymbol{h}}^*]\}$ ] , we rewrite the latter equation as @xmath241 where @xmath242\}$ ] is a real time - averaged poynting vector . equating the real and imaginary parts of , we obtain @xmath243 in a lossy medium the right - hand side of is always positive .
therefore , the projection of @xmath21 onto @xmath234 is also always positive : @xmath244 note that , according to , inhomogeneous waves in lossless media ( where @xmath35 ) have @xmath21 perpendicular to @xmath234 . [
[ sec : poynting ] ] in this appendix we find the condition under which the energy flux of the transmitted wave is directed away from an interface if the incident wave is p - polarized . using
we can write the complex time - averaged poynting vector of the transmitted wave @xmath245 $ ] as @xmath246,\ ] ] where @xmath247 denotes complex conjugate .
despite the fact that there are two incidence planes in our problem , we still can formally represent the complex amplitude vector @xmath248 as a sum of s- and p - polarized components @xcite : @xmath249,\end{aligned}\ ] ] where @xmath250 and @xmath251 are the amplitudes of s- and p - polarized components , respectively , while @xmath252 .
using , we can express the complex amplitude vector @xmath253 $ ] in terms of amplitudes @xmath254 and @xmath255 as @xmath256.\ ] ] substituting and into , we obtain @xmath257 + \varepsilon_0\varepsilon_3^*\omega|a_{\rm t\parallel}|^2 [ { \boldsymbol{s}}^*\times[{\boldsymbol{k}}_{\rm t}\times{\boldsymbol{s } } ] ] \right . \\ & \quad \left .
+ \varepsilon_0\varepsilon_3^*\omega a_{\rm t\perp}a_{\rm t\parallel}^ * [ { \boldsymbol{s}}\times{\boldsymbol{s}}^ * ] - \frac{a_{\rm t\perp}^*a_{\rm t\parallel}}{\mu_0\mu_3^*\omega } [ [ { \boldsymbol{k}}_{\rm t}\times{\boldsymbol{s } } ] \times [ { \boldsymbol{k}}_{\rm t}^*\times{\boldsymbol{s}}^ * ] ] \right\}.\end{aligned}\ ] ] in case of p - polarized incident wave , when @xmath258 , we have @xmath259.\ ] ] taking the real part of @xmath260 , we obtain a real time - averaged poynting vector @xmath104 : @xmath261 ) ( \varepsilon_3'{\boldsymbol{s}}''-\varepsilon_3''{\boldsymbol{s } } ' ) \vphantom{\frac{1}{2}}\right\},\ ] ] where @xmath262 and @xmath263 are the normals to the planes of incidence for the phase and attenuation vectors , respectively .
note that vectors @xmath104 and @xmath143 are non - collinear , meaning the directions of the energy flux and phase velocity in the transmitted wave are different .
from we find that the projection @xmath264 is given by @xmath265 since the factor in front of the brackets in the right - hand side of is always positive , we find that the condition @xmath103 is equivalent to @xmath266 equation is the condition under which the energy flux of the transmitted wave is directed away from the interface in the case of a p - polarized incident wave .
99 veselago v g 1968 _ sov . phys .-
* 10 * 50914 zhang y and mascarenhas a 2005 _ mod .
* 19 * 2133 shelby r a , smith d r and schultz s 2001 _ science _ * 292 * 779 lezec h j , dionne j a and atwater h a 2007 _ science _ * 316 * 4302 xu t , agrawal a , abashin m , chau k j and lezec h j 2013 _ nature _ * 497 * 4704 kosaka h , kawashima t , tomita a , notomi m , tamamura t , sato t and kawakami s 1998 _ phys . rev .
_ b * 58 * r10096 cubukcu e , aydin k , ozbay e , foteinopoulou s and soukoulis c m 2003 _ nature _ * 423 * 604 parimi p , lu w , vodo p , sokoloff j , derov j and sridhar s 2004 127401 zhang y , fluegel v and mascarenhas a 2003 157404 chen x
l , he m , du y x , wang w y and zhang d f 2005 _ phys .
_ b * 72 * 113111 hoffman a j , alekseyev l , howard s s , franz k j , wasserman d , podolskiy v a , narimanov e e , sivco d l and gmachl c 2007 _ nature mat . _ * 6 * 94650 sanz m , papageorgopoulos a , egelhoff w f , nieto - vesperinas m , and garca n 2003 _ phys . rev . _ e * 67 * 067601 wu y - h , gu w , chen y - r , li x - f , zhu x
- s , zhou p , li j , zheng y - x and chen l - y 2008 _ phys . rev .
_ b * 77 * 035134 cai w and shalaev v 2010 _ optical metamaterials _
( new york : springer ) chen h c 1983 _ theory of electromagnetic waves _
( new york : mcgraw - hill ) fedorov f i 2004 _ optics of anisotropic media _
( moscow : urss ) ( in russian ) dupertuis m a , proctor m and acklin b 1994 _ josa _ a * 11 * 115966 rautian s g 1994 _ opt .
spectrosc . _ * 76 * 677 - 81 garcia - pomar j l and nieto - vesperinas m 2004 _ opt
* 12 * 208195 landau l d , lifshitz e m and pitaevskii l p 1984 _ electrodynamics of continuous media _ ( oxford : butterworth - heinemann ) agranovich v m , shen y r , baughman r h and zakhidov a a 2004 _ phys .
_ b * 69 * 165112 agranovich v m and gartstein y n 2006 _ phys .- usp . _ * 49 * 102944 garca - meca c , hurtado j , mart j , martnez a , dickson w and zayats a v 2011 067402 fedorov v yu and nakajima t 2011 143903 rudolph s m and grbic a 2012 _ ieee trans .
antennas propag .
_ * 60 * 366169 mackay t g and lakhtakia a 2009 _ phys .
_ b * 79 * 235121 | we theoretically study negative refraction of inhomogeneous waves at the interface of lossy isotropic media .
we obtain explicit ( up to the sign ) expressions for the parameters of a wave transmitted through the interface between two lossy media characterized by complex permittivity and permeability .
we show that the criterion of negative refraction that requires negative permittivity and permeability can be used only in the case of a homogeneous incident wave at the interface between a lossless and lossy media . in a more general situation ,
when the incident wave is inhomogeneous , or both media are lossy , the criterion of negative refraction becomes dependent on an incident angle .
most interestingly , we show that negative refraction can be realized in conventional lossy materials ( such as metals ) if their interfaces are properly oriented . |
in many combinatorial optimization problems we seek an object composed of some elements of a finite set whose total cost is minimum .
this is the case , for example , in an important class of network problems where the set of elements consists of all arcs of some network and we wish to find an object in this network such as a path , a spanning tree , or a matching whose total cost is minimum . in general , the combinatorial optimization problems can often be expressed as 0 - 1 programming problems with a linear objective function , where a binary variable is associated with each element and a set of constraints describes the set of feasible solutions . for a comprehensive description of this class of problems
we refer the reader to @xcite .
the usual assumption in combinatorial optimization is that all the element costs are precisely known .
however , the assumption that all the costs are known in advance is often unrealistic . in practice , before solving a problem , we only know a set of possible realizations of the element costs .
this set is called a _ scenario set _ and each particular realization of the element costs within this scenario set is called a _
scenario_. several methods of defining scenario sets have been proposed in the existing literature .
discrete _ and _ interval _ uncertainty representations are among the most popular ( see , e.g. , @xcite ) . in the former
, scenario set contains a finite number of explicitly given cost vectors . in the latter one , for each element an interval of its possible values is specified and scenario set is the cartesian product of these intervals . in the discrete uncertainty representation , each scenario can model some event that has a global influence on the element costs . on the other hand , the interval uncertainty representation is appropriate when each element cost may vary within some range independently on the values of the other costs . a modification of the interval uncertainty representation was proposed in @xcite , where the authors assumed that only a fixed and a priori given number of costs may vary .
more general scenario sets which can be used in mathematical programming problems were discussed , for example , in @xcite . in this paper
we assume that no additional information , for example a probability distribution , for scenario set is provided .
if scenario set contains more than one scenario , then an additional criterion is required to choose a solution . in _ robust optimization _
( see , e.g. , @xcite ) we typically seek a solution minimizing the worst case behavior over all scenarios .
hence the min - max and min - max regret criteria are widely applied .
however , this approach to decision making is often regarded as too conservative or pessimistic ( see , e.g. , @xcite ) .
in particular , the min - max criterion takes into account only one , the worst - case scenario , ignoring the information connected with the remaining scenarios .
this criterion also assumes that decision makers are very risk averse , which is not always true . in this paper
we wish to investigate a class of combinatorial optimization problems with the discrete uncertainty representation .
hence , a scenario set provided with the input data , contains a finite number of explicitly given cost scenarios . in order to choose a solution we propose to use the _ ordered weighted averaging _ aggregation operator ( owa for short ) introduced by yager in @xcite .
the owa operator is widely applied to aggregate the criteria in multiobjective decision problems ( see , e.g. , @xcite ) , but it can also be applied to choose a solution under the discrete uncertainty representation .
it is enough to treat the cost of a given solution under @xmath1th scenario as a @xmath1th criterion .
the key elements of the owa operator are weights whose number equals the number of scenarios .
the @xmath1th weight expresses an importance of the @xmath1th largest cost of a given solution .
hence , the weights allow a decision maker to take his attitude towards a risk into account and use the information about all scenarios while computing a solution .
the owa operator generalizes the traditional criteria used in decision making under uncertainty such as the maximum , minimum , average , median , or hurwicz criterion .
so , by using owa we can generalize the min - max approach , typically used in robust optimization .
let us also point out that the owa operator is a special case of _ choquet integral _
, a sophisticated tool for aggregating criteria in multiobjective decision problems ( see , e.g. , @xcite ) .
the choquet integral has been recently applied to some multicriteria network problems in @xcite .
unfortunately , the min - max combinatorial optimization problems are almost always harder to solve than their deterministic counterparts , even when the number of scenarios equals 2 . in particular , the min - max versions of the shortest path , minimum spanning tree , minimum assignment , minimum @xmath2 cut , and minimum selecting items problems are np - hard even for 2 scenarios @xcite .
furthermore , if the number of scenarios is a part of the input , then all these problems become strongly np - hard and hard to approximate within any constant factor @xcite .
since the maximum criterion is a special case of owa , the general problem of minimizing owa is not easier .
however , it is not difficult to show that some other particular cases of owa , such as the minimum or average , lead to problems whose complexity is the same as the complexity of their deterministic counterparts .
it is therefore of interest to provide a characterization of the problem complexity depending on various weight distributions . in this paper
we provide the following new results . in section [ sec2scen ] , we study the case when the number of scenarios equals 2 .
we give a characterization of the problem complexity depending on the weight distribution . in section [ secfptas ] ,
we show some sufficient conditions for the problem to admit a fully polynomial time approximation scheme ( fptas ) , when the number of scenarios is constant .
finally , in section [ secunb ] , we consider the case in which the number of scenarios is a part of the input .
we discuss different types of weight distributions .
we show that for nonincreasing weights ( i.e. when larger weights are assigned to larger solution costs ) and for the hurwicz criterion , the problem admits an approximation algorithm whose worst case ratio depends on the problem parameters , in particular on the number of scenarios . on the other hand
, we show that if the weights are nondecreasing , or the owa criterion is median , then the problem is not at all approximable unless p = np .
let @xmath3 be a finite set of _ elements _ and @xmath4 be a set of _ feasible solutions_. in the deterministic case , each element @xmath5 has a nonnegative _ cost _ @xmath6 and we seek a solution whose total cost is minimum .
namely , we wish to solve the following optimization problem : @xmath7 this formulation encompasses a large class of combinatorial optimization problems . in particular , for the class of network problems
@xmath8 is the set of arcs of a given network @xmath9 and @xmath10 contains the subsets of the arcs forming , for example , @xmath2 paths , spanning trees , assignments , or @xmath2 cuts in @xmath11 . in practice ,
problem @xmath12 is often expressed as a 0 - 1 programming one , where binary variable @xmath13 is associated with each element @xmath14 , @xmath15 , and a system of constraints describes the set @xmath10 in a compact form .
before we discuss the uncertain version of problem @xmath12 , we recall the definition of the owa operator , proposed by yager in @xcite .
let @xmath16 be a vector of reals .
let us introduce a vector @xmath17 such that @xmath18 $ ] , @xmath19 $ ] ( we use @xmath20 $ ] to denote the set @xmath21 ) , and @xmath22 .
let @xmath23 be a permutation of @xmath20 $ ] such that @xmath24 .
then @xmath25 } w_i f_{\sigma(i)}.\ ] ] the owa operator has several natural properties which easily follow from its definition ( see , e.g. @xcite ) . since it is a convex combination of @xmath26 it holds @xmath27 .
it is also _ monotonic _ , i.e. if @xmath28 for all @xmath29 $ ] , then @xmath30 , _ idempotent _ , i.e. if @xmath31 , then @xmath32 and _ symmetric _ , i.e. its value does not depend on the order of the values @xmath26 .
it generalizes several traditional criteria used in decision making under uncertainty and we will describe this fact later .
assume that the costs in problem @xmath12 are uncertain and they are specified in the form of scenario set @xmath33 .
hence @xmath34 contains @xmath0 distinct _ cost scenarios _ , where @xmath35 for @xmath29 $ ] .
the cost of a given solution @xmath36 depends on scenario @xmath37 and will be denoted by @xmath38 . in this paper
we will aggregate the costs by using the owa operator .
namely , given a weight vector @xmath17 , let us define @xmath39 } w_j f(x,\pmb{c}_{\sigma(j)}),\ ] ] where @xmath23 is a permutation of @xmath20 $ ] such that @xmath40 .
we will consider the following optimization problem : @xmath41 we now discuss several special cases of the min - owa @xmath12 problem ( see also table [ tabsc ] ) .
if @xmath42 and @xmath43 for @xmath44 , then owa becomes the maximum and the corresponding problem is denoted as min - max @xmath12 .
this is a typical problem considered in the robust optimization framework .
if @xmath45 and @xmath43 for @xmath46 , then owa becomes the minimum and the corresponding problem is denoted as min - min @xmath12 . in general ,
if @xmath47 and @xmath43 for @xmath29\setminus\{k\}$ ] , then owa is the @xmath48-th largest cost and the problem is denoted as min - quant@xmath49 @xmath12 . in particular , when @xmath50 , then the @xmath48-th largest cost is median and the problem is denoted as min - median @xmath12 .
if @xmath51 for all @xmath29 $ ] , i.e. when the weights are _ uniform _ , then owa is the average ( or the laplace criterion ) and the problem is denoted as min - average @xmath12 .
finally , if @xmath52 and @xmath53 , for some fixed @xmath54 $ ] , and @xmath43 for the remaining weights , then we get the hurwicz pessimism - optimism criterion and the problem is then denoted as min - hurwicz @xmath12 .
our goal has been to provide general properties of min - owa @xmath12 , which follow only from the type of the weight distribution in the owa operator .
we have not taken into account a particular structure of an underling deterministic problem @xmath12 .
thus , the results obtained may be additionally refined if some properties of @xmath12 are taken into account .
s. mittal and a. s. schulz . a general framework for designing approximation schemes for combinatorial optimization problems with many objectives combined into one . in a.
goel , k. jansen , j. d. p. rolim , and r. rubinfeld , editors , _ approx - random _ , volume 5171 of _ lecture notes in computer science _ , pages 179192 .
springer - verlag , 2008 . | in this paper a class of combinatorial optimization problems with uncertain costs is discussed .
the uncertainty is modeled by specifying a discrete scenario set containing @xmath0 distinct cost scenarios .
the ordered weighted averaging ( owa for short ) aggregation operator is applied to choose a solution .
some well known criteria used in decision making under uncertainty such as the maximum , minimum , average , hurwicz and median are special cases of owa .
furthermore , by using owa , the traditional robust ( min - max ) approach to combinatorial optimization problems with uncertain costs can be generalized . the computational complexity and approximability of the problem of minimizing owa for the considered class of problems
are investigated and some new positive and negative results in this area are provided .
these results remain valid for many basic problems , such as network or resource allocation problems .
combinatorial optimization ; owa operator ; robust optimization ; computational complexity ; approximation algorithms |
precision measurements of the top quark properties offer an excellent opportunity to explore indirect effects of new physics beyond the standard model ( sm ) .
their theoretical interest is motivated by the large top quark mass , which leads to the common belief that this fermion may be quite sensitive to new physics effects . and , on the experimental side , top quark studies are greatly facilitated by the short lifetime of this quark , @xmath6 s , which prevents complications from hadronisation effects and allows to study in detail the properties of a `` bare '' quark .
thus , for example , the @xmath1 helicity fractions @xcite have been precisely measured at the tevatron and the large hadron collider ( lhc ) @xcite , namely the relative fractions of @xmath1 bosons with helicity @xmath7 produced in the decay @xmath8 .
new physics can enter both the production and decay of the top quark .
new production mechanisms may be difficult to spot directly , as is the case of wide @xmath9 resonances @xcite , superpartners @xcite or non - resonant @xcite contributions , the latter including @xmath10-channel flavour - changing processes @xcite .
but the presence of such contributions ( arising for example in models addressing the anomalous tevatron @xmath9 asymmetry ) would generally result in a top polarisation or @xmath9 spin correlation @xcite different from the sm predictions @xcite .
these are not measurable quantities , however , and can only be probed by analysing angular distributions in the top decay @xmath0 , with @xmath11 .
( hadronic @xmath1 decays and leptonic decays to taus are sensitive to the top polarisation too , but their experimental measurement is much more difficult . ) for example , a well - known method to probe the top polarisation is through the angular distributions of its decay products in the top quark rest frame .
these take the form @xmath12 being @xmath13 the angle between the momentum @xmath14 of the decay product @xmath15 in the top quark rest frame , and an arbitrary direction @xmath16 chosen to quantise the top spin . in the above equation , @xmath17 is the top polarisation along this direction and @xmath18 are constants called `` spin analysing power '' of the particle @xmath19 , which can be affected by top anomalous couplings @xcite .
( radiative corrections to these quantities have been computed in @xcite . ) hence , eq .
( [ ec : dist1 ] ) clearly shows a production - decay interplay in the @xmath13 distributions : the measurable quantities are the products @xmath20 , which depend on the production ( @xmath17 ) and decay properties ( @xmath18 ) of the top quark . this is a general feature : since the top polarisation ( as well as the @xmath9 spin correlation ) can only be measured through top decay distributions , the resulting observables are also sensitive to anomalous contributions to the @xmath21 vertex .
boson and the charged leptons is the sm one , as implied by low energy measurements . ]
a non - trivial but important issue is then to disentangle new physics in production and decay .
this , of course , would become crucial in case that a deviation from the sm predictions was found .
previous literature @xcite has attempted to get rid of the dependence on the top decay vertex by noting that for the charged lepton the spin analysing power @xmath22 depends on @xmath21 anomalous couplings only quadratically , so @xmath22 _ should be _ less sensitive to new physics .
this solution is not satisfactory , however , not only because the new physics affecting top production may modify @xmath23 at quadratic level too , but also because anomalous @xmath21 couplings are not sufficiently constrained from other sources so as to imply that their quadratic contributions are small . in
@xcite it has been shown that a global fit to several top decay observables ( including @xmath1 helicity fractions , @xmath22 and @xmath24 ) can be used to extract @xmath17 from single top and @xmath9 measurements . in this letter
we focus on a more direct measurement of the top polarisation and introduce a `` doubly forward - backward '' top decay asymmetry @xmath25 where @xmath26 stands for the number of events ; @xmath27 is the angle between the @xmath1 momentum in the top rest frame @xmath28 and a chosen top spin quantisation axis @xmath16 ; @xmath29 is the angle between the charged lepton momentum in the @xmath1 rest frame , @xmath30 , and @xmath28 .
we show that this asymmetry is related to the top polarisation along the @xmath16 direction and the @xmath1 helicity fractions @xmath31 by @xmath32 in full generality .
since the @xmath1 helicity fractions can be ( and have actually been ) measured in a model - independent fashion , @xmath33 provides a model - independent measurement of the top polarisation along a chosen axis , in any process of top production at hadron or lepton colliders .
in addition , we present here an inequality involving @xmath34 and @xmath1 helicity fractions , which can be used to obtain lower bounds on @xmath17 from the measurement of the @xmath35 distribution .
we use the jacob - wick helicity formalism @xcite ( see also @xcite ) to describe the decay of the top quark and @xmath1 boson using general arguments of angular momentum conservation .
let us fix a @xmath36 coordinate system in the top quark rest frame , with the positive @xmath37 axis along the direction in which we want to quantise the top spin .
the most general spin state of an ensemble of top quarks can be described by a density matrix @xmath38 with @xmath39 .
we do not specify the orientation of our @xmath40 and @xmath41 axes , which is not relevant for our discussion .
the amplitudes for the decay @xmath8 , for a top quark having third spin component @xmath42 and the @xmath1 boson and @xmath43 quark having helicities @xmath44 , @xmath45 , respectively , can be written as @xmath46 being @xmath47 the polar and azimuthal angles of @xmath28 in the @xmath36 coordinate system , @xmath48 and @xmath49 a wigner function for a rotation @xmath50 parameterised by its euler angles ( explicit expressions for low @xmath51 can be found in @xcite ) .
hence , we see that all the dependence on @xmath52 and the direction of @xmath28 is encoded in the @xmath53 function , while @xmath54 only depend on the helicities , invariant under rotations .
there are only eight non - zero amplitudes , corresponding to @xmath42 and @xmath55 because the two remaining helicity combinations imply a total angular momentum @xmath56 of the @xmath57 pair along the direction of @xmath28 , which is forbidden for a spin-@xmath58 decaying top quark .
the decay @xmath59 can also be described in a similar fashion , introducing a @xmath60 coordinate system in the @xmath1 boson rest frame , with the @xmath61 axis in the direction of @xmath28 .
then , the full decay amplitude can be written as @xmath62 with @xmath63 ( @xmath64 ) the helicity of the charged lepton ( neutrino ) and @xmath65 ; @xmath66 are the polar and azimuthal angles of the charged lepton momentum @xmath30 in the @xmath1 boson rest frame , using the @xmath60 coordinate system .
( we denote quantities in the @xmath1 boson rest frame with asterisks , as opposed to quantities in the top quark rest frame . )
notice the coherent sum over @xmath1 boson helicities @xmath67 .
in the case of a @xmath68 boson decay , the left - handed structure of the vertex implies @xmath69 for the positively charged lepton ( anti - fermion ) and @xmath70 for the neutrino , both taken massless . from eqs .
( [ ec : rho ] ) and ( [ ec : a ] ) , the fully differential decay width is @xmath71 with @xmath72 a constant phase - space factor and @xmath73 .
integrating over the azimuthal angles @xmath74 , @xmath75 gives factors @xmath76 and @xmath77 , respectively , so that the differential width in the two polar angles reads @xmath78 ^ 2 \ , , \label{ec : dg2}\ ] ] with @xmath79 ( this distribution has already been obtained explicitly @xcite within the sm , including radiative corrections . ) the total width for @xmath8 is obtained by integration over the remaining angles , @xmath80 we can identify the helicity fractions @xmath81 , as the relative widths for @xmath8 with @xmath82 , respectively . denoting for brevity the sum between brackets in eq .
( [ ec : g ] ) as @xmath83 , we have @xmath84 / d \ , , \notag \\[1 mm ] f_- & = & |a_{-1\ , -1/2}|^2 / d \ , .
\label{ec : f}\end{aligned}\ ] ] then , integrating eq .
( [ ec : dg2 ] ) in the four quadrants @xmath85 , @xmath86 and dividing by @xmath87 we obtain an explicit expression for the asymmetry in eq .
( [ ec : adef ] ) , @xmath88 / d = \frac{3}{8 } p_z \left [ f_+ + f_- \right ] \ , .
\label{ec : rel2}\ ] ] for anti - top decays @xmath89 for the negatively charged lepton and @xmath90 for the neutrino , so @xmath91 and the resulting asymmetry is @xmath92 / d = - \frac{3}{8 } p_z \left [ \bar f_+ + \bar f_- \right ] \,,\ ] ] with the helicity fractions for anti - top decays ( denoted with bars , as the corresponding anti - top decay amplitudes ) satisfying @xmath93 , @xmath94 @xcite
. a by - product of our analysis is obtained by integrating eq .
( [ ec : dg2 ] ) over the full @xmath29 range to obtain the @xmath1 boson angular distribution , namely eq .
( [ ec : dist1 ] ) for @xmath95 .
the spin analysing power of the @xmath1 boson is found to be @xmath96 / d \,.\ ] ] this implies , given eqs .
( [ ec : f ] ) for the helicity fractions , that @xmath97 this inequality is practically saturated in the sm because amplitudes with @xmath98 are suppressed due to the left - handed @xmath99 interaction . finally , by integrating eq .
( [ ec : dg2 ] ) over @xmath27 we obtain the well - known @xmath29 distribution @xmath100 in particular , the forward - backward ( fb ) asymmetry in the @xmath1 rest frame @xcite is @xmath101 \ , .
\label{ec : afb}\ ] ]
these results help clarify the relation ( [ ec : rel ] ) : the subtraction of events with @xmath102 and @xmath103 removes the contribution from the `` symmetric '' amplitudes @xmath104 , @xmath105 entering @xmath106 , while the subtraction of events with @xmath107 and @xmath108 removes the polarisation - independent term .
we have introduced a fb asymmetry using two angles @xmath27 , @xmath29 in the top quark and @xmath1 boson rest frames , respectively , showing that it is related to the @xmath1 helicity fractions by eq .
( [ ec : rel ] ) .
its value in the sm can be computed using previous calculations for the helicity fractions @xcite .
the most general effective @xmath21 interaction arising from dimension - six operators can be parameterized as @xcite @xmath109 being @xmath110 and @xmath111 in the sm .
for this general vertex , we obtain @xmath112 with @xmath113 \left(1 - x_w^2 \right ) + \left [ |{g_l}|^2 + |{g_r}|^2 \right ] \left(1 - x_w^2 \right ) \notag \\ & - 4 x_b \ , { \text{re}}\left [ { v_l}{v_r}^ * + { g_l}{g_r}^ * \right ] - 2 \frac{m_t}{m_w } { \text{re}}\ , \left[{v_l}{g_r}^ * + { v_r}{g_l}^ * \right ] \left(1 - x_w^2 \right ) \notag \\ & + 2 \frac{m_t}{m_w } x_b \,{\text{re}}\ , \left[{v_l}{g_l}^ * + { v_r}{g_r}^ * \right ] \left(1 + x_w^2 \right ) \ , , \notag \\ b_0 & = \left [ |{v_l}|^2 + |{v_r}|^2 \right ] \left(1 - x_w^2 \right ) + \frac{m_t^2}{m_w^2 } \left [ |{g_l}|^2 + |{g_r}|^2 \right ] \left(1 - x_w^2 \right ) \notag \\ & - 4 x_b \ , { \text{re}}\left [ { v_l}{v_r}^ * + { g_l}{g_r}^ * \right ] - 2 \frac{m_t}{m_w } { \text{re}}\ , \left[{v_l}{g_r}^ * + { v_r}{g_l}^ * \right ] \left(1 - x_w^2 \right ) \notag \\ & + 2 \frac{m_t}{m_w } x_b \,{\text{re}}\ , \left[{v_l}{g_l}^ * + { v_r}{g_r}^ * \right ] \left(1 + x_w^2 \right ) \ , , \label{ec : ab}\end{aligned}\ ] ] @xmath114 , @xmath115 .
a naive combination of the current helicity fraction measurements @xcite , not taking into account correlations between different experiments and data sets , gives @xmath116 in good agreement with the sm tree - level prediction @xmath117 , @xmath118 , @xmath119 for @xmath120 gev .
this implies a relatively small asymmetry @xmath121 .
( higher - order corrections @xcite slightly modify these values , but the differences are well below the experimental uncertainty . )
when extracting the top polarisation from @xmath33 there will be , in addition to the experimental uncertainty associated to the measurement of this asymmetry , an uncertainty associated to the precision in the determination of helicity fractions . with the values in eqs .
( [ ec : fexp ] ) , this uncertainty is @xmath122 , while with the foreseen lhc precision @xcite it will be reduced by a factor of two , @xmath123 .
this can be compared to the `` theoretical '' uncertainty on @xmath17 that arises when extracting it from the distribution in eq .
( [ ec : dist1 ] ) , which is associated to possible new physics in the decay . for illustration
, we show in fig .
[ fig : alab ] the estimated allowed region for @xmath124 at 68.3% confidence level ( cl ) .
these limits are obtained with a combined fit to the four @xmath21 couplings in eq .
( [ ec : lagr ] ) , using the code topfit @xcite and taking as experimental data the helicity fractions in eqs .
( [ ec : fexp ] ) , and the measured lhc @xmath10-channel single top cross section at 7 tev , @xmath125 pb .
then , for @xmath21 couplings within the range imposed by these measurements , the possible variation of the spin analysing powers is given by fig .
[ fig : alab ] .
these limits are compatible with previous sensitivity projections @xcite made under more optimistic assumptions and using additional observables not yet measured .
we observe that there is a large uncertainty on the spin analysing powers associated to possible new physics in the decay , that arises from ( i ) a number of parameters ( four ) in the effective vertex greater than the number of independent measurements ( three ) ; ( ii ) an approximate cancellation between @xmath126 and @xmath127 contributions to helicity fractions @xcite .
thus , even when the precision in the measurement of helicity fractions and single top cross sections is improved , this trend will persist .
for example , using the charged lepton as spin analyser , the resulting uncertainty on the polarisation measurement is @xmath128 \cup [ p_z^\text{exp},1 ] \,,\ ] ] being @xmath129 the `` observed '' top quark polarisation ( taken positive ) , extracted from eq .
( [ ec : dist1 ] ) assuming @xmath130 .
( the first of these intervals is empty if @xmath131 . ) clearly , the larger the polarisation , the smaller will be this uncertainty .
but , in any case , any measurement of @xmath33 will provide relevant information , and the best precision in top polarisation measurements will be achieved from combination of all observables available . at the lhc , single top quarks produced in the @xmath10-channel process
are highly polarised in the direction of the spectator quark @xcite , with @xmath132 for centre - of - mass energies @xmath133 tev @xcite , so the expected asymmetry @xmath134 is measurable .
lhc statistics are excellent and this measurement may eventually be dominated by systematics . in that case , for a precise determination of the single top polarisation it may be convenient to measure instead the ratio @xmath135 , being @xmath136 the well - known lepton fb asymmetry in the @xmath1 rest frame , see eq .
( [ ec : afb ] ) . given the present helicity fraction measurements , which imply @xmath137 ( this ratio is @xmath138 in the sm ) , one has @xmath139 to a good approximation .
alternatively , @xmath140 can also be measured .
besides , we note that the inequality ( [ ec : ineq ] ) may also be used to obtain relevant bounds on @xmath17 in processes , such as single top production , where it is large .
the product @xmath141 can be determined from the @xmath35 distribution , see eq .
( [ ec : dist1 ] ) . then
, if the experimental measurement is consistent with the sm prediction , say @xmath142 , the inequality ( [ ec : ineq ] ) implies here , in which case eq .
( [ ec : ineq ] ) implies @xmath143 .
this assumption can explicitly be tested with the measurement of the sign of @xmath17 from @xmath33 and the sign of @xmath144 . ]
@xmath145 which is a very stringent bound since @xmath146 by definition .
note , however , that the same result can be achieved with the measurement of @xmath147 from the @xmath148 distribution , since @xmath149 . finally , at a future @xmath150 international linear collider top quarks are produced in pairs with a small but non - zero polarisation , @xmath151 in the helicity axis for @xmath152 gev @xcite .
the precision expected for asymmetry measurements is excellent @xcite , and therefore the measurement of this asymmetry may be very useful to complement top - spin - independent observables to probe anomalous contributions to @xmath153 independently of the decay vertex @xcite . in any case , it is clear that the new asymmetry @xmath33 introduced here provides a new handle to measure the top polarisation in any process , and test the presence of new physics in the top sector .
this work has been supported by micinn by projects fpa2006 - 05294 and fpa2010 - 17915 , junta de andaluca ( fqm 101 , fqm 03048 and fqm 6552 ) and fundao para a cincia e tecnologia ( fct ) project cern / fp/123619/2011 .
g. l. kane , g. a. ladinsky and c. p. yuan , phys .
d * 45 * ( 1992 ) 124 .
v. m. abazov _ et al . _
[ d0 collaboration ] , phys . rev .
d * 83 * ( 2011 ) 032009 [ arxiv:1011.6549 [ hep - ex ] ] .
t. aaltonen _ et al . _
[ cdf collaboration ] , arxiv:1205.0354 [ hep - ex ] ; cdf note 10855 .
r. barcelo , a. carmona , m. masip and j. santiago , phys .
b * 707 * ( 2012 ) 88 [ arxiv:1106.4054 [ hep - ph ] ] ; g. m. tavares and m. schmaltz , phys . rev .
d * 84 * ( 2011 ) 054008 [ arxiv:1107.0978 [ hep - ph ] ] ; e. alvarez , l. da rold , j. i. s. vietto and a. szynkman , jhep * 1109 * ( 2011 ) 007 [ arxiv:1107.1473 [ hep - ph ] ] ; j. a. aguilar - saavedra and m. perez - victoria , phys .
b * 705 * ( 2011 ) 228 [ arxiv:1107.2120 [ hep - ph ] ] .
k. -i .
hikasa , j. m. yang and b. -l .
young , phys . rev .
d * 60 * ( 1999 ) 114041 [ hep - ph/9908231 ] ; m. perelstein and a. weiler , jhep * 0903 * ( 2009 ) 141 [ arxiv:0811.1024 [ hep - ph ] ] . c. degrande , j. -m .
gerard , c. grojean , f. maltoni and g. servant , jhep * 1103 * ( 2011 ) 125 [ arxiv:1010.6304 [ hep - ph ] ] ; c. zhang and s. willenbrock , arxiv:1008.3155 [ hep - ph ] ; phys . rev .
d * 83 * ( 2011 ) 034006 [ arxiv:1008.3869 [ hep - ph ] ] .
j. a. aguilar - saavedra , nucl .
b * 843 * ( 2011 ) 638 [ erratum - ibid .
b * 851 * ( 2011 ) 443 ] [ arxiv:1008.3562 [ hep - ph ] ] ; j. a. aguilar - saavedra and m. perez - victoria , jhep * 1105 * ( 2011 ) 034 [ arxiv:1103.2765 [ hep - ph ] ] . s. jung , h. murayama , a. pierce and j. d. wells , phys .
d * 81 * ( 2010 ) 015004 [ arxiv:0907.4112 [ hep - ph ] ] ; j. shu , t. m. p. tait and k. wang , phys . rev .
d * 81 * ( 2010 ) 034012 [ arxiv:0911.3237 [ hep - ph ] ] ; a. arhrib , r. benbrik and c. -h .
chen , phys .
d * 82 * ( 2010 ) 034034 [ arxiv:0911.4875 [ hep - ph ] ] ; i. dorsner , s. fajfer , j. f. kamenik and n. kosnik , phys . rev .
d * 81 * ( 2010 ) 055009 [ arxiv:0912.0972 [ hep - ph ] ] ; j. a. aguilar - saavedra and m. perez - victoria , jhep * 1109 * ( 2011 ) 097 [ arxiv:1107.0841 [ hep - ph ] ] ; k. blum , y. hochberg and y. nir , jhep * 1110 * ( 2011 ) 124 [ arxiv:1107.4350 [ hep - ph ] ] .
t. han , m. hosch , k. whisnant , b. -l .
young and x. zhang , phys . rev .
d * 58 * ( 1998 ) 073008 [ hep - ph/9806486 ] ; p. m. ferreira , o. oliveira and r. santos , phys .
d * 73 * ( 2006 ) 034011 [ hep - ph/0510087 ] ; p. m. ferreira and r. santos , phys .
d * 73 * ( 2006 ) 054025 [ hep - ph/0601078 ] ; r. a. coimbra , p. m. ferreira , r. b. guedes , o. oliveira , a. onofre , r. santos and m. won , phys .
d * 79 * ( 2009 ) 014006 [ arxiv:0811.1743 [ hep - ph ] ] ; j. gao , c. s. li , j. j. zhang and h. x. zhu , phys . rev .
d * 80 * ( 2009 ) 114017 [ arxiv:0910.4349 [ hep - ph ] ] ; j. a. aguilar - saavedra , nucl .
b * 837 * ( 2010 ) 122 [ arxiv:1003.3173 [ hep - ph ] ] .
j. cao , l. wu and j. m. yang , phys .
d * 83 * ( 2011 ) 034024 [ arxiv:1011.5564 [ hep - ph ] ] ; d. -w .
jung , p. ko and j. s. lee , phys .
b * 701 * ( 2011 ) 248 [ arxiv:1011.5976 [ hep - ph ] ] ; d. choudhury , r. m. godbole , s. d. rindani and p. saha , phys . rev .
d * 84 * ( 2011 ) 014023 [ arxiv:1012.4750 [ hep - ph ] ] ; d. krohn , t. liu , j. shelton and l. -t .
wang , phys .
d * 84 * ( 2011 ) 074034 [ arxiv:1105.3743 [ hep - ph ] ] ; j. cao , k. hikasa , l. wang , l. wu and j. m. yang , phys .
d * 85 * ( 2012 ) 014025 [ arxiv:1109.6543 [ hep - ph ] ] ; s. fajfer , j. f. kamenik and b. melic , jhep * 1208 * ( 2012 ) 114 [ arxiv:1205.0264 [ hep - ph ] ] ; e. l. berger , q. -h . cao , j. -h .
yu and h. zhang , phys . rev .
* 109 * ( 2012 ) 152004 [ arxiv:1207.1101 [ hep - ph ] ] .
g. mahlon and s. j. parke , phys .
d * 55 * ( 1997 ) 7249 [ hep - ph/9611367 ] .
g. mahlon and s. j. parke , phys .
b * 476 * ( 2000 ) 323 [ hep - ph/9912458 ] .
g. mahlon and s. j. parke , phys .
d * 53 * ( 1996 ) 4886 [ hep - ph/9512264 ] .
w. bernreuther , a. brandenburg , z. g. si and p. uwer , nucl .
b * 690 * ( 2004 ) 81 [ hep - ph/0403035 ] .
g. mahlon and s. j. parke , phys . rev .
d * 81 * ( 2010 ) 074024 [ arxiv:1001.3422 [ hep - ph ] ] .
w. bernreuther and z. -g .
si , nucl .
b * 837 * ( 2010 ) 90 [ arxiv:1003.3926 [ hep - ph ] ] .
m. jezabek and j. h. kuhn , phys .
b * 329 * ( 1994 ) 317 [ hep - ph/9403366 ] .
j. a. aguilar - saavedra , j. carvalho , n. f. castro , f. veloso and a. onofre , eur .
j. c * 50 * ( 2007 ) 519 [ hep - ph/0605190 ] .
a. brandenburg , z. g. si and p. uwer , phys .
b * 539 * ( 2002 ) 235 [ hep - ph/0205023 ] .
s. groote , w. s. huo , a. kadeer and j. g. korner , phys .
d * 76 * ( 2007 ) 014012 [ hep - ph/0602026 ] .
b. grzadkowski and z. hioki , phys .
b * 557 * ( 2003 ) 55 [ hep - ph/0208079 ] .
b. grzadkowski and z. hioki , phys .
b * 529 * ( 2002 ) 82 [ hep - ph/0112361 ] .
s. d. rindani , pramana * 54 * ( 2000 ) 791 [ hep - ph/0002006 ] .
r. m. godbole , s. d. rindani and r. k. singh , jhep * 0612 * ( 2006 ) 021 [ hep - ph/0605100 ] .
r. m. godbole , k. rao , s. d. rindani and r. k. singh , jhep * 1011 * ( 2010 ) 144 [ arxiv:1010.1458 [ hep - ph ] ] .
r. m. godbole , l. hartgring , i. niessen and c. d. white , jhep * 1201 * ( 2012 ) 011 [ arxiv:1111.0759 [ hep - ph ] ] .
j. a. aguilar - saavedra , pos ichep * 2010 * ( 2010 ) 378 [ arxiv:1008.3225 [ hep - ph ] ] .
j. a. aguilar - saavedra and j. bernabeu , nucl .
b * 840 * ( 2010 ) 349 [ arxiv:1005.5382 [ hep - ph ] ] .
m. jacob and g. c. wick , annals phys .
* 7 * ( 1959 ) 404 [ annals phys .
* 281 * ( 2000 ) 774 ] .
s. u. chung , cern-71 - 08 .
j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * ( 2012 ) 010001 .
m. fischer , s. groote , j. g. korner , m. c. mauser and b. lampe , phys .
b * 451 * ( 1999 ) 406 [ hep - ph/9811482 ] .
m. fischer , s. groote , j. g. korner and m. c. mauser , phys .
d * 65 * ( 2002 ) 054036 [ hep - ph/0101322 ] .
b. lampe , nucl .
b * 454 * ( 1995 ) 506 .
f. del aguila and j. a. aguilar - saavedra , phys .
d * 67 * ( 2003 ) 014009 [ hep - ph/0208171 ] .
j. drobnak , s. fajfer and j. f. kamenik , phys .
d * 82 * ( 2010 ) 114008 [ arxiv:1010.2402 [ hep - ph ] ] .
j. a. aguilar - saavedra , nucl .
b * 812 * ( 2009 ) 181 [ arxiv:0811.3842 [ hep - ph ] ] ; nucl .
b * 821 * ( 2009 ) 215 [ arxiv:0904.2387 [ hep - ph ] ] .
a. czarnecki , j. g. korner and j. h. piclum , phys .
d * 81 * ( 2010 ) 111503 [ arxiv:1005.2625 [ hep - ph ] ] .
j. a. aguilar - saavedra , j. carvalho , n. f. castro , a. onofre and f. veloso , eur .
j. c * 53 * , 689 ( 2008 ) [ arxiv:0705.3041 [ hep - ph ] ] .
j. a. aguilar - saavedra , nucl .
b * 804 * ( 2008 ) 160 [ arxiv:0803.3810 [ hep - ph ] ] .
g. aad _ et al . _ [ atlas collaboration ] , arxiv:1205.3130 [ hep - ex ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , arxiv:1209.4533 [ hep - ex ] .
b. grzadkowski and z. hioki , nucl .
b * 484 * ( 1997 ) 17 [ hep - ph/9604301 ] .
p. doublet , f. richard , r. poschl , t. frisson and j. rouene , arxiv:1202.6659 [ hep - ex ] .
j. a. aguilar - saavedra , m. c. n. fiolhais and a. onofre , jhep * 1207 * ( 2012 ) 180 [ arxiv:1206.1033 [ hep - ph ] ] . | we introduce a new asymmetry in the decay @xmath0 , which is shown to be directly proportional to the polarisation of the top quark along a chosen axis , times a sum of @xmath1 helicity fractions .
the latter have already been precisely measured at the tevatron and the large hadron collider .
therefore , this new asymmetry can be used to obtain a model - independent measurement of the polarisation of top quarks produced in any process at hadron or lepton colliders . * model - independent measurement + of the top quark polarisation * j. a. aguilar
saavedra@xmath2 , r. v. herrero - hahn@xmath3 + _ @xmath4 departamento de fsica terica y del cosmos , universidad de granada , granada , spain _ + _ @xmath5 instituto de fsica de cantabria ( csic - uc ) , santander , spain _ + _ @xmath3 departamento de fsica aplicada , universidad de granada , granada , spain _ |
henry norris russell ( 1948 ) once delivered a lecture here in cambridge entitled `` the royal road of eclipses , '' about the determination of accurate parameters for eclipsing binary stars , and the promise that such systems held for progress in stellar astrophysics .
given the rapid progress on display at this meeting , it is clear that exoplanetary science too has its royal road : the royal road of transits .
figure 1 illustrates the happy situation in which the planet s orbit is viewed nearly edge - on , and the planet undergoes transits and occultations .
\(1 ) winn et al .
( 2 ) queloz et al . ( 2000 ) .
( 3 ) winn et al .
( 4 ) johnson et al .
( 5 ) bouchy et al .
( 6 ) narita et al .
( 7 ) cochran et al .
( 8) winn et al .
( 2008c ) .
( 9 ) winn et al .
( 2007d ) .
( 10 ) loeillet et al .
( 11 ) wolf et al .
( 12 ) hebrard et al .
( 13 ) joshi et al .
( 14 ) narita et al .
note : where more than one reference is listed , the value in column 2 is taken from the starred reference . to date , rm observations have been reported for 11 different exoplanetary systems , since the pioneering detections by queloz et al .
( 2000 ) and bundy & marcy ( 2000 ) .
this field is at the transition point between a mere handful of results to a large enough sample for meaningful general conclusions to be drawn and for rare surprises to emerge .
the results are summarized in table 2 , which is organized in order of measurement precision , with the most precise result listed first .
figure 8 shows some more recent data for hd 189733 and hat - p-2 , both of which display the antisymmetric waveform indicative of a well - aligned spin and orbit ( at least in projection ) .
both stars have candidate stellar companions , naturally raising the possibility of kozai cycles , but in both cases the projected spin - orbit angle is consistent with zero . in almost all cases ,
the results are consistent with good alignment , with measurement precisions ranging from @xmath0 to @xmath1 .
there are two possible exceptions .
the first is hd 17156 , for which narita et al .
( 2008 ) reported @xmath2 .
however , shortly after this meeting , cochran et al .
( 2008 ) presented new data giving @xmath3 .
the reason for the apparently significant disagreement between the data sets is not yet clear .
the other possible exception is xo-3 , for which hbrard et al .
( 2008 ) found @xmath4 deg , i.e. , a transverse rm effect in which only an anomalous blueshift was seen .
those authors expressed some caution about the result due to possible systematic errors .
further observations of both of these systems are warranted .
broadly speaking , the results show that large spin - orbit misalignments are fairly rare .
the time is now ripe for a statistical analysis of the ensemble results .
this would not only overcome the sky - projection limitation that is inherent in individual results , but could ultimately place constraints on the fraction of systems that have migrated via different channels .
a possible confounding factor is tidal coplanarization : even if the migration process resulted in a large value of @xmath5 , could the system have been eventually realigned by star - planet tidal interactions ?
the simple and widely - used model of tidal interactions , in which the equilibrium tidal bulge is shifted by a constant time lag , suggests that the timescale for inclination damping is much longer than the timescale for eccentricity damping , and that tidal spin - orbit realignment can be neglected ( winn et al .
however , more theoretical work is needed to go beyond this order - of - magnitude argument and define the scope for more complicated tidal and evolutionary models to affect the interpretation of the rm results .
it is possible , for example , that tidal dynamics were more important when the system was younger , or that the coplanarization timescale is shorter than expected ( mazeh 2008 ) .
special thanks are due to d. fabrycky for producing fig . 6 , to a. pl for providing the numerical results in fig . 4 ( which are more complete than the results i showed in my presentation ) , and to k. de kleer for help in producing fig .
i have enjoyed and benefited from collaborations with m. holman , j. johnson , g. marcy , n. narita , y. suto , d. fabrycky , e. turner , s. gaudi , t. mazeh , j. carter , m. nikku , and others too numerous to mention , on the topics presented here .
i am grateful to the soc for the opportunity to speak at the woodstock of transiting planets .
charbonneau , d. , brown , t. m. , noyes , r. w. , & gilliland , r. l. 2002 , _ apj _ , 568 , 377 charbonneau , d. , et al .
2005 , _ apj _ , 626 , 523 chatterjee , s. , ford , e. b. , matsumura , s. , & rasio , f. a. 2007 , arxiv : astro - ph/0703166 deming , d. , seager , s. , richardson , l. j. , & harrington , j. 2005 , _ nature _ , 434 , 740 deming , d. , harrington , j. , laughlin , g. , seager , s. , navarro , s. b. , bowman , w. c. , & horning , k. 2007 , _ apj _ ( letters ) , 667 , l199 everett , m. e. , & howell , s. b. 2001 , _ pasp _ , 113 , 1428 winn , j. n. , et al . 2005 , _ apj _ , 631 , 1215 winn , j. n. , et al .
2006 , _ apj _ ( letters ) , 653 , l69 winn , j. n. , holman , m. j. , & roussanova , a. 2007a , _ apj _ , 657 , 1098 winn , j. n. , et al . 2007b , _ aj _ , 134 , 1707 winn , j. n. , et al .
2007c , _ aj _ , 133 , 1828 winn , j. n. , et al . 2007d , _ apj _ ( letters ) , 665 , l167 winn , j. n. , holman , m. j. , shporer , a. , et al . , 2008a , _ aj _ , 136 , 267 winn , j. n. , et al .
2008b , _ apj _ , 683 , 1076 winn , j. n. , et al . 2008c , _ apj _ , 682 , 1283 wolf , a. s. , laughlin , g. , henry , g. w. , fischer , d. a. , marcy , g. , butler , p. , & vogt , s. 2007 , _ apj _ , 667 , 549 | by observing the transits of exoplanets , one may determine many fundamental system parameters .
i review current techniques and results for the parameters that can be measured with the greatest precision , specifically , the transit times , the planetary mass and radius , and the projected spin - orbit angle . |
in this article we study the coupling of the free dirac operator with measure - valued potentials developed in @xcite .
the aim of the present work is to give spectral properties of these couplings and to investigate the phenomenon of confinement .
given @xmath4 , the free dirac operator in @xmath2 is defined by @xmath5 where @xmath6 , @xmath7 is the family of _ pauli matrices_. although one can take @xmath8 in the definition of @xmath9 , throughout this article we always assume @xmath10 .
the work @xcite contains some results concerning @xmath0 for quite general singular measures @xmath11 in @xmath2 and suitable @xmath12-valued potentials @xmath3 .
however , in that article , most of the interest is focused on the case that @xmath11 is the surface measure on the boundary of a bounded regular domain in @xmath2 , and this is our setting for the present paper . in order to make an understandable exposition of the main results ,
let us first recall the basic ideas used in @xcite .
the ambient hilbert space is @xmath13 with respect to the lebesgue measure , and @xmath9 is defined in the sense of distributions . given a bounded regular domain @xmath14 with boundary @xmath15 and surface measure @xmath11 , in @xcite we find domains @xmath16 in which @xmath17 is an unbounded self - adjoint operator , where @xmath3 is a suitable @xmath12-valued potential .
the construction of @xmath18 relies on the following simple argument : by assumption , @xmath3 is @xmath12-valued .
thus , given @xmath19 , we can write @xmath20 in the sense of distributions for some @xmath21 . moreover , since @xmath22 , we can also write @xmath23 for some @xmath24 . therefore , @xmath25 in the sense of distributions , and so @xmath26 should be the convolution @xmath27 , where @xmath28 is a fundamental solution of @xmath9 .
in particular , @xmath29 to guarantee that @xmath0 is self - adjoint on @xmath30 , in @xcite we impose some relations between @xmath31 and @xmath32 with the aid of bounded self - adjoint operators @xmath33 .
that is to say , given suitable @xmath34 s , following ( [ intro 1 ] ) we find domains @xmath35 where @xmath0 is self - adjoint .
we consider the potential @xmath3 given by ( [ intro 1 ] ) as a `` generic '' potential since it seems to be prescribed from the begining , i.e. , @xmath20 for all @xmath36 , so @xmath3 is independent of @xmath34 . for an a priori given potential @xmath37 , the key idea in @xcite to construct a domain where @xmath38 is self - adjoint is to find a precise bounded self - adjoint operator @xmath39 so that @xmath40 for all @xmath41 , and the self - adjointness of @xmath38 on @xmath42 follows directly from the one of @xmath0 .
the generic potential @xmath3 given by ( [ intro 1 ] ) reflects the following rough idea : if we know that the gradient of a function @xmath26 has an absolutely continuous part @xmath31 and a singular part @xmath32 supported on @xmath15 ( in our setting , @xmath43 and @xmath22 ) , then @xmath26 must have a jump across @xmath15 , and this jump completely determines the singular part of the gradient ( in our setting , the jump determines the value @xmath44 ) . for a given potential @xmath37 ,
one manages to define a suitable domain @xmath18 such that , for any @xmath19 , the singular part which comes from the gradient on the jump of @xmath26 across @xmath15 agrees with @xmath45 .
as outlined above , in @xcite we show that any @xmath46 has non - tangential boundary values @xmath47 when we approach to @xmath15 from @xmath48 , where @xmath49 .
this enables us to consider , for example , the electrostatic shell potential @xmath50 for @xmath51 .
following the argument above , we construct a domain @xmath52 where @xmath53 is self - adjoint for all @xmath54 .
other similar potentials are also treated in @xcite . at this point
a remark is in order . in @xcite
( see also ( * ? ? ?
* section 2 ) ) the author provides , in a very general framework , all self - adjoint extensions of symmetric operators obtained by restricting a self - adjoint operator @xmath55 with domain @xmath56 to a dense ( and closed with respect to the graph norm ) subspace @xmath57 .
typically , @xmath55 is a differential operator and @xmath58 is the kernel of some trace operator along a null set . in particular , @xcite can be used to provide all the domains where @xmath0 is self - adjoint and , in this sense , some of the results in @xcite follow from the ones in @xcite . however , as regards applications , we make use of some explicit layer potentials on @xmath15 derived from our approach in @xcite which turn out to be very fruitful both in the description of the domains and in the study of the properties of the couplings under consideration . concerning the results in this article , in section [ s spec ] we show a general criterion for the existence of eigenvalues in @xmath59 for @xmath0 , namely proposition [ spec p1 ] , which has strong connections with ( * ? ? ?
* theorem 2.5 ) .
this criterion relates the existence of eigenvalues , which is a problem in the whole @xmath2 , with a spectral property of certain bounded operators in @xmath12 , which is a problem settled exclusively on @xmath15 .
afterwards , we show some applications to the case of electrostatic shell potentials @xmath60 .
in particular , theorem [ spec t1 ] shows that @xmath53 and @xmath61 have the same eigenvalues in @xmath59 for all @xmath62 , which is a special case of the more general isospectral transformation , and that @xmath53 has no eigenvalues in @xmath59 if @xmath63 is too big or too small .
this is an interesting feature , since it shows that there are lower and upper thresholds on the possible values of @xmath63 in order to have non - trivial eigenvalues in @xmath59 for @xmath53 , unlike what happens to the coupling of @xmath9 with similar potentials ( see remark [ threshold ] ) .
theorem [ symm a -a ] is another consequence of the general criterion , where we show that , under a symmetry assumption on @xmath11 , @xmath53 has an eigenvalue @xmath64 if and only if @xmath65 has @xmath66 as an eigenvalue . for completeness
, we also show in theorem [ spec t2 ] that if @xmath67 is connected then @xmath53 has no eigenvalues in @xmath68 $ ] .
section [ s sphere ] is devoted to the spectral study of @xmath53 when @xmath15 is the sphere @xmath69 our interest is to characterize the eigenvalues finding sharp constants of some appropriately chosen inequalities . in principle
, this is far from obvious for dirac hamiltonians because they are not semibounded operators .
however , this procedure has been successfully used in @xcite with hamiltonians with coulombic singularities . in that case , the sharp constants appear as strictly positive lower bounds for the absolute value of the commutator of two operators , being @xmath9 one of them .
therefore , this approach can be seen as another use of the uncertainty principle . in the current article
we proceed in the same vein and we give an uncertainty principle that concerns some @xmath70 bounded operators also related to @xmath9 . in theorem [ thmineq ] we obtain a sharp inequality on @xmath71 which turns out to be connected to the eigenvalues of @xmath53 . for its proof , we strongly use the classical spherical harmonics , so a generalization to other surfaces @xmath15 seems an interesting and challenging question . as a consequence of the above - mentioned inequality , we recover a well - known sharp lower bound for the @xmath72-dimensional riesz transform on the sphere ( see corollary [ riesz ] and @xcite ) . in lemma [ eigenfunction ]
we provide a specific criterion ( based on proposition [ spec p1 ] ) to generate eigenvectors of @xmath53 .
section [ further com ] contains some comments on the relation between the uncertainty principle and the eigenvectors of @xmath53 , positive results on the existence of eigenvalues , and an open question ( as far as we know ) and some consequences of an affirmative answer
. finally , in section [ s confinement ] we deal with the confinement phenomenon on regular surfaces . roughly speaking
, one says that an @xmath12-valued potential @xmath3 generates confinement with respect to @xmath9 and @xmath15 if the particles modelized by the evolution operator associated to @xmath0 never cross @xmath15 over time .
that is , if a function @xmath73 verifies the equation @xmath74 and @xmath75 is supported in @xmath48 , that @xmath3 generates confinement means that @xmath76 is supported in @xmath48 for all @xmath77 , so @xmath15 becomes impenetrable for the particles .
similarly to section [ s spec ] , in section [ s confinement ] we first show a general criterion on @xmath0 to generate confinement , namely theorem [ conf t1 ] .
this criterion is stated in terms of an algebraic property of certain bounded operators in @xmath12 , so as before we convert a problem in the whole @xmath2 to a problem exclusively settled in @xmath15 .
an application to electrostatic and lorentz scalar shell potentials is also shown .
in particular , in theorem [ conf appl ] we prove that , for @xmath78 , the potential @xmath79 generates confinement if and only if @xmath80 . this
was already known for the case of the sphere ( see @xcite ) , and we generalize it to sufficiently regular surfaces . for the reader only interested on confinement , we mention that section [ s confinement ] can be read independently of sections [ s spec ] and [ s sphere ] .
finally , we also want to recall @xcite , where the authors studied the confinement phenomenon for singular perturbations of general self - adjoint hamiltonian operators .
it is worth mentioning that , although all the applications in this article are concerned to the potentials @xmath60 and @xmath81 above - mentioned , the general criteria stated in proposition [ spec p1 ] and theorem [ conf t1 ] can be used as a first step to study the spectrum and confinement for the coupling of @xmath9 with other shell potentials . in a sense , once a potential @xmath37 is given , one must find the suitable operator @xmath34 ( mentioned in the beginning of the introduction ) so that @xmath40 for all @xmath82 , and then one must check the criteria for that specific @xmath34 .
since this article is a continuation of the study developed in @xcite , we assume that the reader is familiar with the notation , methods and results in there . however , in this section we recall some basic rudiments for the sake of completeness . given a positive borel measure @xmath83 in @xmath2 , set @xmath84 and denote by @xmath85 and @xmath86 the standard scalar product and norm in @xmath87 , i.e. , @xmath88 and @xmath89 for @xmath90 .
we write @xmath91 or @xmath92 interchangeably to denote the identity operator on @xmath87 .
we say that @xmath83 is @xmath93-dimensional if there exists @xmath94 such that @xmath95 for all @xmath96 , @xmath97 .
we denote by @xmath98 the lebesgue measure in @xmath2 .
let @xmath15 be the boundary of a bounded lipschitz domain @xmath14 , let @xmath11 and @xmath58 be the surface measure and outward unit normal vector field on @xmath15 respectively , and set @xmath49 , so @xmath99 .
note that @xmath11 is 2-dimensional .
since we are not interested in optimal regularity assumptions , for the sequel we assume that @xmath15 is of class @xmath100 ( see remark [ r regularity ] for the lipschitz case ) .
finally , we introduce the auxiliary space of measures @xmath101 observe that @xmath9 , which is symmetric and initially defined in @xmath102 ( @xmath103-valued functions in @xmath2 which are @xmath104 and with compact support ) , can be extended by duality to the space of distributions with respect to the test space @xmath102 and , in particular , it can be defined on @xmath105 .
the following lemma is concerned with the resolvent of @xmath9 , which will be very useful for the results below . as usual
, we denote by @xmath106 the complex conjugate of the transpose of @xmath107 , that is , @xmath108 and @xmath109 for all @xmath110 .
[ resolvent ] given @xmath64 , a fundamental solution of @xmath111 is given by @xmath112 i.e. , @xmath113 in the sense of distributions , where @xmath114 denotes the dirac measure on @xmath115 .
furthermore , @xmath107 satisfies @xmath116 , @xmath117 , and @xmath118 of ( * ? ? ? * section 2.2 ) , that is , * @xmath119 for all @xmath110 , * @xmath120 for all @xmath121 such that @xmath122 , * there exist @xmath123 such that * * @xmath124 for all @xmath125 , * * @xmath126 for all @xmath127 , * * @xmath128 where @xmath129 denotes the fourier transform in @xmath2 .
it is well known that @xmath130 is a fundamental solution of the helmhotz operator @xmath131 in @xmath2 .
note that @xmath132 , so @xmath133 and if we set @xmath134 then @xmath135 in the sense of distributions .
the explicit formula for @xmath107 follows by a straightforward computation .
the rest of the proof is analogous to ( * ? ? ?
* lemma 3.1 ) . given a positive borel measure @xmath83 in @xmath2 , @xmath136 , and @xmath96 , we set @xmath137 whenever the integral makes sense . actually , by lemma [ resolvent ] and ( * ? ? ?
* lemma 2.1 ) , if @xmath83 is a @xmath93-dimensional measure in @xmath2 with @xmath138 , then there exists some constant @xmath94 such that @xmath139 for all @xmath140 . in what follows we use a non standard notation , @xmath141 , to define the convolution of measures in @xmath105 with the fundamental solution of @xmath111 , @xmath107 .
capital letters , as @xmath142 or @xmath31 , in the argument of @xmath141 denote elements of @xmath143 , and the lowercase letters , as @xmath144 or @xmath32 , denote elements in @xmath12 . despite that this notation is non standard , it is very convenient in order to shorten the forthcoming computations .
given @xmath145 , we define @xmath146 then , the above inequality shows that @xmath147 for some constant @xmath94 and all @xmath145 , so @xmath148 .
moreover , following ( * ? ? ?
* section 2.3 ) one can show that @xmath149 in the sense of distributions .
this allows us to define a `` generic '' potential @xmath3 acting on functions @xmath150 by @xmath151 so that @xmath152 in the sense of distributions . for simplicity of notation ,
we write @xmath153 . in order to construct a domain of definition where @xmath0 is self - adjoint , in @xcite we had to consider the trace operator on @xmath15 . for @xmath154 ,
one defines the trace operator on @xmath15 by @xmath155 .
then , @xmath156 extends to a bounded linear operator @xmath157 ( see ( * ? ? ?
* proposition 2.6 ) , for example ) , where @xmath158 denotes the sobolev space of @xmath103-valued functions such that all its components have all their derivatives up to first order in @xmath159 . from lemma [ resolvent ]
, we have @xmath160 for some @xmath94 and all @xmath161 ( see ( * ? ? ?
* lemma 2.8 ) ) , thus we can define @xmath162 and it satisfies @xmath163 for all @xmath161 .
note that , for @xmath164 , the above definitions recover the ones in ( * ? ? ?
* section 2.3 ) .
the next lemma , which is a generalization of ( * ? ? ? * lemma 3.3 ) , will be used in the sequel .
[ l jump ] given @xmath165 , @xmath166 and @xmath167 , set @xmath168 where @xmath169 means that @xmath170 tends to @xmath166 non - tangentially .
then @xmath171 and @xmath172 are bounded linear operators in @xmath12 .
moreover , the following holds : * @xmath173 ( plemelj sokhotski jump formulae ) , * @xmath174 for @xmath175 $ ] .
the proof of the lemma is completely analogous to the one of ( * ? ? ?
* lemma 3.3 ) , so we omit it . concerning the second term in @xmath117 , once we know that @xmath176 , then , multiplying the equation by @xmath177 from the left and from the right and using that @xmath178 , we obtain @xmath179 . in accordance with the notation introduced in @xcite , for the case @xmath164 , we write @xmath180 , @xmath181 , @xmath182 and @xmath183 instead of @xmath184 , @xmath185 , @xmath186 and @xmath187 , respectively .
finally , we recall our main tool to construct domanis where @xmath0 is self - adjoint , namely ( * ? ? ? * theorem 2.11 ) .
actually , the following theorem is a direct application of ( * ? ? ?
* theorem 2.11 ) to @xmath0 , and we state it here in order to make the exposition more self - contained . given an operator between vector spaces @xmath188 , denote @xmath189 [ pre t1 ] let @xmath33 be a bounded operator .
set @xmath190 where @xmath191 and @xmath23 for all @xmath192 .
if @xmath34 is self - adjoint and @xmath193 is closed , then @xmath194 is an essentially self - adjoint operator . in that case , if @xmath195 is closed in @xmath143 , then @xmath196 is self - adjoint .
in particular , if @xmath34 is self - adjoint and semi - fredholm ( see for example ( * ? ? ?
* definition 1.40 ) ) , then the operator @xmath196 in theorem [ pre t1 ] is self - adjoint .
recall also that any bounded , semi - fredholm and self - ajoint operator on a hilbert space is actually fredholm .
[ r regularity ] all the results in this article which are proved without the use of fredholm s theorem are valid when @xmath15 is just lipschitz ( but not necessarily of class @xmath100 ) , or when it is the graph of a lipschitz function from @xmath197 to @xmath198 .
actually , the smoothness and boundedness assumptions on @xmath15 are exclusively required for compactness purposes , in order to use fredholm s theory .
in this section , we show a criterion for the existence of eigenvalues in @xmath59 for @xmath0 , namely proposition [ spec p1 ] .
this criterion relates the eigenvalues with a spectral property of certain bounded operators in @xmath12 .
afterwards , we show some applications to the case of electrostatic shell potentials .
[ spec p1 ] let @xmath196 be as in theorem [ pre t1 ] . given @xmath64 , there exists @xmath192 such that @xmath199 if and only if @xmath200 and @xmath201 .
therefore , @xmath202 if and only if @xmath203 .
let @xmath64 and assume that @xmath199 for some @xmath192 . then , @xmath204 and therefore @xmath205 in the sense of distributions .
this yields @xmath206 , and applying @xmath141 we conclude that @xmath201 . in particular , we have seen that @xmath207 , and thus lemma [ l jump]@xmath116 yields @xmath208 summing these equations , we obtain @xmath209 , which is equivalent to @xmath200 when @xmath210 . for the case
@xmath164 , from ( [ spec p1 eq1 ] ) we have that @xmath211 and , since @xmath212 , @xmath213 . on the contrary , assume that there exists @xmath165 such that @xmath200 .
set @xmath201 and @xmath214 . as we did in ( [ spec p1 eq1 ] ) , @xmath201 implies @xmath206 , so @xmath215 and @xmath216 in particular , @xmath217 , thus lemma [ l jump]@xmath116 once again gives @xmath218 for @xmath210 , this and the assumption on @xmath32 imply @xmath219 , thus @xmath212 . as before , the case @xmath164 can be easily treated separately . the proposition is proved . in ( * ? ? ?
* theorem 3.8 ) we proved that , if @xmath220 and @xmath196 is the operator defined by @xmath221 where @xmath222 and @xmath223 for @xmath192 , then @xmath224 is self - adjoint .
moreover , we also showed that @xmath225 on @xmath226 for all @xmath62 , so the self - adjointness was a consequence of theorem [ pre t1 ] .
[ spec l1 ] set @xmath227 then , we have @xmath228 we write @xmath229 note that @xmath230 and by the mean value theorem we have the same estimate for @xmath231 . using ( [ spec l1 eq1 ] ) , that @xmath11 is 2-dimensional and rather standard arguments ( essentially , that @xmath15 is bounded and the generalized young s inequality ) , it is easy to show that the convolution operator with kernel @xmath232 is bounded in @xmath12 uniformly on @xmath64 .
finally , the @xmath12 boundedness of the singular intergal operator with convolution kernel @xmath233 follows , for example , by ( * ? ? ?
* theorem 20.15 ) , working component by component .
note that this last operator does not depend on @xmath234 , so the lemma is proved .
[ spec t1 ] let @xmath235 , let @xmath236 be as in @xmath237 , and @xmath64 .
then @xmath202 if and only if @xmath238 .
in particular , @xmath53 and @xmath61 have the same eigenvalues in @xmath59 .
moreover , if @xmath239 $ ] then @xmath234 is not an eigenvalue of @xmath53 , and if @xmath240 $ ] then @xmath53 has no eigenvalues in @xmath59 , where @xmath241 . that @xmath202 if and only if @xmath238 is a direct consequence of the definition of @xmath34 and proposition [ spec p1 ] .
assume that @xmath238 .
then there exists a non - trivial @xmath165 such that @xmath242 . using lemma [ l jump]@xmath117
we deduce that @xmath243 .
this easily implies that @xmath244 for @xmath245 , so @xmath246 . the same arguments actually show that @xmath238 if and only if @xmath246 , and by the first part of the theorem , @xmath53 and @xmath61 have the same eigenvalues in @xmath59 . for the last part of the theorem , since @xmath178
, we easily have @xmath247 for all @xmath165 . combining this estimate with lemma [ l jump]@xmath117
, we obtain @xmath248 hence , @xmath249 if @xmath250 .
by the first part of the theorem , if there exists @xmath212 such that @xmath251 for some @xmath64 , then there exists a non - trivial @xmath165 such that @xmath242 .
thus , ( [ spec t1 eq1 ] ) easily implies @xmath252\subset[1/c,4c]$ ] , and the theorem follows . [ threshold ] theorem [ spec t1 ] shows that the coupling of the free dirac operator @xmath9 with electrostatic shell potentials @xmath60 relative to @xmath15 does not generate eigenvalues either for big or small values of @xmath63 . recall that the coupling of @xmath9 with the coulomb potential @xmath253 generates eigenvalues for any small @xmath63 ( see ( * ? ? ?
* theorem 1 ) , for example ) and is not essentially self - adjoint if @xmath63 is big enough . on the other hand , it is not hard to see that there exists a sequence @xmath254 with @xmath255 for @xmath256 such that the coupling of @xmath9 with the potentials @xmath257 generates eigenvalues .
[ lambda - a ] if we define @xmath258 , by lemma [ l jump]@xmath117 we have @xmath259 where @xmath260 and @xmath261 . following the arguments of ( * ? ? ?
* lemma 3.5 ) , one can show that @xmath262 is a compact operator , as well as @xmath263 .
moreover , @xmath263 is easily seen to be self - adjoint , and hence it has a non - trivial eigenfunction .
therefore , for any @xmath64 there exists some @xmath264 such that @xmath238 by ( [ spec t1 eq2 ] ) , so the second part of theorem [ spec t1 ] is meaningful .
note that ( [ spec t1 eq1 ] ) yields @xmath265 for all @xmath64 .
in particular , this lower bound of @xmath266 does not depend on @xmath15 .
for an upper bound , this type of result may not be expected because , roughly speaking , the abruptness of @xmath15 may play a role in questions concerning upper bounds for the norm of singular integral operators on lipschitz surfaces ( see for example ( * ? ? ?
* chapter 20 ) for related topics ) .
the following theorem generalizes some results of ( * ? ? ?
* theorem 3.8@xmath117 ) .
[ symm a -a ] assume that @xmath267 , where @xmath268 for @xmath96 and @xmath269 is the image measure of @xmath11 with respect to @xmath270 .
let @xmath235 and @xmath53 be as in @xmath237 . if @xmath53 has some eigenvalue @xmath64 , then @xmath65 has @xmath66 as an eigenvalue . from theorem [ spec t1 ]
we see that , if @xmath53 has @xmath234 as an eigenvalue , there exists a nontrivial @xmath165 such that @xmath271 .
set @xmath272 , where @xmath273 and @xmath274 denotes the identity operator in @xmath70 .
obviously , @xmath275 and , since @xmath267 , we have @xmath276 .
it is straightforward to check that @xmath277 for all @xmath278 .
therefore , using the assumptions on @xmath11 and on @xmath32 , @xmath279 that is , @xmath280 for some @xmath275 . by theorem [ spec t1 ]
once again , @xmath65 has @xmath66 as an eigenvalue .
[ spec t2 ] let @xmath235 and let @xmath236 be as in @xmath237 .
if @xmath67 is connected , then @xmath196 has no eigenvalues in @xmath68 $ ] .
let @xmath281 $ ] and @xmath212 such that @xmath199 .
we will see that @xmath282 .
note that @xmath283 in @xmath67 and hence @xmath284
. therefore , @xmath285 in @xmath67 for @xmath286 . since @xmath212 , then @xmath214 for some @xmath145 such that @xmath287 .
using that @xmath10 , it is not hard to show that @xmath288 where @xmath289 and @xmath290 denotes the surface measure on @xmath291 . therefore , rellich s lemma yields @xmath282 in a neighbourhood of infinity ( see @xcite , for example ) , and thus @xmath282 in @xmath67 by unique continuation and the connectivity assumption . in particular @xmath292 in @xmath15 , and
so lemma [ l jump]@xmath116 and the definition of @xmath34 give @xmath293 this means that @xmath294 , so @xmath295 and then @xmath296 in @xmath15 .
it only remains to check that @xmath282 in @xmath297 , but this goes along well known lines . since @xmath199 , then @xmath283 in @xmath297 .
if one integrates by parts on the identities @xmath298 where @xmath299 and @xmath300 is the ball centered at @xmath301 and with radius @xmath302 , and then one takes @xmath303 , one can show that @xmath304 where @xmath305 since @xmath296 in @xmath15 , we conclude from ( [ spec t2 eq1 ] ) that @xmath26 vanishes in @xmath297 , and thus @xmath282 .
although the definition of @xmath34 does not make sense for @xmath306 , one can replace @xmath287 by @xmath307 in the definition of @xmath226 in ( [ domain ] ) , in order to recover the free dirac operator @xmath9 when @xmath306 .
it is well known that @xmath9 does not have eigenvalues in @xmath68 $ ] , so theorem [ spec t2 ] also holds for @xmath306 . with the same arguments , one can check that theorem [ spec t2 ] also holds for other @xmath12-valued potentials different from @xmath60 , as far as they yield a suitable @xmath34 for which a relation like ( [ spec t2 eq2 ] ) implies that @xmath296 in @xmath15 .
this section is focused on the coupling @xmath53 given in section [ ss electro ] ( see ( [ domain ] ) ) , and mostly in the case that @xmath15 is the sphere .
however , the following two lemmata hold for any @xmath15 and @xmath11 as in section [ s preliminaries ] .
first , we need some definitions . for @xmath175 $ ] and @xmath308 , where the @xmath309 s are the family of pauli matrices ( see ( [ freedirac ] ) ) , define @xmath310 for @xmath311 .
given @xmath312 and @xmath313 , set @xmath314 that @xmath315 and @xmath316 are bounded operators in @xmath70 can be verified similarly to the case of @xmath317 in @xmath12 , we omit the details .
moreover , note that @xmath318 [ sphere l1 ] for @xmath175 $ ] , the following hold : * the anticommutator @xmath319 vanishes identically , * @xmath320 . from lemma
[ l jump]@xmath117 and ( [ sphere eq1 ] ) we have @xmath321 and the lemma follows . [
sphere l2 ] @xmath315 is a positive operator in @xmath70 for all @xmath64 .
we want to verify that @xmath322 for all @xmath312 .
if we set @xmath323 for @xmath96 , it is not hard to check that @xmath324 and that it satisfies @xmath325 moreover , since @xmath326 for all @xmath311 , a proof analogous to the one of ( * ? ? ?
* lemma 3.3@xmath116 ) shows that @xmath327 where @xmath328 denote the boundary values of @xmath329 when one approaches to @xmath15 non - tangentially from @xmath48 .
therefore , using ( [ sphere l2 eq1 ] ) , ( [ sphere l2 eq2 ] ) , and the divergence theorem , we conclude @xmath330 throughout this section we set @xmath331 , @xmath332 , @xmath11 denotes the surface measure on @xmath71 , and @xmath333 for @xmath334 .
we also use the notation of ( * ? ? ?
* section 4.6.4 ) .
let @xmath335 be the usual spherical harmonics .
they are defined for @xmath336 , and @xmath337 , and they satisfy @xmath338 , where @xmath339 denotes the usual spherical laplacian .
moreover , @xmath335 form a complete orthonormal set in @xmath340 . for @xmath341 and @xmath342 , set @xmath343 then @xmath344 form a complete orthonormal set in @xmath70 , and @xmath345 where @xmath346 ( see ( * ? ? ? * equation ( 4.121 ) and the remark in page 127 ) ) .
[ sphere l3 ] given @xmath64 , there exist positive numbers @xmath347 and purely imaginary numbers @xmath348 for all @xmath341 and @xmath342 , such that : * @xmath349 and @xmath350 .
moreover , @xmath351 * @xmath352 and @xmath353 .
moreover , @xmath354 for any @xmath355 and @xmath356 , we identify the spherical harmonic @xmath335 with its homogeneous extension of degree @xmath115 to @xmath357 .
that is , @xmath358 for all @xmath311 .
in particular , @xmath359 is a homogeneous polynomial of degree @xmath355 which is harmonic .
we use the same identification for @xmath360 .
_ proof of @xmath116_. in order to prove the first identity in @xmath116 , fix @xmath361 and @xmath362 and set @xmath363 given @xmath302 , we define @xmath364 .
it is easy to verify that @xmath365 converges to @xmath11 in the weak@xmath366 topology when @xmath367 .
in particular , since @xmath11 and @xmath365 have compact support and @xmath368 is continuous in @xmath357 and has exponential decay at infinity , it is not hard to show that actually @xmath369 in @xmath370 , where @xmath371 the term on the right hand side of last equality in ( [ uncert l1 eq6 ] ) denotes the usual convolution of ( matrix and vectors of ) functions in @xmath159 . applying the fourier
transform to ( [ uncert l1 eq6 ] ) and using that @xmath368 is a fundamental solution of @xmath372 , we obtain @xmath373 note that , for any @xmath374 , @xmath375 is a bounded radial function with compact support , thus ( * ? ? ?
* corollary in page 72 ) shows that @xmath376 for some radial function @xmath377 depending on @xmath378 but not on @xmath362 .
hence , @xmath379 since @xmath380 is also a radial function , we can use that @xmath381 is the identity operator and ( * ? ? ?
* corollary of page 72 ) in ( [ uncert l1 eq7 ] ) to deduce that @xmath382 for some radial function @xmath383 depending on @xmath378 but not on @xmath362 . finally , using that @xmath369 and ( [ uncert l1 eq8 ] )
, we conclude that @xmath384 for some radial function @xmath385 .
we already know that @xmath315 is a bounded operator in @xmath70 , and since @xmath386 for @xmath387 , one can check that @xmath388 which implies that @xmath389 is continuous at @xmath390 . then , by setting @xmath391 , ( [ uncert l1 eq9 ] ) shows that @xmath392 concerning the second statement in @xmath116 , since @xmath386 for @xmath387 , it is easy to check that @xmath315 is a compact operator in @xmath70 , so the eigenvalues of @xmath315 form a bounded sequence which has @xmath393 as the only possible accumulation point ( see ( * ? ? ?
* fredholm s theorem ( 0.38)@xmath394 ) , for example )
. therefore , @xmath350 .
let us now prove the last statement in @xmath116 . from lemma
[ sphere l2 ] , @xmath315 is a positive operator , which implies that @xmath395 for all @xmath361 .
moreover , @xmath396 . following ( * ? ? ?
* generalized young s inequality ( 0.10 ) ) and since @xmath71 is invariant under rotations , it is easy to see that @xmath397 where @xmath398 ( we have identified the matrix @xmath368 with its scalar version ) .
consider the change of variables to polar coordinates in @xmath71 @xmath399 then , we have @xmath400 where we used the change of variables @xmath401 in the last equality above .
to finish the proof of @xmath116 , it only remains to check that @xmath402 .
note that @xmath403 where @xmath404 is some constant
. therefore , @xmath405 and we are done . _
proof of @xmath117_. fix @xmath361 and @xmath362 .
recall from ( [ uncert l1 eq11 ] ) and ( [ uncert l1 eq10 ] ) that , for @xmath96 , we have @xmath406 for some function @xmath407 , where @xmath408 and @xmath409 .
furthermore , @xmath410 and , by similar arguments to the ones that prove that @xmath389 is continuous at @xmath390 , one can show that @xmath411 and @xmath412 exist ; we omit the details . since @xmath413 , similarly to ( [ sphere l2 eq2 ] )
one can check that @xmath414 where @xmath415 denote the boundary values of @xmath416 when one approaches to @xmath15 non - tangentially from @xmath48 .
it is well known and quite easy to see that @xmath417 , where @xmath418 and @xmath346 . combining this with ( [ uncert l1 eq2 ] ) , ( [ uncert l1 eq1 ] ) and ( [ uncert eq1 ] )
, we obtain @xmath419 i.e. , @xmath352 for some @xmath420 .
by @xmath421 , lemmata [ sphere l1]@xmath117 and [ sphere l3]@xmath116 , @xmath422 which yields @xmath423 from the last statement in @xmath116 , we have @xmath424 thus @xmath425 by ( [ uncert l1 eq3 ] ) , and that means that @xmath348 are purely imaginary numbers .
the last statement in @xmath117 follows by ( [ uncert l1 eq3 ] ) and ( [ uncert l1 eq4 ] ) , so it only remains to prove that @xmath353 .
for that purpose , we use the first identity in @xmath117 and that @xmath426 and @xmath316 are symmetric operators to see that @xmath427 since @xmath428 and we already know that @xmath348 are purely imaginary , we obtain from ( [ uncert l1 eq5 ] ) that @xmath429 , which implies that @xmath353 .
the lemma is finally proved .
note that , if we know that @xmath430 for all @xmath361 , the relation @xmath353 also follows from lemma [ sphere l1]@xmath116 .
the following theorem is based on an uncertainty principle on the sphere and goes on the lines of ( * ? ? ?
* theorem 1 ) .
[ thmineq ] given @xmath431 and @xmath64 , the operator @xmath432 is invertible in @xmath70 . furthermore , for any @xmath433 and any @xmath434 , we have @xmath435 where @xmath436 if @xmath437 is such that @xmath438 , then equality in @xmath439 holds by taking @xmath440 for any @xmath441 and @xmath442 recall that @xmath315 is a positive operator by lemma [ sphere l2 ] , thus @xmath432 is positive and invertible in @xmath70 for all @xmath431 by ( * ? ? ?
* theorem 12.12@xmath443 ) , and the inverse is also a positive operator .
since @xmath395 for all @xmath361 and @xmath350 by lemma [ sphere l3]@xmath116 , lemma [ sphere l3]@xmath117 shows that @xmath444 for all @xmath361 and @xmath445 .
therefore , there exists some @xmath437 such that @xmath446 , and thus @xmath447 is well defined .
the estimate of @xmath447 from below is stated in lemma [ sphere l3]@xmath117 . given @xmath361 and @xmath362
, we define @xmath448 we will prove ( [ ineq ] ) by estimating @xmath55 from above and from below . from lemma [ sphere l3 ] we know that @xmath449 are purely imaginary .
therefore , arguing as in ( [ uncert l1 eq5 ] ) , we have @xmath450 to estimate @xmath55 from above , we use lemma [ sphere l3 ] , ( [ uncert eq1 ] ) , cauchy - schwarz inequality , and that @xmath451 for all @xmath452 , to deduce that @xmath453 from ( [ eq2 ] ) and ( [ eq3 ] ) , we see that ( [ ineq ] ) holds for @xmath454 .
the functions @xmath344 with @xmath341 and @xmath342 form a complete orthonormal system in @xmath70 . hence , to prove ( [ ineq ] ) in full generality , we first write any @xmath312 as a linear combination of the @xmath344 s and we expand the left and right hand side of ( [ ineq ] ) in terms of this basis . then , using the orthogonality and , for the right hand side of ( [ ineq ] ) , that @xmath455 from lemma [ sphere l3 ] , we conclude that ( [ ineq ] ) holds for all @xmath312 if and only if it holds for all @xmath344 .
therefore , ( [ ineq ] ) is finally proved .
it only remains to check the last statement of the theorem .
let @xmath437 be such that @xmath438 ( we already know that such @xmath437 exists ) .
then , for the functions @xmath456 the inequality in ( [ eq2 ] ) becomes an equality .
furthermore , for @xmath457 satisfying ( [ delta ] ) , we have @xmath458 which implies that the inequality in ( [ eq3 ] ) is an equality for @xmath459 .
recall from lemma [ sphere l3]@xmath117 that @xmath460 for all @xmath361 .
hence , for any @xmath437 such that @xmath438 , we have two possible elections of the subindex , say @xmath461 and @xmath462 , and therefore two possible values of @xmath457 for which equality in ( [ ineq ] ) holds . hence , we get two ( a priori different ) sharp inequalities
. the same observation applies if such @xmath437 is not unique .
theorem [ thmineq ] has an interesting consequence concerning a lower bound for the @xmath72-dimensional riesz transform on the sphere .
given a finite borel measure @xmath83 in @xmath2 , @xmath463 and @xmath96 , one defines the @xmath72-dimensional riesz transform of @xmath464 as @xmath465 whenever the limit makes sense .
it is well known that @xmath466 is a bounded operator for 2-dimensional uniformly rectifiable ad regular measures @xmath83 in @xmath2 ( see @xcite for a deep study on this subject ) .
in particular , @xmath466 is a bounded operator when @xmath83 is the surface measure of a bounded lipschitz domain .
this means that @xmath467 for some constant @xmath94 and all @xmath463 .
much less is said about lower @xmath468-bounds for the riesz transform .
however , in the case of the sphere , from the results in @xcite ( see also ( * ? ? ?
* equation ( 4.6.9 ) ) ) one can easily show that the riesz transform ( multiplied by a suitable constant ) is an isometry , providing sharp constants to the above - mentioned inequalities .
the following corollary of theorem [ thmineq ] yields the sharp constant of the inequality from below on @xmath71 .
set @xmath469 for @xmath311 , and @xmath470 [ riesz ] the following inequalities hold and they are sharp : * @xmath471 for all @xmath312 . * @xmath472 for all real - valued @xmath473 .
set @xmath164 and @xmath474 in theorem [ thmineq ] . given @xmath312 , ( [ ineq ] )
yields @xmath475 where @xmath476 .
notice that , by lemma [ sphere l3 ] , @xmath347 are uniformly bounded and @xmath477 uniformly in @xmath361 when @xmath478 .
in particular , @xmath479 when @xmath478 .
recall that @xmath480 and @xmath481 are defiend by means of the convolution kernels @xmath482 we define @xmath483 for @xmath311 , and @xmath484 that @xmath263 and @xmath485 are bounded operators in @xmath70 follows essentially as in the case of @xmath480 and @xmath481 .
moreover , it is not hard to show that @xmath486 ( see the proof of lemma [ spec l1 ] for a related argument ) . roughly speaking , in @xmath71 , @xmath480 and @xmath481
are compact perturbations of @xmath263 and @xmath485 which depend on @xmath487 continuously .
therefore , if we take @xmath478 in ( [ sphere c1 eq1 ] ) and we use that @xmath488 , we obtain @xmath489 for all @xmath434 .
minimazing in @xmath457 , i.e. , taking @xmath490 , we get @xmath491 which is the inequality in corollary [ riesz]@xmath116 .
corollary [ riesz]@xmath117 follows from ( [ sphere c1 eq2 ] ) by taking @xmath492 for any real - valued @xmath473 . that the inequalities are sharp is a consequence of the fact that ( [ sphere c1 eq1 ] ) is sharp for @xmath457 as in ( [ delta ] ) .
since @xmath479 and @xmath493 for @xmath478 ( recall that we have set @xmath164 and @xmath474 ) , using ( [ sphere c1 eq1 ] ) and ( [ pirris ] ) one can check that ( [ sphere c1 eq2 ] ) is sharp , and the corollary follows .
the following lemma gives a specific criterion based on proposition [ spec p1 ] to generate eigenvectors of @xmath53 .
[ eigenfunction ] let @xmath494 be as in @xmath237 with @xmath332 .
if @xmath431 and @xmath64 satisfy @xmath495 then , for any @xmath362 , @xmath344 gives rise to an eigenfunction of @xmath494 with eigenvalue @xmath234 .
let @xmath264 , @xmath234 and @xmath361 be as in the lemma .
since @xmath431 , @xmath496 is invertible by theorem [ thmineq ] . hence we can define @xmath497 in particular , we have the relation @xmath498 using lemma [ sphere l3 ] , we have @xmath499 if @xmath431 and @xmath64 satisfy ( [ a ] ) , then @xmath500 which by ( [ eigenfunction eq1 ] ) implies that @xmath501 finally , combining ( [ sphere eq1 ] ) , ( [ eigenfunction eq2 ] ) and ( [ eigenfunction eq3 ] ) , we obtain @xmath502 which means that @xmath503 . following proposition [ spec p1 ] ( see also theorem [ spec t1 ] ) , if we set @xmath201 and @xmath214 , then @xmath504 .
the lemma is proved . for the case @xmath505
, one can develop results analogous to theorem [ thmineq ] and lemma [ eigenfunction ] by repeating the arguments involved in the proofs but by using the invertibility of @xmath506 instead of @xmath496 .
we leave the details for the reader .
from lemma [ spec l1 ] we have that @xmath241 , and in theorem [ spec t1 ] we proved that if @xmath507\ ] ] then @xmath53 has no eigenvalues in @xmath59 ( recall that @xmath508 by ( [ spec t1 eq1 ] ) ) .
furthermore , in remark [ lambda - a ] we showed that , for any @xmath64 , there exists some @xmath264 such that @xmath53 has @xmath234 as an eigenvalue . from these results ,
it is not clear what can be said positively about the set of @xmath264 s for which there exist an eigenvalue of @xmath53 .
thanks to lemma [ eigenfunction ] , we can give a bit more of information in the case of the sphere , but first we need to do some computations .
lemma [ sphere l3]@xmath116 gives a precise value for @xmath509 in terms of @xmath10 and @xmath64 , that is , @xmath510 following a similar argument , we are going to prove that @xmath511 where @xmath512 corresponds to @xmath513 with @xmath514 . from ( [ uncert eq1 ] ) and ( [ d0 eq ] ) , we have @xmath515 therefore , if we identify the matrix @xmath368 with its scalar version and we set @xmath398 , lemma [ sphere l3]@xmath116 yields @xmath516 that @xmath517 follows from ( [ d1 eq1 ] ) , but it can be also verified using the change of variables ( [ polar ] ) and noting that the resulting integrals contain a @xmath518 or a @xmath519 , which integrated on @xmath520 $ ] vanish . on the other hand , using ( [ polar ] ) , @xmath521 where we used the change of variables @xmath401 and integration by parts in the last equality above . combining ( [ d1 eq1 ] ) and ( [ d1 eq2 ] ) , we get ( [ d1 eq ] ) , as desired . recall that lemma [ sphere l3]@xmath116 states that @xmath395 for al @xmath361 .
it is an exercise to check directly from ( [ d1 eq ] ) that @xmath522 for all @xmath64 .
we now turn to lemma [ eigenfunction ] , which can be used to provide eigenfunctions of @xmath53 under some assumptions on @xmath431 and @xmath64 . for @xmath514 , ( [ a ] )
reads as @xmath523 using ( [ d0bis eq ] ) and ( [ d1 eq ] ) , it is not difficult to see that if @xmath264 is very big or very small , then ( [ pepaa ] ) and ( [ pepab ] ) do not hold for any @xmath64 , so lemma [ eigenfunction ] can not be used .
this agrees with the above - mentioned result on non - existence of eigenvalues given by ( [ pepa2 ] ) .
let us take @xmath524 for simplicity .
figure [ fig1 ] shows the set of points @xmath525 such that ( [ pepaa ] ) and ( [ pepab ] ) hold .
we see that for any @xmath526 we can take a @xmath264 for which ( [ pepaa ] ) holds , hence there exists an eigenfunction with eigenvalue @xmath234 , by lemma [ eigenfunction ] .
this agrees with remark [ lambda - a ] .
however , from figure [ fig1 ] we also see that for any @xmath264 in some interval there exists an @xmath526 such that ( [ pepaa ] ) or ( [ pepab ] ) hold .
more precisely , using the computer one can show that if @xmath527 then there exists @xmath526 such that ( [ pepaa ] ) holds , and if @xmath528 then there exists @xmath526 such that ( [ pepab ] ) holds . in particular , if @xmath529 then there exists @xmath526 such that either ( [ pepaa ] ) or ( [ pepab ] ) hold . by lemma [ eigenfunction ] , this means that the set of @xmath264 s for which there exists an eigenfunction of @xmath53 contains the interval @xmath530 ( recall that we have set @xmath524 for these calculations ) . the combination of theorem [ thmineq ] and lemma [ eigenfunction ] yields an interesting relation between the minimizing functions of ( [ ineq ] ) and some eigenfunctions of @xmath53 .
more precisely , let @xmath437 be such that @xmath531 ( recall that @xmath460 for all @xmath361 ) .
assume that the given @xmath431 and @xmath64 satisfy @xmath532 i.e. , the election of the first sign in ( [ a ] ) holds for this particular @xmath437 .
then lemma [ eigenfunction ] shows that @xmath533 gives rise to an eigenfunction of @xmath494 with eigenvalue @xmath234 , for any @xmath534 .
fix @xmath457 as in ( [ delta ] ) , but we chose the first sign on the possible definitions of @xmath457 , i.e. , @xmath535 then the functions @xmath533 are minimizers of ( [ ineq ] ) , that is , they attain the equality in ( [ ineq ] ) . once @xmath264 , @xmath234 and @xmath457 are fixed depending on @xmath437 , let @xmath536 be set of @xmath361 s such that @xmath537 and that one sign election in ( [ delta ] ) is satisfied for @xmath361 .
in particular , @xmath538 .
given @xmath539 , since @xmath540 and both @xmath437 and @xmath361 satisfy ( [ delta ] ) , we easily deduce that either @xmath541 an inspection of the proof of theorem [ thmineq ] shows that , * if @xmath542 then @xmath543 are minimizers of ( [ ineq ] ) , and * if @xmath544 , then @xmath545 are minimizers of ( [ ineq ] ) . actually , because of the orthogonality , it can be seen that any minimizer of ( [ ineq ] ) must be a linear combination of these functions indexed by @xmath539 and by a choose of the sign in @xmath378 depending on ( [ cases ] ) .
similarly , for any @xmath539 , ( [ cases ] ) and ( [ a1 ] ) show that @xmath546 holds for a suitable election of the sign in @xmath378 , and the corresponding functions @xmath547 give rise to eigenfunctions of @xmath494 with eigenvalue @xmath234 . combining the above - mentioned arguments we can conclude that , once @xmath264 , @xmath234 , and @xmath457 are properly chosen
, then any function which attains the equality in ( [ ineq ] ) give rise to an eigenfunction of @xmath494 with eigenvalue @xmath234 .
roughly speaking , the minimizers of ( [ ineq ] ) provide eigenfunctions of @xmath494 . from lemma
[ sphere l3](i ) , we know that @xmath548 for all @xmath549 and @xmath350 , so it is to be expected that the following question has a positive answer .
[ ques ] is it true that @xmath550 for all @xmath551 ?
if so , from lemma [ sphere l3]@xmath117 we see that the minimum in the definition of @xmath447 in ( [ ineq ] ) would be attained only at one particular @xmath361 , namely @xmath514 . actually , @xmath447 could be calculated explicitely using ( [ d0bis eq ] ) and ( [ d1 eq ] ) , that is , @xmath552 the same could be said about the two possible values of @xmath457 in ( [ delta ] ) , say @xmath553 if question [ ques ] has a positive answer , the argument of section [ delib ] becomes much more transparent , since in this case @xmath554 .
furthermore , it would yield the following result : _ let @xmath64 and @xmath431 . then , for any @xmath312 , @xmath555 where @xmath447 is given by @xmath556 .
the equality in @xmath557 is only attained at linear combinations of @xmath558 for @xmath559 . moreover , if @xmath560 then the minimizers of @xmath557 give rise to eigenfunctions of @xmath53 .
_ these conclusions also hold if we exchange the roles of @xmath509 and @xmath512 in @xmath557 and @xmath561 and we replace @xmath558 by @xmath562 ( that is , we exchange the roles of @xmath563 and @xmath564 for @xmath514 ) .
in this section , we show a criterion on @xmath0 to generate confinement , namely theorem [ conf t1 ] . this criterion is stated in terms of an algebraic property of certain bounded operators in @xmath12 .
an application to electrostatic and lorentz scalar shell potentials is also shown .
but before , we need some auxiliary lemmata .
if @xmath567 is a function which is regular in @xmath568 and @xmath569 denote the boundary values of @xmath567 on @xmath15 when we approach from @xmath48 , then @xmath570 in the sense of distributions .
the proof of this formula , which follows essentially by stoke s theorem , is an easy exercise left for the reader . for any @xmath161 , since @xmath571 in the sense of distributions , by ( [ conf l1 eq1 ] ) we have @xmath572 which implies that @xmath573 .
recall that @xmath574 in @xmath575 , so the boundary values @xmath576 vanish on @xmath15 . combining this fact with lemma [ l jump]@xmath116
, we obtain @xmath577 given @xmath192 , let @xmath578 denote the boundary values of @xmath26 on @xmath15 . since @xmath579 in @xmath568 and @xmath287 , using ( [ conf l1 eq1 ] ) and lemma [ l jump]@xmath116
we obtain @xmath580 where @xmath581 .
this implies that @xmath582 .
hence , @xmath583 if and only if @xmath584 which , by ( [ conf l1 eq2 ] ) , the definition of @xmath585 and that @xmath287 , is equivalent to @xmath586 . therefore , @xmath583 for all @xmath212 if @xmath587 which is equivalent to @xmath588 ; the lemma is proved .
note that , since @xmath226 is a vector space , in the statement of the lemma one only needs to require that @xmath589 for all @xmath212 .
[ rem ult ] from the last part of the proof of lemma [ conf l1 ] , we actually see that @xmath583 for all @xmath212 if and only if @xmath588 holds on the set of functions @xmath165 such that there exists @xmath161 with @xmath590 .
let us prove @xmath597 . since @xmath598 are bounded operators , @xmath599 .
by @xmath116 , @xmath600 thus @xmath601 . if @xmath192 then @xmath602 , but from ( [ conf l1 eq4 ] ) we have seen that @xmath582 for some @xmath603 , so @xmath604 therefore , @xmath605 on @xmath606 , and @xmath117 is proved .
the implication @xmath607 is straightforward . in order to prove @xmath608 ,
recall the well known fact that , if @xmath196 is self - adjoint , then @xmath609 is invertible for all @xmath77 .
this assertion can be easily verified using the arguments in the proof of ( * ? ? ?
* theorem viii.3 ) , for example . by definition ,
@xmath117 is equivalent to @xmath610 which , by writting @xmath611 , is further equivalent to @xmath612 in conclusion , @xmath613 it is well know that , if @xmath614 is a solution of @xmath615 then one can write @xmath616 ( see ( * ? ? ? * theorem 7.9 ) for a similar result ) .
assume that @xmath617 , so @xmath618 . by ( [ conf l2 eq1 ] )
, we have @xmath619 therefore , we have proved that if @xmath117 holds then @xmath620 for all @xmath77 , which is a restatement of @xmath118 .
the implication @xmath621 is left for the reader .
that @xmath0 makes @xmath15 impenetrable means that the particles under consideration which are initially confined either in @xmath297 or @xmath67 at time @xmath623 , remain confined in @xmath297 or in @xmath67 for all @xmath77 under the evolution given by @xmath624 , i.e. , that lemma [ conf l2]@xmath118 holds .
thus , the theorem is a straightforward application of lemmata [ conf l1 ] and [ conf l2 ] .
[ conf appl ] assume that @xmath15 is @xmath100 .
given @xmath78 such that @xmath625 , let @xmath196 be the operator defined by @xmath626 and @xmath627 on @xmath226 , where @xmath628 and @xmath223 for @xmath192 .
if @xmath629 then @xmath196 is self - adjoint . in that case , @xmath196 makes @xmath15 impenetrable if and only if @xmath80 . for proving the self - adjointness of @xmath196 when @xmath629 , we follow the arguments of the proof of ( * ? ? ?
* theorem 3.8 ) . set @xmath632 so @xmath633 , and observe that @xmath634 are self - adjoint on @xmath12 because @xmath635 and @xmath183 also are . using that @xmath636 and lemma
[ l jump]@xmath117 we have @xmath637 where @xmath638 and @xmath639 . in ( * ? ? ?
* lemma 3.5 ) we proved that @xmath640 is a compact operator in @xmath12 , and since @xmath635 anticommutes with the @xmath641 s , we easily have @xmath642 thus @xmath643 is a compact operator by ( * ? ? ?
* proposition 3.11 ) and hence @xmath263 is also compact .
if @xmath629 then @xmath644 and fredholm s theorem applies to @xmath645 ( see , ( * ? ? ?
* theorem 0.38@xmath443 ) ) .
if for example @xmath646 , using ( [ conf t2 eq2 ] ) we can follow the proof of ( * ? ? ? * lemma 3.7 ) to show that @xmath34 has closed range .
moreover , as we did in the first part of the proof of ( * ? ? ?
* theorem 3.8 ) , fredholm s theorem also shows that @xmath647 is closed , we omit the details . in any case , that @xmath193 and @xmath195 are closed for all @xmath629 follows by ( [ conf t2 eq2 ] ) , fredholm s theorem and ( * ? ? ?
* theorem 1.46@xmath117 ) , so the restriction @xmath646 is not necessary . these properties of @xmath34 together with ( [ conf t2 eq1 ] ) allow us to apply theorem [ pre t1 ] , which proves that @xmath196 is self - adjoint for @xmath629 .
let us finally check the impenetrability condition relative to @xmath196 and @xmath15 . by theorem [ conf t1 ]
, @xmath196 makes @xmath15 impenetrable for the particles if @xmath34 satisfies @xmath588 . by a straightforward computation using lemma [ l jump]@xmath117 ,
that @xmath636 , and that @xmath648 and @xmath635 anticommute , we have @xmath649 therefore , @xmath34 satisfies @xmath588 if and only if @xmath80 , and in such case @xmath15 becomes impenetrable .
furthermore , since the right hand side of ( [ kpopkosa ] ) is a constant times @xmath91 , from remark [ rem ult ] and lemma [ conf l2 ] we actually deduce that @xmath196 makes @xmath15 impenetrable if and only if @xmath80 .
we have seen that @xmath650 makes @xmath15 impenetrable if and only if @xmath651 , which is precisely equation @xmath652 of ( * ? ? ? * section v ) .
hence , for the potentials @xmath81 , our results on confinement generalize the ones stated in @xcite to regular surfaces .
s. hofmann , e. marmolejo - olea , m. mitrea , s. prez - esteva and m. taylor , _ hardy spaces , singular integrals and the geometry of euclidean domains of locally finite perimeter , _ geometric and functional analysis ( forthcoming ) . | spectral properties and the confinement phenomenon for the coupling @xmath0 are studied , where @xmath1 is the free dirac operator in @xmath2 and @xmath3 is a measure - valued potential .
the potentials @xmath3 under consideration are given in terms of surface measures on the boundary of bounded regular domains in @xmath2 .
a criterion for the existence of point spectrum is given , with applications to electrostatic shell potentials .
in the case of the sphere , an uncertainty principle is developed and its relation with some eigenvectors of the coupling is shown . furthermore , a criterion for generating confinement is given .
as an application , some known results about confinement on the sphere for electrostatic plus lorentz scalar shell potentials are generalized to regular surfaces . |
the solar wind is a rarefied , highly variable flow of charged particles originating from the sun .
this medium is composed mainly by electrons and protons , with a small fraction of alpha particles ( helium nuclei ) and a lower percentage of heavier ions . due to its low particle density ,
collisions are essentially negligible , and the solar wind plasma is in a state of non - thermodynamic equilibrium @xcite .
one of the most puzzling aspects of the solar wind dynamics consists in the empirical evidence @xcite that it is hotter than expected from adiabatic expansion @xcite .
understanding the mechanisms of energy dissipation and particle heating in such a collisionless system represents a real challenge for space plasma physics . ` in situ ' spacecraft measurements reveal that the solar wind is in a state of fully - developed turbulence @xcite .
the energy injected by large scale solar dynamical features into the heliosphere , in the form of long - wavelength fluctuations , cascades towards small scales via nonlinear interactions until it can be transferred to the plasma as heat . in the inertial range
, the power spectrum of the solar wind fluctuations manifests a behavior reminiscent of the kolmogorov phenomenology for fluid turbulence @xcite .
the turbulent cascade extends to smaller spatial scales , down to a range of wavelengths where kinetic effects start playing a non - negligible role . at the typical ion characteristic scales ( larmor radius and/or inertial length ) ,
different physical processes come into play , leading to changes in the spectral shape @xcite . here , the dynamics of the solar wind ions gets extremely complicated , being dominated by complex kinetic processes such as resonant wave - particle interactions , particle heating , generation of temperature anisotropy , production of beams of accelerated particles , etc . in this range of scales
the particle velocity distribution is generally observed to deviate from the typical maxwellian shape of thermodynamic equilibrium .
several models have been developed to understand the kinetic scale phenomenology , focusing specifically on heating and acceleration processes . to explain the above phenomena , several physical mechanisms have been proposed . among them
, the ion - cyclotron resonance can produce heating @xcite , where ions interact resonantly with ion - cyclotron waves generated along the cascade .
an alternative model is the non - resonant stochastic heating @xcite , where the fluctuations at scales comparable with the ion gyro - scales produce significant distortions of the ion orbits that become stochastic in the plane perpendicular to the magnetic field , violating the magnetic moment conservation .
finally , it has been clearly observed that the interaction of particles with coherent structures can locally produce heating and non - maxwellian features @xcite .
one of the most relevant aspects of the ion heating process in the solar wind is that the heavy ions are preferentially heated and accelerated with respect to protons .
measurements of solar wind hydrogen and helium temperatures @xcite show compelling evidence of alfvn - cyclotron dissipation mechanisms .
a more recent observational study @xcite shows that for the heavy ion component ( @xmath0 amu ) the temperature displays a clear dependence on mass , probably reflecting the physical conditions in the solar corona .
again , the interpretation of these observations is mostly based on the physical processes of wave - particle interactions @xcite , stochastic ion heating @xcite , and interaction between particles and coherent structures @xcite . despite a significant theoretical effort , there is still no definitive solution to the problem of heavy ions differential heating .
it is important to note that the physical processes of the differential heating of heavy ions are also important for laboratory plasmas research , aiming at heating a confined plasma to initiate fusion reactions . in this paper
, we study the kinetic dynamics of protons and alpha particles in typical conditions of the interplanetary medium , by employing multi - component eulerian hybrid vlasov - maxwell ( hvm ) simulations @xcite .
the multicomponent version of the hvm code integrates the vlasov equation numerically for the distribution function of both protons and alpha particles , while treating electrons as a massless , isothermal fluid .
we present the numerical results of hvm simulations of decaying turbulence with guide field , in a 2d-3v phase space domain ( two dimensions in physical space and three dimensions in velocity space ) . during the turbulent cascade ,
the ion species depart locally from the condition of thermodynamic equilibrium @xcite , exhibiting temperature anisotropy , differential kinetic behavior as well as preferential heavy ion heating .
interestingly , the preferential heating process occurs in thin filaments of the typical size of few inertial lengths , preferentially located in the proximity of vorticity structures .
in the same regions , significant deviations from maxwellian distributions are observed for ion species . moreover , as we will discuss in detail below , we have analyzed the distribution of the numerical data in the fire - hose and mirror stability plane , as dependent on the multi - ion composition of the interplanetary medium , showing a good agreement between the numerical results and the observational evidences discussed in a recent paper by chen et al .
@xcite .
finally , through a top - hat simulator @xcite using the output of the hvm simulations , we show that the cold solar wind ( csw ) instrument @xcite on board the turbulence heating observer - thor spacecraft @xcite , a candidate for the next m4 space mission of the european space agency , is able to provide adequate high resolution measurements of the ion velocity distributions for the study of turbulent ion heating . in this regard , the thor mission will allow to measure the velocity distributions of the solar wind protons and alpha particles , with unprecedented phase space resolution .
to perform a numerical study of the microphysics of a multi - ion turbulent plasma , we employ an eulerian hybrid vlasov - maxwell ( hvm ) numerical code @xcite which solves the vlasov equations for the distribution functions of protons and alpha particles . in this
model electrons are treated as an isothermal massless fluid .
vlasov equations for ions are coupled to a generalized ohm s law , in which both the hall term and the electron pressure contribution are retained , and to the ampre and faraday equations .
the displacement current is neglected and quasi - neutrality is assumed .
the dimensionless hvm equations can be written as : [ eq : vlas ] + * v * + _ s ( * * e**+ * v * * b * ) = 0 , [ eq.1 ] + [ eq : ohm ] * e * = - ( * u*_e * b * ) - p_e + * j * , + [ eq : induction ] = - * e*. in the above equations , masses and charges have been scaled by the proton mass @xmath1 and charge @xmath2 respectively , time by the inverse proton - cyclotron frequency ( @xmath3 , @xmath4 being the ambient magnetic field and @xmath5 the speed of light ) , velocity by the alfvn speed ( @xmath6 , @xmath7 being the equilibrium proton density ) , in which we only considered the contribution of the dominant proton species , and length by the proton skin depth ( @xmath8 ) . in eq .
( [ eq : vlas ] ) , @xmath9 is the ion distribution function ( the subscript @xmath10 refers to protons and alpha particles , respectively ) , @xmath11 and @xmath12 are the electric and magnetic fields , and @xmath13 is a constant that depends on the charge to mass ratio of each ion species ( @xmath14 and @xmath15 ) . in eq .
( [ eq : ohm ] ) , the electron bulk velocity @xmath16 is defined as @xmath17 , where @xmath18 is the ion charge number ( @xmath19 and @xmath20 ) , the density @xmath21 and the ion bulk velocity @xmath22 are zero - th and first order velocity moment of the ion distribution function , respectively .
the electron density is derived from the quasi - neutrality equation @xmath23 .
furthermore , @xmath24 is the total current density and @xmath25 is the electron pressure ( @xmath26 is assumed to be constant in time and space ) .
a small resistive term has been added as a standard numerical laplacian dissipation ( @xmath27 ) , in order to remove any spurious numerical effects due to the generation of strong magnetic field gradients during the development of turbulence .
we solve the multi - ion hybrid vlasov - maxwell equations in a 2d-3v phase space domain .
the system size in the spatial domain is @xmath28 in both @xmath29 and @xmath30 directions , where periodic boundary conditions have been implemented , while the limits of the velocity domain are fixed at @xmath31 in each velocity direction , @xmath32 being the ion thermal speed at equilibrium and @xmath33 the equilibrium temperature . in each direction in the velocity domain we set @xmath34 .
the 5d numerical box has been discretized by @xmath35 grid - points in the 2d spatial domain and @xmath36 grid - points in both proton and alpha - particle 3d velocity domains .
a higher velocity resolution has been adopted here with respect to previous works @xcite , in order to ensure a significantly improved conservation of the vlasov invariants and to provide a better description of the details of the ion distribution function in velocity space .
the time step , @xmath37 , has been chosen in such a way that the courant - friedrichs - lewy condition for the numerical stability of the vlasov algorithm is satisfied @xcite . in order to control numerical accuracy , a set of conservation laws
is monitored during the simulations : typical relative variations of the total energy is @xmath38 , of the entropy is @xmath39 and of the mass is @xmath40 for protons and @xmath41 for alpha particles . at @xmath42 ,
both ion species have maxwellian velocity distributions and homogeneous constant densities . to simulate physical conditions close to those of the pristine solar wind , at equilibrium we set the alpha particle to proton density ratio @xmath43 and temperature of each species such that @xmath44 .
the plasma is embedded in a uniform background out of plane magnetic field , @xmath45 ( @xmath46 ) .
we perturb the initial equilibrium with a 2d spectrum of fourier modes , for both proton bulk velocity and magnetic field .
the energy is injected with random phases and wave numbers in the range @xmath47 , where @xmath48 , with @xmath49 .
neither density disturbances nor parallel variances are imposed on the initial equilibrium , namely @xmath50 .
we performed three different simulations whose relevant parameters are summarized in table [ tab ] .
below , we will discuss the results of run 1 .
run 2 and run 3 are qualitatively similar to run 1 and have been considered to have significant statistics to perform a direct comparison with solar wind data .
@xmath51 & @xmath52 + 1 & 0.5 & 1/3 & 49 + 2 & 2.0 & 1/3 & 51 + 3 & 1.0 & 2/3 & 25 +
we numerically investigate the kinetic evolution of protons and alpha particles in a situation of decaying turbulence .
the large scale fluctuations imposed on the initial equilibrium produce a turbulent cascade down to kinetic scales . in analogy with fluid models , by looking at the time evolution of the spatially
averaged mean squared out of plane current density @xmath53 , it is possible to identify an instant of time @xmath54 at which the turbulent activity reaches its maximum level @xcite . the value of @xmath54 in units of @xmath55 is reported in the right column of table [ tab ] , for each simulation . at this time
@xmath54 we perform our analysis @xcite . in panel ( a ) of figure [ fig : spec_prot ] , we show the omni - directional magnetic @xmath56 ( blue solid - square line ) and electric @xmath57 ( red dashed - triangle line ) spectra at @xmath58 for run 1 .
the kolmogorov expectation , @xmath59 ( black dashed line ) , is reported as a reference , while the vertical black dashed line indicates the proton skin depth characteristic wavenumber . in the range of small wavenumbers
the magnetic activity is dominant , while the situation clearly changes for @xmath60 , where the electric energy becomes larger than the magnetic one . to highlight this point , in panel ( b ) of figure
[ fig : spec_prot ] we report the electric to magnetic energy ratio as a function of the wavenumber .
the break point between the two regimes is located around the proton skin depth , where dispersive hall effects come into play @xcite .
these results present several similarities to spacecraft measurements in the solar wind @xcite . as already noticed in previous works @xcite , ion temperature anisotropy
is generated along the development of the turbulent cascade .
we focus , at first , on the temperature anisotropy with respect to the direction of the local magnetic field @xmath61 for each ion species .
the temperature of each ion species has been computed as the second - order velocity moment of the ion distribution function : t_s = ( * v * - * u*_s)^2 f_s d^3 v s = p , . in figure
[ fig1 ] , we show the 2d contours of the temperature anisotropy of protons ( left ) and alpha particles ( right ) at @xmath58 , for run 1 ( the same behavior is observed for run 2 and run 3 ) ; as can be noticed from the comparison of both panels of this figure , the global bi - dimensional patterns of the temperature anisotropy display similar features , with larger values of anisotropy being concentrated in thin filaments ( of the typical size of few ion skin depths ) for both species .
however , a significant differential behavior is observed , as the alpha particles appear consistently more anisotropic than the protons . with the aim of showing that the numerical results of the ensemble of simulations in table [ tab ] reproduce
closely the typical behavior observed in the solar wind multi - ion plasma , we consider here a recent paper by chen et al .
@xcite , in which solar wind ion temperature anisotropy from the _ wind _ spacecraft at 1 au has been analyzed .
the authors proposed a detailed study of the plasma stability , as dependent on the simultaneous presence of protons , electrons and alpha particles , as main constituents of the solar wind plasma .
specifically , they made a comprehensive analysis based on three years of data and gave clear evidence that the fire - hose and mirror instability thresholds , calculated in refs .
@xcite for a multi - species plasma system , well constrain the whole data distribution to the stable side and that the contours of the distribution follow the shape of the thresholds . for the ensemble of simulations in table [ tab ] .
solid ( black ) line represent the instability threshold .
panels ( b ) and ( c ) : @xmath62 and @xmath63 , respectively , as a function of @xmath64.,title="fig:",width=283 ] for the ensemble of simulations in table [ tab ] .
solid ( black ) line represent the instability threshold .
panels ( b ) and ( c ) : @xmath62 and @xmath63 , respectively , as a function of @xmath64.,title="fig:",width=283 ] + for the ensemble of simulations in table [ tab ] .
solid ( black ) line represent the instability threshold . panels ( b ) and
( c ) : @xmath62 and @xmath63 , respectively , as a function of @xmath64.,title="fig:",width=283 ] in order to compare with these observations , we have considered the conditions for the long - wavelength fire - hose and mirror instabilities in a multi - species plasma @xcite : @xmath65 where @xmath66 stands for protons , alphas and electrons respectively ( @xmath67 ) , @xmath68 , @xmath69 is the mass density of the species @xmath66 , @xmath70 is the total mass density , @xmath71 is the local alfvn speed , @xmath72 is the difference between the bulk velocity of the species @xmath66 and the center of mass velocity . in figure [ fig2 ] [ panel ( a ) ]
we report a scatter plot of the first and second term of @xmath73 ( logarithmic axes ) , for all the cases in table [ tab ] .
data are clearly constrained in the stable region ( delimited by the solid line ) .
these numerical results are in agreement with the solar wind analysis , as can be seen comparing our study with figure 2 of ref .
this further confirms that the hvm model can describe the basic dynamics of the solar wind ion species .
however , in this direct comparison one has to take into account the limitations of the approximation of isothermal ( massless fluid ) electrons , and limitations in the statistical convergence ( data produced by three numerical runs can not fully recover the statistics of three years of spacecraft measurements ) . in panels ( b ) and
( c ) of figure [ fig2 ] , we show @xmath74 and @xmath63 respectively , as a function of @xmath64 , @xmath75 being the total plasma beta . here , one can easily see that the instability thresholds ( black - dashed lines ) are reached for @xmath76 , again in good agreement with figure 5 of ref .
during the development of the turbulence cascade , temperature increase is observed for both ion species , alpha particles being preferentially heated with respect to protons .
moreover , the generation of ion temperature anisotropy discussed in the previous section is associated to the appearance of coherent structures , such as current sheets , filaments , strongly sheared flows , etc . ,
observed in the 2d patterns in physical space . in particular , panel ( a ) of figure [ fig : contour ] displays the shaded contours of the out of plane current density @xmath77 at @xmath58 for run 1 , together with the contour lines of the magnetic potential @xmath78 of the in - plane magnetic field @xmath79 .
magnetic flux tubes with different polarizations ( or magnetic islands in 2d ) are identified by closed contours of @xmath78 , as it can be seen from panel ( a ) of this figure .
as it is clear from this contour plot , current density becomes very intense in between adjacent magnetic islands .
we point out that the mirror and fire hose unstable points in figs .
[ fig2 ] ( b)-(c ) ( obtained in run 3 , with increased initial magnetic perturbations ) are located in spatial positions very close to strong current sheets . in panel ( b ) of figure [ fig : contour ] we show the contour map of the proton to alpha particle temperature ratio . even though at @xmath42 ions have been initialized with equal temperature ( @xmath80 ) , during the evolution of turbulence , the alpha particles are heated preferentially with respect to protons .
moreover , the 2d map shows that the differential heating is not uniform , but is strongly inhomogeneous , displaying a pattern similar to the 2d map of @xmath77 and being significantly influenced by the topology of the magnetic field @xcite .
moreover , the physical mechanism responsible for the ion heating process is evidently more efficient for the alpha particles than for the protons . as the hvm code allows for a clean and almost noise - free description of the ion distribution function in phase space
, we can analyze the kinetic dynamics of the ion species associated to the development of the turbulent cascade and to the differential ion heating .
several ` in situ ' observations ( see , for instance , refs .
@xcite ) and previous numerical simulations @xcite have shown that the ion distribution functions in the turbulent solar wind ( especially in the less collisional fast wind ) are typically far from thermodynamical equilibrium and that kinetic effects manifest through the appearance of field - aligned beams of accelerated particles , generation of ring - like modulations in the particle velocity distributions and , in general , through complicated non - maxwellian deformations . in order to quantify the deviation of the ion velocity distributions from the maxwellian shape in our simulations , we introduce the following non - maxwellian measure @xcite , for each ion species : [ eq : eps ] _
s = , where @xmath81 is the associated equivalent maxwellian distribution computed from the parameters of @xmath82 @xcite .
we point out that at @xmath42 , @xmath83 is null , since both ion species have been initialized with a maxwellian velocity distribution .
we observe that , simultaneously with the evolution of turbulence , the ion distributions dramatically deviate from the maxwellian shape .
panels ( c ) and ( d ) of figure [ fig : contour ] display the contour maps of @xmath83 for protons and alpha particles , respectively , at @xmath58 for run1 . as it is clear from this figure ,
both ion species significantly deviate from thermodynamic equilibrium and the spatial distribution of @xmath83 is not uniform , displaying features quite similar to the 2d contours of the out of plane current density [ panel ( a ) ] and of the ion temperature ratio [ panel ( b ) ] .
in particular , kinetic effects seem to be localized near the peaks of the current density in the shape of thin filaments .
it is worth noting that deviations from maxwellian are significantly stronger for the alpha particles than for the protons ; that is , larger values of @xmath84 are recovered with respect to @xmath85 and are achieved in spatial positions close to peaks of the current density , where differential heating is also observed .
ion differential heating usually occurs close to thin current sheets .
panels ( a ) and ( b ) of figure [ fig : taglio ] report the @xmath30 profile of @xmath86 ( red - dashed curves ) , in two different ranges @xmath87d_p$ ] and @xmath88d_p$ ] , computed at @xmath89 and @xmath90 , respectively .
two current sheets of typical width of a few proton skin depths are visible here .
the black - solid curves in the same panels represent the corresponding @xmath30 profile of @xmath91 .
as can be seen in the two panels , differential heating is observed close to each strong current sheet .
however , the behavior of the alpha to proton temperature ratio looks different in the two cases : in panel ( a ) , alpha particles are preferentially heated in the center of the current sheet , while in panel ( b ) the differential heating occurs on both edges of the current sheet .
this is a quite general behavior , in the sense that , close to each coherent current structure identified in the spatial domain , the alpha to proton temperature ratio displays a peak in the middle of the current sheet ( as in panel ( a ) ) or two peaks at its boundaries ( as in panel ( b ) ) .
panels ( c ) and ( d ) of the same figure show similar plots for the squared vorticity component @xmath92 associated to the proton velocity , @xmath93 , and to the alpha particle velocity , @xmath94 where @xmath95 , at the same two positions as for the current @xmath86 .
the correlation observed for the vertical current is also clearly present for the vorticity .
the robustness of the correlation observed between the quantities shown in figure [ fig : taglio ] can be visualized by means of the joint probability distributions @xmath96 and similarly for the vorticity @xmath97 , as shown in figure [ fig : correlation ] . in the left panel , it is evident that there is actually no statistical correlation between the current and the differential heating , because of the spatial shift discussed above . on the contrary , an higher correlation is visible between the proton vorticity and the differential heating ( central panel of figure [ fig : correlation ] ) .
even a stronger correlation is observed with the alpha - particle vorticity ( right panel ) .
a more quantitative assessment can be obtained by estimating the spearman correlation coefficients between the alpha - proton temperature ratio and the current , @xmath98 , and similarly for the two vorticity fields , as reported in each panel of figure [ fig : correlation ] .
this observation suggests that the vortical motion of the alpha particles may play a significant role in , or may be significantly affected by , the differential heating process @xcite .
this correlation can be understood also in terms of magnetic reconnection .
topologically , magnetic reconnection might form quadrupolar vortical structures located near the current peak , as clearly shown in @xcite . in these regions , both magnetic and velocity
shears can trigger non - maxwellian and thermal processes .
a deeper analysis of the relationship of current and vortex structures with collisionless plasma heating can be found in ref .
@xcite . at this point , it is important to better quantify the connection between the differential particle heating and the level of kinetic activity . in figure
[ fig : relation ] we show the joint distributions of @xmath84 and @xmath85 ( left panel ) and of @xmath84 and @xmath91 ( right panel ) for run 1 at @xmath58 .
both panels show that ( i ) alpha and proton velocity distributions tend to deviate from maxwellian at similar spatial locations , and ( ii ) such a deviation is correlated with the differential heating . in figure
[ fig : conditioned ] , the joint distribution of @xmath85 and @xmath84 has been conditioned to specific ranges of values of the temperature ratio @xmath91 , divided into five bins as indicated in each panel .
each distribution is normalized to its peak value . as the temperature ratio increases ( from top to bottom in the figure ) , an evident shift of the distribution towards larger values of both @xmath85 and @xmath84 is observed ; this shift can be clearly appreciated by looking at the location of the average value of @xmath91 , indicated by a red dot in each panel of this figure .
these results indicate that a larger deviation from a maxwellian distribution is observed for larger temperature ratios @xmath91 . on the other hand ,
the conditioned joint distributions are relatively symmetric , so that there is no evidence of different increase in @xmath85 or @xmath84 when the alpha particles are preferentially heated .
temperature variations of the ions seem to be related to their kinetic dynamics and , therefore , to the complex deformations of the particle velocity distributions .
future space missions , such as thor @xcite , designed to provide measurements of the three - dimensional ion distribution functions with very high phase space resolution , will be able to capture the non - maxwellian deviations of the velocity distributions , so as to provide relevant clues on the long - standing problem of particle heating . as discussed previously , close to the sites of enhanced magnetic activity ,
the non - maxwellian measure @xmath83 achieves large values for both ion species . in order to show what the three - dimensional ion velocity distributions look like in turbulence , in figure [ fig : vdf ] we report the iso - surfaces of the proton [ panel ( a ) ] and alpha particle [ panel ( b ) ] velocity distributions , computed in the spatial point where @xmath83 is maximum .
the red tube in both panels indicates the direction of the local magnetic field .
the proton velocity distribution displays a field - aligned beam and the generation of ring - like modulations in planes perpendicular to the direction of the local field , presumably signatures of cyclotron resonance . on the other hand , for the alpha particles we recovered more evident deformations , with the appearance of multiple blobs and elongations ; the alpha particle velocity distribution seems to lose any property of symmetry with respect to the direction of the local magnetic field .
it is clear from these plots that kinetic effects , working along the turbulent cascade at typical ion scales , can efficiently drive the plasma away from thermodynamic equilibrium .
moreover , when the particle velocity distributions are not constrained by limiting assumptions on their shape in velocity space , they can be distorted in a very complicated way and lose their symmetry .
an alternative way to describe the deviations from thermodynamic equilibrium consists in defining the so - called minimum variance frame of the three - dimensional velocity distribution . we compute the preferred directions of the velocity distribution of each ion species in velocity space @xcite , in each spatial position , from the stress tensor : @xmath99 this tensor can be diagonalized by computing its eigenvalues @xmath100 ( ordered in such way that @xmath101 ) and the corresponding normalized eigenvectors @xmath102
. we point out that @xmath103 are proportional to temperatures and @xmath104 are the anisotropy directions of the velocity distribution . the quantity @xmath83 gives general information on the departure of the ion species from thermodynamic equilibrium , while the eigenvalues @xmath103 provide insights into the properties of symmetry of the ion velocity distributions .
therefore , for run 1 at @xmath58 , we computed the pdf of the ratios @xmath105 ( @xmath106 and @xmath107 ) , conditioned to the values of @xmath108 .
note that each of these ratios is equal to unity for a maxwellian velocity distribution . in figure
[ fig : lambda ] , we show the pdf of @xmath109 ( left panels ) , @xmath110 ( middle panels ) and @xmath111 ( right panels ) for protons ( top row ) and alpha particles ( bottom row ) ; these pdfs have been computed for three different ranges of values of @xmath83 , @xmath112 ( black curve ) , @xmath113 ( red curve ) and @xmath114 ( blue curve ) , where @xmath115 represents the maximum value of @xmath83 over the two - dimensional spatial domain at @xmath58 ( @xmath116 , @xmath117 ) and @xmath118 , @xmath119 , @xmath120 , @xmath121 . from this latter figure
it can be noticed that , in the range of small @xmath83 , the pdfs ( for both protons and alphas ) have a peak close to unity ( they are not exactly centered around @xmath122 , since the minimum value of @xmath83 is not zero at @xmath58 ) . as @xmath83 increases , high tails appear in the pdfs suggesting that , in the case of significant deviations from maxwellian , the ion velocity distribution loses its properties of isotropy and gyrotropy .
this observation suggests that the use of reduced models , based on restrictive approximations on the shape of the ion velocity distribution , might not be appropriate , and more complete models , describing the evolution of the velocity distributions in a full 3d velocity space , should be employed . with the aim of characterizing the nature of the deformation of the particle velocity distributions and to identify the spatial regions which are the sites of kinetic activity , we computed for each ion species two indices of departure from maxwellian , i. e. the temperature anisotropy index and the gyrotropy index , in two different reference frames , namely the mvf and the local magnetic field frame ( lmf ) .
we define the anisotropy measures @xmath123 ( mvf ) and @xmath124 ( lmf ) , where @xmath125 and @xmath126 are the temperatures with respect to the local magnetic field , and the gyrotropy indicator in the mvf @xmath127 .
the gyrotropy index in the lmf @xmath128 can be computed by using the normalized frobenius norm of the nongyrotropic part @xmath129 of the full pressure tensor @xmath130 @xcite : @xmath131 where @xmath132 are the components of the tensor @xmath129 , and @xmath133 .
it is worth to point out that all indices defined above are identically zero if the particle velocity distribution is maxwellian . at the maximum of the turbulent activity
, these indices significantly differ from zero for both ion species , meaning that the ion velocity distributions are anisotropic and non - gyrotropic in both mvf and lmf . in figure
[ fig : index_taglio ] , we show the spatial profiles of all these indices inside the region where the current sheet shown in figures [ fig : taglio ] ( b ) is located ( indicated by vertical black - dashed lines ) .
the indices for protons are reported in the left panels , while for alpha particles in the right panels . as it is evident from this figure ,
the non - maxwellian measures of both protons and alpha particles are not null in the regions close to the peak of the current density , in both reference frames ; specifically , the ion velocity distributions become anisotropic and non - gyrotropic , as can be seen from both the mvf and lmf indices .
it is worth noting that the indices for the alpha particles are significantly larger than for the protons , additional evidence that alpha particles are more efficiently driven out of equilibrium with respect to protons .
in the previous sections we have shown that kinetic effects produce highly non maxwellian distributions , particle beams and ring - like modulations . an instrument designed to study these effects should have proper phase space ( and temporal ) resolution , since a poor resolution could impede the observations of such fine features . in near - earth space
it is possible to perform high resolution in situ plasma observations .
measurements of particle velocity distribution functions are usually performed by top - hat electrostatic analyzers @xcite .
such instruments are composed of two concentric hemispheres set at different voltage , with an aperture on the outer hemisphere .
the electric field between the plates allows for particles within a specific energy - per - charge ( e / q ) ratio to pass through the gap and reach the detector . varying the voltage permits to sort particles according to their e / q .
parallel incident particles will be focused on a specific sector of the detector at the analyzer exit , each sector identifying the particle velocity azimuthal direction .
analyzers of this type measure incoming particles with a @xmath134 disk shaped field of view . in order to sample the entire @xmath135 solid angle , either the spacecraft rotation is used or electrostatic polar deflectors are employed .
usually analyzers for solar wind measurements are characterized by a restricted field of view , since the solar wind at 1 au is a cold beam in velocity space .
a schematic illustration of such a top - hat analyzer is given in figure [ fig : tophat ] .
as already said , the energy - angular resolution of the top - hat analyzer is extremely important when aiming to study the signatures of kinetic - scale processes , since low phase space resolution would smooth out the fine structures of the particles distribution functions .
the cold solar wind ( csw ) @xcite on board the thor mission is a top - hat analyzer conceived to measure solar wind protons and alpha particles distribution functions with unprecedented temporal and phase space resolutions @xcite . in particular , it has an angular and energy resolution that will permit to observe the fine structure characterizing the particle distribution functions related to the effects of turbulence . in this respect ,
we use a _ top - hat simulator _ to find out how the proton and alpha velocity distributions characterized by large @xmath83 , and shown in figure [ fig : vdf ] , would be detected by csw .
such distribution functions are meaningful examples since they are characterized by non maxwellian shapes and a high degree of complexity as has been shown in the previous sections . moreover , we compare the moments and the @xmath83 of the hvm and csw distributions in proximity of large @xmath83 sites , showing that csw measurements capture the complexity present in the hvm distributions .
csw will detect particles in 96 energy - per - charge intervals with @xmath136 .
it will have a restricted field of view of @xmath137 in both azimuth and elevation , focused on the cold solar wind population , and will have an unprecedentedly high angular resolutions of @xmath138 , in both azimuth and elevation .
the csw response is further determined by its geometric factor @xmath139 and the time needed for the accumulation of particle counts in one energy channel and elevation angle @xmath140 . as a first step ,
the hvm distributions are expressed in physical units by rescaling the normalized variables of the simulation to physical quantities . in order to adapt to a real solar wind , a proton density of 8.2 @xmath141 , a proton thermal velocity of @xmath142 km / s , which corresponds to a thermal energy of 12.8 ev , and a @xmath143 have been used . moreover ,
a bulk velocity of @xmath144 km / s has been taken into account .
then the distributions functions are reassigned to the field of view of csw by means of an averaging process . at this point ,
the counts that the sensors would detect in each energy - angular interval for each species are computed @xcite : @xmath145 where @xmath146 refer to the energy , elevation and azimuth intervals , @xmath147 is the velocity in the interval , @xmath148 the distribution function , @xmath149 and @xmath150 the accumulation time and the geometric factor , respectively .
@xmath148 is normalized in such a way that its integral in velocity space gives the particle number density .
afterwards , we determine the corresponding moments . in figure [
fig : tophatsim ] proton ( top panels ) and alpha ( bottom panels ) velocity distributions are shown at the spatial point where @xmath151 is maximum , as recovered from the hvm simulations ( left panels ) and as sampled by csw ( right panels ) .
in particular , slices of the distribution functions in the @xmath152 plane are presented .
looking at the protons , it can be seen that all the features that characterize the hvm distribution are revealed by csw .
in fact , the ring - like modulations , that can be identified in the hvm distribution along the @xmath153 direction , are present also in the csw distribution , although they are less readily identifiable . moreover ,
this distribution function spans only over one order of magnitude and will be easily observed by csw for regular or dense solar wind .
looking at the alpha particles , it has to be noted that the alphas population appears in the csw distribution at @xmath154 times the bulk velocity recovered from the simulation , since electrostatic analyzers like csw detect particles with respect to their energy - per - charge . regarding the features characterizing the hvm and csw alpha distributions ,
it can be seen that the shape of the main alpha population , at about @xmath155 , is correctly reproduced .
in fact , the elongation visible in the hvm is clearly unveiled also in the csw distribution , and also discernible is the faint widening on the left part of the core population . moreover , the faint blob at negative @xmath156 in the hvm distribution , is recognizable in the csw distribution as well , and its shape is well captured . in figure
[ fig : moments ] we show the hvm ( red lines ) and csw ( black lines ) density , velocity and temperature , for protons ( left panel ) and alpha particles ( right panel ) , computed for distribution functions along the @xmath30 and @xmath29 direction at @xmath157 and @xmath158 , respectively . along these directions ,
the maximum @xmath85 and @xmath159 are encountered .
results show that the velocity , and to a lesser extent the density , for hvm and csw are extremely well in agreement .
small differences can be noted between hvm and csw temperature .
the different behaviors of the temperature and lower order moments was already observed in @xcite . in general , larger errors are found between the hvm and csw moments for alpha particles than for protons .
this can be explained in terms of the alpha particles lower counts statistics with respect to the protons .
finally , the non maxwellian measure @xmath83 , computed for the hvm ( red line line ) and csw ( black line ) distributions , is presented in figure [ fig : cswepsilon ] .
the @xmath83 evolution is the same for the hvm and csw distributions .
in particular , the two hvm @xmath83 peaks , of the protons and alphas cases , are well pronounced in the csw @xmath83 evolution as well .
the small difference between the peaks of the hvm and csw @xmath84 can probably be explained in terms of the alphas lower counts statistics , like in the case of moments .
it must be noted here that , starting from the hvm simulation , the moments for the alpha particles can be readily computed . in the real case , although the alpha population can be easily identified in the csw distribution functions , part of the protons and alpha particles populations can overlap , especially when the plasma temperature is high , and the moments computation can be hard ( we recall that csw selects particles according to their energy - per - charge ) . in such cases it is extremely useful to have a particle instrument that is able to identify the different ion species .
such an instrument will be on board thor : it is the ion mass spectrometer ( ims ) , a top - hat analyzer complemented with a time of flight section .
we have employed a multi - component eulerian hybrid vlasov - maxwell code to analyze the differential kinetic dynamics of the dominant ions in the turbulent solar wind .
we have performed simulations of decaying turbulence in a five - dimensional phase space , varying relevant parameters , such as the level of turbulence and the plasma beta at equilibrium , in order to explore a large portion of the parameter space and to compare our numerical results with direct solar wind measurements . as the numerical results do not change qualitatively for the runs summarized in table [ tab ] , after showing that our simulations reproduce with a good degree of realism
the phenomenology recovered in solar wind observations , we restricted our analysis to a single simulation .
we focused in particular on the departure of the plasma from thermodynamic equilibrium along the cascade and on the role of kinetic effects coming into play when the cascading energy reaches the typical ion scales .
as turbulence develops , we observed the generation of ion temperature anisotropy with respect to the direction of the local magnetic field as well as a significant heating for both ion species .
also , the non - maxwellian index @xmath83 , which is a measure of the deviations from maxwellian of each ion species , gets different from zero ; at the same time , the generation of coherent structures ( current sheets , vortices , etc . ) of the typical width of few ion skin depths is observed in the contour map of the out of plane current density and of the ion vorticity .
generation of temperature anisotropy , particle heating and deformation of the ion velocity distributions clearly occur in the form of thin filaments at scales of the order of the ion kinetic scales .
a detailed statistical analysis of the simulation results provides evidence that the distribution of the numerical data is constrained to the stable region with respect to the fire - hose and mirror stability thresholds , evaluated by taking into account the multi - ion composition of the solar wind plasma .
these results appear in good agreement with the observational evidences , recently discussed in ref .
@xcite .
the generation of temperature anisotropy and the process of particle heating are significantly more efficient for the alpha particles than for the protons and also the non - maxwellian index gets larger values for the heavy ions .
this means that not only a differential heating occurs , but also a differential kinetic behavior is observed for the dominant ion species in the solar wind : in fact , as shown in figure [ fig : vdf ] , the velocity distribution of the alpha particles appears much more distorted than that of the protons .
interestingly , the process of differential ion heating takes place close to the peaks of the ion vorticity and in spatial regions where the ion velocity distributions appear significantly distorted and far from thermodynamic equilibrium . in order to characterize the deviations from maxwellian
, we analyzed the ion velocity distribution both in its minimum variance frame and in the local magnetic field frame , by defining the anisotropy and the gyrotropy indicators in both frames .
the study of these indicators reveals that , when kinetic effects are at work and @xmath83 gets large , the shape of the velocity distributions of each ion species loses its symmetry , developing anisotropic and non - gyrotropic features .
moreover , close to the peaks of @xmath83 , sharp velocity gradients and complicated distortions are recovered .
if , as suggested by our numerical results , temperature variations of the ion species are related to the kinetic dynamics at short spatial scales , then the details of the velocity distribution , which keep the memory of the particle kinetic dynamics , are clearly crucial ingredients for the understanding of the complex process of particle heating .
in fact , as we discussed above , different ion species , even though experiencing the same turbulent fields , exhibit a deeply different response , being heated and shaped differentially . by feeding a virtual top - hat analyzer with the ion velocity distributions obtained from the kinetic simulations discussed here , we have demonstrated that the energy and angular resolutions of the cold solar wind instrument on board the thor mission should allow to obtain unprecedentedly high resolution measurements of both proton and alpha velocity distributions and will therefore provide important information for unveiling the puzzling aspects of solar wind heating . the results discussed in this paper support the idea that future space missions , designed to provide measurements of particle velocity distributions and fields with very high resolution , will allow to gain relevant insights into the dynamics of the solar wind plasma at kinetic scales .
this work has been supported by the agenzia spaziale italiana under the contract no .
asi - inaf 2015 - 039-r.o `` missione m4 di esa : partecipazione italiana alla fase di assessment della missione thor '' .
the numerical simulations discussed in the present paper have been performed on the newton parallel machine at the department of physics at university of calabria ( italy ) .
work at irap was supported by cnrs and cnes .
ds and ss aknowledge support from the faculty of the european space astronomy centre ( esac ) . | the solar wind plasma is a fully ionized and turbulent gas ejected by the outer layers of the solar corona at very high speed , mainly composed by protons and electrons , with a small percentage of helium nuclei and a significantly lower abundance of heavier ions .
since particle collisions are practically negligible , the solar wind is typically not in a state of thermodynamic equilibrium .
such a complex system must be described through self - consistent and fully nonlinear models , taking into account its multi - species composition and turbulence .
we use a kinetic hybrid vlasov - maxwell numerical code to reproduce the turbulent energy cascade down to ion kinetic scales , in typical conditions of the uncontaminated solar wind plasma , with the aim of exploring the differential kinetic dynamics of the dominant ion species , namely protons and alpha particles .
we show that the response of different species to the fluctuating electromagnetic fields is different . in particular , a significant differential heating of alphas with respect to protons is observed .
interestingly , the preferential heating process occurs in spatial regions nearby the peaks of ion vorticity and where strong deviations from thermodynamic equilibrium are recovered .
moreover , by feeding a simulator of a top - hat ion spectrometer with the output of the kinetic simulations , we show that measurements by such spectrometer planned on board the turbulence heating observer ( thor mission ) , a candidate for the next m4 space mission of the european space agency , can provide detailed three - dimensional ion velocity distributions , highlighting important non - maxwellian features .
these results support the idea that future space missions will allow a deeper understanding of the physics of the interplanetary medium .
february 2014 |
in the seventies of the last century , the concept of list colorings was introduced independently by erds , rubin and taylor @xcite and by vizing @xcite . since then
this topic has been studied extensively by many authors , including @xcite-@xcite . in particular ,
list colorings of planar graphs have received and continue to receive enormous amounts of attention ; see , e.g. , the surveys @xcite .
let @xmath20 be a graph , let @xmath21 be a function , and let @xmath22 be an integer . a _ list assignment _
@xmath0 of @xmath1 is a function that assigns to every vertex @xmath3 of @xmath1 a set ( list ) @xmath2 of colors ( usually each color is a positive integer ) .
we say that @xmath0 is an _
@xmath23-assignment _ or a _ @xmath14-assignment _ if @xmath24 for all @xmath25 or @xmath26 for all @xmath25 , respectively .
a _ coloring _ of @xmath1 is a function @xmath27 that assigns a color to each vertex of @xmath1 so that @xmath28 whenever @xmath29 .
an _ @xmath0-coloring _ of @xmath1 is a coloring @xmath27 of @xmath1 such that @xmath30 for all @xmath25 . if @xmath1 admits an @xmath0-coloring , then @xmath1 is _
@xmath0-colorable_. when @xmath31 the corresponding terms become _ @xmath14-coloring _ and _ @xmath14-colorable _ , respectively . the graph @xmath1 is said to be _
@xmath23-list colorable _ if @xmath1 is @xmath0-colorable for every @xmath23-assignment @xmath0 of @xmath1 .
when @xmath32 for all @xmath25 , the corresponding term becomes _
@xmath14-list colorable_. erds , rubin and taylor @xcite asked , among other problems , the following two questions .
are there planar graphs that are not 4-list colorable ?
is every planar graph 5-list colorable ? both questions were answered in 1993 . in @xcite voigt
gave the first example of a non 4-list colorable planar graph .
in @xcite thomassen answered the second question with a beautiful proof of the following result . [
five ] every planar graph is 5-list colorable .
krekovski @xcite extended this result to @xmath15-minor - free graphs .
[ skre ] every @xmath15-minor - free graph is 5-list colorable . in 2008
hutchinson @xcite published results on list colorings of subclasses of planar graphs , where , for every @xmath11 , the function @xmath33 is the minimum of the vertex degree @xmath9 and a given integer .
if @xmath1 is a 2-connected outerplanar bipartite graph and @xmath34 for all @xmath35 , then @xmath1 is @xmath23-list colorable .
if @xmath1 is a 2-connected outerplanar near - triangulation and @xmath36 for all @xmath25 , then @xmath1 is @xmath23-list colorable except when the graph is @xmath37 with identical 2-lists . in the same paper hutchinson mentioned the following problem posed by bruce richter .
[ p1 ] let @xmath1 be a planar , 3-connected graph that is not a complete graph and let @xmath38 for all @xmath25 . is @xmath1 @xmath23-list colorable ? in this paper
we give partial results concerning the above problem . here
we study the class of @xmath15-minor - free graphs , which contains the class of planar graphs as a subset .
we also investigate the analogous question for non 3-connected @xmath15-minor - free graphs . in that case
both the complete graphs and the so - called gallai trees play a special role .
a _ gallai tree _ is a graph @xmath1 such that every block of @xmath1 is either a complete graph or an odd cycle .
let @xmath39 denote the _ connectivity _ of @xmath1 , that is , the cardinality of a smallest vertex cut set of @xmath1 .
[ p2 ] let @xmath40 and @xmath41 be integers .
let @xmath1 be a @xmath15-minor - free graph with @xmath42 , such that @xmath1 is not a gallai tree .
is @xmath1 @xmath23-list colorable when @xmath43 ?
an important tool for our investigations is the following theorem .
[ gallai ] let @xmath1 be a connected graph and let @xmath0 be a list assignment with @xmath44 for all @xmath35 .
if @xmath1 has no @xmath0-coloring , then the following three conditions hold .
1 . @xmath45 for every vertex @xmath11 .
g is a gallai tree .
3 . let @xmath46 be the set of blocks of @xmath1 and @xmath47 the set of blocks of @xmath1 containing a specified vertex @xmath3 .
there exist color sets @xmath48 for all @xmath49 such that @xmath50 whenever @xmath51 and @xmath52 have a common vertex and for all @xmath11 we have @xmath53 . in this paper
we investigate problem [ p2 ] , considering subsets of planar graphs that fulfill special requirements .
let @xmath54 be the sets of vertices with small degree in @xmath1 and with big degree in @xmath1 , respectively .
the smallest distance between components of @xmath17 $ ] in @xmath1 is denoted by @xmath19 , where @xmath17 $ ] is the subgraph of @xmath1 induced by @xmath55 .
if @xmath17 $ ] has at most one component , then let @xmath56 .
we may always assume that @xmath1 is connected since otherwise we can consider seperately each component of @xmath1 .
we will answer the question of problem [ p2 ] for many cases .
our results for @xmath15-minor - free graphs are summarized in the following tables .
* table 1 : * @xmath57 + 5 & & & & ? + 6 & & / ? & ? & + + 7 & & / ? & ? & + + @xmath58 & & + & + & + + * table 2 : * @xmath59 + 5 & & & & ? + 6 & & ? & ? & + + 7 & & + & + & + + @xmath60 & & + & + & + + in section 2 we give some results for @xmath61 that can be obtained by simple observations including the solution for @xmath62 .
section 3 contains our main results for connected graphs , whereas section 4 deals with graphs with @xmath59 . in section 5
we consider the case @xmath63 and in section 6 we mention open problems .
note that the original problem asked for planar graphs with @xmath59 . in that case
the answer for @xmath64 and @xmath61 is still unknown .
all other entries of the above tables are valid also for planar graphs .
in this section we collect some immediate results .
let @xmath65 , that is , let @xmath66 if @xmath1 is a @xmath15-minor - free graph and @xmath67 for all @xmath25 , then @xmath1 is @xmath23-list colorable .
in this case we have @xmath68 for all @xmath25 and we are done since every @xmath15-minor - free graph is 5-list colorable @xcite . if @xmath1 is 5-connected , then the degree of each vertex is at least 5 .
so our next observation follows immediately . if @xmath1 is a 5-connected @xmath15-minor - free graph , then @xmath1 is @xmath23-list colorable .
the next observation follows from theorem [ gallai ] .
if @xmath1 is a @xmath15-minor - free graph that is not a gallai tree and @xmath69 for all @xmath25 , then @xmath1 is @xmath23-list colorable .
let @xmath1 be a @xmath15-minor - free graph .
if the vertices of degree at most 5 have pairwise distance at least 3 in @xmath1 , then @xmath1 is @xmath23-list colorable . [ dist3 ] to prove observation [ dist3 ] , we color each vertex @xmath70 with an arbitrary color from its list .
we delete the color used on each @xmath3 from the lists of the neighbors of @xmath3 and remove the colored vertices from @xmath1 , obtaining @xmath71 .
because of the hypothesis , every remaining vertex has at most one neighbor @xmath72 in @xmath1 . thus we have @xmath73 for the reduced lists of the vertices of @xmath71 , so @xmath71 is @xmath74-list colorable by theorem [ skre ] .
if @xmath1 is a @xmath15-minor - free graph with @xmath75 , then @xmath1 is @xmath23-list colorable .
* let @xmath1 be a @xmath15-minor - free graph with @xmath75 and let @xmath0 be a list assignment with @xmath76 for all @xmath35 .
let @xmath77 be the components of @xmath78 $ ] . for each @xmath79
, choose a vertex @xmath80 that has at least one neighbor @xmath81 in @xmath79 .
color each @xmath82 by @xmath83 .
this is always possible since @xmath84 .
delete @xmath85 from the lists of the neighbors of @xmath82 in @xmath1 , obtaining a list assignment @xmath86 .
form subgraph @xmath87 from @xmath1 , by removing the vertices of @xmath88 and the vertices @xmath82 .
since the vertices @xmath82 have distance at least 3 to each other in @xmath1 , each vertex @xmath3 of @xmath87 still has at least 5 colors available in @xmath86 .
thus , we can list color @xmath87 from the list assignment @xmath86 by theorem [ skre ] . now consider each @xmath79 .
for every vertex @xmath89 , delete the colors from its list used on its colored neighbors .
we obtain a new list assignment @xmath90 .
note that @xmath91 for all @xmath89 and @xmath92 . by theorem [ gallai
] , each @xmath79 is @xmath90-list colorable .
this completes the @xmath0-list coloring of @xmath1 .
@xmath93 note that an analogous proof works if @xmath78 $ ] has only one component .
now we consider @xmath43 for arbitrary @xmath94 .
if @xmath1 is at most 2-connected , then there are graphs that are not @xmath23-list colorable .
the first examples , with @xmath95 , were given in @xcite .
here we give an example with @xmath96 and minimum degree 3 .
let @xmath94 be an integer .
there are planar 2-connected non - complete graphs @xmath1 with @xmath64 and @xmath97 and list assignments @xmath0 with @xmath98 such that @xmath1 is not @xmath0-list colorable .
* let @xmath0 be the following list assignment for the graph @xmath1 in figure 1 , with @xmath99 .
* @xmath101 , where @xmath102 1.3 mm ( 60,50 ) since the vertex pair @xmath103 has to be colored by one of the pairs @xmath104 with @xmath105 , it follows that one of the vertex pairs @xmath106 with @xmath107 is not colorable , since the only remaining available color is @xmath108 .
@xmath93 the above proposition shows that in the original problem 1 we can not replace the assumption of 3-connectivity by 2-connectivity and/or @xmath97 . furthermore , hutchinson @xcite pointed out that in problem 1 , even if we keep 3-connectivity , the requirement of planarity can not be dropped ; what is more , it can not be relaxed to the restriction of being @xmath15-minor - free .
[ c3,d2 ] let @xmath94 be an integer .
there are 3-connected , non - complete , @xmath15-minor - free graphs @xmath1 with @xmath64 and list assignments @xmath0 with @xmath98 such that @xmath1 is not @xmath0-list colorable .
* let @xmath109 .
let @xmath110 be the graph with @xmath111 such that @xmath112 , @xmath113 , and @xmath114 are pairwise adjacent and every @xmath81 is adjacent to each of @xmath112 , @xmath113 , and @xmath114 .
let @xmath0 be the following list assignment .
* @xmath115 * @xmath116 , where @xmath117 the three vertices @xmath112 , @xmath113 , @xmath114 have to be colored by one of the triples @xmath118 with @xmath119 .
it follows that one of the vertices @xmath120 with @xmath107 is not colorable , since the three colors in its list are used on @xmath112 , @xmath113 , and @xmath114 .
if @xmath121 and @xmath43 , then there are planar non - complete graphs @xmath1 with @xmath122 that are not @xmath23-list colorable . * proof .
* let @xmath87 be a minimal non-4-list colorable planar graph and @xmath86 a 4-assignment such that @xmath87 is not @xmath86-list colorable .
thus @xmath123 .
let @xmath124 .
take @xmath125 copies of @xmath37 where the vertices of the @xmath126-th copy are denoted by @xmath127 .
build @xmath1 by joining each @xmath81 with @xmath128 and @xmath129 .
thus @xmath130 .
let @xmath131 be colors that are not contained in the lists @xmath132 for @xmath133 .
let @xmath134 and @xmath135 .
thus @xmath136 and @xmath137 .
it is easy to see that @xmath1 is not @xmath0-list colorable .
obviously the same construction with shorter lists works if @xmath63 or @xmath65 .
@xmath93 note that our example is a graph @xmath1 with @xmath95 .
we do not know an analogous example for @xmath96 .
however , the example shows that , at least for small @xmath14 , the additional assumption @xmath138 is not sufficient if we do not have an assumption on the degree of connectivity .
in contrast , we show next that if @xmath139 , then a list coloring is always possible . [
eight ] let @xmath139 be an integer and let @xmath43 for all @xmath25 . if @xmath1 is a @xmath15-minor - free graph with @xmath138 such that no component of @xmath1 is a gallai tree , then @xmath1 is @xmath23-list colorable . *
* we prove the theorem for @xmath140 .
assume the theorem is false .
let @xmath1 be a smallest counterexample and let @xmath0 be a list assignment such that @xmath1 is not @xmath0-list colorable .
since @xmath1 is a smallest counterexample it must be connected .
if all vertices of @xmath1 have degree at most 8 , then @xmath141 for all @xmath25 .
theorem [ gallai ] shows that such a @xmath1 is @xmath0-list colorable if it is not a gallai tree .
this is a hypothesis of the present theorem , so we are done .
if all vertices of @xmath1 have degree at least 5 , then we are also done , since every @xmath15-minor - free graph is 5-list colorable @xcite .
thus , we may assume that both @xmath142 and @xmath143 are nonempty .
now we would like to apply the following strategy .
let @xmath144 be a component of @xmath145 $ ] .
if @xmath144 is not a gallai tree , then we remove it from the graph . if @xmath144 is a gallai tree
, then we color some vertices of @xmath144 and delete the used colors from the lists of all corresponding neighbors in @xmath1 ( see the case distinction below ) . if such a neighbor @xmath146 belongs to @xmath143 , then we shall delete at most three colors from its list .
after that , we remove @xmath144 .
if all components of @xmath145 $ ] are removed , then the remaining subgraph is @xmath15-minor - free and all its lists have cardinality at least @xmath147 , since @xmath148 in @xmath1 . by theorem [ skre ]
the remaining graph @xmath149 $ ] is list colorable from its reduced lists .
now we re - insert all uncolored vertices and we delete from their lists all colors used for corresponding neighbors .
let @xmath150 denote the subgraph of @xmath144 induced by the set of all uncolored vertices of @xmath144 .
we shall show that each @xmath150 is list colorable from the reduced lists @xmath86 .
for all uncolored vertices , we know that @xmath151 .
of course , it suffices to show that either @xmath1 is not a smallest counterexample or @xmath152 . by theorem [ gallai ]
, it also suffices to show in each case that @xmath150 is connected and that it satisfies one of the following three conditions : * @xmath150 is not a gallai tree or * there is a @xmath3 in @xmath153 such that @xmath154 or * there are adjacent vertices @xmath3 and @xmath146 in @xmath150 that are not cut vertices and that satisfy @xmath155 . if this strategy works , then the graph @xmath1 is @xmath0-list colorable , which contradicts the assumption that it is a counterexample to the theorem .
if @xmath144 is not a gallai tree , then we let @xmath156 , and we are done by ( i ) .
if @xmath144 is a gallai tree , then its blocks are odd cycles and/or complete graphs on at most four vertices , since @xmath1 is @xmath15-minor - free .
we distinguish the following seven cases . *
( 1 ) @xmath144 is a complete graph @xmath157 , @xmath158 . *
the absence of a @xmath15-minor implies that each vertex in @xmath159 has at most 3 neighbors in @xmath144 .
since @xmath1 is connected , at least one vertex @xmath3 of @xmath144 satisfies the inequality @xmath160 .
thus we can color @xmath144 properly from its lists .
note that we use at most 3 colors from the list of each neighbor in @xmath1 .
now @xmath152 , so we are done . * ( 2 ) @xmath144 is an odd cycle @xmath161 , @xmath162 .
* denote the vertices of the cycle by @xmath163 .
if there are adjacent vertices @xmath164 on the cycle such that there exists a color @xmath165 , then we use color @xmath166 on @xmath81 .
now @xmath150 is a path , where @xmath167 is an end vertex , and @xmath168 , so we are done by ( ii ) .
otherwise we have @xmath169 . since @xmath1 is connected , @xmath170 for all @xmath126 .
we color the cycle with 3 colors from the lists .
thus @xmath152 . *
( 3 ) @xmath144 has an end block with a vertex @xmath3 that is a cut vertex of @xmath1 .
* let @xmath171 be the end block . since @xmath3 is a cut vertex of @xmath1 and @xmath172 for all @xmath173 , we can color all the vertices of @xmath174 before we color @xmath149 $ ] . since @xmath1 is a smallest counterexample , we can color @xmath175 from its lists . *
( 4 ) @xmath144 has an end block that is an odd cycle @xmath161 , @xmath162 .
* denote the vertices of the cycle by @xmath176 , and let @xmath177 be the cut vertex .
since we are not in case ( 3 ) or ( iii ) , we know that for all @xmath178 we have @xmath179 and @xmath180 . thus , we can color @xmath181 through @xmath182 with 3 colors , so that @xmath181 and @xmath182 get the same color .
now @xmath183 and @xmath150 is connected , so we are done by ( ii ) . *
( 5 ) @xmath144 has an end block that is a @xmath184 .
* denote the vertices of the end block by @xmath177 and @xmath181 , where @xmath177 is the cut vertex .
since we are not in case ( 3 ) , we have @xmath185 .
choose two colors , say @xmath186 and @xmath187 , from @xmath188 and delete them from the lists of all neighbors of @xmath181 belonging to @xmath149 $ ] .
remove @xmath144 from @xmath1 . after the coloring of @xmath149 $ ] , add @xmath156 and assign @xmath189 . since @xmath190 , we are done by ( ii ) . * ( 6 ) @xmath144 has an end block that is a @xmath37 .
* denote the vertices of the end block by @xmath191 and @xmath192 , where @xmath177 is the cut vertex .
since we are not in case ( 3 ) , at least one of @xmath181 and @xmath192 has a list of cardinality at least 3 , say @xmath193 .
choose 3 colors from this list for @xmath181 and continue as in the previous case .
note that the same approach would work for endblocks @xmath194 if we wanted to prove the theorem only for @xmath195 , rather than @xmath196 .
however , to prove the theorem for @xmath140 , we must consider this final case . *
( 7 ) all end blocks of @xmath144 are @xmath194 . *
fix one of the end blocks and denote its vertices by @xmath197 and @xmath198 , where @xmath177 is the cut vertex . if there is a color @xmath199 where @xmath200 then color @xmath81 by @xmath186 and continue as prescribed in the strategy .
it follows that @xmath150 is connected and @xmath201 , so we are done by ( ii ) .
thus @xmath202 .
if @xmath203 for @xmath204 then we are done by case ( 3 ) .
so we know that @xmath205 for all @xmath126 , and each of @xmath206 and @xmath198 has at least one neighbor belonging to @xmath143 . *
( 7a ) there are @xmath207 and a vertex @xmath208 such that @xmath146 is a neighbor of @xmath81 , but not a neighbor of @xmath167 .
* remove @xmath144 from @xmath1 and join @xmath146 with all neighbors of @xmath167 belonging to @xmath143 .
note that the new graph is still @xmath15-minor - free .
after removing all components of @xmath145 $ ] , color the remaining graph from the corresponding lists .
this is possible since the graph is @xmath15-minor - free and all lists have cardinality at least 5 . remove the additional edges , add @xmath144 and delete the colors of the neighbors of the vertices of @xmath144 from their lists .
let @xmath186 be the color used on @xmath146 .
if @xmath209 , then we have @xmath210 and we are done by ( ii ) .
otherwise @xmath211 . but
no neighbor of @xmath167 is colored by @xmath186 since all of them were joined to @xmath146 by the additional edges .
since @xmath212 and @xmath213 , we have @xmath214 , so we are done by ( iii ) . * ( 7b ) each vertex @xmath215 that is adjacent to at least one of @xmath216 is adjacent to all of them . * note that @xmath216 must have some neighbors in @xmath143 since we are not in case ( 3 ) .
let @xmath215 be such a neighbor .
then there is no path from @xmath177 to @xmath114 that does not use @xmath206 or @xmath198 , since otherwise we have a subdivision of @xmath15 .
this statement holds also for all further neighbors of @xmath216 in @xmath143 .
so by removing the edges @xmath217 and @xmath218 , we get two components @xmath219 containing @xmath177 and @xmath220 containing @xmath216 and @xmath114 .
if @xmath219 is a gallai tree , then we can color it since @xmath221 .
otherwise we can color it , since it satisfies the hypothesis of the theorem and @xmath1 is a smallest counterexample to it .
delete the color of @xmath177 from the lists of @xmath206 and @xmath198 and consider @xmath220 with the reduced list assignment @xmath86 .
we need to prove the following claim to argue that @xmath220 satisfies the hypothesis of the theorem .
* claim : * @xmath220 _ is not a gallai tree .
_ assume to the contrary that @xmath220 is a gallai tree .
note that @xmath220 has at least one vertex of degree at least 8 , namely @xmath114 .
let @xmath222 be a vertex of degree at least 8 that has the largest distance from @xmath181 in @xmath220 .
this @xmath222 belongs to at least 3 blocks of @xmath220 , since it has at most 3 neighbors in each block ( as @xmath1 is @xmath15-minor - free ) .
moreover , @xmath222 can have neighbors of degree less than 8 in at most one of these blocks , since @xmath148 .
thus there are vertices of degree at least 8 that have a greater distance from @xmath181 than @xmath222 , contradicting the assumption for @xmath222 .
hence the claim is proved .
thus @xmath220 satisfies the hypothesis of the present theorem and we can color it from the reduced lists , since @xmath1 is a smallest counterexample . by combining the colorings of @xmath219 and @xmath220 , we obtain an @xmath0-list coloring of @xmath1 , which is a contradiction .
this completes the proof of the theorem .
[ seven ] let @xmath223 be an integer and let @xmath1 be a @xmath15-minor - free , non - complete , 3-connected graph . if @xmath138 , then @xmath1 is @xmath23-list colorable when @xmath224 .
* we use a strategy similar to the proof of theorem [ eight ] .
let @xmath1 be a smallest counterexample to the theorem .
since @xmath1 is 3-connected and non - complete , @xmath1 is not a gallai tree .
so , if all vertices of @xmath1 have degree at most 7 , then we are done by theorem [ eight ] .
similarly , if all vertices of @xmath1 have degree at least 5 then we are done by theorem [ skre ] . for each component @xmath144 of @xmath225 $ ]
, we would like to color one or more vertices of @xmath144 such that for each adjacent vertex in @xmath226 we have to delete at most two colors from its list .
if we do this for all components of @xmath225 $ ] , then we can color @xmath227 $ ] from the reduced lists .
this is possible because @xmath228 , which ensures that all reduced lists have cardinality at least 5 .
note that we do not need a connectivity assumption for this argument . for a component @xmath144 of @xmath225 $ ] , let @xmath150 be the subgraph induced by the uncolored vertices of @xmath144 . for every vertex @xmath229
, we obtain a reduced list @xmath230 by deleting all colors from @xmath2 that were used for the already colored neighbors of @xmath3 .
finally , we try to color every @xmath150 from the list assignment @xmath86
. it will suffice to show , for each @xmath150 , that @xmath150 is connected and satisfies one of conditions ( i ) , ( ii ) and ( iii ) , from theorem [ eight ] .
if @xmath144 is not a gallai tree , then let @xmath156 , and we are done by ( i ) .
thus we can assume that @xmath144 is a gallai tree .
* ( 1 ) @xmath144 is a single vertex @xmath3 .
* color @xmath3 . since @xmath152 , we are done . *
( 2 ) @xmath144 is @xmath184 or @xmath144 has an end block @xmath37 . *
if @xmath144 is @xmath184 , then denote the vertices of @xmath144 by @xmath177 and @xmath181 .
if @xmath144 has an end block @xmath37 , then denote the vertices of the end block by @xmath231 and @xmath192 , where @xmath192 is the cut vertex . if there is an @xmath199 ( @xmath232 ) , then color @xmath81 by @xmath186 . in this case , @xmath233 , so we are done by ( ii ) .
so instead , we may assume @xmath234 .
* ( 2a ) there is a vertex @xmath235 that is adjacent to @xmath81 and not adjacent to @xmath167 ( @xmath232 ) .
* remove @xmath144 and join @xmath114 with all neighbors of @xmath167 .
the new graph is still @xmath15-minor - free .
let @xmath156 , with the lists reduced by the coloring of @xmath227 $ ] with the additional edges .
let @xmath186 be the color of @xmath114 in this coloring .
if @xmath209 , then we have @xmath236 , so we are done by ( ii ) . otherwise we have @xmath212 and @xmath213 , so we are done by ( iii ) . * ( 2b ) every @xmath235 that is a neighbor of @xmath177 or @xmath181 is a neighbor of both . *
since @xmath1 is 3-connected , there have to be at least 3 such neighbors ( including @xmath192 if @xmath144 is an end block @xmath37 ) .
let @xmath237 be the set of all these neighbors .
consider @xmath238 $ ] .
this subgraph is connected since @xmath1 is 3-connected .
thus any vertex pair @xmath239 is joined by a path in @xmath87 .
consider an auxiliary graph @xmath240 with @xmath241 , where two vertices @xmath129 and @xmath242 are joined by an edge if and only if there is path from @xmath129 to @xmath242 in @xmath87 not containing another vertex @xmath243 .
note that @xmath240 has to be a tree since otherwise there is a subdivision of @xmath15 in @xmath1 .
let @xmath244 be a leaf of @xmath240 and @xmath245 its neighbor in @xmath240 .
since @xmath246 or @xmath244 is a cut vertex of @xmath144 , it must have a neighbor @xmath247 .
now consider a path @xmath248 from @xmath146 to @xmath249 in @xmath1 . since @xmath244 is a leaf in @xmath240 , every path from @xmath146 to @xmath249 must contain a vertex @xmath250 as an internal vertex .
if @xmath248 contains @xmath251 , then , in addition to @xmath249 , @xmath248 must contain another neighbor of @xmath112 that is in @xmath252 ; call it @xmath129 .
thus every path from @xmath146 to @xmath249 contains some vertex @xmath253 as an internal vertex .
since there have to be at least three internally disjoint paths from @xmath146 to @xmath249 , we have a path from @xmath244 to some @xmath254 that has no internal vertices in @xmath255 .
however , now @xmath240 is not a tree .
this again gives us a subdivision of @xmath15 , which contradicts the @xmath15-minor - freeness of @xmath1 .
* ( 3 ) @xmath144 is an odd cycle @xmath161 , @xmath256 , or @xmath144 has an end block that is an odd cycle @xmath161 , @xmath256 .
* denote the vertices of @xmath161 by @xmath257 , where the cut vertex ( if it exists ) is @xmath182 .
if there is an @xmath199 ( @xmath258 ) , then color @xmath81 by @xmath186 . in this case ,
@xmath233 , and we are done by ( ii ) .
so instead , we have @xmath259 for all @xmath260 . if there is a vertex @xmath261 that is adjacent to @xmath81 and not adjacent to @xmath167 ( @xmath262 ) , then there exist @xmath263 and @xmath264 and @xmath265 , which are adjacent on the odd cycle , such that @xmath266 is adjacent to @xmath264 and not adjacent to @xmath265 .
remove @xmath144 and join @xmath266 with all neighbors of @xmath265 .
the new graph is still @xmath15-minor free .
let @xmath156 , with the lists reduced by the coloring of @xmath227 $ ] with the additional edges .
let @xmath186 be the color of @xmath266 in this coloring .
if @xmath267 , then @xmath268 , so we are done by ( ii ) .
otherwise @xmath269 and @xmath270 , so we are done by ( iii ) .
thus we may assume that every vertex @xmath261 that is adjacent to at least one @xmath81 , @xmath271 , is adjacent to all of them .
since @xmath1 is 3-connected , there must be at least two such neighbors that lie in the same component of @xmath227 $ ] . to build a @xmath15-minor
, we take these two neighbors together with @xmath272 and @xmath192 . *
( 4 ) @xmath144 is @xmath37 or @xmath194 or has an end block that is @xmath194 . *
if @xmath144 is @xmath37 , then denote the vertices of @xmath144 by @xmath231 and @xmath192 .
otherwise , denote the vertices by @xmath273 and @xmath198 , where the cut vertex ( if it exists ) is @xmath198 .
if there is an @xmath199 ( @xmath274 ) , then color @xmath81 by @xmath186 . in this case ,
@xmath233 , and we are done by ( ii ) .
so instead , we have @xmath275 . if there is a vertex @xmath261 that is adjacent to @xmath81 and not adjacent to @xmath167 ( @xmath276 ) , then remove @xmath144 and join @xmath146 with all neighbors of @xmath167 .
the new graph is still @xmath15-minor free .
let @xmath156 , with the lists reduced by the coloring of @xmath227 $ ] with the additional edges .
let @xmath186 be the color of @xmath146 in this coloring .
if @xmath209 , then @xmath236 , so we are done by ( ii ) . otherwise @xmath212 and @xmath213
, so we are done by ( iii ) .
thus we may assume that every vertex @xmath261 that is adjacent to @xmath191 or @xmath192 is adjacent to all of them .
since @xmath1 is 3-connected , there must be at least 3 such neighbors that all lie in the same component of @xmath277 . to build a @xmath15-minor , we take two of these neighbors and @xmath191 and @xmath192 . *
( 5 ) all end blocks of @xmath144 are @xmath184 .
* denote the vertices of an end block by @xmath177 and @xmath181 , where @xmath177 is the cut vertex .
since @xmath1 is 3-connected , @xmath193 .
choose two colors @xmath186 and @xmath187 from @xmath188 .
delete @xmath186 and @xmath187 from the lists of all neighbors of @xmath181 in @xmath226 and let @xmath189 .
after the coloring of @xmath227 $ ] we have @xmath278 , so we are done by ( ii ) .
this completes the proof of the theorem . @xmath93
* remark : * the above proof also works for @xmath65 , except in the final case .
let @xmath63 . there is a planar , 3-connected , non - complete graph @xmath1 with @xmath279 and a list assignment @xmath0 with @xmath280 for all @xmath25 such that @xmath1 is not @xmath0-list colorable .
( 148,76 ) * proof . * consider the subgraph @xmath144 in figure 2a with list assignment @xmath281 , @xmath282 , and @xmath283 , for @xmath284 .
we claim that three colors can be assigned to each of the triangles @xmath285 such that in every coloring of @xmath144 at least one of the triangles is colored by its assigned colors ; note that the order of the colors in each assignment matters .
the assignments are @xmath286 note that the prescribed coloring of @xmath287 or @xmath288 occurs if @xmath289 is colored by 1 .
analogously , the prescribed coloring of @xmath290 or @xmath291 occurs if @xmath292 is colored by 1 .
thus we may assume that @xmath289 and @xmath292 are not colored by 1 .
but now the prescribed coloring of @xmath293 or @xmath294 occurs .
now assume that inside of each of @xmath295 we have the structure in figure 2b . for each triangle @xmath296 ,
let @xmath297 be the colors assigned to @xmath296 .
let @xmath298 and @xmath299 and @xmath300 .
thus @xmath144 is not @xmath86-list colorable for the given list assignment . now build a graph @xmath1 by taking 25 copies of @xmath144 and identifying all 25 @xmath112-vertices to a vertex @xmath301 and all 25 @xmath113-vertices to vertex @xmath302 .
join vertex @xmath303 in the @xmath126-th copy with vertex @xmath304 in the @xmath305-th copy , for @xmath306 .
define a list assignment @xmath0 , as follows .
let @xmath307 and @xmath308 .
for all other vertices let @xmath309 , where each pair of @xmath310 corresponds with the color pair @xmath311 of one copy of @xmath144 . finally , observe that @xmath1 is a planar , 3-connected non - complete graph with a list assignment @xmath0 such that @xmath312 for all @xmath25 , but @xmath1 is not @xmath0-list colorable .
@xmath93 note that @xmath279 for the example in the proof above .
in fact , it is almost the same example as in @xcite .
despite our progress in this paper , two essential problems remain open . | a _ list assignment _ @xmath0 of a graph @xmath1 is a function that assigns a set ( list ) @xmath2 of colors to every vertex @xmath3 of @xmath1 .
graph @xmath1 is called _ @xmath0-list colorable _ if it admits a vertex coloring @xmath4 such that @xmath5 for all @xmath6 and @xmath7 for all @xmath8 .
the following question was raised by bruce richter .
let @xmath1 be a planar , 3-connected graph that is not a complete graph .
denoting by @xmath9 the degree of vertex @xmath3 , is @xmath1 @xmath0-list colorable for every list assignment @xmath0 with @xmath10 for all @xmath11 ? more generally , we ask for which pairs @xmath12 the following question has an affirmative answer .
let @xmath13 and @xmath14 be integers and let @xmath1 be a @xmath15-minor - free @xmath13-connected graph that is not a gallai tree ( i.e. , at least one block of @xmath1 is neither a complete graph nor an odd cycle ) .
is @xmath1 @xmath0-list colorable for every list assignment @xmath0 with @xmath16 for all @xmath11 ?
we investigate this question by considering the components of @xmath17 $ ] , where @xmath18 is the set of vertices with small degree in @xmath1 .
we are especially interested in the minimum distance @xmath19 in @xmath1 between the components of @xmath17 $ ] .
[ lem]observation [ theo]conjecture [ lem]proposition |
the study of bubble nucleation and growth is a broad field of study with rich physics@xcite . some of the more recent examples include experimental and theoretical studies of superfluid helium in a superheated state@xcite , and the theoretical study of nucleation of magnetic bubbles@xcite . in this paper
we look at the problem of the quantum description of bubble growth in superheated liquid helium .
this is just one example of many systems whose classical nature is well known but whose quantum representation is not .
the fact that not all functions can be quantized uniquely into operators in the hilbert space has been known for some time .
the groenewald - van hove theorem has explicitly shown that dirac s program of assigning operators to all functions by having the poisson brackets change into commutators works only for polynomials of quadratic form or less@xcite . in the case of superheated liquid helium ,
the difficulty in quantization appears in the form of a position dependent mass ( pdm ) , whereby the kinetic term in the classical hamiltonian is space dependent .
in fact , many systems in nature have the pdm structure ; for instance , free carriers , such as electrons , in semiconductors of non - uniform chemical composition are often described by a pdm hamiltonian@xcite .
other examples of pdm appear in various instances of nuclear many - body problems , quantum dots , and nano - mechanical systems just to name a few@xcite . as a result
there have been a number of attempts to address the issue of operator ordering ambiguities due to pdm , using the gallileian transformation and other methods@xcite .
we aim in the present paper to provide insight into several existing techniques in quantization , their convergence and possible future usage .
in particular , we show in this paper how to apply the supersymmetric quantum mechanics ( susy qm ) formalism@xcite which have been previously applied to solve simpler pdm problems@xcite .
the main difference is that , unlike the usual examples of susy qm , our hamiltonian contains a specific potential which is not easily factorizable ; the method presented in this paper shows how one may apply susy qm to a more general situation .
on the other hand , we apply to our system a more formal quantization procedure which takes the heisenberg commutation relations as primary , the weyl quantization otherwise referred to as the `` weyl transform''@xcite .
this allows one to quantize a pdm hamiltonian but does not provide explicit operator ordering . by combining the two approaches namely the susy qm method and the weyl transform , we are able to identify specifically a couple of correct operator orderings out of infinitely many possible orderings .
the paper is divided as follows : in section 2 , we give a brief description of nucleation in superheated liquid helium , which makes the problem of pdm obvious . in section 3
, we describe the details of using the susy qm formalism , to address the pdm problem . in section 4
, we introduce weyl quantization and use it to obtain an operator form for our classical hamiltonian .
this resulting weyl quantum hamiltonian is converted to a point mass representation to compare with the hamiltonian obtained through susy methods . in section 5
we study the effective potential using realistic parameters , and provide a conclusion .
a superheated ( or supercooled ) fluid@xcite is typically metastable since it can not make a direct , uniform transition to the stable phase throughout its volume .
the transition to the stable phase occurs via the nucleation of a droplet of stable fluid , such that once the nucleus reaches a critical size it grows quickly , ultimately converting the entire liquid from metastable to stable phase .
nucleation can occur due to fluctuations in pressure and volume within a liquid , whereby small quantities of a new phase ( `` bubbles '' ) form within a homogeneous phase .
if the liquid were stable these bubbles would disappear ; but superheated liquid helium is metastable , so these bubbles can grow and eventually transform the system from the old liquid phase onto the new vapor phase .
this growth process is driven by pressure gradients across the surface of the bubble .
a recent study@xcite provided a comprehensive classical description of such a bubble nucleation process that takes into account the fluctuations in both the radius and the pressure .
the helmholtz free energy corresponding to this process is : @xmath0 where @xmath1 is the radius of the bubble and @xmath2 the pressure outside of the bubble .
@xmath3 is the pressure inside the bubble where @xmath4 is the surface tension , @xmath5 is the internal equilibrium pressure of the bubble , and @xmath6 is the isothermal compressibility of the stable phase at the equilibrium pressure
. the first term does not affect the rate of nucleation since it describes the pressure fluctuations which average out to a constant .
the second and third terms of eq .
( [ freeenergy ] ) provide the barrier , with respect to the radius , over which the nucleus must pass in order to expand and fill the volume with the stable phase .
the critical radius for vapor nucleation is then @xmath7 . to establish the ( classical ) hamiltonian for bubble nucleation
, we note that the kinetic energy of a growing nucleus is given by@xcite @xmath8 where the variable mass is given by : @xmath9 with @xmath10 being the density of the liquid and @xmath11 is the vapor density . the potential term based on eq .
( [ freeenergy ] ) leads to the hamiltonian for the bubble nucleation problem @xcite @xmath12 where @xmath13 is the radius of the bubble scaled to the critical radius and @xmath14 and @xmath15 .
returning to our particular case in question , we note that if we were to quantize our classical pdm hamiltonian naively by replacing @xmath16 and @xmath17 by the corresponding position and momentum operators @xmath18 and @xmath19 , an infinite set of operator forms are possible .
for example , @xmath20 , @xmath21 $ ] , @xmath22 , to list just some of the simplest cases ; we show how to identify the correct specific operator ordering below .
while our potential is not super - symmetric , the supersymmetric quantum mechanics ( susy qm ) formalism permits the re - writing of the hamiltonian in terms of the familiar creation - annihilation operators by introducing an effective potential ( `` superpotential '' ) . the kinetic part of of a general pdm hamiltonian ( with @xmath23 , and without any confining potential ) can be written @xmath24 where @xmath25 ( and @xmath26 ) can take any value as long as @xmath27 .
one can then write the creation - annihilation operators @xmath28 as @xmath29 where @xmath30 is known as superpotential and the corresponding harmonic hamiltonian is given by @xmath31 where @xmath32 are the kinetic terms and @xmath33 are the corresponding potential terms that effectively play the role of the quadratic confining potential of a standard harmonic oscillator .
the form of @xmath34 is fixed by demanding that @xmath35 obey the heisenberg algebra i.e. @xmath36 = 1 $ ] .
this condition leads to @xmath37 which then implies @xmath38 we emphasize here that the procedure is independent of the form of the mass the mass @xmath39 is completely general .
the essence of the susy procedure is therefore the introduction of an extra potential @xmath40 which now contains all the effect of operator ordering via the parameter @xmath25 . for our specific pdm corresponding to the case of superheated liquid helium ,
@xmath41 so that : @xmath42 and @xmath43 it is noted that when @xmath44 @xmath45 which are identical , respectively , to the equivalent superpotential and potential obtained in the classical formalism using poisson brackets in place of commutators@xcite . to solve our problem , we need to write the hamiltonian corresponding to our situation in terms of the operators @xmath28 .
writing out the `` momentum '' part of @xmath46 using our pdm @xmath47 , one can show that @xmath48 which does not depend on the parameter @xmath25 i.e independent of ordering . using this result ,
one is able to write down the potentials @xmath33 and @xmath49 in terms of @xmath50 in a straightforward manner . in terms of these operators , the hamiltonian for our system
is given by subtracting @xmath51 from @xmath52 and building the @xmath49 potential by remembering that @xmath16 is unitless in the original hamiltonian this is equivalent in this case to multiplying our @xmath16 by a factor @xmath53 to get : @xmath54 + \frac{1}{2 } .
\label{hq}\end{aligned}\ ] ] the hamiltonian of eq .
( [ hq ] ) is our pdm hamiltonian when @xmath28 is defined as in eqs .
( [ am]-[ap ] ) with the superpotential @xmath34 of eq .
( [ wa ] ) .
effectively , we have re - written our hamiltonian solely in terms of the new raising and lowering operators @xmath55 and @xmath56 .
we note again that in eq .
( [ hq ] ) , the kinetic energy term is still implicitly with an indeterminate operator ordering i.e. @xmath57 of eq .
( [ tplus ] ) .
we note that the operators @xmath58 and @xmath59 function as raising and lowering operators and are constructed such that they abide to the heisenberg algebra , @xmath60=1 $ ] , in the same way as the well - known raising and lowering operators for the standard quantum harmonic oscillator ( qho ) .
guided by the definitions of the standard qho , we therefore create an ansatz whereby we define new `` position '' and `` momentum '' variables : @xmath61 and @xmath62 . @xmath63 and @xmath64 then play the role of position and momentum operators with the usual commutation relation , @xmath65 = i$ ] .
it is noted that the commutation relation for @xmath63 and @xmath64 implies @xmath66 via the stone - von neumann theorem@xcite which states that given @xmath67 = i$ ] , there is a representation such that @xmath68 and @xmath69 . in these definitions ,
@xmath70 and @xmath71 correspond to the mode energy and the mass of the bubble . scaling our hamiltonian by multiplying throughout by @xmath72 and changing the variable to @xmath73 for a later comparison ( note that the variable @xmath74 in this section relates to @xmath63 . here
@xmath74 is _ not _ a coordinate transform of the original position variable @xmath16 as done in the next section ) , we are led to the following hamiltonian by a straightforward substitution of @xmath63 and @xmath64 into eq .
( [ hq ] ) : @xmath75 where @xmath76\ ] ] and @xmath77 + \frac{1}{2 } .\ ] ] here @xmath78 is modified to take into account the scaling of variables .
the effective potential , @xmath79 , is fairly complicated compared to the original potential ; in particular it contains a singularity at @xmath80 such is the `` price '' of removing the pdm from the kinetic term .
the energy spectrum of the effective hamiltonian of eq .
( [ heff ] ) is identical to that of the original pdm hamiltonian ( up to a constant factor ) .
on the other hand , because the position variable @xmath74 is now in a different space , the eigenstate of eq .
( [ heff ] ) is not the eigenstate of the original hamiltonian , although there are ways to map the eigenstate of eq .
( [ heff ] ) back onto the eigenstate of the original pdm hamiltonian .
once the parameter @xmath25 is determined , the unique operator ordering is decided ; we now suggest a way to select the value of this parameter .
we note here that the steps involved to derive eq .
( [ heff ] ) is that of `` picking a particular basis '' for the hamiltonian of eq .
( [ hq ] ) rather than a coordinate transformation from one position coordinate system to another position coordinate system .
in particular , by choosing ansatz based on the well - known quantum harmonic oscillator , the resulting hamiltonian is assured of being of unit measure and hermitian .
these requirements of unit measure and hermiticity is explained in more detail in the next section , where a coordinate transformation is directly involved .
there exist in the literature several ways to formally convert classical function into operator form .
weyl quantization otherwise referred to as the `` weyl transform '' is a well - known way of doing so .
there are many qualities of the weyl transform that make it an optimum way to perform this conversion@xcite .
the weyl transform is defined as : @xmath81(x ) & = & \frac{1}{2 \pi } \int \int [ ft](\alpha , \beta ) e^{i\hbar \alpha \beta/2 } e^{i \beta x } \nonumber \\ & & \times \phi(x + \hbar \alpha ) \omega(\alpha,\beta ) \/ d\alpha
\/ d \beta .
\label{wt}\end{aligned}\ ] ] here @xmath82(\alpha , \beta)= \int t(p , x ) \exp [ -i(p \alpha + x \beta ) ] \/ dp \/ dx\ ] ] denotes the fourier transform of the classical function , @xmath83 and @xmath84 is the wave function upon which the weyl operator acts .
the weighting factor , @xmath85 , characterizes the type of quantization to be utilized .
weyl quantization , in particular , is defined when @xmath86 .
therefore , the weyl transform of the kinetic part of our classical hamiltonian is @xmath87(x ) = \frac{1 } { 2 \pi } \int \int f \left ( \frac{p^2}{2 m_0 x^3 } \right ) \nonumber \\ & & \times e ^{i \hbar \alpha \beta/2 } e^{i \beta x } \phi ( x + \hbar \alpha ) \/ d\alpha
\/ d\beta .\end{aligned}\ ] ] using the result @xmath88 \left [ -i \sqrt { \frac{\pi}{2 } } \left ( - \frac{\beta^2}{2 } sgn \beta \right ) \right ] , \ ] ] the weyl transform for our case is @xmath89 = - \frac{\hbar^2}{2 m_0 r_c^2 } \nonumber \\ & & \times \left [ \frac{1}{x^3 } \frac{\partial^2}{\partial x^2}\phi(x ) - \right .
\frac{3}{x^4}\frac{\partial}{\partial x } \phi(x ) + \frac{3}{x^5 } \phi(x ) \right ] .
\label{weyleq.}\end{aligned}\ ] ] we note here that the hermiticity of the hamiltonian must be ensured .
one can show that , given a general hamiltonian of the form @xmath90 the hermiticity condition , @xmath91^ { * } { \phi}(x ) dx \label{herm}\ ] ] imposes the following conditions after integrating various terms by parts : ( i ) @xmath92 i.e. @xmath93 is real ; ( ii ) @xmath94 $ ] ; ( iii ) @xmath95 = \frac{1}{2 } { \rm im}[b'(x)]$ ] where @xmath96 $ ] and @xmath97 $ ] denote real and imaginary parts respectively .
this general result shows that the weyl transform such as the result of eq .
( [ weyleq . ] ) preserves the hermiticity of the hamiltonian . in order to compare this result with the hamiltonian obtained in susy , we begin with eq .
( [ weyleq . ] ) and note that the associated kinetic term in the corresponding lagrangian is @xmath98 we introduce a coordinate transformation @xmath99 where @xmath100 and @xmath101 are real numbers transforming @xmath102 to @xmath103 . in order for the @xmath104 factor in the first term to disappear , one needs to choose @xmath105 and @xmath106 i.e. @xmath107 , so that the mass term becomes a constant , @xmath108 and eq .
( [ lpdm ] ) is transformed to @xmath109 with this substitution , eq.([weyleq . ] ) is transformed into @xmath110 . \label{pmr}\ ] ] once the coordinate transform is carried out , the measure ( or the jacobian of integration @xmath111 ) associated with the hamiltonian of eq.([pmr ] ) is no longer unity i.e. coordinate transformation does not preserve the inner product of the hilbert space , which is problematic . in general , unit measure can be restored through the comparison of the inner products between two hamiltonians @xmath112 and @xmath113 where @xmath112 is the hamiltonian for the inner - product space with unit measure and @xmath113 corresponds to that with non - unit measure @xmath114 so as to preserve the hilbert space : @xmath115 noting that @xmath116 where @xmath117 or @xmath118 , eq .
( [ meas ] ) implies the relationship between the two hamiltonians as @xmath119 . writing a general unit measure hamiltonian @xmath112 as @xmath120 and the non - unit measure hamiltonian as one without the tildes ,
the above relation between @xmath121 and @xmath113 implies @xmath122 \nonumber \\ & & - b(z ) \left [ \frac{1}{2 } \frac{\mu'(z)}{\mu(z ) } \right ] + c(z ) .\end{aligned}\ ] ] we note that these results are general results for any hamiltonian of the form eq .
( [ unith ] ) ; it is not a property of a specific hamiltonian , or any transforms involving differential operators , but is rather a self - consistent result following from preserving the inner product , and ensuring hermiticity .
a corollary of these results is that for a non unit measure hamiltonian , the hermiticity condition implies @xmath123 i.e. the hamiltonian of eq .
( [ pmr ] ) is still hermitian .
this relation also means that a point - mass , unit measure hamiltonian will always have @xmath124 due to hermiticity condition
i.e. the first order derivative term always vanishes , which is what we normally see in `` ordinary '' hamiltonians . in our case , since @xmath125 , the measure is calculated to be @xmath126 .
the hamiltonian with unit measure emanating from eq.([pmr ] ) is then @xmath127 + u_0 z^{4/5 } ( 1 - z^{2/5 } ) .
\label{hqz}\ ] ] equating this result of eq .
( [ hqz ] ) with the hamiltonian obtained through susy qm , eq . ( [ heff ] ) , we require @xmath128 which implies that specific operator orderings that corresponds to the weyl quantization rule are given by @xmath129 . therefore , the quantum hamiltonian for the bubble nucleation problem that obeys the weyl quantization rule is either @xmath130 + u_0 x^2(1-x)\ ] ] or @xmath131 + u_0 x^2(1-x ) .\ ] ] we have therefore narrowed down the possible operator orderings to two ( out of infinity ) .
this is the central result of our paper .
we note here that it seems on first sight that one could avoid the operator ordering ambiguity from the outset by transforming the classical hamiltonian to a point mass form first , and then quantizing it .
however , the proper procedure is quantization and then coordinate transform as we have done in this paper .
this is because one loses more information than necessary otherwise the extra information that a weyl transform gives us as a result of having pdm is lost by carrying out the coordinate transformation first . more specifically
, one would not obtain the term proportional to @xmath132 if the point mass transform was done first .
this would also be inconsistent with the susy procedure of section 3 which independently gave the term proportional to @xmath132 .
finally , we note the deeper connection between the methods of section 3 and this section .
the fact that the relatively simple procedure based on susy qm formalism matches the result of section 4 that involves weyl transform , coordinate transform for finding a point mass representation and re - scaling to unit measure is , indeed , a demonstration of mathematical self - consistency . at a deeper level ,
the match is not surprising since the relationship between @xmath16 and @xmath63 of section 3 is indeed the same as the coordinate transform of section 4 , namely @xmath133 .
however , the method of section 3 is somewhat more efficient than that of section 4 since it bypasses the need to find the point mass representation and to restore unit measure .
this efficiency is possible since the quantization takes place not in the position space but in the space of raising and lowering operators ; picking a useful position representation related to well known quantum harmonic oscillator then ensures hermiticity and unit measure owing to the stone - von neumann theorem .
the goal of this paper was in finding the correct operator ordering for the bubble nucleation problem .
a full quantum mechanical treatment of the bubble nucleation process that addresses tunneling through the barrier will be presented elsewhere@xcite . in this section , before we conclude the paper , we attempt to gain some physical insight into this problem by briefly examining the effective hamiltonian of eq .
( [ heff ] ) [ and eq .
( [ hqz ] ) ] with realistic parameters .
we consider physically reasonable parameters for this system so that we can get an order of magnitude estimate for the parameters .
the typical values for superfluid helium at temperature @xmath134 is @xmath135 , @xmath136 , @xmath137 . for zero applied pressure ,
the classical critical radius is @xmath138 . at @xmath139 , the thermal de broglie wavelength of a single helium atom
is @xmath140 , and hence thermal momentum of @xmath141 over all range of applied pressure @xmath2 .
typically a single bubble contains an order of 100 helium atoms , or momentum of roughly @xmath142 . for pressure @xmath143 ( dashed line ) and @xmath144 ( solid line ) .
( b ) same as ( a ) but for @xmath145 instead .
( c ) @xmath146 for the same values of pressure .
( d ) @xmath146 for several values of pressure @xmath147 $ ] . in this figure ,
the position @xmath74 is in scaled units ( in units of @xmath148 ) and the potential is in ev.,height=264 ] with these realistic numbers , we have calculated and plotted the shape of @xmath149 , and @xmath150 in fig .
[ potential ] for various values of applied pressure @xmath151 .
the non - confining shape of the potential makes it clear why the numerical solution to the eigenvalue problem is difficult and non - convergent .
the changing shape of @xmath145 near @xmath80 due to changes in pressure and subsequently @xmath148 is also shown in fig . [ potential ] . on the other hand
, the potential for larger @xmath74 corresponding to @xmath152 looks simpler .
it is clear from fig .
[ potential ] demonstrating unbounded potential that the energy spectrum is expected to be continuous , and the bubble growth is similar to a ball rolling downhill with the `` gradient '' dependent on the applied pressure .
we note that had we not transformed the hamiltonian and left the pdm in the kinetic energy term we would not be able to come up with such an intuitive understanding of the system . to conclude
, we have studied the bubble nucleation problem in a superheated liquid helium and provided a quantum mechanical description .
although the system seems rather complicated at the outset and furthermore involves an exotic hamiltonian that diverges at the origin due to the specific form of the pdm , we were able to extract useful information through the methods of susy qm and the weyl transform .
we have presented a convergence between two methods by converting to a point mass description of the quantum hamiltonian .
further analysis will need to be performed in regards to comparison to the microscopic description of the problem at hand , and these will be presented elsewhere@xcite .
g. bastard , _ wave mechanics applied to semiconductor heterostructures _ editions de physique , les ulis , ( 1998 ) ; p. ring and p. schuck , _ the nuclear many body problem _ springer new york ( 1980 ) ; a. dutra , c. a. s. almeida , phys .
a. 275 , ( 2000 ) , 25 ; a. d. alhaidari , phys . rev . a 66 , ( 2002 ) , 66 ; c .- y .
cai , z .- z .
ren , and g .- x .
ju , commun .
43 , ( 2005 ) , 1019 | we discuss a quantum description of bubble growth in a superheated liquid helium by addressing the problem of operator ordering ambiguities that arise due to the presence of position dependent mass ( pdm ) in this system . using a supersymmetric quantum mechanics formalism along with the weyl quantization rule , we are able to identify specific operator orderings for this problem .
this is a general method which should be applicable to other pdm systems . |
cosmic microwave background ( cmb ) experiments such as the wilkinson microwave anisotropy probe ( _ wmap _ ) have made significant progress in the study of the primordial temperature fluctuations .
their best fitting power spectra strongly support a spatially flat , @xmath4 , universe .
this model requires relatively few parameters , yet apparently manages a compelling concordance between a variety of other cosmological data ; snia , large scale structure and big bang nucleosynthesis .
although the statistical errors on these power spectra are small , this precision does not necessarily imply accuracy and there remains the potential for systematic errors to alter these conclusions .
indeed , several anomalies between @xmath4 and the _ wmap _ data have been discussed .
typically these have involved the large - scale temperature multipoles eg : @xcite . however ,
other anomalies in the cmb at smaller scales have also been detected , connected in particular with radio sources @xcite and sz decrements from galaxy clusters @xcite radio sources are sometimes regarded as a contaminant in cmb temperature maps . however , radio point sources prove particularly interesting because they provide a complementary check of the beam measured by the _ wmap _
team from observations of jupiter @xcite .
jupiter has a flux of @xmath5 1200jy which is @xmath6 orders of magnitude higher than radio source fluxes or cmb fluctuations .
this high flux has advantages in terms of defining the wings of the beam profile but has the disadvantage that the calibrating source is much brighter than typical cmb fluctuations .
furthermore , jupiter only checks the beam on the ecliptic whereas radio sources are spread over the sky . @xcite
made a stacked analysis of radio point sources and found evidence for a wider beam than _ wmap _ measured using jupiter .
a tentative detection of a non - linear relation between _ wmap _ fluxes and ground based radio telescope fluxes was also found .
a thorough analysis of possible systematics did not find an explanation and we return to these issues later in this paper .
the beam profile of a cmb telescope like _ wmap _ is critical because it smoothes the temperature anisotropies and therefore needs to be known accurately to produce the final power spectrum from temperature maps @xcite .
various authors have noted small - scale anomalies with respect to the sz decrements measured by _
sz decrements are created when cmb photons inverse compton scatter off hot electrons in galaxy clusters . @xcite
first stacked _
wmap _ data at the positions of galaxy clusters and suggested that the profiles were more extended than expected . @xcite and
@xcite then found that the sz decrements from _ wmap _ were reduced compared to x - ray predictions , possibly due to the _ wmap _ beam being wider than expected .
@xcite also found that the _ wmap _ decrements were significantly lower than the ground - based sz measurements by @xcite in 38 x - ray luminous clusters . in their esz sample ,
the _ planck _
team find excellent agreement with the self - similar x - ray estimates of the sz decrement @xcite .
this is corroborated by the ground based south pole telescope collaboration with their blind sz selected cluster sample @xcite .
this compounds the question of why _ wmap _ sz analyses from @xcite and @xcite failed to find such an agreement . in this paper
we use the recent _ planck _ early data release and other radio source data to re - investigate both the _ wmap _ radio source beam profile and sz anomalies .
the _ planck _ early release compact source catalogue ( ercsc ) is of particular interest and provides the basic parameters of radio sources and sz clusters from the _ planck _ cmb maps .
although , the corresponding temperature maps from which these were estimated have not been released , both radio source fluxes and sz profile parameters are available as measured by _ planck_. we can therefore use these to compare _ wmap _ and _ planck _ radio source fluxes directly and also to make _ wmap _ stacks centred now on the new radio source and sz cluster lists from _
planck_. from these stacks , the _ wmap _ beam profile can be inferred and the sz results from _ wmap _ and _ planck _ compared .
given the higher angular resolution , lower noise and different calibration strategy for _ planck _ , this comparison will allow new insight into the robustness of the _ wmap _ cmb analysis .
the _ planck _ team have recently made their first release of data collected by the _ planck _
satellite between 13 august 2009 and 6 june 2010 ( amounting to @xmath7 full sky surveys ) .
this early data release is concerned solely with the foreground contamination in the cmb maps .
the two sets of catalogues relevant to this paper form the early release compact source catalogue ( ercsc ) .
these are the radio source catalogues and the sz catalogue .
the early release compact source catalogue ( ercsc ) lists all the high reliability radio sources with accurate flux determinations .
the ercsc has been quality controlled so that @xmath8 of the reported sources are reliable , @xmath9 , detections and that the fluxes are determined within @xmath1030% accuracy .
the catalogues are band specific and for the bands of interest ( @xmath11 ) are created using the ` powellsnakes ' method , a bayesian multi - frequency algorithm for detecting discrete objects in a random background .
flux estimates were obtained by use of aperture photometry within a circle of the beam s fwhm . for the case of unresolved and potentially faint point sources ,
the _ planck _
team recommend the use of the parameter flux and its corresponding error , flux_err @xcite .
we reject any extended objects from the catalogue to maintain an unresolved sample with which to test the _ wmap _ data .
to do this we have used the _ planck _
quality tag ` extended ' .
this is defined by comparing the source areal profile with the 2-d _ planck _ beam . an additional quality flag ` cmbsubtract '
has also been provided , which reflects on the quality of the source detection in a map with the best estimate of the cmb removed .
we minimise cmb contamination by using only cmbsubtract=0 sources . when measuring the beam profile in section [ subsec : planckradioresults ] we further cut the catalogue to ensure the best quality sample .
@xcite did suggest that their faintest _ wmap _ source samples were probably affected by @xcite bias . to ensure the robustness of our results against eddington bias
, we have used a @xmath12jy flux cut , the same limit as previously used by @xcite .
we have additionally rejected sources within @xmath13 of the lmc , sources at low galactic latitude , @xmath14 and any sources flagged by _
planck _ as having high astrometric error .
finally , we tightened the _ planck _ ` extended ' flag to remove any sources intrinsically wider than the _ wmap _ beam .
the _ planck _ ` extended ' flag excludes sources with @xmath15 .
we now ensure that the _ planck _
sources are unresolved in the _ wmap _ maps by imposing cuts in both the major and minor axis so that both the fitted gaussian profiles ( gau ) and the local psf ( beam ) fwhm estimates are less than the fwhm of the _ wmap _ beam in the band being studied .
band and colour corrections for the _ wmap _ and _ planck _ fluxes have been ignored .
this factor is in any case small due to the typically flat spectral indices considered @xcite .
the full details of the catalogue construction and composition are described by @xcite and briefly overviewed in table [ tb : bands - planckpara ] .
the early sz ( esz ) catalogue lists all the robust and extensively verified sz detections in the first data release .
as described by @xcite , the _ planck _
team extract the integrated sz signal , the _ y _ parameter , using a multifrequency matched filter ( mmf3 ) method @xcite .
the algorithm is run blindly on all - sky maps , assuming the characteristic sz spectral signature and self - similar cluster profile . in the early release of the _ planck _ sz catalogue ,
only data from the 100ghz frequency channel or higher has been used to study the sz effect .
this is to avoid the detrimental effect on s / n from beam dilution caused by the larger beam sizes of the lower frequency channels . at the higher frequencies , the _
beam fwhm is typically @xmath16 .
the full details of the catalogue construction and composition are described by the @xcite .
the catalogue provides estimates of the sz flux , extent , redshift and position .
it consists of 189 clusters , all detected at high s / n ( @xmath17 ) with 95% reliability . whilst the sample is primarily composed of known clusters ( 169/189 ) , it provides a wealth of new information as it gives the first sz measurements for @xmath18 of the clusters . in this paper
we only make use of clusters which have been pre - detected in the x - ray and have redshifts .
we therefore , after masking , consider 151 clusters , including coma .
for this sample the redshift range spans @xmath19 $ ] with a mean redshift of @xmath20 . we will be using the 7-year _ wmap _ temperature maps obtained from the lambda cmb resource .
we work with the @xmath21 ` healpix ` maps resulting in a pixel scale of @xmath22 .
we use the foreground unsubtracted temperature band maps for q , v and w. our default _ wmap _ datasets are the co - added maps in q ( = q1+q2 ) , v ( = v1+v2 ) and w ( = w1+w2+w3+w4 ) .
however , particularly in the w band , the increased s / n for radio source profiles obtained by using all the da s can be regarded as a trade - off with the precision of just using the narrowest w1 ( and w4 ) beams as previously used by @xcite . in using the co - added data ,
the jupiter beams have to be combined before comparison with the data .
we estimate the jupiter beam in each band by averaging the 7-year beam profiles from the various detector assemblies , assuming the appropriate correction for pixelisation @xcite . when working with radio point sources we use the point source catalogue mask ( ` wmap_point_source_catalog_mask ` ) . to avoid galactic contamination for the sz analyses
we have instead used the extended temperature mask ( ` wmap_ext_temperature_analysis_mask ` ) which admits 71% of the sky .
we have used the 7-year _ wmap _ 5-band point source catalogue @xcite .
these sources are detected at least the 5@xmath23 level in one _ wmap _ band .
for a flux density to be stated , the detection must be above the 2@xmath23 level in that band .
following @xcite we ensure that the sources are genuinely point sources by matching to the 5ghz ( @xmath24 resolution ) catalogues from the greenbank northern sky survey ( gb6 , @xcite ) , or parkes - mit - nrao ( pmn , @xcite ) , surveys .
wmap _ team also provide a 7 yr cmb - free catalogue as described by @xcite .
this catalogue has been created with the objective of detecting point sources free of boosting by cmb fluctuations .
we proceed with the raw 5-band catalogue with 471 sources and a cmb - free catalogue with 417 sources .
.summary of the _ planck _ bandpass parameters and the flux range of the sample we use from the ercsc , @xcite . [ cols="^,^,^,^ " , ]
the choice of coordinates to use for the radio source positions and cluster centres is important . scatter or an offset in the centroid used in the cross - correlation could cause the stacked result to appear artifically broad or induce artifacts .
however , the only sample used in this paper for which astrometric errors are appreciable are the _ wmap _ derived radio - source catalogues where the astrometric error in both longitude and latitude is 4@xmath25 @xcite . we mitigate for this effect by using the position of the corresponding matched 5ghz source , since these have negligble astrometric error @xmath26 @xcite .
we find no evidence for an offset between the _ wmap _ and 5ghz sources and hence we are confident that astrometry error will not cause new broadening of the beam .
we also note that the stacking procedure we use is dominated by the brighter objects , which typically have better astrometry . for the _ planck _ radio source catalogues we have used the _ planck _
positions since these are of high astrometric quality @xcite . in our sample selection
we have rejected sources which the _ planck _ team estimate to have relatively high astrometric errors .
the effect of this selection in the 100ghz band is to ensure that @xmath27 for the @xmath12jy sample . for the _ planck _ esz objects we have taken the _ planck _ estimated positions rather than the x - ray derived positions as the cluster centres .
we do this to avoid the complications associated with rare but potentially large offsets between the sz and x - ray signals which are likely caused by merging events @xcite . our cross - correlation / stacking procedures for measuring both radio point source profiles and sz decrements are similar to those of @xcite , @xcite and then as updated by @xcite .
ultimately , we shall be stacking / cross - correlating _ wmap _ data around radio source positions and cluster centres from catalogues , particularly from the _ planck _ ercsc . to estimate a temperature profile for an individual source _ j _ we use , @xmath28 where the sum is over the pixels , denoted _
, within a circular annulus of radius @xmath29 .
here @xmath30 represents the number of pixels within the annulus and @xmath31 is the temperature recorded for the pixel _
i _ and source _
j_. @xmath32 is the average background temperature which can either be estimated locally in a surrounding annulus in a ` photometric method ' or globally ( see @xcite ) .
these two background estimates make no difference in the stacked results but can make a difference for individual sources ( see section [ sec : szresults ] ) .
we then stack the _ _ wmap__7 data by averaging @xmath33 over the sources that have pixels within the annulus @xmath29 using the statistical average , @xmath34 .
@xmath35 is the number of sources that have pixels within the annulus @xmath29 and is usually constant for all except the @xmath36 bins .
we have followed @xcite in using jack - knife errors , for both the radio and sz sources , based on 6 equal area sub - fields defined by lines of constant galactic longitude and split by the galactic equator .
for @xmath37 fields denoted _ k _ , the errors are , @xmath38^{2 } \big\rangle , \label{eq : errdefintion}\ ] ] where @xmath39 is the average of the fields excluding field @xmath40 . we have experimented with both alternative sub - fields and methods such as bootstrap resampling finding approximately equivalent results . in section [ sec : sim ] we have used simulations to test whether our method can robustly recover the beam profile , in doing so we find that our jack - knife errors are reasonable .
we first compare _ _ wmap__7 sources at q , v , w from @xcite to their counterparts in the _ planck _ ercsc at 100 , 70 and 44 ghz .
we also compare the _ planck _ fluxes in the 100ghz band to the ground - based atca and iram source fluxes previously used by @xcite . in fig .
[ fig : wmapdirectfluxcompare ] we first focus on the comparison of the _
_ wmap__7 fluxes to _ planck _ and also the ground - based atca and iram sources .
we only consider the matches with separation less than @xmath41 to avoid any possible systematic errors associated with sources that have poor astrometry .
however , our results are independent of this cut up to separations of @xmath42 . at high fluxes
we see evidence for a systematically lower _ wmap _ flux , @xmath43 above 2jy .
this non - linearity is particularly prominent in the w band , the band with the greatest angular resolution . since there are errors in both variables we have used the numerical recipes @xcite _ fitexy _ as our fitting routine .
we find best fit logarithmic slopes of @xmath44 $ ] for the [ 100ghz - w,70ghz - v,44ghz - q ] comparisons . to obtain realistic errors on these fits
we have linearly scaled the flux errors until we obtained a @xmath45 probability of 0.5 as recommended by @xcite , for data with a dominant intrinsic dispersion . clearly , _
planck _ and
_ wmap _ fluxes for sources were measured at different times . since at least @xmath46% of the _ _
wmap__5 radio sources exhibit some level of variability @xcite , we expect and observe much larger scatter than accounted for by the estimated flux uncertainty .
however , we note that the brighter _ wmap _ sources are fainter than the one - to - one relation , this is in the opposite sense expected if variability was biasing faint _
sources into the _ wmap _
catalogue when in a bright phase .
we investigate whether variability is alternatively causing a bias due to _ planck _ dropouts by limiting the _ wmap _ sample to @xmath47 sources .
the advantage is that the _ planck _ team have investigated all the _ wmap _ @xmath47 objects that are not in the ercsc .
they conclude that for the 100ghz - w comparison the missing objects are ` all ' spurious and can be explained by the object having a weak or missing 5ghz i d @xcite .
the resulting @xmath47 _ wmap _ w - band sample of 48 sources ( with @xmath48 ) is therefore complete in the sense of being 100% represented in the _ planck _ sample with no bias due to a _
planck _ dropout population .
when we then repeat the _ wmap_-_planck _
100ghz - w flux comparison , we find a logarithmic slope of ( @xmath49 ) , consistent with the original result and therefore strong evidence against a highly variable source population causing dropouts that bias the _ wmap_-_planck _ comparison .
the disagreement between _ planck _ and _ wmap _ is in contrast to direct comparisons between _ planck _ and ground - based atca / iram data .
these instead show good agreement , as shown in fig .
[ fig : planckwdirectfluxcompare ] for the _ planck _ 100ghz radio point sources .
the best fit logarithmic slope of @xmath50 $ ] is statistically consistent with the one - to - one relation .
however , comparing _
wmap _ w - band and the ground - based atca / iram data we find evidence for non - linearity with a best fit logarithmic slope of @xmath51 $ ] .
these contrasting fits are particularly significant because the greatest _
planck_-_wmap _ non - linearity comes from the 100ghz - w flux comparison .
given the agreement between _
planck _ and the ground - based atca / iram observations , we interpret the flux disagreement as being due to _ wmap_. in response to the referee we have made additional bootstrap and jack - knife re - sampling tests of the _ wmap_*_planck _ flux - flux comparison and its error . after 1000 bootstrap resamplings
we estimate logarithmic slopes of @xmath52 $ ] for the [ 100ghz - w,70ghz - v,44ghz - q ] comparisons .
we also perform jacknife resamplings of the _ wmap_*_planck _ flux - flux comparison , we estimate a logarithmic slope of @xmath53 $ ] for the [ 100ghz - w,70ghz - v,44ghz - q ] comparisons .
these resampling results are consistent with our original samples and support the accuracy of our previous error analysis .
we have also made monte - carlo simulations of the flux - flux comparison .
we generated samples with the same number of sources as in the real flux - flux comparison using the _ wmap _ q - band power - law distribution , @xmath54 , @xcite .
these fluxes are then scaled to the respective _ wmap _ and _ planck _ band centres on the basis of a gaussian distribution in spectral indices , @xmath55 , with mean -0.09 and standard deviation 0.176 , @xcite .
realistic gaussian measurement errors are then assigned as a function of flux in a manner consistent with the original _ wmap _ and _ planck _ samples . to include variability we start from the @xcite analysis of the _ _ wmap__5 data that measured a median rms flux variability for the 25 brightest q - band objects of @xmath56 and which we therefore additionally apply to all our sources , assuming a gaussian distribution .
we then impose detection limits corresponding to the faintest source in the given band for the _ wmap _ and _ planck _ fluxes respectively . finally , we compare these two flux types by measuring the best fit relation in the same way as was originally done for the _
planck_-_wmap _ comparison .
after 10,000 simulations of the 100ghz - w comparison we find average logarithmic slopes and intercepts of @xmath57 $ ] and @xmath58 $ ] .
these results are not only in agreement with a one - to - one relation but support the errors found in our original _ wmap_-_planck _ comparison .
we therefore conclude we are able to robustly recover the expected one - to - one result and hence that our comparison may be unbiased .
@xcite have noted that the 100ghz bandpass contains the @xmath59 rotational co emission line .
this is a potential explanation for the flux non - linearity we report between the _ wmap _ 94ghz w - band and _ planck _ 100ghz bands .
however , such an explanation would imply that _ wmap _ and _ planck _ are in agreement away from the galactic plane where co emission is lower .
however , we see no evidence for such a distinction , with galactic latitude cuts of @xmath60 , @xmath61 and @xmath62 we find 100ghz - w logarithmic slopes of @xmath63 $ ] , @xmath64 $ ] , and @xmath65 $ ] respectively .
@xcite have made a comparison of _ wmap _ fluxes of bright radio source fluxes from ground - based telescopes and claim that they support the _ wmap _ flux scale .
some of the sources used are planets and have not been through the same reduction procedures as the cmb maps but five other sources , cyg a , cas a , tau a , 3c58 and 3c274 have gone through the same procedures . @xcite selected these sources on the basis that they were the brightest and least variable of the sources with adequate background contrast and a history of prior observation .
we now expand our flux comparisons by including these @xcite sources in fig .
[ fig : wmapdirectfluxcompare ] .
we use the _ wmap _ fluxes quoted by these authors and the independent ground - based fluxes that are mostly those quoted by these authors .
we see that in the w band at least , _ wmap _ also underestimates the flux of these sources ( blue points ) and indeed cyg a , 3c274 and 3c58 appear not inconsistent with our fitted relation . however
, the underestimation for cas a is less than predicted by extrapolating the fit to the brighter radio fluxes .
if this result were to be more highly weighted then there would still be evidence for a _ wmap _ flux problem , but one which now looked more like a constant offset than a scale error .
however , we note that there are differences between the two ground - based observations of cas a. furthermore , cas a lies close to the galactic plane ( @xmath66 ) and hence contamination might be an issue
. it may therefore be too early to infer a flux offset on the basis of this source .
when we include the celestial sources from @xcite with independent ground - based measurements into the _
planck_-_wmap _ comparison we find logarithmic slopes of @xmath67 $ ] for the [ 100ghz - w,70ghz - v,44ghz - q ] comparisons .
after jack - knife and bootstrap resampling we find logarithmic slopes of @xmath68 $ ] and @xmath69 $ ] respectively for the [ 100ghz - w,70ghz - v,44ghz - q ] comparisons .
hence , whilst including the celestial source data changes the degree of the non - linearity , the results are still in significant disagreement with a one - to - one relation .
we finally add the _ planck _ ercsc measurements of these 5 sources to fig .
[ fig : wmapdirectfluxcompare ] and fig .
[ fig : planckwdirectfluxcompare ] .
this complicates the picture further since they appear to agree with the _ wmap _ results more than the ground - based results .
as far as we can see , the _ planck _
fluxes are not calibrated via _
wmap_. if we then fit the full _ planck_-_wmap _ w band comparison we now find less evidence for a discrepancy between the two finding a 100ghz - w logarithmic slope of @xmath70 $ ] . but just making the 100ghz - w comparison in the 3 - 400jy region , the result might then again look more like a constant offset with a logarithmic slope and intercept of @xmath71 $ ] and @xmath72 $ ] .
we conclude that the _ wmap _ fluxes in the @xmath73jy region show systematically lower fluxes than _ planck _ and we have argued that this discrepancy is unlikely to be explained by variability , underestimated errors or inaccurate flux estimation . at lower and higher fluxes the _ wmap_-_planck _
agreement seems better , implying some non - linearity in their relative scales . if _
wmap _ data are compared to ground - based data rather than _
, the same discrepancy is seen at @xmath73jy and a small but significant discrepancy is seen at brighter fluxes , which would more imply a linear offset rather than a non - linearity .
similar effects are seen at q and v but at a lower level .
we now apply the stacking analysis of @xcite to the co - added _ _
wmap__7 maps , centring on the _ planck _ radio point sources .
_ planck _ sources are selected at multiple wavebands which may be advantageous in avoiding spurious sources etc .
fig : radioproftempandbsplanck ] ( a),(b),(c ) are shown for completeness because these raw temperature plots demonstrate the main uncertainty in this analysis which is the accuracy of the background subtraction .
we note that there is some difference between the global background and the background local to the source samples but generally this effect appears smaller in the _ _ wmap__7 data ( eg at w ) than it was in the _
_ wmap__5 datasets used by @xcite . figs .
[ fig : radioproftempandbsplanck ] ( d),(e),(f ) show the same profiles now background subtracted and scaled to unity at the origin to produce @xmath74 .
we have used the ` photometric ' subtraction to produce the radio point source temperature profile , @xmath75 . for the _ _
wmap__7 dataset there is very little difference in the profiles resulting from global or local / photometric background subtractions .
these @xmath74 are now compared to the _ wmap _ jupiter beam and the best fit beam to the bright _ wmap _ radio source profiles found by @xcite ( dashed orange line in their fig .
2 ) . there is again evidence that the _ planck _ selected radio sources suggest a wider beam than the jupiter beam , particularly in the w band , although the _ planck _ sources lie slightly below the profile fits from @xcite .
we further note that the statistical significance of the deviations from the jupiter beam for the _ planck _ selected sources at @xmath76 is only modest at @xmath77 for the w band .
the normalisation of @xmath74 to unity at small scales forms a further uncertainty in these beam comparisons . in figs .
[ fig : radioproftempandbsplanck ] ( g),(h),(i ) we have applied the formalism of section [ subsec : model - radiofluxes ] and attempted to make absolute normalisations of the various model profiles , using the _ planck _ ercsc listed fluxes .
we assume in turn the jupiter profile and then the radio source profile of @xcite in calculating the resulting @xmath78 factor . these give
respectively the blue and red lines .
hence , if the radio sources followed the jupiter profile , for example , we should see the same peak temperature for the stacked model profile and the stacked data .
we see that the _ planck _ peak temperatures , particularly in the w band , tend to lie between the jupiter profile and the previous _ wmap _ bright radio source fits .
these results suggest that the previous radio source fit may be too wide at @xmath79 where it is essentially an extrapolation , unconstrained by the data , and this will affect the accuracy of its absolute normalisation i.e. there is a large error in @xmath80 . otherwise , the conclusion is similar to that from figs . [
fig : radioproftempandbsplanck ] ( d),(e),(f ) in that the _ planck _
data is suggesting that the jupiter beam is a poor fit to the radio source profiles particularly at w. the question of eddington bias was discussed by @xcite and has also been suggested by @xcite as a possible explanation of the wide radio profiles . in terms of the _ planck _ sources an eddington bias of @xmath50.02mk is required to explain our results .
however our pre - selection of these sources as being point - like at _ planck _ resolution and our rejection of both faint ( @xmath81jy ) and cmb - contaminated sources mean that it is difficult to see how eddington bias could be affecting these results . in fig .
[ fig : radioproftempandbsplanck ] ( d),(e),(f ) we have also presented the source sample without the @xmath12jy flux cut .
the consistency of the full source and brighter source samples indicates that eddington bias is not significantly affecting these samples .
we next repeat the analysis of @xcite using the co - added _ _
wmap__7 maps and source catalogue @xcite .
the results are shown in fig .
[ fig : radioproftempandrawwmap ] .
we see that the results again tend to lie between the jupiter profile and the previous _ wmap _ bright radio source fits by @xcite
. this may in part be due to the _ _
wmap__7 profiles returning to zero at large scales more uniformly than _ _
wmap__5 , making the differences between the photometric and global profile estimates more marginal . however , we also found that using the _ _ wmap__5/__wmap__7 catalogue in the _
_ wmap__7/__wmap__5 temperature maps gives profiles more consistent with the @xcite fits .
we therefore attribute the difference between fig . [
fig : radioproftempandrawwmap ] and the results of @xcite to a possible systematic difference between _ _ wmap__5 and _ _ wmap__7 , with perhaps a contribution from statistical fluctuations .
following @xcite we minimised any effect of eddington bias for this sample by pre - selecting only sources that appear in the 5ghz gb6 and pmn radio samples .
we have only used 5ghz coordinates for the radio sources , with their sub-30@xmath82 accuracy to minimise any positional error in our analysis .
although eddington bias may well be affecting the faintest _ wmap _ sources , as was also noted in @xcite , we have used a flux limit of @xmath12jy . in section
[ sec : sim ] we shall check for the presence of eddington bias in this particular dataset using random simulations . in
the ` cmb - free ' method @xcite,_wmap _ sources are selected using the q , v , w bands simultaneously to form an internal linear combination map ( ilc ) with weights chosen to cancel out the cmb anisotropy signal . again , any eddington bias due to cmb fluctuations should be reduced in the case of this point source catalogue .
we therefore repeated our stacking analysis with the 417 qvw sources from the @xcite _ _ wmap__7 ` cmb - free ' catalogue ( see fig .
[ fig : radioproftempandbscmbfree ] ) .
overall we again see wider - than - expected profiles at w , broadly consistent with the results in figs .
[ fig : radioproftempandbsplanck ] , [ fig : radioproftempandrawwmap ] .
finally , we have also presented these results without the @xmath12jy flux limit , we note that the result is unchanged .
this consistency is evidence for robustness of the result to eddington bias .
point source catalogues made at significantly lower frequencies than the _ wmap _ bands are unlikely to be affected by eddington bias due to cmb fluctuations , if identification is done independently of the _ _ wmap__7 point source catalogue . for example , point sources selected at 1.4 ghz will have rayleigh - jeans temperature @xmath83 higher than a source with similar flux density selected at w - band ( @xmath84 ghz ) , i.e. @xmath85 , whereas the rms rayleigh - jeans temperature due to the cmb fluctuations stays roughly the same between the two frequency bands , @xcite .
therefore , we now stack co - added _
_ wmap__7 temperature data centred around the positions of the 1147 @xmath86 jy nvss point sources . fig .
[ fig : radioproftempandbsnvss ] shows the resulting q , v and w profiles .
we see that they are consistent with those measured using _ _
wmap__5 total / bright sources in fig . 2 of @xcite .
however , the profiles do not appear as wide as the _ _ wmap__5 faintest subsample despite the average flux of the nvss sample at _ wmap _ bands being @xmath87 lower .
many of the nvss sources are resolved into multiple components @xcite . however , this is unlikely to cause the widening of the beam beyond @xmath88 . here , as a precautionary measure , we shall test the beam profile measured using the nvss by excluding any source that has neighbouring source(s ) within @xmath89 .
this extra condition reduces the number of @xmath90 jy sources outside the _ _
wmap__7 ` point source catalogue ' mask to 967 .
the resulting co - added beam profiles are also shown in fig .
[ fig : radioproftempandbsnvss ] .
we see that the beam profiles are in good agreement with the previous results .
we made 100 monte carlo simulations to check our method and the robustness of the results .
these simulations are due to sawangwit ( 2011 ) who made them in the context of his test of the w1 beam in the _ _ wmap__5 dataset .
thus they are conservative in terms of both the errors they imply and the test of eddington bias they make in our new context of the averaged da s ( w1-w4 in the w - band case ) and the _ _ wmap__7 dataset .
we followed the procedures described by @xcite ( see also * ? ? ? * ) . for each set of simulations ,
@xmath91 point sources are generated with a power - law distribution , @xmath92 , at _ wmap _ q - band @xcite .
their spectral indices , @xmath55 , are drawn from a gaussian distribution with a mean -0.09 and standard deviation 0.176 as characterised by the _ _ wmap__5 point source catalogue @xcite . the flux density for each object
is scaled to the centre of the other four bands using the relation @xmath93 .
the source positions are then randomly distributed on the sky and each source is assigned to a pixel in a healpix @xmath94 map . for a source with flux density @xmath95 , the peak rayleigh - jeans temperature difference , @xmath96 ,
is given by eqs.([eq : gammatemprel ] , [ eq : gammadef ] ) , but with the @xmath97 replaced by @xmath98 sr , solid angle of a @xmath94 pixel .
the publicly available _ wmap _ maps ( section [ sec : wmap_data ] ) are given in thermodynamic temperature @xcite . for a direct comparison with our results , we thus convert the simulated source s @xmath96 to @xmath99 using eq.([eq : anttothermo ] ) .
five temperature maps , one for each band , are then smoothed with the corresponding _ wmap _ beam transfer function @xcite before being downgraded to @xmath100 . the simulated cmb temperature map ( smoothed with an appropriate beam transfer function ) constructed from _ _
wmap__5 best - fit @xmath101 and pixel noise are then added to the source temperature maps .
the pixel noise is modelled as a gaussian distribution with zero mean and standard deviation @xmath102 , where @xmath103 is the number of observations in each pixel and @xmath104 is given for each da and frequency band @xcite . here
, we use the _ _
wmap__5 @xmath105 map to generate pixel noise for its corresponding band map .
next , we applied the five - band detection technique following procedures utilised by _ wmap _
team @xcite .
firstly , the temperature maps are weighted by the number of observations in each pixel , @xmath106 .
the weighted map is then filtered in harmonic space by @xmath107 ( e.g. * ? ? ?
* ; * ? ? ?
* ) where @xmath108 is the cmb power spectrum and @xmath109 is the noise power , and @xmath110 is the beam transfer function @xcite . the filter is designed to suppress fluctuations due to the cmb at large scales and pixel noise at scales smaller than the beam width .
we used the _ _ wmap__5 best - fit @xmath101 for @xmath108 . the @xmath111
are determined from pixel noise maps constructed using @xmath104 and five - year @xmath103 for each band as described above .
we then search the filtered maps for peaks which are @xmath112 .
peaks detected in any band are fitted to a gaussian profile plus a planar baseline in the unfiltered maps for all other bands .
the recovered source positions are set to the best - fit gaussian centres in w - band .
the best - fit gaussian amplitude is converted to rayleigh - jeans temperature , using the relation given in eq.([eq : anttothermo ] ) , and then to a flux density using conversion factors , @xmath113 , given in table 4 of @xcite . in
any given band , we only use sources that are @xmath114 and the fitted source width smaller than 2x the beamwidth , following the _ wmap _ team .
the number of detected sources from 100 realisations are consistent with _ _ wmap__5 point source analyses by @xcite and @xcite .
our simulations also recover the input power - law @xmath115 distribution down to the expected _ _
wmap__5 limit , @xmath116 jy , remarkably well ( see * ? ? ?
* ) . for each simulation we applied our beam profile analysis outlined in section [ sec
: method ] ( including a flux cut of @xmath117jy ) .
the average beam profiles derived from 100 simulations are plotted in fig .
[ fig : sim](a)-(c ) where the error bar represents their standard deviation in each angular bin .
we found that even profiles as narrow as the w1-band jupiter profile can be retrieved remarkably well out to @xmath118 .
the estimated uncertainties using these monte carlo simulations are consistent with the jack - knife error estimations .
note that the monte carlo error converges after @xmath560 - 70 simulations .
the monte carlo simulations we performed here suggests that our method for recovering beam profile by stacking temperature maps around point sources is robust and the jack - knife error estimation is reliable .
although we argued above that sources ( i.e. their identifications and positions ) selected at nvss frequency are robust against the cmb fluctuations compared to _
wmap _ bands ,
our beam analysis is still carried out using _ wmap _ temperature maps .
as we noted , the average flux of the @xmath86jy nvss sources in the _ wmap _ bands is @xmath119x lower than the _ wmap _ sample .
therefore it is important to check whether the _ wmap _ beam profiles can be robustly recovered using these nvss sources .
again the results come from sawangwit ( 2011 ) and were only applied to the w1 detector assembly and use _ _ wmap__5 data .
we again created 100 monte carlo simulations similar to those described above but without the five - band detection procedure since these sources are pre - detected by nvss with high positional accuracy ( @xmath120 , @xcite ) .
the 967 nvss source positions are used and fluxes at 1.4 ghz are extrapolated to _ wmap _
q , v and w bands assuming a mean spectral index , @xmath55 , of -0.45 in order to mimic the average flux density observed in these bands .
the temperature maps are smoothed with the corresponding _ wmap _ ( jupiter ) beam profiles .
the simulated cmb fluctuations and radiometer noise are then added to the source temperature maps as described above . for each _ wmap _ band , we applied our beam profile analysis to each of the 100 simulated maps ( including a flux cut of @xmath121jy ) .
the results are shown in fig .
[ fig : sim](d)-(f ) .
the plot shows that with these nvss radio sources the _ wmap _ beam profiles can be robustly recovered out to @xmath118 and are not affected by the source clustering consistent with the semi - empirical calculation presented in @xcite .
we then take the standard deviation of the 100 simulated results in each angular bin as the @xmath122 error .
the ratio of the monte carlo error to the jack - knife error is shown as the dotted line in fig .
[ fig : sim](d)-(f ) .
the monte carlo and jack - knife errors are in good agreement except at scales @xmath123 where jack - knife errors are somewhat over - estimates in q and v bands .
the simulations suggest that when flux limited at @xmath124 , the _ wmap _ selected source profiles are unaffected by eddington bias .
the simulations also support the accuracy of our empirical errors .
the simulations suggest the same conclusions apply when dealing with flux - limited ( @xmath86jy ) nvss data .
@xmath125 profiles for the _ planck _ , _ _ wmap__7 and cmb - free _
_ wmap__7 catalogues .
also shown are the @xmath125 for the jupiter beam ( blue , solid ) and a model showing the effect of a 25.6ms timing offset on the w1 jupiter beam ( orange , dashed ) @xcite.,width=283 ] [ fig : w1_offset_beam ] we now briefly consider possible explanations for the radio source profiles we have observed .
we start by accepting that in the _ _ wmap__7 data the profiles are less wide than found in the _ _ wmap__5 data discussed by @xcite .
we regard our best current result to come from comparison with the _ planck _ radio sources where we have checked the sources against spatial extension at _ planck _ resolution and also rejected any that are contaminated by the cmb .
this sample still rejects the w - band jupiter beam at @xmath77 significance at @xmath76 , about the same as the rejection of the previous best fit model from @xcite .
therefore it is not outside the bounds of possibility that the previous result may be explained by a statistical fluctuation in the _ _ wmap__5 data .
however , at the suggestion of the referee , we have now evaluated the rejection significance of the jupiter beam using the full covariance matrix from our simulations , fitting in the range @xmath126 . for the jupiter beam in the 7yr coadded maps we find formal rejection significances from the @xmath45 distribution of [ @xmath127 , @xmath128 , @xmath129 , @xmath130 for _ planck _ , _ _ wmap__7 , _ _ wmap__7-cmbfree and nvss respectively .
although we note that the overlap between these samples means that these results can not be simply combined , individually these probabilities correspond to @xmath131 rejections of the jupiter beam .
it is therefore still worth considering whether a wider beam could be related to other possible _ wmap _ data problems .
the first of these is the possible disagreement in _ wmap _ flux comparisons with ground - based and _ planck _ datasets noted by @xcite and also in this paper .
certainly a non - linearity like we first fitted to fig .
[ fig : wmapdirectfluxcompare ] goes in the right direction to explain a flatter than expected profile . indeed ,
if the addition of cas a , cyg a , tau a , 3c274 and 3c58 does imply that _ wmap _ fluxes are simply offset from _ planck _ and ground - based fluxes , then flux comparisons would be consistent with the wide beam .
a logarithmic intercept of @xmath132 implies the _ wmap _ flux is @xmath580% of the corresponding _ planck _ flux .
equally , we find the w - band _ wmap _ jupiter beam solid angle is @xmath580% of the 25.6 ms timing offset derived beam s @xmath97 .
this is in agreement with the expectation from eq.([eq : gammatemprel ] ) that at fixed temperatures ( ie : those provided in the _ wmap _ maps ) @xmath133 .
however , more data at brighter fluxes is needed to check if the _ wmap _ flux is non - linear or simply offset with respect to other datasets .
we note that @xcite has used the cmb dipole to look for non - linearity in the _ wmap _ temperature scale and failed to find any evidence for such an effect . the second possible explanation for the wider than expected radio source profiles focused on the possibility that there was a timing offset between the _ wmap _ antenna pointing and temperature data , as proposed by @xcite . as well as causing effects at large scale due to a wrongly subtracted dipole ,
this scan pattern offset would cause a wider beam profile ( see @xcite ) . @xcite
have calculated the beam pattern that a 25.6 ms timing offset would cause in the w band .
the calculation assumes the w1 jupiter beam and takes into account its initial asymmetry on the sky . after creating simulated _ wmap _ tod that include point sources distributed in ecliptic latitude and then applying mapmaking to these data
, @xcite found the azimuth averaged beam profiles shown in fig .
[ fig : w1_offset_beam ] for both zero timing offset and the 25.6ms timing offset with the latter giving a reasonable fit to the _ planck _ data .
more details are presented by @xcite .
these include further results based on using the dependence of beam shape with ecliptic latitude to try and determine the timing offset which marginally prefer zero timing offset .
we note that @xcite made similar tests based on bright _ wmap _ sources and found no evidence for a timing offset at the map - making stage . on the other hand
, @xcite checked between offsets by minimising dipole residuals and found strong evidence for a non - zero offset , see sawangwit et al ( in prep ) .
we note that the _ wmap _ team have indicated that they use a timing offset of zero in which case the above agreement would simply represent a coincidence .
our final aim is to make a comparison between the _ planck _
esz and _ wmap _ sz results as described in section [ sec : model - sz ] , [ subsec : model_sz_recon ] and [ subsec : model_sz_convol ] .
however this involves reverse engineering the _ planck _ sz @xmath134 profiles .
we therefore first use recently released _ planck
_ sz data to check our reverse engineered _ planck _ profiles .
a series of papers have been released as a follow up to the _ planck _ esz data .
two papers in particular are relevant to corroborating the _ planck _ profiles presented in this paper . in @xcite
the _ planck _ coma sz profile has been published .
additionally , the ` physical ' _ planck _ sz temperature profiles for the 62 local clusters ( lsz ) in the @xcite analysis have been published in @xcite .
below , we compare our reverse engineered profiles to these _ planck _ data . in fig .
[ fig : mazzotta_planckdata_coma_results ] we now compare our _ planck _ coma sz profiles to @xcite .
we have shown two alternative _ planck _ models in order to display the sensitivity of our results to the cluster size estimates .
this is motivated by the significant difference between the value of @xmath135 used in @xcite and the esz value , @xmath136 .
since the value for the integrated sz signal , @xmath137 , is degenerate with cluster size the esz value for @xmath137 can not be assumed .
therefore , in the first instance we do show an expected _ planck _ temperature decrement using the esz values and calculated using eq.([eq : selfsimtemppredfinal ] ) .
however , we also show a model which uses an alternative method for calculating the expected _ planck _ profile . here
, the @xcite value of @xmath138 is used to calculate @xmath139 which can then be used to set the profile amplitude as shown in eq.([eq : finalyscale ] ) .
this method is solely dependent on the cluster size estimate , and is further described in appendix [ append : selfsim ] .
we find agreement between the coma self - similar sz model and the observed _ planck _ temperature profiles .
although the _ planck _ data does seem to have both a lower central amplitude and a slightly wider profile at large angular scales than the self - similar expectation .
we note that corresponding differences between the self - similar model and the _ planck _ data can be seen in the @xcite analysis .
a flatter inner profile may also be expected if any pixelisation effects cause any further smoothing beyond the stated resolution of @xmath42 .
we also find reasonable agreement between our two estimates of the _ planck _ profile that use different cluster size estimates .
although , as expected , the model using the @xcite value of @xmath138 does provide a better fit to the _ planck _ data .
we conclude that the agreement between our _ planck _ expectation and the underlying _ planck _ profile supports the validity of our inversion of the _ planck _ esz data to obtain _
planck _ temperature profiles . in fig .
[ fig : pointecouteau_lsz62_results ] we compare our _ planck _ ` physical ' sz profiles for the 62 @xcite clusters to the ` physical ' _ planck _ profiles presented in @xcite .
as was previously shown in @xcite the _ planck _ [ 100,70,44 ] ghz profiles are in agreement with the self - similar expectation .
we now expand on this by attempting to use these results to verify our method of inverting the _ planck _ ercsc data to obtain _
planck _ temperature profiles . since we again find that the @xcite estimates of cluster size can be significantly different from the esz estimates we have used an alternative method of obtaining expected _ planck _ temperature profiles .
this method replicates @xcite s approach in assuming the @xcite self - similar model for the cluster and directly evaluating the compton - y parameter , as outlined in appendix [ append : selfsim ] .
we have further followed @xcite s section . 4.3 in using the @xcite estimates of @xmath140 and
calibrate the central gnfw pressure , @xmath141 , using the x - ray equivalent of the integrated sz signal , @xmath142 . as shown in fig .
[ fig : pointecouteau_lsz62_results ] the two self - similar models convolved with @xmath42 fwhm gaussian beam profiles are in agreement beyond @xmath138 . however the inner profile of @xcite s model ( black , solid @xmath143 dotted ) is substantially sharper than our model ( green , solid ) .
although our model lies within the @xcite s @xmath144 dispersion , we are comparing the stacked models ( ie : the statistical average ) so the error range is a @xmath145 smaller .
we believe this difference is caused by the different stacking procedure used in @xcite where depending on the noise properties within the bin either logarithmic or linear weightings were used . we have found that using these alternative weightings can accentuate the central peak of the profile , although not to the extent necessary for full agreement with @xcite .
we currently do not have a full explanation for the difference in central amplitude . ) using the esz value of @xmath138 ( black , solid ) and eq.([eq : finalyscale ] ) using the @xcite value of @xmath138 ( green , solid ) .
both models are convolved with a @xmath42 fwhm gaussian beam appropriate to the @xcite data .
, width=283 ] ` physical ' sz decrements for the 62 @xcite clusters compared to the _
temperature decrement from eq.([eq : selfsimtemppredfinal ] ) .
the _ planck _
profile is shown convolved with a @xmath42 fwhm gaussian ( green , solid ) .
we also show the 100ghz _ planck _ profiles presented in @xcite converted into thermodynamic temperature at the _ wmap _ band centre ( red , stripes ) .
the @xcite models ( black , solid ) are plotted with their associated dispersions ( black , dotted ) .
, height=755 ] we next show the stacked _ _ wmap__7 temperature profiles for 151 clusters listed in the _ planck _ esz catalogue .
we are using the ` photometric ' approach to background subtraction , with an annulus from @xmath146 being used in w ( and scaled according to beamwidth in q and v ) .
the final models are based on the statistical average of the clusters .
we see in fig .
[ fig : sztempcompplanck ] that the _ wmap _ data is an excellent fit to the _ planck _ expectation .
the fit between the _ planck _ data and the _ wmap _ data is not only good in all three [ w , v , q ] bands but at all angular scales .
we have further quantified the sz measurements using jack - knife and bootstrap techniques all of which support _
wmap_-_planck _ agreement , however we acknowledge these techniques do nt include covariance . in fig .
[ fig : sztempcompplanck ] we have shown the _ planck _ self - similar models convolved with the power - law beams from @xcite .
we find that in the case of the w band where the radio source profiles are most different from the jupiter beam , there is now disagreement with the _ wmap _ data with a deficit of @xmath5 20% in the centre . in the q and v bands where the radio source profiles are closer to the jupiter beam ,
the wider beams give virtually no change in the agreement with the _ wmap _ data .
we conclude that the _ planck _ sz profiles disagree with the @xcite _ _ wmap__5 radio source profile fits . however , the @xcite timing offset derived beam , which provides an excellent fit to the radio source profiles as shown in fig .
[ fig : w1_offset_beam ] , is significantly less wide than the @xcite beam . as shown in fig .
[ fig : sztempcompplanck ] we find that the the timing offset beam only marginally reduces the central sz temperature .
we therefore conclude that the _ wmap _ sz results are not at sufficient s / n to differentiate between the timing offset derived and the jupiter beams .
sz decrements for 151 _ planck _ sz clusters compared to the stacked _
temperature decrement from eq.([eq : selfsimtemppredfinal ] ) .
the _ planck _
profile is shown convolved with a _
jupiter beam , a beam fitted to the radio source profiles by @xcite and the @xcite timing offset derived beam .
, height=755 ] we have also looked at the _ planck _
model fits for the coma cluster and compared them to _
wmap_. part of the motivation here is that previous authors , @xcite and @xcite , have used coma in investigating the consistency of the _ wmap _ sz signal with x - ray predictions . in fig .
[ fig : mazzotta_coma_results ] we now show the _ planck _ self - similar model for coma ( solid blue line ) and see that it is substantially overestimated by the _ wmap _ data .
an analogous situation was found by @xcite in that the _ wmap _ coma v and w band profiles ( potentially affected by cmb contamination ) showed @xmath147 more sz signal than the optimal combined v and w profiles ( free of cmb contamination ) .
@xcite proposed that coma may sit on @xmath147 downwards cmb fluctuation .
our results are entirely consistent with this interpretation because the _ planck _
mmf method is essentially ` cmb - free ' whereas our _ wmap _ coma data may still be contaminated by cmb fluctuations . on this basis
we also show in fig .
( [ fig : mazzotta_coma_results ] ) a simple alteration to the _ planck _ coma sz self - similar model by including a gaussian with amplitude @xmath148 and @xmath149 to mimic the effect of a downwards cmb contribution centred on coma ( blue , dashed ) .
agreement with the _ wmap _ data is improved when a cmb contamination term is included .
we therefore conclude that the difference between the _ planck _ and _ wmap _ coma sz profiles is the result of cmb contamination .
sz decrements for the coma cluster compared to the _ planck _ temperature decrement from eq.([eq : selfsimtemppredfinal ] ) .
the error within each annulus for this individual cluster is simply the standard deviation of the esz clusters and is therefore only indicative .
the _ planck _
profile is shown convolved with a _
jupiter beam ( blue , solid ) .
we also show a model incorporating a @xmath150 downwards cmb fluctuation ( blue , dashed ) .
, height=755 ] in @xcite a _ wmap _ discrepancy with the sz / x - ray results of @xcite was presented .
this is of particular interest as the @xcite _ wmap _ sz discrepancy was largely associated with the inner profile .
the @xcite results complement this because they are weighted heavily to small scales because of the high resolution of their interferometric observations . in fig .
[ fig : scaledsztempcomp_bonamente ] we have therefore presented a stack of the 36 @xcite clusters that are unmasked in the _ wmap _ temperature maps .
we now find good agreement between the _ wmap _ sz decrements and the @xcite sz / x - ray expectation .
this is in contradiction to the results of @xcite .
we have found this is attributable to @xcite s assumption that the cluster is well resolved with respect to the _ wmap _ beam . as discussed in section
[ subsec : model_sz_convol ] this assumption introduces a systematic error into their 1-d convolution with the _ wmap _ beam profiles .
we therefore now report no evidence for a _ wmap _ sz discrepancy with respect to the @xcite x - ray models .
sz decrements for the 36 @xcite clusters that are unmasked in the _ wmap _ temperature maps .
this is compared to a stacked isothermal model based on the sz / x - ray parameters fitted by @xcite , convolved with the jupiter beam , a beam fitted to the radio source profiles by @xcite and the @xcite timing offset derived beam .
, height=755 ]
the main criticism that was made of the previous results of @xcite was that the wide _ wmap _ radio source profiles may be caused by eddington bias @xcite . essentially , low s / n sources detected in the _ wmap _ data may be contaminated by upwards cmb fluctuations and not balanced by downwards fluctuations .
this could explain the wider than expected profiles , particularly at faint fluxes
. there may be some evidence for eddington bias in the faintest _ _ wmap__5 w band source sub - sample that was initially used by @xcite .
however , it was because of this the faintest sources were not used in @xcite fits of the beam profile and that a flux limit of @xmath12jy has been used in calculating our radio source profiles .
we also note that the _ planck _ sources show the wider beam independent of whether the cmbsubtract flag applies .
we further note that we have restricted the _
planck _ sources to have a fwhm strictly less than the _ wmap _ w band beam profile width and find a wider than expected beam profile for these clearly point sources .
_ wmap _ sources selected from a ` cmb - free ' map and nvss selected sources at low frequency also show the same wider than expected beam . furthermore , we have also run monte carlo re - simulations of the source detection , producing artifical source catalogues extracted from simulated cmb maps . here , after applying the same cross - correlation technique as for the data , the _ wmap _ beam was recovered as input ( see fig . [
fig : sim ] ) , again arguing that these sources are little affected by eddington bias .
the _ planck _ data also confirms the non - linearity of _ wmap _ fluxes , particularly in the w band , in the range previously used .
decreasingly non - linear effects are also seen at q and v. but when ground - based and _ planck _
data for the bright @xcite sources are included in these comparisons the evidence for non - linearity becomes less and it could still be that the discrepancy corresponds more to a constant offset . given that the beam profile widening is smaller in the _
_ wmap__7 data than in _ _
wmap__5 , a scan pattern timing offset as discussed by @xcite becomes a more plausible explanation for this effect .
we have seen that the effect , originally invoked as an explanation for the alignment of the low order multipoles with the ecliptic , also provides a reasonable fit to the w band beam profiles ( see fig .
[ fig : w1_offset_beam ] ) . in our comparison of _ planck_-_wmap
_ sz decrements we have found good agreement .
similarly , our _ wmap _ sz profile results are now in agreement with the x - ray data for the @xcite sample .
this work is now in line with previous authors who when studying the integrated _ wmap _ sz signal @xmath151 ( as opposed to the compton y - parameter ) have found no evidence for a _ wmap _ discrepancy @xcite .
we have no explanation for the @xcite _ wmap _ sz profile discrepancies at this point .
we have also found our _ planck _ profiles are consistent with the _ planck _ results of @xcite and @xcite .
we interpret this as validating our our method of inverting the _ planck _ ercsc data to obtain _ planck
_ sz temperature profiles . in response to a question from a referee
, we note the integrated sachs - wolfe ( isw ) effect is at most a 10@xmath152 effect for clusters / superclusters , @xcite .
this is too marginal to affect the profiles we recover .
the isw is an even more negligible effect for radio sources , as observed by @xcite where it is shown to be @xmath5 0.3@xmath152 effect .
it is therefore highly unlikely to cause any bias to our results in either the sz or radio source analyses .
we have also compared our results to those of @xcite whose paper appeared while this one was being refereed .
we compare our results directly to theirs in fig .
[ fig : radio_schultzcompare ] .
wmap__7 w3 graph they use as an example is significantly wider than any profile shown by @xcite or here .
this is because they have used a sample with no cut at all in terms of significance of detection or flux and clearly these data will be strongly affected by eddington bias .
we repeat that at the flux limits used here , the simulations show no sign of such bias and so we are confident that this criticism does not apply to our results .
we note that there are additional quality cuts that @xcite have made with respect to our work , such as an expanded mask and a culling of close pairs .
however , we find that our results are unchanged when we apply them as well .
we find that their _
_ wmap__7-cmbfree and nvss beam profiles are very comparable to ours for the w3 band and they are wider than the jupiter profile as can be seen .
@xcite suggest that the _ _
wmap__7-cmbfree profiles are wider due to errors on the radio source positions
. however , their assumed positional errors may be overestimates for their stacked radio source profiles since the stacks are weighted towards the brighter radio sources which have more accurate positions . the fact that we are using 5ghz gb6 and pmn positions accurate to sub-@xmath153 accuracy in the _ wmap _ case and obtain _ _
wmap__7-cmbfree profiles consistent with @xcite suggests that positional errors can not be the explanation .
the main difference with the nvss results of @xcite is their larger errors .
our nvss sample is @xmath154x larger than theirs due to our 1.4ghz flux limit of 1jy compared to their 2jy limit , this ( as well as our larger binning ) explains most of the difference in errors .
otherwise the results appear entirely consistent .
result from debeaming with the jupiter beam ( blue , stars ) as compared to the result from debeaming with the timing offset derived beam from @xcite and also shown in fig .
[ fig : w1_offset_beam ] ( red , crosses ) . , width=283 ] [ fig : nl_cl ] we finally show in fig . [
fig : powerspectrafluxcorrectw ] the effect wide _ planck _ radio source profiles ( see fig . [
fig : radioproftempandbsplanck]f ) has on the _ wmap _
we take here the model with the 25.6ms timing offset that gave a reasonable fit to the data in fig .
[ fig : w1_offset_beam ] . otherwise without the model , we would need to extrapolate a fit out to large angles . then debeaming the raw w1/w2 @xmath155 from polspice @xcite via eq.(1,2 ) of @xcite
, we see that the @xmath156 shows a modest increase in amplitude at @xmath157400 , with a larger increase at @xmath158400 .
this reflects where the jupiter and timing offset beam are different . at @xmath157400
there is very little difference between the jupiter and the timing offset beam .
we note that this region is essentially unconstrained by the radio source profiles .
so the lack of change to the first acoustic peak is tied to the specific details of the timing offset model .
a different model could give significantly different results and therefore _ _
wmap__7 first acoustic peak s amplitude and position relies heavily on the accuracy of the observed jupiter beam beyond 30@xmath25 scales , which is untested by our observations .
we have investigated the beam profile of _ wmap _ by comparing beam profiles from radio sources with the jupiter beam profile .
we have compared sources from _ planck _ , _ wmap _ , _ wmap _ cmb - free and nvss catalogues .
we find that in all cases the radio sources show wider profiles than the jupiter beam with little indication of eddington bias or dependence on the method of normalisation .
applying our cross - correlation to realistic simulations strongly supports the accuracy of our beam profile measurements .
however , it must be said that in the _ _ wmap__7 data the w radio source profiles are less wide than previously found by @xcite in the _
_ wmap__5 release .
the rejection of the jupiter beam is now only @xmath1 in the _ planck _ radio source comparison .
but the rejection of the jupiter beam is reasonably consistent between the admittedly overlapping radio source samples from _ planck _ , _ wmap _ and nvss .
we have therefore considered explanations for the wide profiles assuming that they are not statistical fluctuations .
two such possibilities are a non - linearity in the _ wmap _ temperature scale and a timing offset in the _ wmap _ scan pattern as discussed by @xcite .
the narrower profiles measured here compared to the @xcite _ _ wmap__5 profiles increase the possibility of their being explained by a timing offset . we have also found discrepancies between _ wmap _ fluxes compared to _ planck _ and ground - based fluxes .
for @xmath159jy the _ wmap _ fluxes look to have a non - linear relation with _ planck _ fluxes .
however , when the further very bright sources discussed by @xcite with ground - based and _ planck _ measurements are included then this flux - flux discrepancy appears more like a linear than a non - linear offset .
we have compared stacked _ wmap _
sz decrements with those measured by _ planck _ and by ground - based observations .
in contrast to previous reports we now find _ wmap _ agrees with both the _ planck _ and ground - based data .
however this work is not at high enough s / n to distinquish between the timing offset beam of @xcite and the _ wmap _ jupiter beam .
we have shown that transforming the jupiter beam using a model that fits the radio source profiles results in small but significant changes to the _ wmap _ @xmath156 . at the least
, a wider beam would imply a much larger uncertainty in the normalisation and hence the estimate of @xmath160 from _ wmap_. unfortunately , faint radio sources can not check the _
wmap _ beam at scales larger than @xmath118 and a wider beam at these scales could , in principle , change the position , as well as the normalisation , of even the first acoustic peak .
clearly it is important to continue to test the calibration and beam profile of _ wmap _ , particularly in the w band .
jrw acknowledges financial support from stfc .
us acknowledges financial support from the royal thai government .
we acknowledge the use of data from nasa _ wmap _ and esa _ planck _ collaborations .
arnaud m. , pratt g. w. , piffaretti r. , bhringer h. , croston j. h. , pointecouteau e. , 2010 , a&a , 517 , a92 bailey , m. e. , & sparks , w. b. 1983 , , 204 , 53p bennett , c. l. , et al .
2003 , , 148 , 97 bennett c. l. , et al . , 2011 , apjs , 192 , 17 bielby r. m. , shanks t. , 2007 , mnras , 382 , 1196 blake c. , wall j. , 2002 , mnras , 337 , 993 bonamente m. , joy m. k. , laroque s. j. , carlstrom j. e. , reese e. d. , dawson k. s. , 2006 , apj , 647 , 25 cavaliere a. , fusco - femiano r. , 1976 , a&a , 49 , 137 chen x. , wright e. l. , 2009 , apj , 694 , 222 condon , j. j. , cotton , w. d. , greisen , e. w. , et al .
1998 , , 115 , 1693 cotton , w. d. , mason , b. s. , dicker , s. r. , et al .
2009 , , 701 , 1872 eddington a. s. , 1913 , mnras , 73 , 359 gold b. , et al . , 2011 , apjs , 192 , 15 granett b. r. , neyrinck m. c. , szapudi i. , 2008 , apj , 683 , l99 gregory p. c. , scott w. k. , douglas k. , condon j. j. , 1996 , apjs , 103 , 427 griffith m. r. , wright a. e. , 1993 , aj , 105 , 1666 hill r. s. , et al . , 2009 , apjs , 180 , 246 hinshaw , g. , spergel , d. n. , verde , l. , et al .
2003 , , 148 , 135 jarosik n. , et al . , 2003 , apjs , 145 , 413 jarosik n. , et al . , 2011 , apjs , 192 , 14 kenney , j. d. , & dent , w. a. 1985 , , 298 , 644 komatsu e. , et al . , 2011 ,
apjs , 192 , 18 lieu r. , mittaz j. p. d. , zhang s .- n . ,
2006 , apj , 648 , 176 limon m. , et al . , 2008 , wilkinson microwave anisotropy probe ( _ wmap _ ) : five year explanatory supplement , ` http://lambda.gsfc.nasa.gov/data/map/doc/ ` ` map_supplement.pdf ` liu h. , li t .- p .
, 2011 , apj , 732 , 125 liszt , h. , & lucas , r. 1999 , , 347 , 258 lonsdale , c. j. , doeleman , s. s. , & phillips , r. b. 1998 , , 116 , 8 malik , sawangwit u. , shanks t. & whitbourn , j.r . , 2011 , in preparation melin j .- b .
, bartlett j. g. , delabrouille j. , arnaud m. , piffaretti r. , pratt g. w. , 2011 , a&a , 525 , a139 melin j .- b . , bartlett j. g. , delabrouille j. , 2006 , a&a , 459 , 341 moss a. , scott d. , sigurdson k. , 2011 , jcap , 1 , 1 mroczkowski t. , et al . , 2009 , apj , 694 , 1034 myers a. d. , shanks t. , outram p. j. , frith w. j. , wolfendale a. w. , 2004 , mnras , 347 , l67 nagai d. , vikhlinin a. , kravtsov a. v. , 2007 , apj , 655 , 98 nolta m. r. , et al . , 2004 , apj , 608 , 10 page l. , et al . , 2003 , apjs , 148 , 39 page l. , et al . , 2003 , apj , 585 , 566 piffaretti r. , jetzer p. , kaastra j. s. , tamura t. , 2005 , a&a , 433 , 101 _ planck _ collaboration et al , 2011d , the all - sky early sunyaev - zeldovich cluster sample , arxiv , arxiv:1101.2024 _ planck _ collaboration et al , 2011f , statistical analysis of sunyaev - zeldovich scaling relations for x - ray galaxy clusters , arxiv , arxiv:1101.2043 _ planck _ collaboration et al , 2011 g , calibration of the local galaxy cluster sunyaev - zeldovich scaling relations , arxiv , arxiv:1101.2026 _ planck _ collaboration et al , 2011exp , explanatory supplement , ` http://www.sciops.esa.int/sa/planck/docs/ ` ` ercsc_explanatory_supplement.zip ` _ planck _ collaboration et al , 2012v , pressure profiles of galaxy clusters from the sunyaev - zeldovich effect , arxiv , arxiv:1207.4061 _ planck _ collaboration et al , 2012x , physics of the hot gas in the coma cluster , arxiv , arxiv:1208.3611 pratt g. w. , bhringer h. , croston j. h. , arnaud m. , borgani s. , finoguenov a. , temple r. f. , 2007 , a&a , 461 , 71 press w. h. , teukolsky s. a. , vetterling w. t. , flannery b. p. , 1992
, in ` numerical recipes ' , cup : cambridge , pp . 660 refregier a. , spergel d. n. , herbig t. , 2000 , apj , 531 , 31 roukema b. f. , 2010 , arxiv , arxiv:1007.5307 salter , c. j. , reynolds , s. p. , hogg , d. e. , payne , j. m. , & rhodes , p. j. 1989 , , 338 , 171 sawangwit u. , shanks t. , 2010a , mnras , 407 , l16 sawangwit u. , shanks t. , 2010b , in ` 45th rencontres de moriond : cosmology 2010 ' , eds auge , e. , dumarchez , j. & tran thanh van , j. , pp .
53 - 57 , gioi : vietnam , ( arxiv:1006.1270 ) .
sawangwit u. , shanks t. , cannon r. d. , croom s. m. , ross n. p. , wake d. a. , 2010 , mnras , 402 , 2228 sawangwit u. , 2011 , phd thesis .
schultz , k. w. , & huffenberger , k. m. 2011 , arxiv:1111.7311 shanks t. , 1985 , vistas in astr . , 28 , 595 sunyaev r. a. , zeldovich i. b. , 1980 , ara&a , 18 , 537 szapudi i. , prunet s. , colombi s. , 2001 , apj , 561 , l11 tegmark m. , de oliveira - costa a. , 1998 , , 500 , l83 weiland j. l. , et al . , 2011 , apjs , 192 , 19 wright e. l. , et al . , 2009 , apjs , 180 , 283 wright , m. , & birkinshaw , m. 1984 , , 281 , 135 wright , m. c. h. , & sault , r. j. 1993 , , 402 , 546
in the self - similar sz model as employed in the _ planck _ esz , the fundamental parameters of a cluster are @xmath161 , @xmath162 and @xmath138 . using the terminology of @xcite , which can be used as a characteristic sz parameter instead of @xmath161 . in eq.([eq : y500def ] ) the units of @xmath139 are mpc@xmath165 , but are easily convertible to the arcmin@xmath165 units used in the esz and throughout this paper = \frac{1}{60^{2 } } ( \frac{\pi}{180})^{2}(d_{a}[mpc])^{2 } y[arcmin^{2}]$ ] . ] .
this @xmath139 is a distinct quantity from @xmath166 as found by evaluating eq .
( [ eq : rawcylprofile ] ) .
the introduction of @xmath139 is well motivated because , as shown by @xcite , it allows a scale - free description of eq.([eq : rawcylprofile ] ) s @xmath167 and @xmath168 in terms of @xmath169 as follows , finally , we can use the above to calculate @xmath175 and the compton _
parameter , where @xmath176 .
we now describe three methods for doing so . + * 1 . using @xmath137 as an amplitude
* : since @xmath177 , eq.([eq : finalyscale ] ) can be expressed as , @xmath178 this is the method we adopt in this paper , it is dependent on both @xmath137 and @xmath179 . *
2 . using @xmath139 as an amplitude * : @xmath139 can be calculated using @xmath162 and @xmath161 .
we can therefore directly evaluate @xmath175 using eq.([eq : finalyscale ] ) .
this method is independent of the _ planck _ provided @xmath137 and instead solely uses @xmath179 .
* 3 . avoiding the integrated sz signal *
: the compton y - parameter can be expressed as @xcite , @xmath180 we can therefore directly evaluate the compton y - parameter if a self - similar cluster profile is assumed for @xmath181 .
this method is independent of the _ planck _ provided @xmath137 and instead solely uses @xmath179 . | the prime evidence underpinning the standard @xmath0cdm cosmological model is the cmb power spectrum as observed by _
wmap _ and other microwave experiments .
but @xcite have recently shown that the _ wmap _ cmb power spectrum is highly sensitive to the beam profile of the _ wmap _ telescope . here
, we use the source catalogue from the _ planck _ early data release to test further the _ wmap _ beam profiles .
we confirm that stacked beam profiles at q , v and particularly at w appear wider than expected when compared to the jupiter beam , normalised either directly to the radio source profiles or using _ planck _ fluxes .
the same result is also found based on _ wmap_-cmbfree source catalogues and nvss sources .
the accuracy of our beam profile measurements is supported by analysis of cmb sky simulations .
however the beam profiles from _ _ wmap__7 at the w band are narrower than previously found in _ _
wmap__5 data and the rejection of the _ wmap _ beam is now only at the @xmath1 level . we also find that the _ wmap _ source fluxes demonstrate possible non - linearity with _
planck _ fluxes . but
including ground - based and _ planck
_ data for the bright @xcite sources may suggest that the discrepancy is a linear offset rather than a non - linearity . additionally , we find that the stacked sunyaev - zeldovich ( sz ) decrements of @xmath2 galaxy clusters observed by _ planck _ are in agreement with the _ wmap _ data .
we find that there is no evidence for a _ wmap _ sz deficit as has previously been reported . in the particular case of coma
we find evidence for the presence of an @xmath3 downwards cmb fluctuation .
we conclude that beam profile systematics can have significant effects on both the amplitude and position of the acoustic peaks , with potentially important implications for cosmology parameter fitting .
[ firstpage ] cosmology : cosmic microwave background - large - scale structure of universe |
the extensive applications of synchrotron radiation in a wide variety of experiments and in many disciplines ( see @xcite-@xcite and references therein ) , motivate the importance of investigations for various mechanisms of controlling the radiation parameters . in particular , it is of interest to consider the influence of a medium on the spectral and angular characteristics of the radiation .
it is well known that the presence of medium can essentially change the characteristics of the electromagnetic processes and gives rise to new types of phenomena such as cherenkov , transition , and diffraction radiations .
moreover , the operation of a number of devices assigned to production of electromagnetic radiation is based on the interaction of high - energy particles with materials ( see , for example , @xcite ) .
the synchrotron radiation from a charged particle circulating in a homogeneous medium was considered in @xcite , where it has been shown that the interference between the synchrotron and cherenkov radiations leads to remarkable effects .
new interesting features arise in the case of inhomogeneous media .
interfaces of media are widely used to control the radiation flow emitted by various systems .
well - known examples of this kind are the cherenkov radiation of a charge moving parallel to a plane interface of two media or flying parallel to the axis of a dielectric cylinder , transition radiation , smith - purcell radiation . in @xcite
the radiation from charged particles flying over a surface acoustic wave generated on a plane interface between two media is studied . in a series of papers initiated in grig95a , grig95b , grigoryan1995 ,
we have investigated the influence of spherical and cylindrical boundaries between two dielectrics on the characteristics of the synchrotron radiation .
in particular , the investigation of radiation from a charge rotating about / inside a dielectric ball showed @xcite that when the cherenkov condition for the ball material and particle speed is satisfied , there appear high narrow peaks in the spectral distribution of the number of quanta emitted to outer space at some specific values of the ratio of ball - to - particle orbit radii . in the vicinity of these peaks
the rotating particle may generate radiation field quanta exceeding in several dozens of times those generated by particle rotating in a continuous , infinite and transparent medium having the same real part of permittivity as the ball material .
similar problems with the cylindrical symmetry have been discussed in grigoryan1995,kot2000 . it is shown that under similar cherenkov condition for permittivity of cylinder and the speed of particle gyrating about / inside the cylinder , high narrow peaks are present in the spectral - angular distribution for the number of radiated quanta . in the vicinity of these peaks
the radiated energy exceeds the corresponding value for homogeneous medium case by several orders of magnitude . in recent papers @xcite-@xcite ,
we have investigated more general problem of the radiation from a charged particle in a helical trajectory inside a dielectric cylinder . the corresponding problem for a charge moving in vacuum has been widely discussed in the literature ( see , for instance , @xcite and references given therein ) and the influence of a homogeneous dispersive medium is considered in @xcite .
this type of electron motion is used in helical undulators for generating electromagnetic radiation in a narrow spectral interval at frequencies ranging from radio or millimeter waves to x - rays ( see bord99,hofm04,onuk03 ) .
the present paper is devoted to the investigation of the radiation from a charge moving along a helix around a dielectric cylinder immersed in a homogeneous medium .
the paper is organized as follows . in the next section , by using the formulae for the green function from @xcite , we derive expressions for the vector potential and electromagnetic fields in the region outside the cylinder .
the angular - frequency distribution of the radiation intensity in the surrounding medium is considered in section sec : radiation . in section [ sec : properties ] we discuss the features of the radiation intensity and the results of the numerical evaluations are presented .
section [ sec : conc ] concludes the main results of the paper .
consider a point charge @xmath0 moving along the helical trajectory of radius @xmath1 outside a dielectric cylinder with radius @xmath2 and with permittivity @xmath3 .
we will assume that this system is immersed in a homogeneous medium with dielectric permittivity @xmath4 ( magnetic permeability will be taken to be unit ) .
this type of motion can be produced by a uniform constant magnetic field directed along the axis of a cylinder , by a circularly polarized plane wave , or by a spatially periodic transverse magnetic field of constant absolute value and a direction that rotates as a function of the longitudinal coordinate . in helical undulators
the last configuration is used . in the cylindrical coordinate system ( @xmath5 ) with the axis @xmath6 directed along the cylinder axis , the components of the current density created by the charge are given by the formula @xmath7where @xmath8 is the angular velocity of the charge , @xmath9 and @xmath10 are the particle velocities along the axis of the cylinder and in the perpendicular plane , respectively . the solution to maxwell equations for the vector potential is expressed in terms of the green function @xmath11 of the electromagnetic field by the formula @xmath12where the summation over @xmath13 is understood . in accordance with the problem symmetry , for the green function we have the following fourier expansion : @xmath14 .
\label{gf_furieexp}\end{aligned}\]]substituting expressions ( [ hosqxtutjun ] ) and ( [ gf_furieexp ] ) into formula ( [ vecpot ] ) , we present the vector potential as the fourier expansion @xmath15where the coefficients @xmath16 are given in terms of the fourier components of the green function by the formula @xmath17 .
\label{vecpot1new}\]]here and in the discussion below we use the notation @xmath18 by making use of the formula for the green function given earlier in grigoryan1995 , in the lorentz gauge for the corresponding fourier components @xmath19 in the region outside the cylinder , @xmath20 , we find @xmath21 h_{m}(\lambda _ { 1}\rho ) , \notag\end{aligned}\]]where in the first and second equations @xmath22 , @xmath23 , @xmath24 , @xmath25@xmath26 is the bessel function , @xmath27 is the hankel function of the first kind , and @xmath28 in formulae ( [ gfcompd ] ) the coefficients @xmath29 , @xmath30 , are defined by the expressions @xmath31
\notag \\ & & + \frac{p\lambda _ { 0}j_{m+p}(\lambda _ { 0}\rho _ { 1})j_{m}(\lambda _ { 0}\rho _ { 1})}{2\rho _ { 1}\alpha _ { m}v_{m+p}^{h}}\sum_{l=\pm 1}\frac{h_{m+l}(\lambda _ { 1}\rho _ { 0})}{v_{m+l}^{h } } , \label{bm+p}\end{aligned}\]]with the notation @xmath32note that the equation @xmath33 determines the eigenmodes of the dielectric cylinder . in the definition of @xmath34 given by ( [ lambdaj ] )
, one should take into account that in the presence of the imaginary part @xmath35 for the dielectric permittivity ( @xmath36 ) the radiation field in the exterior medium should exponentially damp for large @xmath37 .
this leads to the following relations : @xmath38for @xmath39 the sign of @xmath34 may also be determined from the principle of radiation ( different signs of @xmath40 and @xmath41 in the expressions of the fields ) for large @xmath37 . by taking into account formulae ( [ gfcompd ] ) , from ( [ vecpot1new ] ) we obtain the expressions for the fourier components of the vector potential : @xmath42 h_{m}(\lambda _ { 1}\rho ) , \label{vecpot3c}\end{aligned}\]]with the coefficients @xmath43 the fourier expansions similar to ( [ vecpot2 ] )
may also be written for the electric and magnetic fields . in the case of the magnetic field
the corresponding fourier coefficients have the form @xmath44with the notation @xmath45 , \quad p=\pm 1 .
\label{dm}\]]the corresponding fourier coefficients for the electric field are obtained from the maxwell equation @xmath46 with the result : @xmath47where , as before , @xmath22 .
note that for purely transversal motion ( @xmath48 ) one has @xmath49 . for a particle motion in a homogeneous medium we have @xmath50 and the terms in the expressions of the coefficients @xmath51 and @xmath29 involving
the function @xmath52 vanish . in this case
@xmath53 , @xmath54 , and for the coefficients @xmath55 one finds @xmath56 , \quad \varepsilon _ { 0}=\varepsilon _ { 1}. \label{dmpham}\ ] ] the fourier coefficients for the fields determined by relations ( magneticc ) , ( [ electricc ] ) have poles corresponding to the zeros of the function @xmath52 appearing in the denominators of ( [ bm+p ] ) and ( [ cm+p ] ) .
it can be seen that this function has zeros only under the conditions @xmath57 . in particular , from here , as a necessary condition , we should have @xmath58 . for
the corresponding modes the fourier coefficients of the fields depend on the radial coordinate through the macdonald function @xmath59 with @xmath60 , and they are exponentially damped with the distance from the cylinder surface .
these modes are the eigenmodes of the dielectric cylinder and they correspond to the waves propagating inside the cylinder .
below , in the consideration of the intensity for the radiation to the exterior medium , we will neglect the contribution of the poles corresponding to these modes .
having the electromagnetic fields we can investigate the intensity of the radiation propagating in the exterior medium . as we have mentioned before , for @xmath61 the corresponding fourier components are exponentially damped for large values @xmath37 , and the radiation is present only under the condition @xmath39 .
the average energy flux per unit time through the cylindrical surface of radius @xmath37 , coaxial with the dielectric cylinder , is given by the poynting vector : @xmath62 \cdot \mathbf{n}_{\rho } \rho dz , \label{poyntingvector}\]]being @xmath63 the period for the transverse motion of the charge .
substituting the corresponding fourier expansions of the fields and using the asymptotic expressions of the hankel functions , at large distances from the cylinder for the radiation intensity we find @xmath64 , \label{int10}\]]where the prime over the sum means that the term with @xmath65 should be taken with the weight @xmath66 and the coefficients @xmath55 are defined by formulae ( [ dm ] ) .
first let us consider the special case @xmath67 for a fixed @xmath68 , which corresponds to a charge moving with constant velocity @xmath69 on a straight line @xmath70 parallel to the cylinder axis .
for this case , from ( [ omegam ] ) , @xmath71 and expressions ( [ lambdaj ] ) take the form @xmath72 with the notation@xmath73from the condition @xmath39 it follows that the radiation in the exterior medium is present only under the cherenkov condition for the particle velocity , @xmath9 , and dielectric permittivity @xmath74 for the surrounding medium : @xmath75 . introducing the angle @xmath76 of the wave vector with the cylinder axis , from the relation
@xmath77 it follows that @xmath78 , and the radiation propagates along the cherenkov cone of the external medium .
since for the case under consideration one has @xmath79 , the first term on the right of formula ( [ cm+p ] ) vanishes and the expression for the coefficients @xmath55 is written in the form@xmath80 , \label{dmpm0}\end{aligned}\]]where @xmath81the substitutions @xmath82 should also be made in the formulae for @xmath83 , @xmath84 , and @xmath52 .
changing the integration variable , from ( [ int10 ] ) for the corresponding radiation intensity we find @xmath85 , \label{iom0}\ ] ] where in formula ( [ dmpm0 ] ) for the coefficients @xmath55 the expressions ( [ lambj ] ) should be used with @xmath77 . in the limit @xmath86 the only contribution to the radiation intensity comes from the first term on the right - hand side of formula ( [ dmpm0 ] ) and , after the summation over @xmath87 by using the formula @xmath88 , we obtain the standard expression for the intensity of the cherenkov radiation in a homogeneous medium . for small values of the cylinder radius ,
the part in the radiation intensity induced by the presence of the cylinder vanishes as @xmath89 for the harmonics @xmath90 and as @xmath91 for @xmath65 .
now we return to the general case of the particle motion along the helix .
first we consider the contribution of the mode with @xmath65 to the radiation intensity given by ( [ int10 ] ) .
for this mode one has @xmath71 and , as in the previous case , from the condition @xmath39 it follows that in the exterior region , the corresponding radiation is present only under the condition @xmath92 .
this radiation propagates along the cherenkov cone @xmath93 of the external medium . by using the expressions for the coefficients @xmath55 and introducing a new integration variable @xmath94 , for the radiation intensity at @xmath65 one obtains@xmath95 , \label{intm0}\]]with the notations@xmath96 in formulae ( [ ulom ] ) ,
@xmath97 are defined by relations ( lambj ) with @xmath77 and @xmath98 in the limit @xmath86 , by using the formulae for the bessel functions for small values of the argument , from formula ( [ intm0 ] ) we find @xmath99where @xmath100 is the neumann function , and@xmath101 , \label{intm0ham}\]]is the radiation intensity at @xmath65 in the case of the charge motion in a homogeneous medium with @xmath102 . for the radiation at @xmath103 harmonics , from the condition @xmath104 one obtains the quadratic inequality with respect to @xmath105 : @xmath106it is convenient to write the solution to this inequality in terms of a new variable @xmath76 , @xmath107 , defined in accordance with the relation @xmath108now the function @xmath109 is presented in the form @xmath110and the quantities @xmath105 and @xmath109 are connected by the relation @xmath111 . in terms of the new variable @xmath76 , at large distances from the charge trajectory the dependence of elementary waves on the space - time coordinates has the form@xmath112 , \label{timedep}\]]which describes wave with the frequency @xmath113propagating at the angle @xmath76 to the @xmath6-axis .
formula ( omegamtet ) describes the normal doppler effect in the cases @xmath114 and @xmath75 , @xmath115 , and anomalous doppler effect inside the cherenkov cone , @xmath116 , in the case @xmath75 .
changing the integration over @xmath105 in ( [ int10 ] ) to the integration over @xmath76 in accordance with ( [ kazet ] ) , the radiation intensity at @xmath103 harmonics is presented in the form @xmath117where @xmath118 is the solid angle element , and @xmath119 , \label{int3}\]]is the average power radiated by the charge at a given harmonic @xmath87 into a unit solid angle . in formulae ( [ dm ] ) for the coefficients @xmath120 , the quantities @xmath121 and @xmath34 are expressed in terms of @xmath76 as @xmath122 hence , under the cherenkov condition , @xmath75 , the total radiation intensity is presented in the form @xmath123where the first term on the right - hand side is given by formula ( [ intm0 ] ) and describes the radiation with a continuous spectrum propagating along the cherenkov cone @xmath124 of the external medium .
the second term describes the radiation , which , for a given angle @xmath76 , has discrete spectrum determined by formula ( [ omegamtet ] ) .
if the cherenkov condition is not satisfied ( @xmath125 ) the first term is absent and only the modes with @xmath103 contribute to the radiation intensity . for a charge moving in a homogeneous medium with dielectric permittivity @xmath74 , one
has @xmath102 , and using formula ( [ dmpham ] ) for the coefficients @xmath55 , from formula ( int3 ) we obtain @xmath126 .
\label{intham}\]]in the case @xmath127 this formula is derived in ref .
soko68 ( see also refs .
@xcite ) . for a charge with a purely transversal motion ( @xmath128 ) one has @xmath129 , and from ( [ int3 ] ) as a special case we obtain the formula derived in @xcite
in this section , on the base of the formulae given before , we discuss the characteristic features of the radiation intensity . for a non - relativistic motion , @xmath130 , from the general formula ( [ int3 ] ) we find @xmath131 ^{2}(1+\cos ^{2}\vartheta ) \sin ^{2(m-1)}\vartheta , \label{intnonrel}\]]and the contribution of the harmonics with @xmath132 is small compared to that in the fundamental one , @xmath133 .
note that the part of the radiation intensity with the first term in square brackets of ( [ intnonrel ] ) corresponds to the case of the motion in homogeneous medium with permittivity @xmath134 , and the part with the second term is induced by the presence of the cylinder with permittivity @xmath3 . in the limit @xmath86 , in the general formula ( [ int3 ] ) we use the asymptotic expressions for the bessel functions for small values of the argument .
after long calculations it can be seen that the difference between the radiation intensities in the cases when the cylinder is present and absent , @xmath135 , vanishes as @xmath136 for @xmath90 . in the same limit and for the radiation corresponding to @xmath65 , the part induced by the cylinder
is given in formula ( [ intm0lim ] ) and vanishes like @xmath91 .
now let us consider the behavior of the radiation intensity near the cherenkov angle when @xmath137 . using the asymptotic formulae for the cylinder functions for large values of the argument it can be seen that in this limit @xmath138note that for the radiation intensity in a homogeneous medium with dielectric permittivity @xmath4 we have the same behavior . in accordance with ( [ omegamtet ] ) , near the cherenkov cone the frequencies of the radiated photons are large and the dispersion of the dielectric permittivity @xmath4
should be taken into account .
the corresponding angles are determined implicitly from the condition @xmath139 , by using formula ( [ omegamtet ] ) and frequency dependence of the permittivity @xmath140 . here
@xmath141 is the characteristic frequency of the dispersion .
note that for the charge helical motion inside the dielectric cylinder ( @xmath142 ) the behavior of the radiation intensity near the cherenkov cone is radically different for the cases @xmath143 and @xmath144 ( see @xcite ) . in the first case the intensity behaves like
@xmath145 $ ] , with @xmath146 given by ( [ omegamtet ] ) . by using debye s asymptotic expansions for the bessel and neumann functions , in @xcite , in the case of helical motion inside a dielectric cylinder
, it has been shown that under the condition @xmath147 , at points where the real part of the function @xmath52 , given by formula ( [ bet1 ] ) , is equal to zero , the contribution of the imaginary part of this function into the coefficients @xmath55 can be exponentially large for large values @xmath87 . this leads to the appearance of strong narrow peaks in the angular distribution for the radiation intensity at a given harmonic @xmath87 .
the condition for the real part of the function @xmath148 to be zero has the form:@xmath149 ^{-1}=\frac{2\varepsilon _
{ 0}}{% \varepsilon _ { 1}-\varepsilon _ { 0 } } , \label{peakscondn}\]]where @xmath150 is the neumann function .
this equation is obtained from the equation determining the eigenmodes for the dielectric cylinder by the replacement @xmath151 .
equation ( [ peakscondn ] ) has no solutions for @xmath152 , which is possible only for @xmath153 .
hence , the above mentioned possibility for the appearance of peaks is not realized for the case @xmath152 which , in accordance with ( [ lambondb ] ) , corresponds to the angular region @xmath154 . under the condition ( [ peakscondn ] ) for the coefficient @xmath52
one has@xmath155where @xmath156 is defined by formula ( [ wronsk ] ) with @xmath157 . for large values @xmath87 , from ( [ alphapeaks ] )
we have the estimate @xmath158 $ ] , with@xmath159and this coefficient is exponentially small .
now , by using the asymptotic formulae for the bessel functions , we can see that under the conditions@xmath160for the angles being the solutions of the equation ( [ peakscondn ] ) , one has @xmath161 $ ] and , hence , for the radiation intensity@xmath162 .
\label{npeak}\ ] ] note the when the charge moves inside the cylinder ( @xmath142 ) , for the appearance of the peaks we have two possibilities @xcite . in the first case , corresponding to @xmath163 , at the peaks the radiation intensity behaves as @xmath164 $ ] . by taking into account that the function @xmath165 is monotonically decreasing and comparing this estimate with ( [ npeak ] ) ,
we conclude that in this case the peaks in the radiation intensity for the motion inside the cylinder are stronger .
the second case corresponds to @xmath166 , and the radiation intensity at the peaks behaves like @xmath167\}$ ] . from ( [ peakscondn2 ] ) , by taking into account formulae ( [ lambondb ] ) , as necessary conditions for the presence of the strong narrow peaks in the angular distribution for the radiation intensity one has @xmath168these conditions can be satisfied only if we have @xmath169where @xmath170 is the velocity of the charge image on the cylinder surface .
the second condition in ( [ peakscond4 ] ) is the cherenkov condition for the velocity of the charge image on the cylinder surface and dielectric permittivity of the cylinder . from ( [ peakscond3 ] )
it follows that when the cherenkov condition for the velocity of the charge and dielectric permittivity of the surrounding medium is not satisfied , @xmath171 , the possible strong peaks are located in the angular region defined by the inequality @xmath172with @xmath173 . if the cherenkov condition @xmath174 is satisfied , in addition to this inequality we also need to have the condition @xmath175 in order to estimate the angular widths for the peaks , we note that near them the main contribution to the radiation intensity comes from the terms in the coefficients @xmath176 containing in the denominators the function @xmath52 . expanding this function near the angle corresponding to the peak , @xmath177 , we see that the angular dependence of the radiation intensity near the peak has the form @xmath178where @xmath179 $ ] . hence , the angular widths of the peaks are of the order @xmath180 $ ] . from estimate ( [ npeak ] )
it follows that at the peaks the angular density of the radiation intensity exponentially increases with increasing @xmath87 .
however , one has to take into account that in realistic situations the growth of the radiation intensity is limited by several factors . in particular ,
the factor which limits the increase , is the imaginary part for the dielectric permittivity @xmath181 , @xmath182 .
this leads to additional terms in the denominator of formula ( [ dimnear ] ) , proportional to the ratio @xmath183 , where @xmath184 is the real part of the dielectric permittivity . as a result for @xmath185 the intensity and the width of the peak
is determined by these terms and the saturation of the radiation intensity at the peak takes place . by using the general formula ( [ int3 ] ) ,
we have carried out numerical calculations of the radiation intensity at a given harmonic @xmath87 for various values of the parameters . as an example , in figure [ fig1 ] we have plotted the dependence of the angular density for the number of the radiated quanta @xmath186as a function of the angle @xmath107 for @xmath187 , @xmath188 , @xmath189 .
the full and dashed curves correspond to the cases @xmath190 and @xmath191 ( homogeneous medium ) respectively .
the left panel is plotted for the longitudinal component @xmath192 . in order to compare with the radiation intensity in the case of purely transversal motion , we have plotted on the right panel the corresponding graphs for @xmath193 . in this case the curves are symmetric with respect to the rotation plane @xmath194 . in both cases strong narrow peaks
appear when the dielectric cylinder is present . at the peaks of the right panel one has @xmath195 , @xmath196 , with the width of the peak @xmath197 , and @xmath198 , @xmath199 , @xmath200 .
for the left peak on the left panel one has @xmath201 , @xmath202 , @xmath203 ; @xmath204 , @xmath205 , @xmath206 ; @xmath207 , @xmath208 , @xmath209 .
these numerical data are in good agreement with the analytic estimates given before .
from presented graphs it is seen that , for angles away the peaks the drift essentially amplifies the radiation intensity . [ cols="^,^ " , ]
the unique characteristics of synchrotron radiation , such as high intensity , high collimation , and the wide spectral range , have resulted in its extensive applications . in the present paper , continuing our previous work on the influence of a medium on the parameters of the synchrotron radiation , we have investigated the properties of the radiation from a charged particle moving along a helical orbit around a dielectric cylinder . in order to evaluate the corresponding vector potential and the electromagnetic fields in the exterior medium ,
we have employed the green function , previously investigated in @xcite .
these fields are presented in the form of the fourier expansion ( [ vecpot2 ] ) with the fourier coefficients given by formulae ( [ vecpot3c ] ) , ( [ magneticc ] ) , ( [ electricc ] ) . on the base of these formulae , in section [ sec : radiation ] , we have investigated the spectral - angular distribution of the radiation propagating in the exterior medium . in the case
when the cherenkov condition for dielectric permittivity of the exterior medium and drift velocity of the charge is satisfied , @xmath75 , the radiation intensity is decomposed into two terms .
the first one , given by formula ( [ intm0 ] ) , corresponds to the radiation with continuous spectrum propagating along the cherenkov cone of the external medium .
the second term in the expression for the total radiation intensity , given by formula ( [ int3a ] ) , presents the contribution of the harmonics with @xmath90 .
it describes the radiation , which for a given propagation direction characterized by the angle @xmath76 , has a discrete spectrum determined by formula ( omegamtet ) .
this formula describes the normal doppler effect in the angular region outside the cherenkov cone , @xmath210 , and anomalous doppler effect in the region inside the cone . in the case
@xmath125 , the first term in ( [ totintom ] ) is absent and the angular distribution of the radiation intensity at a given harmonic @xmath87 is given by formula ( [ int3 ] ) . in section [ sec : properties ] we have investigated characteristic features of the radiation . in the non - relativistic limit the general formula for the radiation intensity reduces to ( [ intnonrel ] ) and the main part of the radiation energy
is emitted on the fundamental harmonic @xmath133 . for a fixed radius of the charge orbit and for small values of the cylinder radius , for a given harmonic
the effects induced by the presence of the cylinder vanish as @xmath211 . as in the case of the particle motion inside the dielectric cylinder , under certain conditions on the parameters strong narrow peaks
appear in the angular distribution of the radiation intensity .
these peaks correspond to the values of the angle @xmath76 for which the real part of the function @xmath52 from ( [ bet1 ] ) vanishes .
the corresponding condition ( peakscondn ) is obtained from the equation determining the eigenmodes for the dielectric cylinder replacing the hankel functions by the neumann ones . at the zeros of the real part of @xmath52
the imaginary part of this function is exponentially small for large values @xmath87 and this leads to strong angular peaks in the radiation intensity . by using asymptotic formulae for the cylindrical functions
, we have specified the conditions for the appearance of the peaks and analytically estimated their heights and widths .
in particular , we have shown that the peaks are present only when dielectric permittivity of the cylinder is greater than the permittivity for the surrounding medium and the cherenkov condition is satisfied for the velocity of charge image on the cylinder surface and the dielectric permittivity of the cylinder .
these features are well confirmed by the results of the numerical calculations .
these results show that the presence of the cylinder provides a possibility for an essential enhancement of the radiated power as compared to the radiation in a homogeneous medium .
the authors are grateful to professor l.sh .
grigoryan , s.r .
arzumanyan , h.f .
khachatryan for stimulating discussions .
a.a.s . acknowledges the hospitality of the federal university of paraiba ( joo pessoa , brazil ) . the work has been supported by grant no .
0077 from ministry of education and science of the republic of armenia .
tsytovich , westnik mgu * 11 * , 27 ( 1951 , in russian ) ; k. kitao , progr .
* 23 * , 759 ( 1960 ) ; t. erber , d. white , and h.g .
latal , acta phys .
austriaca * 45 * , 29 ( 1976 ) ; j. schwinger , w - y .
tsai , and t. erber , ann .
phys . * 96 * , 303 ( 1976 ) ; t. erber , d. white , w - y .
tsai , and h.g .
latal , ann . phys . * 102 * , 405 ( 1976 ) ; t.m .
rynne , g.b .
baumgartner , and t. erber , j. appl .
49 * , 2233 ( 1978 ) .
mkrtchyan , l.sh .
grigoryan , a.a .
saharian , and a. n. didenko , acustica * 75 * , 1984 ( 1991 ) ; a.a .
saharian , a.r .
mkrtchyan , l.a .
gevorgian , l.sh .
grigoryan , and b.v .
khachatryan , nucl .
instr . and meth .
b * 173 * , 211 ( 2001 ) .
grigoryan , h.f .
khachatryan , and s.r .
arzumanyan , izv .
, fiz . * 33 * , 267 ( 1998 ) ( engl .
transl . : j. contemp . phys . ) ; l.sh .
grigoryan , h.f .
khachatryan , and s.r .
arzumanyan , izv . nats . akad .
, fiz . * 37 * , 327 ( 2002 ) ( engl .
transl . : j. contemp . phys . ) ; l.sh .
grigoryan , h.f .
khachatryan , s.r .
arzumanyan , and m.l .
grigoryan , nucl .
instr . and meth .
b * 252 * , 50 ( 2006 ) ; s.r .
arzumanyan , l.sh .
grigoryan , h.f .
khachatryan , and m.l .
grigoryan , nucl .
instr . and meth .
b , in press . a.s .
kotanjyan , h.f .
khachatryan , a.v .
petrosyan , and a.a .
saharian , izv . nats . akad .
, fiz . * 35 * , ( 2000 ) ( engl
. transl . : j. contemp .
phys . ) ; a.s .
kotanjyan and a.a .
saharian , izv . nats . akad .
36 * , 310 ( 2001 ) ( engl .
: j. contemp .
phys . ) ; a.s .
kotanjyan and a.a .
saharian , izv .
, fiz . * 37 * , 135 ( 2002 ) ( engl .
transl . : j. contemp . phys . ) ; a.s .
kotanjyan and a.a .
saharian , mod .
a * 17 * , 1323 ( 2002 ) ; a.s .
kotanjyan , nucl .
instr . and meth .
b * 201 * , 3 ( 2003 ) ; a.a .
saharian and a.s .
kotanjyan , izv . nats
, fiz . * 38 * , 288 ( 2003 ) ( engl .
transl . : j. contemp . phys . ) ; a.a .
saharian and a.s .
kotanjyan , nucl .
instr . and meth .
b * 226 * , 351 ( 2004 ) . | in this paper we investigate the radiation emitted by a charged particle moving along a helical orbit around a dielectric cylinder immersed into a homogeneous medium .
formulae are derived for the electromagnetic potentials , electric and magnetic fields , and for the spectral - angular distribution of the radiation in the exterior medium .
it is shown that under the cherenkov condition for dielectric permittivity of the cylinder and the velocity of the particle image on the cylinder surface , strong narrow peaks appear in the angular distribution for the number of quanta radiated on a given harmonic . at these peaks the radiated energy exceeds the corresponding quantity for a homogeneous medium by several dozens of times .
simple analytic estimates are given for the heights and widths of these peaks .
the results of numerical calculations for the angular distribution of the radiated quanta are presented and they are compared with the corresponding quantities for the radiation from a charge moving along a helical trajectory inside a dielectric cylinder .
pacs number(s ) : 41.60.ap , 41.60.bq |
pulsar wind nebulae ( pwne ) are extended sources around a rotation powered pulsar .
they show a broadband spectrum from radio to @xmath2-rays so that they would contain very high energy non - thermal particles @xcite .
the central pulsar releases its rotational energy as the pulsar wind , which is a highly relativistic magnetized electron - positron outflow , and plays a role of the energy source of the pwn .
the strong termination shock , which is formed by the interaction between the pulsar wind and external supernova remnant ( or interstellar medium ) , has been supposed to produce the non - thermal electrons and positrons , and they emit the synchrotron radiation @xcite and the inverse compton emission @xcite .
pwne are characterized by a center - filled morphology caused by confinement in a supernova remnant ( snr ) , which is associated with the progenitor of the central pulsar .
most pwne are detected as spatially extended sources in radio and x - rays . while the radio spectral index is almost spatially homogeneous ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , the x - ray spectral index increases with the distance from the pulsar ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
recently , pwne are found to be very bright sources in also tev @xmath2-rays @xcite .
one - zone models of pwne , which are invented by @xcite and developed by several authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , can reproduce broadband spectra of entire nebulae well .
the one - zone models do not describe the spatial distribution of the emission ( e.g. , * ? ? ?
* ) , thus it is indispensable to invest a model which includes the spatial structure of nebulae . the model of ( * ?
* ; * ? ? ?
* hereafter kc84s ) has been established as a standard model of pwne .
they constructed a steady - state and 1-d magnetohydrodynamic model of the crab nebula . also , assuming the particle acceleration at the termination shock , they calculated the evolution of non - thermal pairs along the flow and the synchrotron emission from advected particles . adopting the model of kc84s , @xcite succeeded in reproducing the entire photon spectrum including the inverse compton component
however , as @xcite suggested , it is unclear whether kc84s can apply to general pwne other than the crab nebula .
furthermore , @xcite showed that the model of kc84s disagrees with the observed radial profile of the x - ray spectral index in 3c 58 .
they suggested that the radial profile of the x - ray spectral index in the model should change more rapidly , and the x - ray nebula size becomes more compact than the observation .
note that the entire photon spectrum was not taken into account as a model constraint in @xcite .
this ignited the revision of kc84s .
@xcite introduced the effect of the spatial diffusion of the particles , and reproduced the x - ray radial profile .
@xcite supported this idea via 3-d magnetohydrodynamic and test - particle simulations . in each of the studies , simultaneous verification of the entire spectrum and the spatial profile of the emission
is not discussed , thus there is no consensus on the spatial structure in pwn models so far . in order to advance the study ,
it is essential to clarify controversial points in the simple steady 1-d model before introducing nontrivial effects such as the particle diffusion .
we choose two objects , 3c 58 and g21.5 - 0.9 , to examine the 1-d steady model .
both of the pwne show the feature that the extent of the x - ray emission is the same as the radio one , in contrast to the crab nebula , in which the observed size shrinks with increasing frequency .
furthermore , the 1-d steady model has been never applied to those two pwne .
our purpose is to make the validity of the 1-d steady model clear for general pwne , so that those two pwne are suitable targets for testing . in this paper
, we revisit the 1-d steady model and calculate the photon spectrum and its radial profile numerically . in section [ sec : model ] , we review the 1-d steady model of pwne based on kc84s . the parameter dependence of this model is investigated in section [ sec : dependence ] .
the application to the two observed sources ( 3c 58 and g21.5 - 0.9 ) is presented in section [ sec : application ] .
in this paper , we adopt the one - dimensional magnetohydrodynamical ( mhd ) model constructed by kc84s .
we numerically solve the energy distribution of the electron
positron pair plasma along the outflow obtained by mhd equations . from the resultant spectral distribution of the pairs
, we calculate the photon emission from those pairs , resultant surface brightness , and volume - integrated spectrum . in this section ,
we review the the 1-d steady model and present the method to calculate the radial evolution of the energy distribution of electrons and positrons . here , we consider that the pwn is a steady outflow and has a radius @xmath3 .
the relativistic magnetized wind emitted from the central pulsar forms a strong termination shock at a radius @xmath0 . assuming that the pre - shock plasma is cold ,
the wind property is regulated by three quantities , number density in the comoving frame @xmath4 , bulk lorentz factor @xmath2 , and magnetic field in the lab frame @xmath5 at the shock .
almost all of the * * * * pulsar spin - down luminosity @xmath6 is converted to the wind luminosity as @xmath7 where @xmath8 , the subscript u denotes values at just upstream of the shock , and @xmath1 is the ratio of the magnetic energy flux to the particle energy flux at the upstream of the shock , @xmath9 the magnetic field of the wind is dominated by the toroidal component ( e.g. , * ? ? ?
* ) , and the upstream plasma is highly relativistic ( @xmath10 ) , which means the downstream temperature is relativistic ( adiabatic index @xmath11 ) .
the rankine - hugoniot jump conditions provide the values in the downstream ( kc84s ) as @xmath12 @xmath13 @xmath14 , \label{eq : kc_rhc}\ ] ] @xmath15 where the subscript d denotes the values at just downstream of the shock , and @xmath16 is the thermal pressure . for @xmath17 and @xmath18 , we obtain @xmath19 , which coincides with the well - known result in the relativistic hydrodynamics . as boundary conditions , we adopt equations ( [ eq : kc_rha])([eq : kc_rh ] ) at the radius @xmath20 , and solve the steady state and spherical symmetric mhd equations . under the toroidal field approximation and the adiabatic assumption , the mhd equations are integrable .
after some algebra with introducing @xmath21 , we obtain ( kc84s ) @xmath22 from which we obtain the radial profile of the four velocity @xmath23 ( or equivalently @xmath24 ) . in the strong shock approximation , @xmath25 is a function of only @xmath1 as shown in equation ( [ eq : kc_rhb ] ) .
if @xmath26 is established , the above flow equation ( [ eq : kc_vel_eq ] ) is depicted by only one parameter @xmath1 independently of @xmath27 and @xmath28 .
we calculate the radial profile of @xmath23 numerically , while kc84s neglected @xmath29 and adopted @xmath30 in the downstream .
then , the mhd conservation laws provide the other quantities as follows : @xmath31 where @xmath32 is the comoving number density in the wind . in figure
[ fig : flow ] , the radial profiles of @xmath23 and @xmath33 in our test calculations are shown for @xmath34 , @xmath35 , @xmath36 , @xmath37 and @xmath38 . the results are almost the same as the behavior shown in kc84s because @xmath26 for all the cases . for @xmath39 , at a small radius , the pressure ratio @xmath40 ( @xmath41 at @xmath20 ) gradually increases with radius as @xmath42 . in that regime , the outflow behaves as @xmath43 ( equivalently @xmath44 ) , @xmath45 and @xmath46 . at the radius @xmath47 @xmath48 becomes unity , namely the magnetic pressure starts to dominate .
outside this radius , the radial four speed is approximately constant with @xmath49 and @xmath50 .
thus , the magnetic field has a maximum value at @xmath51 as shown in figure [ fig : flow ] .
@xmath52{kcv_l } & \includegraphics[width=0.5\textwidth]{br } \end{array}$ ] the energy distribution of electron positron pairs @xmath53 is calculated consistently with the mhd model , @xmath54 at the inner boundary @xmath20 , @xmath55 is assumed to have a broken power - law energy distribution at injection as following : @xmath56 where the parameters are the break energy @xmath57 , minimum energy @xmath58 , maximum energy @xmath59 , and two power - law indices @xmath60 and @xmath61 for low and high energy regions , respectively .
the normalization @xmath62 is determined by equation ( [ eq : bpl : con_number ] ) as @xmath63^{-1}.\end{aligned}\ ] ] while the origin of the radio spectral component may be different from that for the x - ray and optical components as discussed in kc84s , in this paper , we use a broken power - law distribution that has been adopted by the one - zone studies ( e.g. , * ? ? ?
the observed radio spectral index almost uniquely gives the index @xmath60 , which is generally lower than 2 ( e.g. , * ? ? ?
* ) . in this case ,
the particles with energies @xmath64 dominate the particle number .
for simplicity , we fix the minimum energy as @xmath65 and leave the pair density problem ( c.f . * ? ? ? * ) . the particles above @xmath57 may be produced via the shock acceleration @xcite .
the maximum energy @xmath59 is determined by the same method in kc84s ; the energy at which a gyro radius is equal to the shock radius provides @xmath66 the pressure obtained by @xmath55 , @xmath67 should satisfy equation ( [ eq : kc_rhc ] ) .
we thus obtain @xmath68 , \label{eq : nd_bpl}\ ] ] where @xmath69^{-1}.\end{aligned}\ ] ] notify that @xmath70 or @xmath25 is already given as a function of only @xmath1 ( see equation ( [ eq : kc_rhb ] ) ) , so that the quantities in the upstream are written with the six parameters @xmath6 , @xmath1 , @xmath0 , @xmath57 , @xmath60 , and @xmath61 as @xmath71^{-1},\ ] ] @xmath72^{1/2 } , \label{eq : mag_us}\ ] ] and @xmath73 ^ 2.\end{aligned}\ ] ] from the above relations , the functional shape described in equation ( [ eq : bpl : initial ] ) is also written with the six parameters . in this 1-d model
, we have a unique parameter @xmath0 , which is not in the one - zone models , and @xmath0 significantly affects the results as will be shown in section [ sec : dependence ] . the flow solution given by equation ( [ eq : kc_vel_eq ] )
provides the advection time as @xmath74 while the age of the pwn is an important parameter in the one - zone time - dependent models such as @xcite , the parameter @xmath0 in our steady model adjusts the advection time , which may be close to the age of the pwn .
the radial evolution of @xmath53 and photon emission are calculated with the numerical code used in @xcite , which are based on the time - dependent code in @xcite ( see also * ? ? ?
* ) . taking into account the klein
nishina effect on the inverse compton ( ic ) cooling , the code can follow the temporal evolution of the energy distribution along the stream .
transforming the elapsed time into radius as @xmath75 , our calculation is practically equivalent to solve the steady transport equation ( e.g. @xcite ; @xcite ) @xmath76 \nonumber \\ & & + \frac{\partial } { \partial e}\left [ \frac{c e n(e
, r)}{3r^2}\frac{d}{dr}\left(r^2u(r ) \right ) \right ] \nonumber \\ & & -\frac{c}{r^2}n(e , r)\frac{d}{dr}\left(r^2u(r)\right ) , \label{eq : fp}\end{aligned}\ ] ] where @xmath77 and @xmath78 are the energy loss rates due to synchrotron radiation and ic scattering , respectively .
the three terms of the right hand side of equation ( [ eq : fp ] ) represent the effects of the radiative cooling , adiabatic cooling and volume expansion , respectively . here
, we have used the solution for @xmath23 with adiabatic approximation , namely the radiative cooling is assumed not to affect the dynamics of the flow .
this approximation is valid when the cooling time for the particles with @xmath79 is longer than the advection time ( kc84s ) .
most of the results shown in this paper safely satisfy this condition .
the spectral emissivity @xmath80 per unit volume is obtained consistently with the energy distribution @xmath53 , the magnetic field profile @xmath33 , and the interstellar radiation field ( isrf ) with the klein
nishina effect .
the model of the isrf is taken from galprop v54.1 ( * ? ? ?
* and references therein ) , in which the results of @xcite are adopted , as shown in figure [ fig : soft_photon ] . the isrf is assumed as uniform and isotropic in the pwn , and we neglect the synchrotron self - compton , which significantly contribute for only limited cases like the crab nebula @xcite .
we neglect the contribution of bremsstrahlung as well , because the density of pairs is low enough @xcite . and @xmath81 .
the red dashed line is for 3c 58 , which is located at @xmath82 and @xmath83 . ] neglecting the emission from the pre - shock wind , the photon spectrum of the entire nebula @xmath84 is given by @xmath85 where @xmath86 is the distance to the pwn from us . the surface brightness @xmath87 is also given by @xmath88 where @xmath89 is the distance perpendicular to the line of sight from the central pulsar .
our numerical code is checked by reproducing the result of @xcite for the crab nebula .
if we adopt the same assumption as @xcite , the resultant spectrum agree with the observed one .
note that our more conservative model leads to a slightly dimmer flux for the inverse compton component as shown in appendix .
in this section , we discuss how the entire spectrum @xmath90 and the surface brightness @xmath91 depend on the parameters in this 1-d steady model .
there are some previous studies that discussed the parameter dependence of the 1-d model .
kc84s have already discussed how the parameters @xmath28 and @xmath1 change the total synchrotron luminosity @xmath92 ( not the spectral distribution ) . while @xcite have studied the x - ray spatial profile for different @xmath0 assuming that the flow velocity decreases as a power - law of @xmath93 independently of @xmath1 , we investigate the parameter dependence consistently with the mhd flow solution .
we focus on the dependence on the parameters @xmath0 and @xmath1 , which largely affect the spatial structure of the emission . in this section ,
the nebula size @xmath94 is fixed .
the external photon field is taken from the model for g21.5 - 0.9 in figure [ fig : soft_photon ] . for four parameters out of the six parameters in our model ,
we adopt a parameter set as @xmath95 , @xmath96 , @xmath97 , and @xmath98 , and change @xmath0 or @xmath1 below .
first , we introduce some typical particle energies and corresponding photon frequencies , and their dependence on the model parameters are discussed . in this subsection ,
we limit the discussion within the case of @xmath99 ( i.e. , @xmath100 ) . in this case
, the magnetic field has a maximum value @xmath101 at the edge of the nebula .
the cooling effect for pairs with energies @xmath102 is also found to be negligible .
we find the first typical frequency , the intrinsic break frequency , @xmath103 in the case of @xmath45 or equivalently @xmath43 , the energy of pairs injected with @xmath104 decreases with @xmath93 via the synchrotron cooling as @xmath105 where the burn - off energy @xmath106 ( kc84s ) . at @xmath107 , the maximum energy becomes @xmath108 .
then , the cooling frequency at the outer boundary is given by @xmath109 above @xmath110 , the entire spectrum should show the softening behavior .
the maximum particle energy decreases following equation ( [ eq : cut ] ) , while the magnetic field increases as @xmath111 . the typical synchrotron frequency @xmath112 peaks at @xmath113
where we have assumed @xmath114 . with @xmath115 , the cut - off frequency in the synchrotron spectrum
is obtained as @xmath116 above which the flux decreases exponentially .
one - zone models do not include the shock radius as a parameter .
the value @xmath0 is a characteristic parameter in the 1-d model . for the crab nebula @xcite , vela @xcite , and msh 15 - 52 @xcite , a possible shock structure ( inner ring )
is detected with x - ray observations . in most of pwne ,
however , the shock radii are not observationally constrained well . in figure
[ fig : rs_dependence ] , we show the @xmath0-dependences of the entire spectrum @xmath117 and the x - ray surface brightness with @xmath118 .
the synchrotron component dominates below @xmath119 hz ( @xmath120 kev ) and the ic component dominates in @xmath2-rays .
note that in the case for @xmath121 pc , @xmath122 beyond @xmath123 pc , i.e. , the adiabatic approximation is invalid . in table
[ rs_dep_list ] , for various @xmath124 , we summarize the advection time , the volume - averaged magnetic field @xmath125 as , @xmath126 and the maximum magnetic field @xmath127 because we have @xmath128 from equation ( [ eq : req ] ) with @xmath129 . first , as a benchmark case , we take up the case of @xmath130 pc ( the red solid line in figure [ fig : rs_dependence ] ) , which leads to @xmath131 pc .
there are two breaks in the synchrotron spectrum : the intrinsic break at @xmath132 hz corresponding to @xmath57 and the cooling break at @xmath133 . adopting the average magnetic field and the advection time , the cooling break energy of pairs is @xmath134 the corresponding cooling break frequency is written as @xmath135 the above value obtained with a one - zone - like treatment roughly agree with our results . in figure [ fig :
appendix : crab ] , the entire photon spectrum and the x - ray radial profile for the crab nebula are shown .
all the parameters to calculate the spectrum for the crab nebula are same as aa96 without `` obtained parameter '' , and are summarized in table [ crab_parameter ] .
the flux of inverse compton becomes a little smaller than the value observed and calculated by aa96 .
the difference in the ssc flux is due to the additional parameters @xmath136 and @xmath59 in aa96 .
even if we take the smaller @xmath1 to enhance the ssc flux , the maximum energy of synchrotron emission becomes much lower than the cut - off energy of observed spectrum ( see equation ( [ eq : nucut ] ) ) . within our conservative model assumption for the maximum energy of particles , we are not able to fit the spectrum around the cut - off of the spectrum . the right panel of figure [
fig : appendix : crab ] shows the radial profile of surface brightness in 3.0 - 5.0 kev range .
the extent of the x - ray nebula calculated by our model is comparable with observations . comparing with the cases of 3c 58 and g21.5 - 0.9
, we do not have so strong motivation to improve the 1-d steady model in the case of the crab nebula . | the observed radial profiles of the x - ray emission from pulsar wind nebulae ( pwne ) have been claimed to conflict with the standard one - dimensional ( 1-d ) steady model .
however , the 1-d model has not been tested to reproduce both the volume - integrated spectrum and the radial profile of the surface brightness , simultaneously .
we revisit the 1-d steady model and apply it to pwne 3c 58 and g21.5 - 0.9 .
we find that the parameters of the pulsar wind , the radius of the termination shock @xmath0 and magnetization @xmath1 , greatly affect both the photon spectrum and radial profile of the emission .
we have shown that the parameters constrained by the entire spectrum lead to a smaller x - ray nebula than observed one .
we have also tested the case that reproduces only the observations in x and gamma - rays , ignoring the radio and optical components . in this case , there are parameter sets that reproduce both the spectrum and emission profile , but the advection time to the edge of the nebula becomes much smaller than the age .
additionally , we calculate the surface brightness profile of the radio , optical and tev gamma - rays .
the future observations in these wavelengths are also important to probe the spatial distributions of the relativistic plasma and the magnetic field of pwne . |
hp thanks b. allanach , k.s . babu and s. pascoli for discussions and the university of hawaii and desy for kind hospitality .
99 m. doi , t. kotani , h. nishiura and e. takasugi , prog .
phys . * 69 * , 602 ( 1983 ) . c. aalseth _
et al . _ ,
arxiv : hep - ph/0412300 .
p. vogel , aip conf .
proc . * 870 * , 124 ( 2006 ) .
h. v. klapdor - kleingrothaus and h. ps , arxiv : hep - ph/9808350 . j. schechter and j. w. f. valle , phys .
d * 25 * , 2951 ( 1982 ) .
a. ali , a. v. borisov and d. v. zhuridov , arxiv : hep - ph/0606072 .
m. hirsch , k. muto , t. oda and h. v. klapdor - kleingrothaus , z. phys .
a * 347 * , 151 ( 1994 ) .
f. simkovic , m. nowak , w. a. kaminski , a. a. raduta and a. faessler , phys .
c * 64 * , 035501 ( 2001 ) .
h. ps , m. hirsch , h. v. klapdor - kleingrothaus and s. g. kovalenko , phys . lett .
b * 453 * , 194 ( 1999 ) .
h. ps , m. hirsch , h. v. klapdor - kleingrothaus and s. g. kovalenko , phys . lett .
b * 498 * , 35 ( 2001 ) .
m. hirsch , s.g .
kovalenko , h. ps , in preparation . k. s. babu and r. n. mohapatra , phys
. rev .
lett . * 75 * , 2276 ( 1995 ) ; m. hirsch , h. v. klapdor - kleingrothaus and s. g. kovalenko , phys .
b * 372 * , 181 ( 1996 ) [ erratum - ibid .
b * 381 * , 488 ( 1996 ) ] ; h. ps , m. hirsch and h. v. klapdor - kleingrothaus , phys .
b * 459 * , 450 ( 1999 ) .
r. n. mohapatra , phys .
d * 34 * , 3457 ( 1986 ) ; j. d. vergados , phys .
b * 184 * , 55 ( 1987 ) .
m. hirsch , h. v. klapdor - kleingrothaus and s. g. kovalenko , phys .
d * 53 * , 1329 ( 1996 ) .
g. bhattacharyya , h. v. klapdor - kleingrothaus , h. ps and a. pilaftsis , phys .
d * 67 * , 113001 ( 2003 ) .
a. staudt , k. muto and h. v. klapdor- kleingrothaus , europhys .
* 13 * , 31 ( 1990 ) .
k. muto , e. bender and h. v. klapdor , z. phys .
a * 334 * , 187 ( 1989 ) .
f. simkovic , l. pacearescu and a. faessler , nucl .
a * 733 * , 321 ( 2004 ) .
v. a. rodin , a. faessler , f. simkovic and p. vogel , nucl .
phys . a * 766 * , 107 ( 2006 ) .
s. m. bilenky and j. a. grifols , phys .
b * 550 * , 154 ( 2002 ) .
a. faessler , s. kovalenko , f. simkovic and j. schwieger , phys .
* 78 * , 183 ( 1997 ) .
s. r. elliott , arxiv : nucl - ex/0609024 ; k. zuber , acta phys .
b * 37 * , 1905 ( 2006 ) ; a. barabash , arxiv : hep - ex/0608054 ; a. piepke , nucl .
a * 752 * , 42 ( 2005 ) .
v. cirigliano , a. kurylov , m. j. ramsey - musolf and p. vogel , phys .
* 93 * , 231802 ( 2004 )
. v. m. gehman and s. r. elliott , j. phys .
g * 34 * , 667 ( 2007 ) [ arxiv : hep - ph/0701099 ] . | a measurement of neutrinoless double beta decay in one isotope does not allow to determine the underlying physics mechanism .
we discuss the discrimination of mechanisms for neutrinoless double beta decay by comparing ratios of half life measurements for different isotopes .
six prominent examples for specific new physics contributions to neutrinoless double beta decay are analyzed .
we find that the change in corresponding ratios of half lives varies from 60% for supersymmetric models up to a factor of 5 - 20 for extra - dimensional and left - right - symmetric mechanisms . an uncontroversial detection of neutrinoless double beta ( @xmath0 ) decay @xcite will be a discovery of uttermost significance .
most importantly , it will prove lepton number to be broken in nature , and neutrinos to be majorana particles @xcite .
on the other hand , it will immediately generate another puzzle : what is the mechanism that triggers the decay ?
the most prominently discussed mechanism for neutrinoless double beta decay is the exchange of light majorana neutrinos .
but other mechanisms , like the exchange of susy superpartners with r - parity violating or conserving couplings , leptoquarks , right - handed w - bosons or kaluza - klein excitations , among others , have been discussed in the literature as well .
possibilities to disentangle at least some of the possible mechanisms include the analysis of angular correlations between the emitted electrons @xcite or a comparative study of @xmath0 and @xmath1 with electron capture ( @xmath2 ) decay @xcite .
another possibility seems to be the study of double beta decay to excited @xmath3 states @xcite .
unfortunately , the search for @xmath4 decay is complicated due to small rates and the experimental challenge to observe the produced x - rays or auger electrons , and most double beta experiments of the next generation are not sensitive to electron tracks or transitions to excited states . without identification of the underlying mechanism , an experimental evidence for neutrinoless double beta decay will only provide ambiguous information about the concrete physics underlying the decay . for example
, no information about the neutrino mass can be obtained from a measurement of the neutrinoless double beta decay half life . in general , contributions to neutrinoless double beta decay
can be categorized as either long - range or short - range interactions . in the first case ,
the diagram involves two vertices which are pointlike at the fermi scale , and the exchange of a light neutrino in between , and is described by an effective lagrangian of the type @xcite @xmath5 where the sum runs over all lorentz invariant combinations of hadronic and leptonic lorentz currents of defined helicity , @xmath6 and @xmath7 , respectively . here
@xmath8 denotes the corresponding transition operator .
the effective coupling strengths in new physics contributions are denoted as @xmath9 throughout . for short - ranged contributions , on the other hand ,
the interactions are described by a single vertex being pointlike at the fermi scale .
the decay rate therefore results from first order perturbation theory , and is described by the lagrangian @xcite @xmath10 here @xmath11 denotes the proton mass and the sum runs over all lorentz invariant combinations of hadronic , @xmath12 , and leptonic , @xmath13 , currents of defined chirality .
the combination involving two vertices of the first term in ( [ lrlag ] ) leads to the usual neutrinoless double beta decay half life formula for the mass mechanism , @xmath14^{-1}= ( \langle m_\nu \rangle / m_e)^2 g_{01}|{\cal m}^{m_\nu}|^2,\ ] ] where @xmath15 is the effective neutrino mass in which the contributions of individual neutrino mass eigenstates are weighted by mixing matrix elements squared , @xmath16 .
the combination of the first term in ( [ lrlag ] ) with any of the latter terms as well as the short - range lagrangian ( [ lagsr ] ) leads to the expression @xmath17^{-1}=\epsilon_{np}^2 g_{np } |{\cal m } ^{np}|^2.\ ] ] here , @xmath18 and @xmath19 are the nuclear matrix elements for the mass mechanism and alternative new physics contributions , and @xmath20 and @xmath21 denote the corresponding phase space integrals from the list given in @xcite .
we have assumed , that one mechanism dominates the double beta decay rate , and we do not consider interference between different mechanisms .
calculational details and results for the relevant matrix elements involved have been given elsewhere @xcite , and numerical results for all common double beta emitter isotopes will be published soon @xcite . in the present context , we will concentrate on the observation that the combinations of leptonic and hadronic currents specific to different mechanisms result in different nuclear matrix elements .
this fact taken alone is not of much help in order to disentangle the different mechanisms , since e.g. a smaller nuclear matrix element for the mass mechanism as compared to any alternative new physics mechanism can be compensated by a larger value for the neutrino mass , at least within the constraints implied by other observations such as tritium beta decay and cosmology .
however , under the assumption that one mechanism dominates in triggering the decay , the new physics parameter @xmath15 or @xmath9 drops out in the ratio of experimentally determined half lives for two different emitter isotopes , @xmath22 consequently , half life ratios depend on the mechanism of double beta decay , but not on the new physics parameter , and thus can be compared with the theoretical prediction for different mechanisms . moreover , the error in the isotope nuclear matrix element ratio can be reduced compared to the theoretical error in one matrix element , due to cancellations of systematic effects . in the following
we study several prominent examples of specific alternative new physics contributions by calculating the corresponding ratios of half lives @xmath23 where we concentrate on a comparison with @xmath24ge as it constitutes the best tested isotope to date .
we choose the following mechanisms for a detailed discussion : * * susy - accompanied neutrinoless double beta decay : @xmath25 * + this mechanism has been first discussed in @xcite .
the effective lagrangian for the dominant contribution assumes the form @xmath26,\end{aligned}\ ] ] and results from integrating out a heavy @xmath27-squark of the @xmath28-th generation with @xmath29-parity violating couplings @xmath30 and @xmath31 , and exchanging a light neutrino of the @xmath32-th generation between the nucleons .
the new physics parameter is given by @xmath33 where @xmath34 parametrizes the left - right sfermion mixing of the mass eigenstates @xmath35 and @xmath36 .
* * gluino exchange mechanism in r - parity violating susy * : @xmath37 + in this short - range contribution discussed in @xcite , integrating out @xmath38- and @xmath27-squarks and a gluino leads to the effective lagrangian @xmath39 with @xmath40.\ ] ] * * right - handed currents * : @xmath41 and @xmath42 + integrating out right - handed @xmath43-bosons occurring in left - right symmetric models can lead to two types of new contributions with right - handed leptonic currents @xcite , @xmath44 where the new physics parameters are given by @xmath45 and @xmath46 . * * kaluza - klein neutrino exchange in extra - dimensional models * : @xmath47 + in extra - dimensional theories , the double beta observable is given by a sum over contributions from all kaluza - klein excitations with masses @xmath48 , weighted with the mass dependent matrix element @xmath49 @xcite : @xmath50 in this case the effective coupling constant @xmath51 depends on the nuclear matrix element @xmath52 , and therefore the particle physics does not decouple from the nuclear physics .
this is because the masses of the kaluza - klein excitations vary from values much smaller than the nuclear fermi momentum @xmath53 to values much larger than @xmath53 , while the @xmath48-dependence of @xmath49 changes around @xmath53 .
therefore the kaluza - klein spectrum has to be fixed by choosing specific values for the brane shift parameter @xmath54 and the radius of the extra dimension @xmath29 . in the limit of @xmath55 or @xmath56
, @xmath47 approaches @xmath57 .
the matrix elements for the mass mechanism and for the susy accompanied neutrino exchange have been calculated in the pn - qrpa approach of @xcite , in the latter case for the first time . for the other mechanisms , existing numerical values obtained with the same nuclear structure model
have been adopted from the literature .
the values for the phase space integral factors @xmath20 , @xmath21 have been calculated in @xcite .
numerical values for @xmath58 are given in table 1 , and fig .
1 displays the relative change expected from various new physics contributions , compared to the mass mechanism .
an application of the procedure to any other alternative new physics contribution by using the matrix elements listed in @xcite is straightforward .
all isotope ratios have been normalized to the half life of the most extensively studied nucleus @xmath24ge .
moreover , while at present no experiment using a @xmath59te source has been proposed , we included this isotope since it provides a particularly powerful discriminator and thus may encourage future experimental efforts to study this nucleus . .
ratios @xmath60 of half lives for various important double beta decay emitter isotopes , normalized to the half - life of @xmath24ge . for the exchange of kaluza - klein excitations in extra dimensional theories the brane shift parameter and bulk radius do not factorize , and are chosen to be @xmath61 and @xmath62 ev@xmath63 . [ cols="<,^,^,^,^,^,^,^",options="header " , ] , normalized to the half - life of @xmath24ge , compared to the ratio in the mass mechanism @xmath64 .
, scaledwidth=95.0% ] the two supersymmetric contributions show similar deviations , which are rather small for all isotopes .
it is obvious that these mechanisms are most effectively discriminated from the mass mechanism by comparing the half life ratios between @xmath65se and @xmath66xe which vary by 60% . in left - right symmetric models ,
strong deviations can be found for the @xmath67 combinations , while deviations for the @xmath68 combination are rather small . a comparison of half life ratios between @xmath69mo and @xmath66xe yields a variation of 70 % for the @xmath68 contribution with right - handed hadronic currents , while a comparison of measurements in @xmath59te and @xmath70nd will provide a powerful discriminator with a variation of more than a factor of 20 for the @xmath67 contribution with left - handed hadronic currents .
similarly in extra - dimensional neutrino models with a large brane shift parameter , large deviations can be found for @xmath66xe and @xmath70nd , and the half life ratios for @xmath70nd and @xmath69mo vary by more than a factor of 5 .
some caution is necessary when referring to the half life ratio of the heavily deformed @xmath70nd , which is ignored in most qrpa calculations ( compare the discussion in @xcite ) .
finally it should be stressed that not necessarily two positive results are needed - already the comparison of one half life measurement and one upper bound in another isotope could provide non - trivial information on the double beta mechanism .
since the theoretical errors of the nuclear matrix element calculation dominate the experimental errors , it is difficult to determine the confidence level with which either mechanism can be excluded to generate the observed double beta evidence .
if , for example , a statistical distribution of matrix element values is assumed , a relative variation of 60% in @xmath58 with respect to @xmath64 is significant only if the corresponding nuclear matrix elements would be known with an accuracy of 15% , which seems to be unrealistic , if only one pair of isotopes is being analyzed .
indeed , estimates of errors in nuclear matrix elements vary from a factor 3 - 5 , when the spread of published values is used as a measure , to only 30% , according to an assessment of uncertainties inherent in qrpa @xcite . however , the significance of the comparison of two isotopes will increase if a whole set of measurements in different isotopes resembles the expected pattern .
moreover , one would expect that systematical effects , like an overestimation of the nuclear matrix elements due to a too small value for the particle - particle interaction @xmath71 in the pn - qrpa approach , a different value for the axial - vector coupling @xmath72 , the inclusion of higher - order terms or a different model - space would influence calculations for the different isotopes in a similar way , and thereby cancel in the half life ratios discussed .
this expectation is confirmed by the comparison of the results of different qrpa codes in @xcite , and of qrpa and shell model codes in @xcite .
finally it has been pointed out in @xcite that the half life ratios ( [ ratiosdef ] ) can also be used to single out the correct nuclear structure model . in this case the correct combination of mechanism and nuclear structure code can be determined by the best fit of the theoretical half life ratios to half life measurements in various nuclei .
thus the results presented in this letter should be complemented and checked with alternative codes for the nuclear matrix element calculation . moreover , other mechanisms , including pion exchange @xcite , may be dominating in some of the models discussed , and should be discussed as well . in summary , we discussed how different mechanisms of neutrinoless double beta decay would manifest themselves in half life ratios involving different isotopes . we thus conclude that complementary measurements in different isotopes would be strongly encouraged . at present , next - generation experiment proposals exist for @xmath24ge ( gerda , majorana , gem , geh@xmath73 ) , @xmath65se ( super - nemo , dcba , sef@xmath74 ) , @xmath69mo ( moon ) , @xmath75te ( cuore ) , @xmath66xe ( exo , xmass , xe ) , as well as for the isotopes @xmath76ca , @xmath77cd and @xmath78gd not discussed in this letter ( candles , cobra and gso ) ( for recent overviews of the experimental status see @xcite ) .
an experimental study of this kind should be complemented by neutrino mass searches in tritium beta decay experiments and cosmology , as well as searches for effects of the alternative new physics source of lepton number violation in other processes , such as lepton flavor violating decays @xcite . after this paper had been submitted for publication , the paper @xcite appeared , which comes to similar conclusions and estimates the number of required measurements and their precision needed . |
that gauge fields successfully describe the forces between elementary particles is certainly among the greatest discoveries in physics . but gauge theories are not restricted to high energy physics . in condensed matter physics one is often interested in theories with constraints . in a sense
a constraint is just a ( infinitely strong ) force and it is quite natural that these theories also can be formulated as gauge theories .
+ + the gauge field description is in terms of gauge _ potentials _ , which are redundant variables .
the theory is therefore invariant with respect to a change of gauge .
the most important consequence of this gauge symmetry is that it dictates the form of the interaction vertices between the gauge fields and the other fields in the theory .
other than that ; the gauge - invariance is just a reminder that we are working with redundant variables .
this redundancy can be very annoying , in particular when using path - integrals , and it is often necessary to choose a specific gauge .
this choice of gauge _ can not _ affect any physical quantities calculated in the theory .
+ + in this paper we will consider a theory of @xmath1 gauge bosons coupled to nonrelativistic fermions .
this theory is motivated by the composite fermion approach to the half - filled landau level state in the quantum hall effect . + +
some years ago it was observed that the @xmath2 quantum hall state exhibits metal - like features @xcite .
using the idea of jain @xcite , in which two flux quanta are `` attached '' to each electron ; making a composite fermion , halperin , lee and read @xcite formulated the theory of the @xmath2 state as a theory of composite fermions interacting with a fictitious gauge field , in _ zero _ external magnetic field . in the mean field approximation , in which the electrons are uniformly spread out , this theory is just that of free ( composite ) fermions .
therefore , in the mean field approximation , the @xmath2 state is certainly metallic .
striking evidence of the existence of a fermi surface was subsequently observed experimentally @xcite,@xcite,@xcite .
+ + in order to make a consistent theory of the @xmath2 state it is necessary to also include the effects of gauge - fluctuations .
however , when fluctuations of the gauge field are taken into account , it seems at first sight that the mean field picture is destroyed .
as shown in @xcite the self - energy correction due to the transverse gauge fluctuations dominates over the linear in frequency fermi liquid term at low frequencies .
+ + in an attempt to address the role of the gauge field fluctuations nayak and wilczek @xcite showed , using a renormalization - group approach , that for electron - electron interactions of shorter range than @xmath3 , the fluctuations lead to a strong coupling fixed point in the infra - red , different from the fermi liquid fixed point .
their result was based on the calculation of the single - particle green function and gauge - invariance through the ward identity .
a somewhat similar approach was taken by chakravarty _
et al _ @xcite .
they studied the same system using the @xmath4-expansion around @xmath5 , and obtained essentially the same result ; a strong coupling fixed point .
+ + stern and halperin @xcite constructed a boltzman transport equation for the @xmath2 state in the same way as landau did for the fermi liquid . this construction is based on the pole structure of the single - particle green function .
+ + as emphasized in @xcite the single - particle green function is not gauge - invariant , and so any conclusions based on it might not be physical .
this leads naturally to the question whether or not the existence of the strong coupling fixed point found in @xcite,@xcite , or the pole structure used in @xcite is dependent on the choice of gauge ?
+ + we will in this paper show that although the single - particle green function is dependent on the choice of gauge , the @xmath0-function is not , i.e. neither the existence of the strong coupling fixed point found by nayak and wilczek nor the pole structure used by stern and halperin is dependent on the choice of gauge .
the lagrangian density considered in this paper is @xmath6 this lagrangian is motivated by the composite fermion description of the @xmath2 quantum hall state , where two flux tubes are attached to each electron forming composite fermions . since the flux quanta are attached to the electrons , fluctuations in electron - density lead to fluctuations in the gauge field .
this is taken care of by the third term in which the electron - electron interaction @xmath7 controls the gauge fluctuations . here
we will consider electron - electron interactions of the form @xmath8 where @xmath9 is close to @xmath10 .
the last term is the gauge - fixing term .
it is obtained using the faddeev - popov ansatz to restrict @xmath11 to a gauge in which @xmath12 and then averaging over a gaussian distribution , with mean value @xmath13 and width @xmath14 , of such functions @xmath15 .
coulomb gauge corresponds to @xmath16 .
+ + in order to describe the @xmath2 state properly , a chern - simons term and a time - like gauge field ( @xmath17 ) should also be included .
we omit these terms because it is likely , as for coulomb interaction in metals , that the @xmath17-field is screened and does not contribute to the fluctuations at low frequencies .
there are however subtleties associated with the fluctuations of the @xmath17 field @xcite,@xcite which we do not address here .
+ + before we go on to study the lagrangian ( [ lagrangian ] ) , let us understand why the @xmath0-function in qed is gauge - invariant .
the qed - lagrangian density in terms of the bare coupling and fields is @xmath18 by redefining fields and the coupling constant @xmath19 the form of the interaction between the fermions and the gauge - field is dictated by gauge - invariance .
so for gauge - invariance to hold in the renormalized theory : @xmath20 this identity can also be proven using the ward identity , which is nothing but the statement of gauge - invariance .
now consider the correlation function @xmath21 this correlation function is obviously gauge - independent , and therefore @xmath22 should also be independent of gauge .
this and the above ward identity ( [ ward ] ) ensures that the @xmath0-function which is obtained by differentiating @xmath23 is independent of the choice of gauge . + + we see that neither @xmath24 nor @xmath25 play any role in this argument .
let us think about computing the above correlation function ( [ corr ] )
. the exact answer will of course be a very complicated function of @xmath26 and @xmath27 .
this answer should not change as we change the gauge , and so if @xmath25 and @xmath23 were dependent on gauge , amazing cancellations would have to occur .
conceivably the only cancellations that can occur should be due to the ward identity .
since the ward identity has nothing to do with @xmath25 it follows that @xmath25 can not be dependent on gauge either .
@xmath24 is not constrained by the above as the correlation function ( [ corr ] ) can be calculated using the effective action for the gauge fields in which the fermions are integrated out .
@xmath24 will be absorbed into the measure , and there is nothing preventing @xmath24 from being gauge - dependent .
+ + for non - abelian gauge theories the situation is somewhat different from qed .
there the gauge - field renormalization constant is dependent on gauge , but in spite of that , the @xmath0-function is gauge - independent there also .
+ + let us now return to the lagrangian ( [ lagrangian ] ) .
written in fourier space the euclidean action in terms of renormalized couplings and fields is @xmath28 we will assume a circular fermi surface and eventually take the zero- temperature limit such that @xmath29 .
it is of course possible to linearize the energy dispersion around the fermi surface , in that case @xmath25 is naturally named @xmath30 .
this linearization can not be essential for any important physics and will not be used in this paper .
note that there is no renormalization constant associated with @xmath31 .
@xmath31 measures the total number of particles which does not change .
the renormalization of @xmath32 is taken care of by @xmath22 .
+ + by simple dimensional analysis one can verify that the relevant coupling between the gauge field and the fermions is @xmath33 and not @xmath34 as in qed . therefore the relevant @xmath0-function for
this problem does nt involve just @xmath35 , but @xmath35,@xmath25 and @xmath22 . as in qed
, gauge invariance requires @xmath36 .
this can be seen explicitly by writing down the ward identity .
our lagrangian is not gauge - invariant with respect to gauge- transformations which mix space and time .
we must therefore consider the spatial vertex separately in the ward identity , setting the external frequency to zero .
as @xmath22 is obviously gauge - invariant , we need only to find if @xmath25 depends on the choice of gauge , to make statements about the @xmath0-function .
+ + in the mean field approximation the gauge field vanishes .
since the mean field theory describes a metal it is reasonable to proceed as in the theory of metals where one uses the rpa - corrected form of the interaction .
the bare photon propagator extracted from the lagrangian is @xmath37 the rpa - approximation of the propagator is obtained by summing an infinite series of bubble diagrams .
the polarization bubble is shown in figure 1 , and its expression is @xmath38 where @xmath39 and the bare single - particle propagator is @xmath40 summing the infinite series of bubble diagrams we obtain @xmath41 the transverse part is independent of the gauge - fixing parameter @xmath14 , and the longitudinal part of the gauge interaction does not include any information about the electron - electron interaction .
+ + the calculations of @xmath42 and @xmath43 are straight - forward at @xmath44 . in the zero - temperature limit of the imaginary time formalism we get in @xmath45 , for @xmath46 and @xmath47 : @xmath48 having calculated the gauge boson propagator
we can now calculate the self - energy diagrams .
the one - loop diagrams contributing to the self - energy are drawn in figure 2 .
the expression is @xmath49 for convenience we will split the contributions to the self - energy into two pieces , @xmath50 the first part is the self - energy obtained from the diagrams in figure 2 using the transverse part of the gauge propagator , while the second part is obtained using the longitudinal propagator in the same diagrams .
it is obvious from the form of the gauge - propagator ( [ rpagauge ] ) that only @xmath51 will depend on gauge .
let us first show that the logarithmic singularity necessary for the strong coupling fixed point is obtained by calculating @xmath52 . performing the angular integral in @xmath45 , the transverse contribution to the one - loop self - energy
can be written @xmath53 .
\nonumber\end{aligned}\ ] ] we have inserted an arbitrary cutoff @xmath54 in the @xmath55-integration and used the fact that @xmath56 . expanding the square root
the self - energy takes the form @xmath57 .
\label{above}\end{aligned}\ ] ] the ellipses indicate higher order terms in the expansion of the square root .
these terms are not as singular as the leading term , so they can safely be neglected .
let us now expand the renormalization constants @xmath24 and @xmath25 in the coupling constant , @xmath58 using this expansion we have @xmath59 contributions to @xmath60 are distinguished by the fact that they renormalize @xmath61 differently from @xmath62 .
so it is the second term in the above expression ( [ above ] ) which will contribute to @xmath60 .
the second term is @xmath63 when @xmath64 the first term in the last parenthesis will give the leading behaviour .
setting @xmath44 , @xmath65 , @xmath32 and using the result ( [ d2a ] ) for @xmath66 this leading behaviour is @xmath67 in the special case @xmath68 , corresponding to @xmath3-interactions between the fermions , we find in the low - frequency limit : @xmath69 in this case the correction to the @xmath70-behaviour is logarithmic , as observed by nayak and wilczek .
this case is the marginal case which sets the borderline between the fermi liquid theory and a strong coupling theory .
for @xmath71 , corresponding to interactions more long - range than coulomb interactions , the self - energy vanishes faster than @xmath72 and no deviations from fermi liquid are seen .
but for @xmath73 , short - range electron - electron interactions , the self - energy becomes the dominant term at small @xmath72 and the system is not in the fermi liquid universality class .
+ + the question is now whether these conclusions depend on the choice of gauge . at first sight
they do not , since the transverse gauge - propagator does not depend on @xmath14 .
however , one could imagine that the longitudinal gauge - propagator , which is dependent on @xmath14 , could give additional competing singularities .
we will now show , by computing the contribution to the self - energy using the longitudinal gauge - propagator , that there are no @xmath14-dependent competing singularities .
+ + performing the angular integration in the expression for the self - energy contribution coming from the longitudinal gauge propagator we get @xmath74.\end{aligned}\ ] ] the gauge boson propagator is invariant with respect to @xmath75 .
therefore no gap in the fermion spectrum is generated by the self - energy , which of course is expected on general grounds .
in the same way as for the transverse contribution we expand the square root . @xmath76 . \end{aligned}\
] ] the contribution to @xmath25 comes from the second term and is @xmath77 this quantity does not vary fast with @xmath78 so we set @xmath65 .
taking the zero - temperature limit , using the result ( [ d2b ] ) for @xmath79 and keeping just the dominant term as @xmath64 we get @xmath80 which vanishes faster than @xmath72 and therefore does not contribute to the singularities of @xmath25 .
this shows that the singularities of @xmath25 are not dependent on @xmath14 to this order in perturbation theory .
in @xcite the authors used the @xmath4-expansion around @xmath5 , and the free gauge - propagator @xmath81 to analyze the same theory .
we will in this chapter show that the @xmath0-function is independent of gauge also in this case .
the computation in three dimensions follows along the same lines as the computations in @xmath45 .
the polarization diagrams at @xmath44 give for @xmath46 and @xmath47 , @xmath82 the contribution from the transverse gauge boson to the self - energy takes the following form when the angular integrations are performed @xmath83 \right ] .
\end{aligned}\ ] ] for small @xmath84 , @xmath85 and @xmath86 , @xmath47 the logarithm can be expanded ; @xmath87 & = & \mathrm{i } \pi \mathrm{sgn}(\omega_n+\epsilon_m ) \\ & & -{\frac}{2m}{pq } \left ( \mathrm{i } \omega_n-{\frac}{p^2}{2m}+\mu+\mathrm{i } \epsilon_m-{\frac}{q^2}{2 m } \right)+\cdots .
\nonumber \end{aligned}\ ] ] when using this expansion in @xmath52 we can write @xmath88.\end{aligned}\ ] ] using the expression ( [ d3a ] ) for @xmath42 all terms odd in @xmath89 will vanish , and so no mass term is generated . the most singular term contributing to @xmath25 is @xmath90 taking the zero - temperature limit and setting @xmath65 we get for the dominant term in @xmath72 as @xmath64 : @xmath91 where we have used the expression ( [ d3a ] ) for @xmath66 . for the free gauge - propagator
( [ freegauge ] ) , @xmath92 , and the leading behaviour as @xmath64 is @xmath93 this logarithmic behaviour leads to a logarithmic correction to the specific heat of the electron gas @xcite .
this result is not dependent on the choice of gauge as can be seen by again computing the contribution to the self - energy using the longitudinal part of the gauge - propagator , @xmath94 \right ] .
\nonumber \end{aligned}\ ] ] expanding the logarithm we get @xmath95.\end{aligned}\ ] ] the most singular term contributing to @xmath25 is @xmath96 taking the zero - temperature limit and using the expression ( [ d3b ] ) for @xmath79 , we get in the limit @xmath64 @xmath97 this vanishes faster than @xmath72 as @xmath64 , so there are no gauge - dependent corrections to the singularities of @xmath25 in this case either .
there have been many attempts @xcite-@xcite to go beyond the mean field approximation for the @xmath2 quantum hall state .
some of these attempts @xcite,@xcite,@xcite are based on the behaviour of the singe - particle green function .
one criticism put forward @xcite against arguments based on the single particle green function is that conclusions based on it might not be physical since the single - particle green function is not gauge - invariant .
+ + in this paper we have shown that the dominant singular behaviour determining the pole - structure of the single - particle green function is not affected by the choice of gauge , when the gauge choices are restricted to a particular family of gauges .
this is true both in @xmath45 and in @xmath5 .
+ + this implies in particular that the @xmath0-functions obtained in @xcite,@xcite are independent of gauge , and that the pole - structure used to construct the ( singular ) fermi liquid description in @xcite is also gauge - invariant .
+ i wish to thank sudip chakravarty for his continuous intellectual support during this work , and chetan nayak for helpful comments on the manuscript .
this work was supported by the national science foundation , nsf - dmr-92 - 20416 .
et al . _ ,
b 40 ( 1989 ) 12013 . j.k .
jain , phys .
63 ( 1989 ) 199 .
halperin , p.a .
lee and n. read , phys .
b 47 ( 1993 ) 7312 . r.l .
71 ( 1993 ) 3846 .
phys . rev .
71 ( 1993 ) 3850 .
goldman , b. su and j.k .
jain , phys .
72 ( 1994 ) 2065 .
c.nayak and f.wilczek , nucl .
b 417 ( 1994 ) 359 ; + c.nayak and f.wilczek , nucl
b 430 ( 1994 ) 534 ; + m.onoda , i.ichinose and t.matsui , nucl . phys .
b 446 ( 1995 ) 353 .
s.chakravarty , r.e.norton , o.f.syljusen , phys .
74 ( 1995 ) 1423 ; + phys .
lett . 75 ( 1995 ) 3584 .
a. stern and b.i .
halperin , phys .
b 52 ( 1995 ) 5890 . y.b .
kim , a. furusaki , x .-
wen and p.a .
lee , phys.rev .
b 50 ( 1994 ) 17917 . b.l .
altshuler , l.b .
ioffe and a.j .
millis , phys .
b 50 ( 1994 ) 14048 . y.b .
kim , p.a .
lee and x .- g .
wen , phys .
b 52 ( 1995 ) 17275 . d.v .
khveshchenko and p.c.e .
stamp , phys .
71 ( 1993 ) 2118 .
j. gan and e. wong , phys .
71 ( 1993 ) 4226 .
kwon , a. houghton and j.b .
marston , phys .
73 ( 1994 ) 284 ; + phys .
b 52 ( 1995 ) 8002 .
lee , y. krotov , j. gan and s.a .
kivelson , cond - mat/9607195 .
t. holstein , r.e .
norton and p. pincus , phys .
b 8 ( 1973 ) 2649 . | recently there has been much interest in gauge theories applied to condensed matter physics .
i show that for a system of nonrelativistic electrons coupled to a u(1 ) gauge field in the presence of a fermi surface , the @xmath0-function to one - loop order is , for a particular family of gauge - choices , independent of the gauge - choice . |
in a study of the kinematics and abundances of the outer regions of the galactic bulge ibata , gilmore & irwin ( 1994 , 1995 ) discovered a group of stars with heliocentric radial velocity around 140 kms@xmath3 and a low velocity dispersion , which could not be accounted for by any model of the galactic bulge .
the finding was readily interpreted as a dwarf spheroidal galaxy ( dsph ) which is a satellite of our own galaxy . at a distance of about 25
kpc sagittarius is the nearest external galaxy .
it is undergoing tidal disruption and is probably in the process of merging with the galaxy .
this could in fact be one of the many merger events which have contributed to shape the present day galaxy .
the colour - magnitude diagram of sagittarius suggests a spread in metallicity , which may result from different bursts of star formation .
we have analyzed low resolution spectra ntt+emmi / mos to obtain radial velocities and crude estimates of abundances .
the method we have developed is quite powerful but the low s / n obtainable with ntt on these v@xmath4 stars does not allow any firm conclusion on the chemical history of sagittarius .
the observations described here were taken at eso la silla on june 19th 1996 with the ntt telescope and emmi in multi - object - spectroscopy ( mos ) mode .
we used grism # 5 , which provides a resolving power of about @xmath5 .
the range covered varies slightly depending on the position of the star relative to the center of the field , but the range 480 - 550 nm was available for all the stars .
the spectra are not useful towards shorter wavelengths due to the falling stellar flux and the decreased efficiency of the ccd .
the log of observations is given in table 1 .
details on the observed field can be found in marconi et al ( 1998 ) .
cccc date & # of slits & field & @xmath6 + 1998/06/19 & 27 & sgr 1 & 1800 + 1998/06/19 & 27 & sgr 1 & 2400 + 1998/06/19 & 27 & sgr 1 & 2400 + 1998/06/19 & 27 & sgr 1 & 1380 + 1998/06/19 & 30 & sgr 1 & 3600 +
the spectra were calibrated in wavelength using a he - ar lamp spectrum and zero - point shifts were determined from atmospheric emission lines . to measure radial velocities we used cross - correlation technique in the 480 - 530 nm range . as templates we used synthetic spectra of appropriate temperature and [ fe / h]@xmath7 computed with the synthe code ( kurucz 1993 ) .
we selected as likely sgr members all the stars with @xmath8 ( ibata et al 1997 ) .
this left us with a sample of 23 stars out of 57 observed .
the velocity dispersion of these stars is 23 kms@xmath3 .
since the internal velocity dispersion of sagittarius is only 10 kms@xmath3 ( ibata et al 1995 ) , we assume the dispersion to be dominated by the measurement error of which it may be taken as a somewhat conservative estimate .
the radial velocities are given in table 2 , the stars are identified by their number in the marconi et al ( 1998 ) paper .
for each star we determined the effective temperature from the @xmath9 colour of marconi et al ( 1998 ) and the calibration of alonso et al ( 1996 ) .
although the calibration is only valid for dwarfs , @xmath10 depends little on gravity ; based on theoretical colours computed by castelli ( 1998 ) we expect to make an error less than 100 k , negligible in this context .
gravities were determined from the isochrones of straniero , chieffi & limongi ( 1997 ) of 8 gyr and [ fe / h]@xmath7 .
we developed an iterative procedure for the estimate of abundances .
for each star we started from a guess of the metallicity and computed a synthetic spectrum .
then both the observed and computed spectra were pseudo - normalized using the same pseudo - continuum windows , the observed spectrum was then normalized by multiplying it by the ratio normalized / pseudo - normalized obtained for the synthetic spectrum . on this normalized spectrum
we measured four line indices , two of which measure the mgb triplet ( m1 and m2 ) and four which measure features due to iron - peak elements ( f1,f2,f3,f4 ) .
placement of the indices is illustrated in figure 1 .
the indices were defined so as to work efficiently on the sun .
then for each index we iteratively computed its value on the synthetic spectrum , changing the input abundance until it matched the measured index .
we first iterated only on the f ( iron - peak ) indices at each iteration scaling all of the abundances of the input model .
next we took as scaling the median of the four f indices and keeping this value fixed we iterated on the m indices changing only the mg abundance .
next we computed a new synthetic spectrum using this scaling and mg abundance and re - normalized the observed spectrum on this new synthetic spectrum and iterated the whole procedure until the change in metallicity was less than 0.2 dex .
typically three iterations were enough to meet this criterion .
the procedure is quite cpu intensive and each iteration takes about three hours on the dec alpha of the oat .
the effectiveness of the procedure was tested on the solar spectrum using both the kurucz et al ( 1984 ) solar atlas , degraded at the resolution of our grism spectra and twilight spectra .
we were able to reproduce the solar ( fe / h ) and ( mg / fe ) to within 0.1 dex .
the values determined from the solar atlas for each index were used as zero point , so as to make our analysis strictly differential with respect to the sun .
three matters are of main concern in assessing the results obtained in this way : 1 ) the reliability of the line list we are using .
the good results obtained for the sun do not grant the completeness and correctness of the list , since our program stars are about 1000 k cooler and of much lower gravity .
we shall try to quantify this issue in the future by analyzing the spectrum of a cool giant ; 2 ) the placement of the continuum .
the emmi slitlets have ragged edges which do not allow a flux calibration , this , in conjunction with the low s / n of our spectra may induce an error in the continuum which is difficult to asses ; 3 ) the microturbulent velocity has to be assumed , the results presented here are obtained for @xmath11 .
clearly at this resolution we have no handle on microturbulence what we shall do in a forthcoming paper is to quantify the effect of microturbulence on abundances .
while these factors may affect our analysis in a systematic way ( i.e. we may have a zero point uncertainty in the derived metallicities ) they will not affect the relative abundances of the analyzed stars since the effective temperatures and gravities are quite similar .
ccccccc # & @xmath12 & @xmath13 & @xmath14 & @xmath15 & @xmath16\hfill$ ] & @xmath17\hfill$ ] + & & kms@xmath3 & k & + 105 & 17.55 & 112 & 5041 & 2.59 & @xmath18 & @xmath19 + 115 & 17.84 & 144 & 4953 & 2.36 & @xmath20 & @xmath21 + 124 & 17.00 & 114 & 4891 & 2.21 & @xmath22 & @xmath23 + 128 & 17.40 & 145 & 4778 & 1.93 & @xmath24 & @xmath18 + 139 & 17.78 & 147 & 4891 & 2.21 & @xmath25 & @xmath26 + 141 & 17.45 & 161 & 4977 & 2.42 & @xmath22 & @xmath26 + 142 & 17.55 & 149 & 5118 & 2.81 & @xmath27 & @xmath28 + 201 & 17.54 & 135 & 5003 & 2.49 & @xmath19 & @xmath29 +
the results for eight of the 23 candidate members of sgr are presented in table 2 .
two facts stand out : 1 ) we do not find any star as metal poor as [ fe / h]@xmath30 , as was suggested by the photometry ; 2 ) three out of eight stars are metal rich .
our metallicity scale may still suffer from a zero - point uncertainty which could make the most metal poor stars we observe close to @xmath31 , it is however interesting to note that a mismatch between photometric and spectroscopic metallicities in the same sense as the one we have here has been found by francois et al ( 1997 ) for the young halo globular cluster ruprecht 106 .
investigation of abundances of the whole available sample might allow to establish if we are in the presence of some systematic effect in abundance determinations or if what we see is merely a selection effect .
the presence of the metal - rich stars is disturbing , although they may in fact just be bulge interlopers , three out of eight seems a rather high percentage .
furthermore the metal rich stars do not show sign of alpha element enhancement as that displayed by the bulge k giants analyzed by mcwilliam & rich ( 1994 ) , but are more like the smr stars in the solar neighbourhood
. an independent investigation of metallicities of sgr giants has been carried out by smecker - hane , mcwilliam & ibata ( 1998 ) based on high - resolution spectra obtained at the keck telescope , their results indicate a metallicity @xmath32 for the metal poor stars , lower than our lowest metallicity stars , however two out of seven stars they have analyzed are metal rich and alpha elements o and ca are underabundant with respect to solar , in keeping with our results . probably the best way to decide whether these metal - rich stars belong to the bulge or to sagittarius is to determine the proper motions of the stars in their field .
bulge stars should show larger proper motions while sagittarius stars should show a small and coherent proper motion .
such an investigation is within the capabilities of a diffraction limited instrument , such as hst , over a time baseline of just few years . at the distance of 8.5 kpc a star with a transverse motion of 100 kms@xmath3
would show a proper motion of 2.48 mas / yr . the future for abundance determinations for sagittarius lies in the use of 8 m class telescopes .
the 8.2 m vlt with the fors1 instrument will allow to obtain spectra of these stars with s / n @xmath33 , albeit with a slightly lower resolution .
moreover the slitlets of fors1 in multi - object mode are sharp and will allow flux calibration of the spectra .
our technique of abundance estimation applied to such spectra should then be capable of producing abundances accurate to the 0.3 dex level .
we wish thank f. castelli for providing the theoretical colours and constant support in using the model atmosphere and spectrum synthesis codes and a. chieffi for providing the isochrones used in this work .
part of this work was done while p.b .
was at eso - garching as a visiting scientist .
alonso a. , arribas s. , martinez - roger c. 1996 , aas , 117 , 227 castelli f. , 1998 ( private communication ) francois p. , danziger j.i . , buonanno r. , perrin n.m . , 1997 ,
aa , 327 , 121 ibata r.a . , gilmore g. , irwin m.j . , 1994 , nat , 370 , 194 ibata r.a .
, gilmore g. , irwin m.j .
, 1995 , mnras 277,781 ibata r.a .
, wyse r.f.g . , gilmore g. , irwin m.j . , suntzeff n.b . , 1997 ,
aj , 113 634 kurucz r.l . , 1993 , cd - rom no .
13 , 18 kurucz r.l .
, furenlid i. , brault j. , testerman l. , 1984 , solar flux atlas from 296 to 1300 nm , nat .
obs . , atlas no . 1 marconi g. , buonanno r. , castellani m. , iannicola p. , molaro p. , pasquini l. , pulone l. , 1998 , aa , 330 , 453 mcwilliam a. , rich r.r : , 1994 , apjs , 91 , 749 smecker - hane t. , mcwilliam a. , ibata r.a . , 1998 aas meeting , 192 6613 straniero o. , chieffi a. , limongi m. , 1997 , apj , 490 , 425 | the photometric study of the sagittarius dwarf galaxy by marconi et al ( 1998 ) has suggested the presence of a spread in metallicity ( -0.7@xmath0 [ fe / h ] @xmath0 -1.6 ) , which may result from different bursts of star formation .
we present the results from a multi object slit spectroscopy program carried out at the ntt with the aim to determine spectroscopic abundances of sagittarius giants selected from the ccd photometry . from our intermediate resolution ( r@xmath11500 ) spectra radial velocities are determined to confirm the membership and metallicities are derived by using spectral synthesis codes for stars down to v@xmath118 , i.e. below the rgb clump . out of 57 observed stars 23 have a radial velocity consistent with sagittarius membership , here we present results for 8 of these . no star with
[ fe / h]@xmath2 is found , three stars are found to be metal - rich . |
in the area of nanoelectronics , single electron transistor ( set ) is well known due to its quantised nature of transport since its discovery @xcite that has found application for sensitive charge detections @xcite .
the packing density of the mosfet devices in the present day cmos technology is limited by the gate lengths accessible using various lithography techniques and the tunnelling issues . as a result of which the industry is constantly looking for useful alternatives in next generation technology to achieve faster switching performance and speed .
set s can be useful in this direction due to its small size and low power operation and performance that can be easily integrated towards large scale fabrication in future .
while the conventional metallic set s were made using a metallic island artificially placed between the s / d electrodes , in the recent times growing interest was found in replacing the metallic island by an organic molecule and hence the area of molecular electronics was developed that has made significant progress towards developing single molecular devices ( smt ) with the organic molecules as the active component of the circuit @xcite .
operation of such devices have been in presence for the past years which was first demonstrated by reed @xmath1 @xcite using benzene-1,4-dithiolate as the active molecule .
experimental realisations of such devices were made successfully in carbon based systems @xcite , benzene @xcite , oligophenylenevinylene ( opv ) @xcite , fullerene @xcite , dipyridylamide @xcite and to design logic circuits @xcite etc .
however , understanding the detailed nature of electronic transport in such devices in still under study and has not been well understood in details . in the strong coupling ( sc )
limit , the conduction behaviour is usually dominated by the coherent electron transport that often overestimates the current / conductance and in the weak coupling ( wc ) limit , the sequential nature of transport is governed by the orthodox theory of coulomb blockade that underestimates the value of the energy gap @xcite . while the transport behaviour is usually estimated semi - empirically using a combination of non - equilibrium green s function formalism ( negf ) and density functional theory ( dft ) @xcite in the sc limit , the incoherent transport behaviour in the wc limit was formalised by a recent approach proposed by kaasberg @xmath2 @xcite and strokobro @xmath1 @xcite has introduced this within a dft framework for the estimation of various energy levels .
this approach has been used computationally in the recent times to estimate the charging energy of systems based on benzene @xcite , fullerene @xcite , napthalene @xcite , dna chain @xcite etc . which has found excellent agreement with the experimental results .
due to the incoherent nature of transport , a smt is incredibly sensitivity to its charge state similar to a set .
while an incoming charge is capacitively coupled to the smt molecule , any change in the charge state can be sensed through the charge stability diagram . in a conventional smt
, the gate electrode is placed in very close proximity of the island ( or the molecule ) to control the chemical potential independently . here in this present case
, we have modified this geometry to introduce an additional gate electrode on top of the molecule to achieve better electrostatic control over the device and to find its influence on the energy levels of the molecule and to demonstrate its usefulness as a sensitive charge detector .
in this present work , the active component of the the system under investigation is an 1,3-cyclobutadiene molecule working as the ` island ' in the smt devices .
the molecule has a square central structure with 4 hydrogen atoms placed symmetrically at each corners of the carbon atoms as illustrated in fig .
[ fig.1](a ) .
the small and planar structure of cyclobutadiene is advantageous in such a device as the influence of electrical polarisation is significantly less than the direct gate - molecule coupling that determines the performance of a transistor . for the single and double gated smt @xcite as illustrated in fig .
[ fig.1](b),(d ) , the molecule was placed symmetrically between the source ( s ) and drain ( d ) electrodes on top of a dielectric slab with the molecular plane lying parallel to the dielectric surface . for the single gated device ,
the dielectric layer is of thickness ( @xmath3 ) = 3.7 and with a dielectric constant of 10@xmath4 which is connected to a metallic backgate electrode of thickness 1 as illustrated in fig . [
fig.1](b ) . in the case of the double gated device
( see fig .
[ fig.1](d ) ) , an additional top gate of thickness 1 was introduced to the device between the top part of the source / drain(s / d ) electrodes connected to the top of an additional dielectric slab of thickness ( @xmath5 ) = 4 .
the spatial separation between the molecule and both the dielectric layers is @xmath6 1 .
gold was chosen as the generic electrode material to ensure minimum contact resistance and good electrical conductivity at this temperature of operation .
the equivalent capacitance network for the single and double gated devices were illustrated in fig .
[ fig.1](c ) and fig .
[ fig.1](e ) respectively which describes the nature of electrostatic coupling between the molecule and different electrodes .
details of the capacitance estimation methods were described at a later stage of this article .
the charging energies of this system was estimated by performing ab - initio calculations based on stokbro s method @xcite using the commercial atomistic toolkit package ( atk 11.2.3 ) made by the quantum wise division @xcite .
this self - consistent calculations were performed using a nonspin - polarised dft framework under the local density approximation ( lda ) where the wave functions were expanded in double-@xmath7 polarised ( dzp ) basis set .
considering the metal electrodes as equipotential surfaces , neumann boundary conditions were used while solving the poisson s equation assuming the perpendicular component of the electric field stays at zero at the interfaces , details of which could be found here @xcite .
the island - s / d electrode separation was chosen to be large enough so that calculations can be performed in the weak coupling limit as it is possible to estimate the molecular energy states accurately when incoming electron gets sufficiently enough time to stabilise on the molecule , hence carrying no information about its initial or final states . in such a device , electronic transport is only possible when an electron can be moved from the highest occupied molecular orbital ( homo ) level to the lowest occupied molecular orbital ( lumo ) level and the minimum energy needed for this is called as the additional energy ( @xmath8 ) which can be expressed by , @xmath9 where @xmath10 is the number of electrons in the neutral state of the molecule .
in fig . [ fig.2](a ) , the total energy of the molecule in the smt environment has been estimated for different charge states ( @xmath11 ) as function of the backgate voltage ( @xmath12 ) for a fixed top - gate voltage @xmath13 = -8v .
a part of the total energy ( @xmath14 ) is contributed by the metallic reservoir potential (= @xmath15 ) where @xmath16 is the work function of the metal electrode , which for the present case of au electrode is 5.28 ev .
for all the charge states , @xmath14 showed monotonic dependence with the gate excitation . for the positive charge states ,
the total energy gets reduced at negative gate voltages and the situation is opposite for negative charge states where positive gate voltage reduces @xmath14 .
this happens due to the stabilisation of the positive charge states at @xmath17 and vice - versa for the negative charge states at @xmath18 which can be explained from the relative orientation of the molecular energy levels for different backgate voltages . under the application of a @xmath18
, the lumo level goes below the fermi level ( @xmath19 ) of the electrode and this allows the transfer of an additional electron to the molecule and it becomes more negatively charged .
however for @xmath17 , the homo level gets shifted towards a higher energy level and in this case , an electron moves from the molecule to the electrode making it more positively charged . for investigating further dependence of @xmath14 with @xmath12 ,
the energy can be analytically fitted by a @xmath20 order polynomial as , @xmath21 where coefficients @xmath22 and @xmath23 are the fitting parameters which for the present case were estimated to be @xmath22 = 0.224 and @xmath23= 0.01 ev@xmath24 .
the @xmath20 term in the eqn .
[ eqn.2 ] is proportional to @xmath11 which is due to the strong coupling of the backgate electrode to the molecule and the value of @xmath22 is a measure of the strength of coupling between them .
the final term in eqn . [ eqn.2 ] indicates the influence of the electrical polarisation of the molecule and is independent of the charge states .
since the molecule stays flat parallel to the dielectric surface and hence the difference of the electric field experienced by different atoms is minimal and this term has a smaller contribution compared to the @xmath20 term as supported by @xmath25 .
the values of @xmath22 and @xmath23 stay almost constant for different charge states and the dependence of @xmath26 with the charge states was illustrated in fig .
[ fig.2](b ) . in the absence of any @xmath12
, @xmath26 decreases with an increase in @xmath11 initially for @xmath27 as removing an electron at @xmath28 is energetically more expensive comparerd to the situation when @xmath29 .
however , the rate of this change in @xmath26 , @xmath30 which can be estimated from the slope as plotted in fig . [ fig.2](c ) is linear in @xmath11 .
the linear slope indicates that only in the presence of a @xmath13 , adding or removing a charge from the molecule at a specific state will have identically opposite influence on the zero - term in energy . for investigating further dependence of the total energy on @xmath12 and the conduction behaviour , the charge stability diagrams ( also called as coulomb diamond plot )
have been plotted in fig .
[ fig.3 ] .
as the smt operates in the weak coupling regime , electron transport between the s / d electrodes is only possible when the molecular energy levels are accessible within the applied bias ( @xmath31 ) window .
details of this conduction regime could be found from the charge stability diagram as illustrated in fig .
[ fig.3](a ) which was calculated for @xmath13 = -8v .
the colourbar on the right represents the number of available energy levels for conduction for different @xmath31 and @xmath12 .
experimentally the charge stability diagram is measured by measuring the source - drain current and the @xmath32-axis is represented by the current , conductivity or differential conductance .
the diamonds indicate the accessible regions of conduction to separate the non - conducting and conducting regions for different values of @xmath31 and @xmath12 . in fig .
[ fig.3](a ) , the central region of the large diamond surrounded by points ` a ' , ` b ' , ` c ' , ` d ' marks the key region of conduction within which sequential tunnelling between different charge states occurs in this device . in fig .
[ fig.3](b ) , this region was enlarged to describe the behaviour of the ground state excitation of the molecule .
the central diamond in fig .
[ fig.3](b ) corresponds to the neutral charge state of the molecule in its ground state with a hypothetical @xmath33 electrons in it . within each of these diamonds ,
the population does not change on the molecule , however , moving from one of these regions to its left or right will result in changing the population . the charging energy of the ground state can be estimated from the height of the central diamond ( marked by dotted arrow in fig .
[ fig.3](b ) ) which is @xmath34 ev and the height of its ` left ' and ` right ' diamonds are @xmath35 ev and @xmath36 ev respectively which refers to the ` cationic ' and the ` anionic ' states of the molecule respectively .
the additional energy needed for creating the ` cationic ' and the ` anionic ' states can also be estimated from eqn .
[ eqn.1 ] . the smaller value of the charging energy of these two excited states can be accounted for their lower occupation in their frontier orbitals indicating that adding / removing an electron from the excited states are much energetically favourable than the neutral state .
the slight difference between the charging energies of the two excited states are due to the differences in their homo - lumo gaps which contributes a part in addition to the coulomb energy . from the charge stability diagram
, further information about the detailed electrostatics of the system could be obtained about the smt .
the values of the junction capacitances as indicated in fig .
[ fig.1](c),(e ) could be obtained from the ` abcd ' region of the diamond as indicated in fig .
[ fig.3](a ) and their estimated values were listed in table . [ tab.1 ] .
the source and drain capacitances i.e. @xmath37 and @xmath38 respectively were estimated from the slopes of the lines ` ad ' and ` ac ' .
the slope of the line ` ad ' is @xmath39 and the slope of ` ac ' = @xmath40 where @xmath41 is the capacitance between the ` dot ' and the ` backgate ' .
the gate capacitances were estimated analytically by considering a planar approximation of the gate electrodes in a parallel plate geometry which can be expressed by : @xmath42 and @xmath43 where @xmath44 is the relative permeability of the dielectric layer , @xmath45 is the capacitance between the ` dot ' and the ` topgate ' , @xmath3 and @xmath5 are the thicknesses of the ` bottom ' and ` top ' dielectric layers respectively and @xmath46 , @xmath47 are the areas of contacts between the ` bottom ' and ` top ' gates with their respective dielectric layers in contacts . in fig .
[ fig.3](c ) , a horizontal linescan was taken along the dotted line ` cd ' as represented in fig .
[ fig.3](a ) .
the appearance of sharp peaks represent the points when the conduction window changes form a certain diamond to the neighbouring one and hence the change of the ` dot ' population by 1 between each of the neighbouring diamonds , from left to right for increasing @xmath12 . at low bias voltage ( @xmath31 )
, the excitations only occur from the ground state to the first excited state which is why the peaks arise at values @xmath48 for all values of @xmath12 .
another vertical line scan taken along ` ab ' ( as illustrated in fig .
[ fig.3](d ) ) showed occurrence of periodic plateaus symmetrically placed on both sides of @xmath49 v. the step like structure arises only at finite values of @xmath11 as a result of charging between different charge states and the charging energies can be estimated from the width of the respective steps . in the coulomb blockade regime ,
the minimum occours at @xmath31 = 0 v in the absence of any excitation and subsequent enhancement of @xmath31 results in moving to the higher excited states .
the occurrence of the plateaus can be understood in details from the differential ( @xmath50 ) charge plot which is similar to the behaviour observed from the differential conductance ( di / dv ) measured in an experiment as illustrated in the bottom panel of fig .
[ fig.3](d ) .
the peak positions observed in fig .
[ fig.3](d ) [ bottom panel ] represent the values of @xmath31 at which charging to an excited state occurs in the absence of continuous transport when the tunnelling rate ( @xmath51 ) is very low .
the separation between two neighbouring peaks in fig .
[ fig.3](d ) [ bottom panel ] represents the charging energy of the molecule between subsequent excited states for a given @xmath12 which is not uniform between all the charge states . to find the systematic dependence of the @xmath13 in such devices , the charging energy ( @xmath52 ) for the neutral case in the ground state of the molecule
was compared for different values of @xmath13 .
the minimum in @xmath52 occours at @xmath13 = -8v and it increases with an increase in the @xmath13 .
reduction of the @xmath52 occurs when the molecule is bought from its gas phase to a smt environment due to the polarisation of its charge states in the presence of metallic electrodes . under the addition
a top electrode , this effect gets enhanced and the non - linear dependence of @xmath53 indicates the influence of additional polarisation in the presence of the top electrode and a positive @xmath13 induces additional polarisation to the molecule in its ground state . it is to be noted that the vertices of the large diamond enclosed by ` abcd ' ( see fig . [ fig.3](a ) ) does not stay at the same position with a change of @xmath13 .
the position of the line ` ab ' can be referenced to investigate this dependence .
since the @xmath12 stays the same for both ` a ' and ` b ' , the @xmath54coordinate which is the value of @xmath12 , can be considered as a reference point . in fig .
[ fig.4](b ) , such values of @xmath12 from the line ` ab ' was plotted as function of @xmath13 which indicates an almost linear dependence between them . in fig .
[ fig.4](b ) , the dependence between the two gate voltages was plotted to estimate the nature of coupling between them through the molecule .
the slope of this line is @xmath55 , which is the same as the ratio of the separation between the two gates from the molecule i.e @xmath56 = 1.063 .
the slope @xmath6 1 of the line indicates the almost identical electrostatic coupling of the gate electrodes to the molecule .
the electrostatic energy landscape was plotted as a 3d surface as functions of @xmath12 and @xmath13 in fig .
[ fig.4](c),(d ) . for @xmath57 ,
the surface is almost symmetrically curved around the red dotted line as illustrated in fig .
[ fig.4](c ) .
when the molecule is in a neutral charge state , maximum of the energy occurs due to the polarisation of the molecule when both the gates are at equal values ( ` red ' dotted line in fig .
[ fig.4](c ) ) and when they are of opposite signs , total energy gets reduced as one of two gate voltages increases the polarisation while the other counter balances it . for this reason , the minimum of the energy can be found at the two extremes ( @xmath13 = + 8v , @xmath12 = -8v ) and ( @xmath13 = -8v , @xmath12 = + 8v ) for @xmath58 . for higher charge states ,
the situation changes significantly which can be seen from fig .
[ fig.4](d ) . for @xmath59 ,
the maximum in energy occurs at ( @xmath13 = -8v , @xmath12 = -8v ) and minimum on the other end at ( @xmath13 = + 8v,@xmath12 = + 8v ) when the charge state is more stabilised . for @xmath60 ,
the situation is opposite which has minimum at @xmath13 = -8v , @xmath12 = -8v and maximum at @xmath13 = + 8v , @xmath12 = + 8v .
the nature of the energy surfaces is different for different charge states and starting from any point of any of these landscapes , energy sharply changes with a change of @xmath12 and @xmath13 which makes it sensitive for detection of an incoming charge state .
the steep slope of the energy surfaces indicates that the energy state of the smt is highly sensitive to a small change of the gate voltages and this property can be used as a sensitive charge detection technique . due to the sequential nature of transport
, the change of energy can be translated directly to the change of current flowing through the device ( @xmath61 ) which in the case of a real experimental device is expected to demonstrate an identical sensitivity .
the difference in the nature of the energy surfaces for different charge states thus allows to detect an incoming charge state uniquely in such a device and the addition of a top gate proves to be extremely useful for investigating such behaviour .
.junction capacitances for single and double gated devices [ cols="^,^,^,^,^",options="header " , ]
in this work , we have performed computational investigation of a smt with an 1,3-cyclobutadiene molecule working as the ` quantum dot ' . unlike the conventional case of a single gated device
, we have found better control of the electrostatics and performance by adding a top gate electrode .
the performance of such a device was investigated by calculating the energy of the molecule for different gate voltages which were used later to construct the charge stability diagram to understand the detailed nature of conduction .
the molecular energy states are highly sensitive to the gate voltages that changes significantly for different charge states and this unique behaviour can be utilised for using it as a sensitive organic charge detector .
these kinds of single molecular devices could be the ideal candidates for future generation nano - electronic devices as charge sensors for faster operational speeds and portability . unlike the set devices , these smt s are operational at room temperature which is a genuine advantage of these molecular devices for operational usefulness and industrial applications .
sr acknowledges the support provided through a research excellence grant ( reg ) within the joint research project `` qu - ampere '' ( sib07 ) supported by the european metrology research programme ( emrp ) .
the emrp is jointly funded by the emrp participating countries within euramet and the european union . for v@xmath63 = -8 v for different charge states of the molecule .
the points are the calculated values and the dotted lines represent the fitted values using eqn .
[ eqn.2 ] , ( b ) @xmath26 estimated from the fits above as a function of the charge state @xmath11 at v@xmath63 = -8 v , ( c ) differential of the zero - term energy ( @xmath30 ) as function of @xmath11 , showing a linear dependence .
the ` green ' line is the fit to the data points in ` red'.,width=321 ] | the electrostatic behaviour of an 1,3-cyclobutadiene ( c@xmath0h@xmath0 ) based single molecular transistor ( smt ) has been investigated using the first principle calculation based on density functional theory and non - equilibrium green s function approach . while the molecule is placed on top of a dielectric layer ( backed by a metallic gate ) and weakly coupled between the source / drain electrodes , the charge stability diagram revealed the presence of individual charge states in the coulomb blockade regime .
this gets affected significantly on addition of an another gate electrode placed on the top of the molecule .
this modified double - gated geometry allows additional control of the total energy of the system that is sensitive to the individual charge states of the molecule which can be used as a charge sensing technique operational at room temperature . |
local galaxies are highly clustered . on large scales
they are organized into a network of sheets and filaments which surround large underdense regions , usually referred to as voids . on smaller scales galaxies
are found in gravitationally - bound groups and clusters . according to the standard theoretical paradigm
, the structures observed today were formed by the gravitational amplification of small perturbations in an initially gaussian dark matter density field .
small scale overdensities were the first to collapse , and the resulting objects subsequently merged under the influence of gravity to form larger structures such as clusters and superclusters .
galaxies formed within dense _ halos _ of dark matter , where gas was able to reach high enough overdensities to cool , condense and form stars .
in this hierarchical formation picture , the clustering of the dark matter , as measured by the amplitude of the matter correlation function @xmath8 , increases monotonically with time .
the precise evolution of @xmath8 with redshift has been studied extensively using both n - body simulations ( e.g. jenkins et al .
1998 ) and analytic methods ( hamilton et al 1991 ; peacock & dodds 1994 ; jain , mo & white 1995 ) .
if @xmath9 were observable , it would be straightforward to use its behaviour to determine @xmath0 , @xmath1 and the power spectrum of linear density fluctuations .
what one measures in practice , however , is the clustering of galaxies , and the interpretation then requires an understanding of how these objects trace the underlying dark matter density field .
if galaxies form at the centre of dark matter halos , considerable insight may be gained by using n - body simulations to study the clustering of halos ( brainerd & villumsen 1994 ; mo & white 1996 ; roukema et al 1997 ; jing & suto 1998 ; wechsler et al 1998 ; bagla 1998a , b ; ma 1998 ) .
mo & white ( 1996 ) tested an approximate analytic theory against their numerical results , and this theory and its extensions can also be used to analyse the evolution of halo clustering with redshift ( matarrese et al 1997 ; coles et al 1998 ) .
an important conclusion from all these studies is that the clustering of halos of galactic mass ( @xmath5 ) evolves much more slowly than the clustering of the dark matter .
this is because at high redshifts , such halos correspond to rare peaks in the initial density field , and are thus more strongly clustered than the dark matter ( kaiser 1984 ) .
another important conclusion is that more massive halos are more strongly clustered than less massive halos .
if the luminosity of a galaxy is correlated with the mass of its halo , more luminous galaxies ought to be more strongly clustered .
a detailed comparison with observational data requires a model for the observable properties of the galaxies present within halos of given mass at each epoch .
precise measurement of the clustering amplitude of galaxies at high redshift has just recently become feasible .
usually this is done by calculating the angular two - point correlation function @xmath10 as a function of apparent magnitude . in order to assess how clustering has evolved
, @xmath10 must be deprojected using limber s equation under the assumption of some specific model for the redshift distribution of the observed galaxies . in future large surveys of faint galaxies with photometric and/or spectroscopic
redshifts will be available ( see , for example , connolly et al 1995 )
. it will be possible to classify the galaxies in these surveys according to absolute magnitude , spectral type , star formation rate and colour , and to investigate how clustering evolution depends on these properties . at present , most of the data indicate that the clustering amplitude of galaxies _ decreases _ from @xmath3 to @xmath4 .
different analyses , however , yield very different estimates for the strength of this decrease . le fvre et al ( 1996 ) analyzed the clustering of 591 galaxies with @xmath11 in the five 10 arcminute fields of the cfrs survey .
they find that clustering has evolved dramatically , quoting a comoving correlation length at redshift 0.5 of @xmath12 h@xmath13 mpc ( @xmath14 ) .
more recently , carlberg et al ( 1998 ) presented a preliminary analysis of clustering in the cnoc2 field galaxy redshift survey .
their sample is spread over four patches of sky with a total area of 1.5 square degrees .
they estimate that the comoving correlation length of galaxies with @xmath15 evolves as @xmath16 h@xmath13 mpc .
this is much weaker than the evolution found for the cfrs galaxies .
the results of carlberg et al ( 1998 ) agree reasonably well with those of postman et al ( 1998 ) , based on 710 000 galaxies with @xmath17 from an imaging survey of a contiguous 4 square degree region of the sky .
the latter authors suggest that the small volumes sampled by the cfrs and other early surveys resulted in their derived correlation lengths being biased low .
the clustering of lyman break galaxies at @xmath18 has now been measured with surprising precision ( steidel et al 1998a ; giavalisco et al 1998 ; adelberger et al 1998 ) .
the comoving correlation length of these objects is comparable to that of @xmath7 galaxies today , implying , as expected from models of halo clustering , that lyman break galaxies are highly biased tracers of the dark matter distribution at these redshifts .
in addition , giavalisco et al ( 1998 ) find that the fainter lyman break galaxies are less strongly clustered .
this accords well with a simple model in which the star formation rates in these objects increase with the mass of their halos .
more detailed theoretical modelling of the observed properties of lyman break galaxies at @xmath19 , including analysis of their abundances , sizes , luminosities , colours , star formation rates and clustering properties , has been carried out by mo & fukugita ( 1996 ) , baugh et al ( 1998 ) , governato et al ( 1998 ) , somerville , primack & faber(1998 ) and mo , mao & white ( 1998 ) in this paper , we combine cosmological n - body simulations and semi - analytic modelling of galaxy formation to study the evolution of galaxy clustering as a function of redshift . our methods for incorporating galaxy formation in the simulations are discussed in detail in kauffmann et al ( 1998 , paper i ) .
two variants of a cold dark matter ( cdm ) cosmology are analyzed here : a high - density model with @xmath20 , @xmath21 and @xmath22 km s@xmath23 mpc@xmath13 ( @xmath24cdm ) , and a low - density flat model with @xmath25 , @xmath26 and @xmath27 km s@xmath13 mpc@xmath13 ( @xmath1cdm ) .
paper i was concerned with the global properties of the galaxy distribution at @xmath3 , including b and k - band luminosity functions , the i - band tully - fisher relation , galaxy two - point correlation functions , colour distributions , star formation rate functions and peculiar velocity distributions . here
we focus on clustering evolution in the two models .
we study the predicted differences in clustering evolution for galaxies of different magnitude , type and star formation rate , and we outline how future observational data will clarify the galaxy formation process .
because galaxy formation is complex and involves many poorly - understood physical processes , for example , star formation and radiative and hydrodynamical feedback , it is worthwhile to ask whether the clustering of dark matter halos can be used to constrain cosmological parameters directly . in figure 1
we plot the correlation length @xmath28 as a function of redshift for halos of different mass in our two simulations . here , as in the rest of the paper ,
all length scales are expressed in comoving units .
since the correlation functions in the models are not exact power laws , we define @xmath28 as the radius where @xmath29 .
the smallest halos resolved in the simulations contain 10 particles and have virial masses @xmath30 .
for these objects , @xmath28 initially decreases with redshift , reaches a minimum , and then increases again .
the redshift of this minimum is different for the two cosmologies : @xmath31 for @xmath24cdm and @xmath32 for @xmath1cdm .
massive halos do not exhibit the same `` dip '' in correlation length ; their @xmath28 remains constant for a while , then increases at high redshift .
once again , the redshift at which the evolution becomes strong is lower for @xmath24cdm than for @xmath1cdm .
this is simply because structure formation occurs _
later _ in the @xmath24cdm model . with 10 metre telescopes , it is now possible to measure the rotation curves of disk galaxies at redshift @xmath33 ( vogt et al 1996 ) .
it will be many years , however , before such samples are both large enough and complete enough for an analysis of the clustering evolution of galaxies as a function of their halo mass . in all likelihood
, we will have to deal with flux - limited surveys of galaxies for some time to come .
let us now make the simplifying assumption that each simulated dark matter halo contains one observable galaxy , and that the luminosity of the galaxy increases with the mass of its halo .
the correlation function of a flux - limited sample of galaxies _ of known abundance _ at redshift @xmath34 may then be calculated by evaluating @xmath35 for the mass - limited set of simulated halos which has the same abundance .
this is illustrated in figure 2 , where we plot the correlation length of halos versus their number density ( in units of h@xmath36 mpc@xmath37 ) at a series of redshifts .
as expected , @xmath28 decreases as the number density increases , because the correlation signal becomes dominated by low - mass halos , which are more weakly clustered .
note that the differences between the @xmath24cdm and the @xmath1cdm models are small at all redshifts .
mo , mao & white ( 1998 ) show a similar plot for halos at @xmath38 for four different cdm cosmologies and find that they all give similar results
. there thus appears to be a `` cosmic conspiracy '' that makes it impossible to infer information about cosmological parameters from the clustering of halos of given abundance . on the other hand
, the uniformity seen in figure 2 can be used as a _ test _ of the entire class of hierarchical models and of the hypothesis that there is a one - to - one correspondence between halo mass and galaxy luminosity . as discussed by steidel et al ( 1998b ) and by mo , mao & white ( 1998 ) , this hypothesis works well for the lyman break population at @xmath18 . at low redshifts , the assumption of a one - to - one correspondence between halos and galaxies must break down , because the abundance of high mass halos is larger , and more and more bright galaxies are grouped together in each such halo .
the values of @xmath28 plotted in figure 2 are then lower limits on the true values .
in this section , we study the evolution of galaxy clustering in the @xmath1cdm and @xmath24cdm simulations .
we use the star formation and feedback recipes that resulted in the best fits to the observational data at @xmath3 .
as discussed in paper i , extremely efficient feedback was required in the @xmath24cdm model in order to obtain a reasonable fit to the correlation function on scales below 1 h@xmath13 mpc and to avoid producing too many galaxies with luminosities below @xmath39 .
even so , the model failed to fit the observed bright end of the luminosity function and the clustering amplitude was too low on large scales .
the @xmath1cdm model with relatively inefficient feedback resulted in a better overall fit to most of the data at @xmath3 . for simplicity
, we do not consider dust extinction in the analysis of this paper because it is very uncertain how the empirical recipes we adopted in paper i should be extended to high redshift .
this neglect has little effect on our @xmath1cdm model but means that our @xmath24cdm model now substantially underpredicts galaxy clustering at @xmath3 .
we concentrate below on the relative evolution of @xmath28 rather than on its absolute value , so this problem does not strongly affect our conclusions . in figure 3 results are shown for galaxies with rest - frame b - band magnitudes brighter than @xmath40 in the @xmath1cdm simulation . at redshift zero , this corresponds to selecting galaxies brighter than @xmath6 .
the first three panels in the plot show the evolution of @xmath35 evaluated at @xmath41 2 , 3 and 8 h@xmath13 mpc ( co - moving units ) .
the fourth panel shows the evolution of the co - moving correlation length @xmath28 . for comparison ,
the dotted line in each panel shows the evolution of the corresponding quantity for the dark matter .
results for @xmath24cdm are given in figure 4 .
in each case the redshift extends to the point at which the abundance of @xmath7 galaxies becomes too low for reliable estimation of the correlation function . in the @xmath1cdm model ,
the clustering amplitude decreases from @xmath3 to @xmath42 , remains approximately constant from @xmath42 to @xmath43 and then increases again at higher redshift .
the dip in clustering amplitude is stronger on small scales : @xmath35 decreases by a factor of 3 at 2 h@xmath13 mpc and by a factor 1.5 at 8 h@xmath13 mpc .
the correlation length @xmath28 decreases from 5.5 h@xmath13 mpc at @xmath3 to 3.9 h@xmath13 mpc at @xmath42 .
this agrees remarkably well with the parametrization of @xmath28 as a function of @xmath34 quoted by carlberg et al ( 1998 ) . in the @xmath24cdm model
the clustering amplitude remains fixed from @xmath3 to @xmath44 and then rises steeply at higher redshifts .
note , as mentioned above , that the correlation length at @xmath3 is low ( @xmath45 h@xmath13 mpc ) in this model . in figure 5
, we plot the evolution of the bias @xmath46 , defined as the square root of the ratio between the galaxy and the dark matter correlation functions : @xmath47 the four lines on the plot show the bias as a function of redshift evaluated at @xmath41 2 , 3 , 5 and 8 h@xmath13 mpc .
the bias does not depend on @xmath48 in either model , except at very high redshifts where @xmath46 is somewhat larger on small scales .
the bias evolves much more rapidly in the @xmath24cdm model than in the @xmath1cdm model .
galaxies in the @xmath1cdm model are unbiased tracers of the mass out to @xmath49 .
galaxies in the @xmath24cdm model have bias values of about 2 at this redshift .
in figures 6 and 7 , we demonstrate that the clustering evolution depends on the way in which galaxies are selected in the simulations .
the first panel compares the clustering evolution of galaxies selected in the rest - frame b - band with that of galaxies selected in the rest - frame i - band .
the second panel compares the clustering evolution of galaxies with @xmath50 with that of galaxies 1.5 magnitudes brighter .
the third panel shows what happens if galaxies are selected by star formation rate rather than by luminosity .
the fourth panel shows the clustering of early - type galaxies with stellar mass greater than @xmath51 ( recall from paper i that these objects form by mergers of two galaxies of similar mass and have @xmath52 mag ) . in the @xmath1cdm model
, we find that the strength of the `` dip '' in clustering between @xmath3 and @xmath42 is sensitive to sample selection .
the dip is weaker for more luminous galaxies and for galaxies selected in the i - band , but stronger for galaxies selected by star formation rate .
the clustering amplitude of early - type galaxies is stronger than that of the population as a whole and evolves very little with redshift . in the @xmath24cdm model ,
clustering evolution is less sensitive to sample selection .
the clustering always remains fixed out to @xmath49 and then rises steeply at higher redshifts .
early - type galaxies are also very strongly clustered in this model , particularly at high redshifts .
we now explain _ why _ the evolution of clustering depends on sample selection and _ what _ can be learned about galaxy formation by studying the observed evolution as a function of morphological type , luminosity , colour and star formation rate . for a given cosmology ,
the clustering amplitude predicted for a sample of galaxies depends on the masses of the dark matter halos they inhabit .
the evolution of clustering depends on how the mass distribution of these halos changes with redshift .
additional relevant and observationally accessible information comes from the variation with redshift of the abundance of galaxies in the sample . as an example , let us suppose that galaxies with fixed star formation rate are found in smaller halos at high redshift than at the present day .
we would expect the clustering amplitude of a sfr - selected sample to show a stronger dip than a sample of galaxies that tracked halos of the same mass at all redshifts .
we would also expect the abundance of galaxies in a sfr - selected sample to increase more strongly with redshift , because there are many more small halos than large ones . as a second example , let us suppose that early type galaxies are found primarily in massive halos at all redshifts . as seen in figure 1 , these galaxies should not exhibit any dip in clustering and their abundances should decrease strongly at high redshifts because massive halos are rare objects at early times .
these points are illustrated in detail in figure 8 , where we plot the evolution of the median halo mass and the comoving number density of galaxies in samples selected in different ways from the @xmath1cdm simulation .
the top 3 panels show results for galaxies selected according to rest - frame b - magnitude , rest - frame i - magnitude and star formation rate . at @xmath3 ,
all three galaxy samples have the same abundance and occur in halos of roughly the same mass .
galaxies selected according to star formation rate move to smaller halos at higher redshift .
this effect is simply a result of the parametrization of star formation in our models .
following kennicutt ( 1998 ) , we have adopted a star formation law of the form @xmath53 , where @xmath54 is the mass of cold gas in the galaxy and @xmath55 is the dynamical time of the galaxy .
since @xmath55 decreases at higher redshifts , the star formation rates are higher in halos of the same cold gas content .
galaxies selected in the b - band exhibit a weaker trend towards low - mass halos . in the case of the i - band selection , galaxies trace halos of roughly the same mass at all redshifts below 2 .
this is because the i - band magnitude of a galaxy is a measure of its total stellar mass , rather than its instantaneous star formation rate .
we thus conclude that sfr - selected samples show the strongest dip in clustering in figure 6 because this selection procedure favours galaxies in lower mass halos at high redshift .
note that galaxies in the sfr - selected samples also exhibit the strongest increase in abundance from @xmath3 to @xmath42 .
the bottom two panels in figure 8 show that very bright galaxies and early - type galaxies in the simulation are found in halos with masses @xmath56 .
as seen in figure 1 , the clustering of these objects evolves very little from @xmath3 to @xmath4 . in figure 9
, we show the evolution of galaxy abundances and median halo masses for samples selected in the same way from the @xmath24cdm simulation .
the results are qualitatively similar to those found for @xmath1cdm .
@xmath7 galaxies occur in halos of roughly the same masses ( @xmath5 ) in both models .
the reason why no dip is seen in the @xmath24cdm model is because halos of these masses are more strongly biased at @xmath4 than in the @xmath1cdm model and the decrease in halo mass with redshift for the sfr - selected sample is less pronounced .
note also that the redshift at which the abundance curves peak is higher for @xmath1cdm than for @xmath24cdm . in the @xmath1cdm simulation ,
the abundance of early - type galaxies only decreases substantially at redshifts greater than 1.5 , whereas in the @xmath24cdm simulation , the abundance of ellipticals has already declined by a factor of 3 by @xmath4 .
figure 10 shows the evolution of the slope @xmath57 of the two - point correlation function for galaxies with rest - frame b - magnitudes brighter than @xmath58 in the @xmath1cdm and @xmath24cdm models .
we have fit a power - law to @xmath35 over three different ranges in scale : @xmath59 h@xmath13 mpc , @xmath60 h@xmath13 mpc and @xmath61 h@xmath13 mpc . in the @xmath1cdm model ,
the evolution of the slope is stronger on small scales . over the range @xmath62 h@xmath13 mpc
, @xmath57 evolves from @xmath63 at @xmath3 to @xmath64 at @xmath4 . on large scales , @xmath57 remains approximately constant . over the range @xmath65 h@xmath13 mpc
, @xmath57 evolves from @xmath66 at @xmath3 to @xmath67 at @xmath4 and then remains constant .
these results appear to be in qualitative agreement with the observations ( postman et al 1998 ) .
these au thors find no dependence of @xmath57 on magnitude for the bright ( @xmath68 ) galaxies in their survey . at fainter magnitudes @xmath57 flattens , reaching a value of @xmath67 at i=22.5 .
they also find that the flattening is stronger on smaller angular scales .
neuschaeffer & windhorst ( 1995 ) find similar results from an independent survey carried out at a different wavelength . in the @xmath24cdm model
, there is very little change in the slope with redshift on any scale .
estimates of dynamical quantities such as @xmath0 or cluster m / l ratios from galaxy data require knowledge not only of the spatial bias in the galaxy distribution , but also of any possible bias in the kinematics of the galaxies relative to those of the dark matter .
in paper i we explored this `` velocity bias '' for our @xmath3 models using pairwise velocity statistics , and in paper iii ( diaferio et al 1998 ) we will do the same using group and cluster velocity dispersions . here we briefly explore the predicted evolution of velocity bias using pairwise statistics .
the thin solid lines in figure 11 show the redshift evolution of the pairwise peculiar velocity dispersion @xmath69 evaluated at relative separations @xmath41 0.5 , 1 and 2 h@xmath13 mpc ( co - moving units ) for galaxies with rest - frame b - band magnitudes less than @xmath70 in the @xmath1cdm and @xmath24cdm simulations .
the thick solid lines show the evolution of @xmath69 for the dark matter . in order to compare the _ relative change _ in @xmath69 as a function of redshift in the two models , we scale the results by dividing by the value of @xmath71 at z=0 .
as shown in figure 13 of paper i , @xmath72 km s@xmath13 ( @xmath73 h@xmath13 mpc ) in both the @xmath1cdm and @xmath24cdm models at the present day
. the pairwise peculiar velocities of the galaxies follow those of the dark matter quite closely in both models .
the galaxy velocities are 10 - 40% lower than those of the dark matter at @xmath3 .
( note that the `` antibias '' in galaxy peculiar velocities at @xmath3 is stronger than that shown in figure 13 of paper i , because the models presented in this paper do not include dust extinction .
dust reduces the contribution of star - forming field galaxies in a b - selected sample , but has little effect on early - type galaxies in rich groups and clusters .
models that include dust extinction thus give values of @xmath71 that are 10 - 25 % larger ) .
the difference between the galaxy and dark matter peculiar velocities decreases at higher redshift .
in contrast to the spatial distributions , galaxy peculiar velocities in our models are never very strongly biased . in the @xmath24cdm model , there is nearly a factor 2 decrease in @xmath71 from z=0 to z=0.5 . in the @xmath1cdm models ,
@xmath71 remains roughly constant out to @xmath74 , before decreasing at higher redshift . to11.cm ' '' ''
in a hierarchical universe , the evolution of galaxy clustering depends on the following : 1 .
cosmological parameters , such as @xmath0 , @xmath1 and @xmath75 , because these determine the rate at which structure grows and the epoch at which halos of given mass change from being rare objects , and thus biased tracers of the dark matter distribution , to being `` typical '' objects with clustering properties similar to that of the mass .
2 . the relationship between the mass of a dark matter halo and the properties of the galaxies that form within it . note that this relationship depends _ only _ on halo mass and is independent of the environment in which the halo finds itself ( lemson & kauffmann 1998 ) .
3 . the evolution of the galaxy population with redshift ( and hence the evolution of the relationship between galaxy properties and halo mass ) . in this paper
, we illustrate how differences in clustering evolution between galaxies of differing luminosity , colour , morphological type and star formation rate may help constrain galaxy formation models , and perhaps even cosmological parameters .
one interesting diagnostic that we highlight is the `` dip '' in correlation amplitude observed between @xmath3 and @xmath4 .
we show that this dip occurs naturally in a @xmath1cdm model , where structure forms early and halos with masses in the range @xmath76 , which contain galaxies of intermediate luminosities , are unbiased tracers of the mass over this redshift range . in the @xmath24cdm model
, bias evolves rapidly , and the clustering amplitude of @xmath7 galaxies remains constant from @xmath3 to @xmath4 .
although it might be possible to `` force '' a dip in clustering in the @xmath24cdm model by requiring that @xmath7 galaxies form in less massive halos , it would be difficult to come up with a physically - motivated scheme for doing this that would not simultaneously produce too many bright galaxies .
we also show that the strength of the dip in the @xmath1cdm model is sensitive to sample selection .
if galaxies are selected according to star formation rate rather than b - band luminosity , objects in low mass halos contribute more to the clustering signal at high redshifts and the dip is stronger . if galaxies are selected in the red rather than the blue , the dip is reduced .
very luminous galaxies and massive early - type galaxies exhibit no dip in clustering between @xmath3 and @xmath4 because they occur in high mass halos ( @xmath77 ) that are already biased at @xmath3 and become substantially more biased at high redshift .
the predictions presented in this paper should be viewed as illustrative rather than quantitative .
as discussed in paper i , the precise relation between the mass of a halo and the properties of the galaxies that form within it depends strongly on the adopted recipes for star formation and feedback ; these are very uncertain .
the exciting prospect is that future observations of galaxy clustering at high redshift will place strong _ empirical _ constraints on these processes .
+ the simulations in this paper were carried out at the computer center of the max - planck society in garching and at the eppc in edinburgh .
codes were kindly made available by the virgo consortium .
we especially thank adrian jenkins and frazer pearce for help in carrying them out .
we are also grateful to john peacock for useful discussions .
is a marie curie fellow and holds grant erbfmbict-960695 of the training and mobility of researchers program financed by the ec . | in hierarchical cosmologies the evolution of galaxy clustering depends both on cosmological quantities such as @xmath0 , @xmath1 and @xmath2 , which determine how collapsed structures dark matter halos form and evolve , and on the physical processes cooling , star formation , radiative and hydrodynamic feedback which drive the formation of galaxies within these merging halos . in this paper , we combine dissipationless cosmological n - body simulations and semi - analytic models of galaxy formation in order to study how these two aspects interact .
we focus on the differences in clustering predicted for galaxies of differing luminosity , colour , morphology and star formation rate , and on what these differences can teach us about the galaxy formation process .
we show that a `` dip '' in the amplitude of galaxy correlations between @xmath3 and @xmath4 can be an important diagnostic .
such a dip occurs in low - density cdm models , because structure forms early and dark matter halos of mass @xmath5 , containing galaxies with luminosities @xmath6 , are unbiased tracers of the dark matter over this redshift range ; their clustering amplitude then evolves similarly to that of the dark matter . at higher redshifts
bright galaxies become strongly biased and the clustering amplitude increases again . in high density models , structure forms
late and bias evolves much more rapidly . as a result ,
the clustering amplitude of @xmath7 galaxies remains constant from @xmath3 to @xmath4 .
the strength of these effects is sensitive to sample selection .
the dip becomes weaker for galaxies with lower star formation rates , redder colours , higher luminosities and earlier morphological types .
we explain why this is the case and how it is related to the variation with redshift of the abundance and environment of the observed galaxies .
we also show that the relative peculiar velocities of galaxies are biased low in our models , but that this effect is never very strong .
studies of clustering evolution as a function of galaxy properties should place strong constraints on models of galaxy formation and evolution .
-0.55 cm -0.55 cm 0.1 mm 16.75 cm clustering of galaxies in a hierarchical universe : ii .
evolution to high redshift + guinevere kauffmann , jrg m. colberg , antonaldo diaferio & simon d.m .
white + + keywords : galaxies : formation ; galaxies : halos ; cosmology : large - scale structure ; cosmology : dark matter |
observational evidence in the sense that the nuclear / circumnuclear starburst coexists with the central active nucleus has increased recently . in the study of a sample of seyfert 2 galaxies , storchi - bergmann et al .
( @xcite ) and gonzalez delgado et al ( @xcite ) have found that about 30 to 50 % seyfert 2 galaxies have nuclear starbursts . since both of them might be related to gas inflow , which could be triggered by the axis - asymmetric perturbation ( such as bars , tidal interactions or mergers ) , there is extensive speculation on the connection between starburst and nuclear activity ( norman & scoville @xcite ; terlevich et al .
@xcite ; heckman et al .
@xcite ; heckman et al .
@xcite ; gonzalez delgado @xcite ; and see the recent review by veilleux @xcite ) .
gu et al . ( @xcite ) studied 51 seyfert 2 galaxies from data available in the literature , and found that seyfert 2 galaxies with no evidence of a hidden seyfert 1 nucleus have characteristics in common with starburst galaxies .
there are several observable spectral signatures to find the nuclear starburst in seyfert 2 galaxies .
for example , in the ultraviolet ( uv ) band , the stellar wind resonance lines , such as n v @xmath01240 , si iv @xmath01400 and c iv @xmath01550 ( heckman et al .
@xcite ; heckman et al .
@xcite ; gonzalez delgado et al .
@xcite ) . in the optical , the broad emission lines of wolf - rayet stars around 4686 (kunth & contini @xcite ) and high order balmer absorption lines originating in the photosphere of o , b and a stars ( gonzalez delgado et al . @xcite ; gonzalez delgado et al .
recently , cid fernandes et al .
( @xcite ) have suggested a series of empirical criteria for the existence of starburst in seyfert 2 galaxies , such as weak caiik @xmath03933 , low excitation lines and large far - infrared luminosity , etc . in this paper , we report the presence of spectral signatures of nuclear starburst in the infrared luminous seyfert 2 galaxy ngc 7679 , with the infrared luminosity , @xmath2 , of @xmath3(veilleux et al .
@xcite ; veron - cetty & veron @xcite ) and the redshift ( _ z _ ) 0.01714 ( corresponding to a distance of 68.5 mpc , for h@xmath4 = 75 km s@xmath5 mpc@xmath5 and q@xmath4 = 0 ) .
ngc 7679 is one member of arp 216 , the other is ngc 7682 , their projected separation is 89.6kpc ( arp @xcite ) .
we suggest that the starburst activity in this galaxy might be related to the tidal interaction with ngc 7682 , as indicated by recent numerical simulations ( barnes & hernquist @xcite ; mihos & hernquist @xcite ) .
this paper is organized as follows . in sect.2
we describe our observations , data reduction and results . in sect .
3 we discuss our results and summarize our conclusions in sec .
spectra of ngc 7679 were obtained on sept .
28 - 29 , 2000 , with the boller & chivens spectrograph attached to the 2.1 m telescope of observatorio astronmico nacional at san pedro martir ( mexico ) .
we used a thomson 2048 @xmath1 2048 ccd and a 600 lines / mm grating , which provides a dispersion of 83 / mm .
the width of the long slit was set to 2 " ( i.e. , 664 pc for ngc 7679 ) .
hear - spectra had been taken before and after the object spectra for wavelength calibration , and bd + 28 4211 was selected from kpno standards for flux calibration .
the total exposure time was 10,800 s ( 2,700s @xmath1 4 ) .
all spectra were reduced using standard iraf procedures . in fig . 1 , we show the central 2 `` @xmath1 2 '' region spectrum of ngc 7679 , and in table 1 , we list all emission and absorption lines that we identify in the spectrum along with their respective equivalent widths ( ews ) .
stellar absorption wings clearly evolve the h@xmath6 and h@xmath7 emission lines .
for these lines , fluxes and ews were obtained using the iraf-_specfit _ routine .
the spectrum also shows the presence of high order balmer absorption lines(up to h12 ) , and several weak absorption lines , such as , caiik @xmath03933 , cn @xmath04200 , g - band @xmath04300 and mgib @xmath05173 . [ cols="<,<,^,^",options="header " , ] @xmath8 corrected for stellar absorption of 6.0 using _
specfit_. @xmath9 corrected for absorption of 5.7 .
@xmath10 means for emission lines .
to estimate the stellar population in the nuclear region of ngc 7679 , we compare our observed spectrum with the evolutionary synthesis model , gissel96 ( bruzual & charlot @xcite ) . following cid fernandes
et al.(@xcite ) , we use three components : an old and an intermediate populations , with ages @xmath11 1 gyr and @xmath12 yr , respectively , and an @xmath13 power - law component to account for either a scattered agn continuum or a young starburst component , since it is impossible to disentangle the power - law agn continuum from the contribution of a young starburst of an age @xmath14 10 myr ( storchi - bergmann et al .
@xcite ; gonzalez delgado et al .
@xcite and cid fernandes et al .
@xcite ) .
we normalized all spectra at 4800 , and we compare parameters of continuum fluxes at 4020 , 4510 and 5500 , as well as ews of the absorption lines for caiik @xmath03933,cn @xmath04200 and g - band @xmath04300 . the relative errors on these quantities are set to less than 5% .
2 shows the observed spectrum and the three component population synthesis fit .
the fractions of the power - law component , the intermediate and the old populations contributing to the total 4800 monochromatic light are 35.4% , 42.9% and 21.7% , respectively . for comparison
, we make use of the empirical formulae by cid fernandes et al.(@xcite ) to derive the fractions of old and young population , the results are for the young , intermediate and old population , the fractions are 33.5% , 48.3% and 18.2% , respectively , which are consistent with our fitting results .
the presence of high order balmer absorption lines indicate clearly the existence of intermediate - age stars in the nuclear region of ngc 7679 .
but the origin of the power - law component , accounting for about 35% of the total 4800 flux , may be the result of a young starburst or a scattered agn continuum .
it is well known that the stellar wind resonance absorption lines ( n v @xmath01240 , si iv @xmath01400 and c iv @xmath01550 ) are unambiguous signatures of young starbursts in the uv band , as suggested by heckman et al .
( @xcite ) ; heckman et al .
( @xcite ) ; and gonzalez delgado et al .
( @xcite ) .
unfortunately , these features can only be observed clearly in the brightest uv galaxies(heckman et al . @xcite ; gonzalez delgado et al .
@xcite ) . is there any indication of young starburst in the uv spectrum of ngc 7679 ?
we retrieved the uv spectrum for this object from the iue public archive ( fig .
3 ) . for comparison
, we also plot the iue and hst ghrs spectra of ngc 5135 .
ngc 5135 is one of seyfert 2 galaxies that presents clear nuclear starburst activity and high order balmer absorption lines ( gonzalez delgado et al .
@xcite ) . comparing iue with hst ghrs spectra of ngc 5135 , we checked that all typical absorption lines(as marked in fig .
3 ) in high resolution and high s / n ghrs spectrum are also present in the iue data , though the s / n is roughly low . at the same time , it is clear that all these stellar wind absorption lines are also present in ngc 7679 , in particular , we could see the clear p - cygni profile of c iv @xmath01550 , which is even more significant than that in ngc 5135 .
although the iue spectrum is marginal , we find the clear signature of a young starburst in the nuclear region of ngc 7679 . that the power - law component is the contribution of young starburst found in this work for ngc 7679 is consistent with a recent work by cid fernandes et al.(@xcite ) .
these authors have found that all @xmath15 sources are confirmed seyfert 2/starburst composites ( where @xmath16 means the contribution from a power - law featureless continuum ) and in these sources , the featureless continuum is dominated by nuclear / circumnuclear starburst , because in seyfert 2 galaxies the scattered light from the central agn can not exceed @xmath17 30 % , otherwise we should observe the reflected broad lines directly and the galaxy would be no longer classified as a seyfert 2 galaxy .
in this paper , we present the unambiguous evidence for the recent starburst activity in the nuclear region of ngc 7679 , which are : the higher order balmer absorption lines and weaker caiik @xmath03933 , cn @xmath04200 , g - band @xmath04300 and mgib @xmath05173 ; and in the uv band , several stellar wind absorption lines . using the simple three - component stellar population synthesis model
, we obtain the nuclear stellar population could be 42.9% intermediate age(@xmath18 yr ) , 21.7% old ( @xmath111gyr ) combining with a 35.4% power - law component , which might be from the young massive nuclear starburst activity .
we would like to thank dr .
rosa gonzalez delgado for sending us their hst ghrs spectra .
qsgu acknowledges support from unam postdoctoral program ( mexico ) and support from national natural science foundation of china and the national major project for basic research of the state scientific commission of china . and
dd - h , jad and eb acknowledge support from grant in 115599 from papiit - unam . this research has made use of nasa s astrophysics data system abstract service and the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
arp h.c . , 1966 ,
apjs , 14 , 1 .
barnes j. e. , hernquist l. 1996 , apj , 471 , 115 .
bruzual a.g .
, charlot s. , 1996 , galaxy isochrone syntheis spectral evolution library , aas , cd - rom series vol .
vii ( gissel 96 ) .
cid fernandes r. , heckman t. , schmitt h. , gonzalez delgado r.m . , storchi bergmann t. , 2001 , apj , in press .
( astro - ph/0104186 ) gonzalez delgado r.m . , heckman t. , leitherer c. , meurer g. , krolik j. , wilson a.s .
, kinney a. , koratkar a. , 1998 , apj , 505 , 174 .
gonzalez delgado r.m.,leitherer c. , heckman t.,1999 , apjs , 125 , 489 .
gonzalez delgado r.m . , heckman t. , leitherer c.,2000 , astro - ph/0001104 . gonzalez delgado r.m . , heckman t. , leitherer c.,2001 , apj , 546 , 845 .
, dultzin - hacyan d. , de diego j.a . , 2001 , rev.mex.astron.astrofis .
heckman t.m . , krolik j. , meurer g. , et al . , 1995 , apj , 452 , 549 .
heckman t.m . ,
gonzalez - delgado r.m .
, leitherer c. , meurer g.r . , krolik j. , wilson a.s .
, koratkar a. , kinney a. , 1997 , apj , 482 , 114 .
kunth d. , contini t. , 1999 , in iau symposium 193 , wolf - rayet phenomena in massive starts and starburst galaxies , ed .
van der hucht , g. koenigsberger & p.r.j .
eenens ( san francisco : asp ) , 725 .
mihos j. c. , hernquist l. 1996 , apj , 464 , 641 .
norman c. , scoville n. , 1988 , apj , 332 , 124 .
storchi - bergmann t. , raimann d. , bica e.l.d .
, fraquelli h. a. , 2000 , apj , 544 , 747 .
terlevich e. , diaz a.i . ,
terlevich r. , 1990 , mnras , 242 , 271 .
veilleux s. , kim d.c .
, sanders d.b . , mazzarella j.m . ,
soifer b.t . , 1995 ,
apjs , 98 , 171 .
veilleux s. , 2000 , astro - ph/0012121 .
veron - cetty m.p . , veron p. , 2000 , | we present our recent spectrophotometric results of the infrared luminous seyfert 2 galaxy ngc 7679 .
we find compelling evidence of the presence of a recent nuclear starburst , revealed by the observations of a ) the spectral features of high order balmer absorption lines , b ) the weak equivalent widths for caiik @xmath03933 , cn @xmath04200 , g - band @xmath04300 and mgib @xmath05173 , and c ) the suggested uv stellar wind resonance lines ( n v @xmath01240 , si iv @xmath01400 and c iv @xmath01550 ) in the iue spectrum .
using the simple stellar population synthesis model , we find that in the nuclear 2 `` @xmath1 2 '' region , the contributions from the old , intermediate and young components are 21.7% , 42.9% and 35.4% , respectively
. these nuclear starburst activities might be triggered by the close encounter with ngc 7682 , as suggested by recent numerical simulations . |
heavy fermions started out as a fertile ground to study strongly correlated fermi liquids and superconductors .
@xcite there was a great deal of amazement at seeing a fermi liquid whose quasiparticle mass is over a hundred times the bare electron mass ; hence the name of the field .
it was also surprising to find superconductors in an inherently magnetic environment : the existence of local magnetic moments in these materials is established through the observation of a curie - weiss susceptibility at intermediate temperatures ( of the order of 100 k ) , and magnetism is supposed to be `` hostile '' to superconductivity .
it was qualitatively understood that the large mass reflects the kondo screening of the magnetic moments , which is necessary to overcome magnetism , @xcite and the proper theoretical description of the fermi liquid state @xcite was subsequently achieved . when high temperature superconductors were discovered in 1986 , the development of the heavy fermion field was naturally interrupted at least partially .
the hiatus proved to be relatively short - lived . as it re - emerged , however , the field acquired a considerably different outlook , with the emphasis now placed on non - fermi liquid behavior and magnetic quantum phase transitions . since the late 1990s
the field has become a focal point @xcite for the general study of quantum criticality .
the interest in qcps is by no means unique to heavy fermions ; it also arises in high temperature superconductors among other materials .
@xcite however , heavy fermions are particularly advantageous in one important regard .
that is , second - order quantum phase transitions are explicitly observed in a growing list of this family of materials .
the qcp in heavy fermions typically separates an antiferromagnetic metallic phase from a paramagnetic metallic phase . near the magnetic qcp
, transport and thermodynamical properties develop anomalies .
the @xmath2 sdw picture @xcite describes the qcp in terms of fluctuations of the magnetic order parameter
the paramagnons both in space and in time .
this theory amounts to a @xmath3 theory with @xmath4 being the paramagnon field in an effective dimensionality of @xmath5 . here
, @xmath6 is the dynamic exponent , and is equal to @xmath7 in the antiferromagnetic case .
so , @xmath8 is above the upper critical dimension , @xmath9 , of the @xmath3 theory , for spatial dimensions @xmath10 .
the corresponding fixed point is gaussian . in this picture , the non - fermi liquid properties of the qcp reflect the singular scattering of electrons by the paramagnons and are unrelated to the process of kondo screening . the most direct indication for the unusual nature of the heavy fermion quantum criticality came from inelastic neutron scattering experiments .
@xcite the magnetic dynamics shows a fractional exponent and @xmath0 scaling over an extended range of momentum space .
these features deviate drastically from the expectations of the sdw theory .
since the sdw critical fixed point is gaussian , spin damping would be controlled by some ( dangerously ) irrelevant coupling , and neither fractional exponent nor @xmath0 scaling would be expected .
one way to resolve this impasse invokes the breakdown of kondo screening at the magnetic quantum critical point . on the paramagnetic side ,
local moments become entangled with the conduction electrons and , in the process , are delocalized and a part of the electron fluid . at the magnetic qcp , the magnetic fluctuations turn soft and act as a source of dissipation that couples to the local moments ; this coupling competes with the kondo interactions and destroys the kondo screening . going from the paramagnetic side to the qcp ,
the electronic excitations depart from those of a fermi liquid and acquire a non - fermi liquid form .
these non - fermi liquid excitations are part of the quantum critical spectrum . unlike the paramagnons , they are characterized by an interacting fixed point . as the control parameter
is tuned further , into the magnetically ordered side , the system is again a fermi liquid , but the kondo effect is completely destroyed . an important corollary of this picture @xcite is that the fermi surface has a sharp jump at the qcp . to microscopically study
the magnetic quantum phase transition requires an approach that can handle not only the heavy fermion and magnetic states but also the dynamical competition between the two on an equal footing .
one suitable approach ( if not the only one so far ) is the extended dynamical mean field theory ( edmft ) .
@xcite in this article , we discuss the edmft studies of the kondo lattice system .
in addition to a brief summary of the earlier analytical works , @xcite we will pay particular attention to two issues .
the first one deals with the magnetic dynamics near the qcp .
a fractional dynamical - spin - susceptibility exponent accompanies the destruction of kondo screening .
@xcite the second one concerns the nature of the zero - temperature transition .
when the edmft is implemented such that there is no double - counting of the static component of the rkky interactions , the transition turns out to be second order .
@xcite we compare these results with those of some recent related works . @xcite
we consider a kondo lattice model , @xmath11 here , the @xmath12-electron component @xmath13 describes the interactions between spin-@xmath14 local moments and @xmath15 are the spin projections .
we have taken the valence fluctuations to be completely frozen , which should be a good description of at least those heavy fermion metals that have a heavy effective mass and that are undergoing a magnetic quantum phase transition . without loss of generality
, we have assumed that a unit cell contains one local moment ( @xmath16 ) , whose spin-@xmath14 nature reflects the projection onto the lowest kramers doublet .
@xmath17 describes the rkky exchange interaction between the local moments . in the physical systems ,
the rkky interaction is generated by the kondo interactions .
here , we have taken it as an independent parameter , for two reasons .
first , it is useful to do so for the purpose of specifying the global phase diagram .
second , as it will be seen below , in the edmft approach to kondo lattice , it is necessary to include this parameter at the hamiltonian level to treat its effects dynamically ( see sections [ paramagnetic ] and [ ordered ] for details ) .
the conduction electron component of eq . ( [ kondo - lattice ] ) is simply @xmath18 it is implicitly assumed that the conduction electrons alone would form a fermi liquid , and the residual interaction ( landau ) parameters for the conduction - electron component alone can be neglected .
we will consider the number of conduction electrons per unit cell , @xmath19 , to be in the range @xmath20 ; in addition , we will assume that it is not too small , so that the physical rkky interaction between the nearest - neighboring local moments is antiferromagnetic , and not too close to @xmath21 , so that the kondo insulating physics does not come into play .
all the phases described below are metallic . finally , the local moments interact with @xmath22 , the spins of the conduction electrons , through an antiferromagnetic kondo coupling @xmath23 : @xmath24 for qualitative considerations , consider the kondo lattice model with a fixed value of @xmath25 and @xmath26 , with a small ratio @xmath27 ; here @xmath25 is the typical ( say , nearest - neighbor ) interaction of @xmath28 , and @xmath26 the conduction - electron bandwidth .
we further assume that @xmath29 has an ising anisotropy . in the antiferromagnetic phase of @xmath28
, the spin excitation spectrum is fully gapped .
an infinitesimal @xmath23 can not lead to any kondo screening .
hence , the fermi surface encloses only the conduction electrons , whose number is @xmath19 per paramagnetic unit cell .
we label this phase af@xmath30 .
on the other hand , when @xmath23 dominates over @xmath25 , the standard kondo screening does occur .
each local moment is converted into a spin-@xmath14 charge @xmath31 kondo resonance .
the fermi surface now encloses not only the conduction electrons but also the local moments , the total number being @xmath32 per unit cell .
we label this phase pm@xmath33 .
while the existence of this phase is well - established , @xcite the easiest physical way to see it is to consider the limit @xmath34 .
[ since the kondo state restores @xmath35 symmetry , we have , without loss of generality , taken the kondo exchange coupling to be spin - isotropic in eq .
( [ h - k ] ) . ] in this limit , there is a large binding energy ( of the order @xmath36 ) for a local singlet between @xmath37 and @xmath22 , and we can safely project to this singlet subspace . in this subspace , @xmath38 becomes special : here each local moment is locally paired up with a conduction electron , and the entire system becomes a kondo insulator . for a system of total @xmath39 unit cells ,
an @xmath40 amounts to creating @xmath41 unpaired local moments , each of which is equivalent to creating a hole in the singlet background .
the kondo lattice model becomes equivalent to an effective single band hubbard model of @xmath42 holes per site , with an infinite on - site repulsion ( it is impossible to create two holes there is only one electron in the singlet to begin with ) .
@xcite in the paramagnetic phase , the luttinger theorem then ensures that the fermi volume contains @xmath42 holes or , equivalently , @xmath43
electrons!. these general arguments show that the af@xmath30 and pm@xmath33 phases are two stable metallic phases .
they differ in two important regards . the af@xmath30 is magnetically ordered while the pm@xmath44 is not . equally important , the pm@xmath33 has the kondo screening while the af@xmath30 does not .
increasing the ratio @xmath45 takes the system from af@xmath30 to pm@xmath44 .
a key question is this : does the destruction of magnetism and the onset of kondo screening occur at the same stage ?
if so , the transition is distinctly different from the @xmath2 sdw picture . if not
i.e. _ , if the destruction of magnetism happens after the kondo screening has already set in then the magnetic transition can be interpreted as an sdw instability of the quasiparticles near the large fermi surface ; the transition goes back to the realm of the @xmath2 sdw transition picture .
microscopical studies provide a way to address this issue .
a suitable method has to capture not only the af@xmath30 and pm@xmath33 phases , due respectively to the rkky exchange interactions and the kondo interactions , but also the dynamical competition between these interactions ; this dynamical interplay is crucial for the transition region . at this stage ,
the edmft method is the only one we are aware of which fits this requirement .
of course , microscopic studies always have limitations .
approximations are inevitably used , in the process of solving a hamiltonian or at the level of the model itself ( or both ) . controlled approximations , nonetheless , provide us with not only ways of understanding experiments but also intuitions that serve as a basis for more macroscopic approaches .
this general philosophy is readily reflected in the edmft approach .
even though it is `` conserving '' ( _ i.e. _ , satisfying thermodynamic consistency requirements ) , it assumes that the @xmath46-dependences of some irreducible single - electron and collective quantities [ @xmath47 and @xmath48 defined in the next section ] are unimportant .
there are several reasons to believe that the edmft approach is useful for the antiferromagnetic quantum transitions at hand .
first , because staggered magnetization is not a conserved quantity , the spin damping does not have to acquire a strong dependence on @xmath49 .
this is especially true in metallic systems , where the dynamic exponent @xmath6 associated with the long - wavelength magnetic fluctuations is larger than 1 . in the sdw case ,
@xcite for instance , @xmath50 and @xmath51 can simply be taken as a linear function of @xmath52 without any singular dependence on @xmath46 . whether the @xmath46-dependence in @xmath51 is singular or regular
can be stated in terms of the anomalous spatial dimension @xmath53 characterizing the long - wavelength fluctuations in space : a non - zero @xmath53 means that the @xmath46 dependence in @xmath51 is more singular than that [ @xmath54 already incorporated in @xmath55 .
so , if @xmath56 , the edmft is expected to fail , at least for the asymptotic behavior . for instance , the classical critical points associated with a finite temperature magnetic transition in @xmath57 must have a finite @xmath53 ; the edmft turns out to produce ( an artificial ) first - order phase transition .
for a quantum critical point , on the other hand , long - wavelength fluctuations occur in @xmath5 dimensions .
there is then a greater likelihood for the vanishing of the spatial anomalous dimension , in which case the @xmath46-dependence of @xmath51 is not singular and neglecting it will not change the universal behavior .
second , as we have already discussed , the different classes of magnetic quantum critical points of a kondo lattice can be classified in terms of whether the fermi surface ( in the paramagnetic zone ) , which is large in the paramagnetic metallic phase , stays large as the qcp is crossed or becomes small by ejecting the local moments .
such large _ vs. _ small fermi surfaces are well described in terms of whether the kondo effect is preserved or destroyed , which , in turn , are readily captured by the edmft approach .
we now turn to the edmft studies of the kondo lattice model .
the edmft approach treats certain intersite ( coherent and incoherent ) collective effects on an equal footing with the local interaction effects .
the edmft equations have been constructed in terms of a `` cavity '' method , @xcite diagrammatics , @xcite and a functional formalism .
@xcite all of these constructions yield the same dynamical equations . in the diagrammatic language , the edmft is seen as entirely different from a systematic expansion @xcite in @xmath58 whose zero - th order would correspond to the dynamical mean field theory ( dmft ) .
@xcite instead , the edmft is a re - summation scheme that incorporates an infinite series of processes associated with the intersite collective effects , in addition to the local processes already taken into account in the dmft . unlike the single - electron properties , the collective modes
do not have a chemical potential .
in other words , the bottom of a `` band '' is important and this provides a means for spatial dimensionality to come into play in the edmft . within the edmft ,
the collective effects are organized in terms of an explicit intersite interaction term at the hamiltonian level . for the kondo lattice model described in the previous section ,
this is the intersite exchange term , @xmath59 of eq .
( [ h - f ] ) .
there are several ways to see the details of this formalism .
one way is to focus on a spin cumulant , @xcite whose inverse , @xmath47 , is colloquially referred to as a spin self - energy matrix . while it can be rigorously defined for any spinful many - body problem , this quantity
is taken as @xmath46-independent in the edmft . the dynamical spin susceptibility , on the other hand
, is @xmath46-dependent and is given by @xmath60 the conduction electron self - energy is still given by @xmath48 , and the conduction electron green s function retains the standard form , @xmath61 the irreducible quantities , @xmath47 and @xmath62 , are determined in terms of a self - consistent bose - fermi kondo model : @xmath63 the self - consistency reflects the translational invariance : @xmath64 \label{self - consistent } \\[-1ex ] g_{{loc } } ( \omega ) & = & \sum_{\bf k } g ( { \bf k } , \omega ) \;. \nonumber\end{aligned}\ ] ] when combined with the dyson equations , @xmath65 and @xmath66 , where @xmath67 $ ] and @xmath68 , are the weiss fields , these self - consistency equations specify the dispersions , @xmath69 and @xmath70 , and the coupling constant @xmath71
. we will focus on the case of two - dimensional magnetic fluctuations , characterized by the rkky density of states @xmath72 .
the first of eq .
( [ self - consistent ] ) becomes @xmath73 \nonumber \\ & = & i + 2i \exp \left [ -2i { \chi_{\rm loc}^a(\omega ) } \right ] + \cdots \;. \label{m - expansion}\end{aligned}\ ] ] where the last equality is an expansion in terms of @xmath74 $ ] , valid when the local susceptibility is divergent .
both the kondo screening and its destruction are encoded in the bose - fermi kondo model , eq .
( [ h - imp ] ) . the antiferromagnetic kondo coupling ( @xmath23 ) is responsible for the formation of a kondo singlet in the ground state and the concomitant generation of a kondo resonance in the excitation spectrum .
the coupling of the local moment to the dissipative bosonic bath ( @xmath71 ) provides a competing mechanism . to see this in some detail
, we first analyze eq .
( [ h - imp ] ) alone without worrying about the self - consistency conditions .
we consider a given spectrum of the bosonic bath , @xmath75 \propto |\omega|^{1-\epsilon } { \rm sgn } \omega \;. \label{bosonic - bath}\end{aligned}\ ] ] the problem can be studied using an @xmath1-expansion .
@xcite for small @xmath71 , the kondo coupling dominates , leading to a kondo screening .
a sufficiently large coupling @xmath71 destroys the kondo screening completely , reaching a local - moment phase .
the transition between these two phases is of second - order , and is described by a qcp where the kondo screening is just destroyed and the electronic excitations have a non - fermi liquid form .
the local spin susceptibility has a pauli form on the kondo side .
the destruction of kondo screening is then manifested in a divergent local susceptibility at the qcp .
an important property that is shared by the bose - fermi kondo model with @xmath35 spin symmetry , @xmath76 spin anisotropy , or ising spin anisotropy , is that @xmath77 , at the qcp . here , @xmath15 , @xmath78 , and @xmath79 for the @xmath35 , @xmath76 , and ising cases , respectively .
correspondingly , @xmath80 while the critical amplitude , @xmath81 , depends on the spin anisotropy , the critical exponent does not ; it is equal to @xmath82 in all cases
. there is an important point that follows from the above analysis which we will use in the following discussion of the numerical results . within the edmft approach to the kondo lattice ,
if the local spin susceptibility of the kondo lattice model is divergent at the magnetic qcp , the corresponding local problem is sitting on the critical manifold .
in other words , a divergent local susceptibility is a signature of the critical kondo screening and its associated non - fermi liquid electronic excitations .
these @xmath1-expansion results for the bose - fermi kondo model were initially used to study the full self - consistent problem , @xcite with the self - consistency conditions specified in the previous section . the fact that the critical exponent for the local susceptibility is equal to @xmath82 [ eq . ( [ chi - qcp ] ) ] turns out to be essential for the existence of a local qcp solution .
the self - consistent solution in the case of two - dimensional magnetic fluctuations has @xmath83 , corresponding [ eq .
( [ chi - qcp ] ) ] to a logarithmically divergent local susceptibility . the edmft equations in the ising case ( taking only the @xmath79 component ) were studied numerically in ref . using the quantum monte carlo method of grempel and rozenberg .
@xcite it was found that , at the magnetic qcp , the local spin susceptibility is indeed logarithmically divergent .
figure [ chi - tau ] shows a log - log plot of the local spin susceptibility @xmath84 _ vs. _ @xmath85 at a relatively low temperature ( @xmath86 ) .
it is seen from the figure that the zero - temperature limit of the local susceptibility is @xmath87 .
this corresponds to a frequency dependence that is logarithmically divergent in the low - frequency limit .
a fit of the data yields the value of the amplitude @xmath88 that is directly related to the critical exponent of the peak value of the lattice susceptibility .
figure [ chi - omega ] shows the logarithmic dependence of @xmath89 directly , for frequencies smaller than the bare kondo scale @xmath90 . as already mentioned in the previous subsection , such a divergent local susceptibility signifies that the bose - fermi kondo model is located on the critical manifold ; correspondingly , there is a destruction of kondo screening at the magnetic qcp of the kondo lattice model . from the divergent local susceptibility
, the self - consistency equation ( [ m - expansion ] ) also determines the spin self - energy and , by extension , the dynamical spin susceptibility of the kondo lattice .
the inverse peak susceptibility , @xmath91 , where @xmath92 is the ordering wavevector , is shown in fig .
[ peak - suscep ] .
a power - law fit yields a dynamical spin susceptibility exponent that is fractional , close to @xmath93 .
the fractional critical exponent is only seen at @xmath94 .
likewise , the fermi - liquid ( linear in @xmath95 ) damping inside the paramagnetic phase is also seen only at frequencies up to at most @xmath90 .
it is instructive to compare the above results with those of ref . , which studied an anderson lattice model .
the lowest temperature studied in ref .
is @xmath96 .
the first non - zero matsubara frequency , @xmath97 , is already larger than @xmath90 . as a result , neither the fractional exponent at the critical coupling @xmath98 nor the linear damping at @xmath99 can be observed .
it is tempting to consider the bose - fermi kondo model as equivalent to a local @xmath3 theory .
one maps the kondo coupling to an ising chain ( along the imaginary time @xmath100 axis ) with @xmath101 interactions .
@xcite in addition , the bosonic bath , with a spectrum of eq . ( [ bosonic - bath ] ) , adds an additional retarded interaction , @xmath102 .
the corresponding local @xmath3 theory is @xmath103 $ ] where @xmath104 with the constraint @xmath105 .
indeed , such a local @xmath3 theory emerges in the large-@xmath106 limit of a certain @xmath107 generalization of the bose - fermi kondo model . in the @xmath108 case ,
two of us @xcite showed that , while the destruction of kondo screening does occur , the fractional exponent is absent .
subsequently , pankov _ et al . _
@xcite demonstrated that this conclusion remains valid to order @xmath109 .
recent works have considerably clarified the limitations of the large @xmath106 limit and demonstrated the failure of the local @xmath3 description of the bose - fermi kondo model , at least for @xmath110 ( including the case of the self - consistent @xmath83 ) . for the local-@xmath3 theory ,
@xmath111 is the `` upper critical dimension '' .
@xcite at @xmath112 , the critical point would then be gaussian , implying violations of both @xmath0 scaling and hyperscaling . a number of recent studies on the bose - fermi kondo and closely related impurity models @xcite ,
have shown that the quantum critical point is interacting over the entire range @xmath113 , obeying @xmath0 scaling and hyperscaling .
these results support the observation @xcite of @xmath0 scaling in the ( self - consistent ) case of @xmath83 .
they also imply that the bose - fermi kondo model is perhaps the simplest model in which the standard description of a qcp in terms of a classical critical point in elevated dimensions fails .
physically , the kondo effect , involving the formation of a kondo singlet , is intrinsically quantum - mechanical . in the language of a path integral representation for spin , the kondo singlet formation necessarily involves the berry phase term .
it is then natural for the destruction of kondo screening to be inherently quantum mechanical and , by extension , for the qcp of the bose - fermi kondo model to be different from its classical counterpart at a higher dimension .
a more in - depth understanding of the underlying mechanism for this effect will surely be illuminating .
we now address whether the above results , derived from the paramagnetic side , are pre - empted by magnetic ordering .
to do so , we approach the transition from the ordered side . the edmft equations for the antiferromagnetically ordered phase require normal ordering @xcite of @xmath28 : @xmath114 , where the normal - ordered operator is @xmath115 . the effective impurity model and the self - consistency conditions are similar to eqs . ( [ h - imp ] ) and ( [ self - consistent ] ) , except for the following modifications .
first , there is a local magnetic field the static weiss field , @xmath116 coupled to @xmath117 .
this local field , arising through @xmath118 , must be self - consistently determined by the magnetic order parameter @xmath119 .
second , the conduction electron propagators are also influenced by magnetism .
it turns out that the second feature has to be treated with care so that there is no double - counting of the rkky interactions between the local moments . in ref . ,
we avoided double - counting the rkky interactions by working with a featureless conduction electron band ; in this case , the magnetism is driven by the interaction @xmath55 already incorporated at the hamiltonian level ( in @xmath29 ) .
we will expand on this issue in the next section .
our phase diagram is shown in fig .
[ phase - diagram ] . at @xmath120 ,
the system is in the paramagnetic metal phase .
the coherence scale of the kondo lattice @xmath121 marks the temperature / energy below which kondo resonances are generated and the heavy fermi liquid behavior occurs . in particular , the landau damping is linear in frequency at @xmath122 . at @xmath123 ,
the system has an antiferromagnetic ground state .
there is a finite - temperature first - order transition at the nel temperature , @xmath124 .
however , @xmath124 continuously goes to zero as the rkky interaction is reduced . within the numerical accuracy , it vanishes as @xmath125 , @xcite the same place where @xmath121 does so as @xmath126 . plotted in fig .
[ phase - diagram ] is the magnetic order parameter , @xmath127 , at the lowest studied temperature , @xmath86 .
again , it is seen to continuously go to zero as @xmath125 .
further support for the second order nature comes from the study of a nominally paramagnetic solution at @xmath123 .
this solution to the paramagnetic edmft equations co - exists with the solution to the ordered edmft equations at @xmath123 . but this `` paramagnetic '' solution is found to contain a curie component @xmath128 in the static local susceptibility with @xmath129 or , equivalently , a jump of magnitude @xmath128 in @xmath130 as @xmath95 goes to zero .
( its spin self - energy at zero frequency and zero temperature tracks @xmath25 . )
what this implies is that the `` paramagnetic '' solution is not the physical one ; instead , the physical ground state corresponds to the magnetic solution .
nonetheless , the study of this unphysical paramagnetic solution is helpful to the determination of the zero - temperature transition .
it is seen in fig .
[ phase - diagram ] that @xmath131 extrapolates to zero at the same value of @xmath132 ( @xmath125 ) where @xmath121 goes to zero ( @xmath126 ) .
this provides an additional consistency check for the critical interaction where the paramagnetic phase terminates . to summarize fig .
[ phase - diagram ] , within numerical accuracy , all relevant scales vanish simultaneously at @xmath132 , and the quantum transition at zero temperature is continuous . to discuss the double - counting issue in some detail , we revisit the procedure by which antiferromagnetism is treated in the standard dmft .
@xcite here , the dynamical equations are constructed entirely in terms of local single - particle quantities ; two - particle responses are derived once the dynamical equations have been solved . on the paramagnetic side
, the two - particle susceptibility satisfies the following bethe - salpeter equation ( in matrix form ) : @xmath133 here @xmath134 is the particle - hole susceptibility bubble of the full single - particle propagators @xmath135 .
the triplet particle - hole irreducible vertex has the following form ( again , in matrix form ) , @xcite @xmath136 where @xmath137 is the particle - hole susceptibility bubble of the full local single - particle propagators @xmath138 . combining this with eq .
( [ chi - bethe - salpeter ] ) implies that , on the paramagnetic side , @xmath139 where @xmath140 for our kondo lattice model , @xmath141 has the meaning of the generated rkky interaction [ after inverting the matrix in the @xmath142 space ] .
how can @xmath141 appear in the ( particle - hole ) spin response and not contribute in the dynamical equations for the single - particle quantities ?
the answer lies in the way the brillouin zone is divided in dmft into special " @xmath49 s and generic " @xmath46 s .
@xcite @xmath143 of eq .
( [ delta - i - q ] ) is non - zero only at `` special '' @xmath46 s .
the dynamical equations are constructed in terms of quantities that are local , _
i.e. _ summed over @xmath46 : the special @xmath46 s , having measure zero , are not important for this summation ; only `` generic '' @xmath46 s have the phase space to contribute to the local quantities . to be more specific ,
consider the hypercubic lattice .
@xmath141 depends on @xmath46 only through the combination @xmath144 .
@xcite the dispersion @xmath145 is @xmath146 ( in the @xmath147 limit ) only for `` special '' @xmath46 s , @xmath148 along the diagonals of the brillouin zone , @xmath149 . on the other hand , for `` generic '' @xmath46 , @xmath145 vanishes [ being formally of order @xmath150 , as can be seen from the central - limit theorem ] . when @xmath145 vanishes , @xmath151 , so @xmath141 vanishes !
the antiferromagnetic wavevector * q * ( q@xmath152 for all @xmath153 ) , belongs to the set of special @xmath46 s , so @xmath154 . and
the antiferromagnetic instability , from the paramagnetic side , is signaled by @xmath155 , at @xmath156 . on the antiferromagnetic side
, the nonzero value for the corresponding @xmath157 , is implemented through the introduction of the doubling of the conduction electron unit cell and different single - particle propagators at the two sub - lattices .
formally , this doubling of the conduction electron unit cell can be described in terms of an effective susceptibility , @xmath158 , @xmath159 here , again , @xmath160 .
the instability of the ordered state is signaled by @xmath161 , also at @xmath156 .
because the effective rkky interaction incorporated on the ordered side , @xmath157 , is the same as its counterpart on the paramagnetic side , @xmath162 , the magnetic transition is in general of second order .
there is a major limitation to this approach .
the rkky interaction , being zero at generic wavevectors , does not have enough phase space to dynamically interplay with the kondo interaction .
so the self - consistent dynamical equations of dmft does not incorporate @xmath141 at all , and the kondo screening is always present including at the magnetic qcp .
the quantum critical behavior falls in the sdw category , the same as in any static mean - field description of kondo lattices .
the edmft is introduced precisely to allow this dynamical interplay . here , an intersite interaction , as given in @xmath28 of eq . ( [ h - f ] ) ,
is elevated to the hamiltonian level .
the bethe - salpeter equation ( [ chi - bethe - salpeter ] ) still applies .
the particle - hole irreducible vertex becomes , @xcite @xmath163 where @xmath55 is the fourier transform of the intersite interactions , @xmath164 , already included at the hamiltonian level .
we have , on the paramagnetic side , @xmath165 where @xmath166 is the spin self - energy . likewise ,
we can write the effective susceptibility that comes into the stability analysis of the ordered phase as @xmath167 it was shown in ref . that the edmft can be rigorously formulated only when all @xmath46 are considered to be generic .
[ otherwise , @xmath55 is formally of order @xmath168 at the special @xmath46 s , and no paramagnetic phase could exist . ]
this implies that @xmath169 for all @xmath49 . from the kondo lattice point of view
, this is equivalent to saying that we will only use @xmath55 to represent the rkky interaction and will not incorporate additional , generated rkky interactions from the fermion bubbles ( illustrated in fig .
7 of ref . ) . in order to be consistent
, one would also need to demand that @xmath170 on the ordered side .
otherwise , we would be counting _ additional _ contributions to the rkky interaction on the ordered side that were absent on the paramagnetic side .
this requirement ( @xmath170 ) was achieved in ref . by using a featureless conduction - electron band .
the latter ensures that all wavevectors are generic in the sense defined earlier . within this procedure ,
the magnetic ordering is entirely driven by @xmath55 , and the instability criteria from the paramagnetic and ordered sides coincide .
therefore , the quantum transition is naturally of second order , as was indeed seen numerically in ref .
; see fig .
[ phase - diagram ] above .
the procedure used in refs . and amounts , in our language , to keeping @xmath171 while @xmath172 . on the paramagnetic side , all wavevectors , including the ordering wavevector @xmath92 , are taken as generic , and @xmath171 , as in all edmft schemes .
on the ordered side , @xmath92 is considered as one of the special wavevectors in the sense defined earlier and , through the conduction electron unit - cell doubling , @xmath172 ( as in dmft ) .
the ordered side then has an added energy gain , and the magnetic quantum transition is of first order .
( that an edmft approach to kondo lattices which incorporates a dmft - like fermion bubble on the ordered side alone yields a first order transition at zero temperature was in fact already recognized in @xcite . ) the procedure would actually lead to a first - order magnetic transition at zero - temperature in any itinerant model , including any @xmath2 sdw transition without any kondo physics .
we close by noting that the different edmft schemes that we have discussed can be equivalently seen as different local approximations to a baym - kadanoff - type functional .
an important manifestation of the destruction of the kondo screening is that @xmath12-electrons participate in the fermi volume on the paramagnetic side but fails to do so on the antiferromagnetic side .
there is a sudden reconstruction of the fermi surface across the magnetic qcp .
fairly direct electronic evidence for this effect has appeared in the recent hall - effect measurement in ybrh@xmath173si@xmath174 .
@xcite the hall coefficient shows a rapid crossover as a function of the control parameter magnetic field in this case .
the crossover sharpens as temperature is lowered , extrapolating to a jump in the zero - temperature limit .
the jump occurs at the extrapolated location of the magnetic phase boundary at zero temperature .
related features have also been observed in ybagge .
@xcite a more direct probe of the fermi surface comes from the de haas - van alphen effect .
recent dhva measurement @xcite in cerhin@xmath175 provides tantalizing evidence for a large reconstruction of the fermi surface , with a divergent effective mass , at a qcp .
specific - heat measurement under magnetic field @xcite points towards the possibility that cerhin@xmath176 undergoes a second - order magnetic quantum transition at the magnetic field of the strength used in the dhva experiment .
if the existence of the magnetic qcp is indeed established , cerhin@xmath176 will provide more insights into quantum criticality than cerh@xmath173si@xmath173 . in the latter system
, a large fermi - surface reconstruction has also been seen in the dhva measurements , @xcite but the zero - temperature transition is likely to be first order with a large jump in the magnetic order parameter across the transition . the fractional exponent and @xmath0 scaling in the magnetic dynamics have been seen , since early on , in the antiferromagnetic qcp of au - doped cecu@xmath177 @xcite ( whose magnetic fluctuations have a reduced dimensionality ) and in some frustrated compounds .
@xcite ( on the other hand , the sdw behavior is observed in the magnetic dynamics of ce(ru@xmath178rh@xmath179)@xmath180si@xmath180 , @xcite which has quasi-3d magnetic fluctuations . )
related non - trivial scaling exponents that are relatively easy to connect with theory have come from the grneisen ratio .
@xcite theoretically , there have also been efforts to study the kondo lattice systems using certain mixed - boson - fermion representations for the local - moment spin operators .
@xcite such auxiliary - particle representations set up the basis for a picture with spin - charge separation .
however , it has been hard to use this formalism to properly capture the kondo - screened fermi liquid phase , @xcite making it difficult to study its destruction as well .
it may also be possible to describe the destruction of kondo screening in terms of the static mean field theories based on slave boson and an rvb order parameter , supplemented by gauge - field fluctuations .
the corresponding phase diagram has recently been studied in some detail .
@xcite the magnetic transition and destruction of kondo screening are found to occur at different places in the zero - temperature phase diagram , @xcite so the magnetic quantum transition is still of the sdw type .
we believe that this is a reflection of the static nature of the mean field theory . finally , it is instructive to put in the present context the qcp proposed for the transition from an antiferromagnet to a valence - bond solid in frustrated quantum magnets .
@xcite dubbed a `` deconfined '' qcp , it has certain properties that may be qualitatively compared with the local quantum criticality : the qcp containing exotic excitations is surrounded by two conventional phases , and the corresponding energy scales of both vanish as the qcp is approached .
hence , it would be enlightening to explore the concrete connections ( if any ) of this approach with the physics of the destruction of kondo screening . for this purpose
, it would be necessary to either construct microscopic spin models for the deconfined qcp or reformulate the kondo screening beyond microscopic approaches .
to summarize , we have discussed some of the microscopic approaches underlying the local quantum critical picture . beyond the initial studies based on an @xmath181expansion renormalization group method ,
the most extensive investigations have been carried out in kondo lattice models with ising anisotropy .
the latter have allowed the study of both the destruction of kondo screening and the concomitant fractional exponent and @xmath0 scaling in the magnetic dynamics .
we have also discussed the magnetic phase diagram and summarized the evidence for the second order nature of the magnetic quantum phase transition .
the edmft studies of kondo lattice models with continuous spin symmetry [ @xmath35 or @xmath76 ] are mostly confined to the @xmath1-expansion studies .
efforts to access the quantum critical point in these systems , beyond the @xmath1-expansion , are still underway . a dynamical large-@xmath106 limit , for instance ,
has recently been shown to be promising .
@xcite the microscopic approaches described here have shown that critical modes beyond the order parameter fluctuations exist , and the modes are associated with the destruction of kondo screening .
these insights have a number of phenomenological consequences not only in magnetic dynamics but also in the fermi surface properties and in thermodynamics which have been supported by the existing and emerging experiments .
the insights will also help the search for the field theory that describes quantum critical heavy fermions .
finally , they may very well be broadly relevant to the exotic quantum critical behavior in other strongly correlated systems such as doped mott insulators .
we would particularly like to thank s. burdin , m. grilli , k. ingersent , s. kirchner , e. pivovarov , and l. zhu for collaborations on related work , and a. georges , g. kotliar , and t. senthil for discussions .
one of us ( qs ) would like to thank c. ppin for her organization of an informal workshop at saclay ( dec .
2004 ) , where some of these issues were discussed .
we are especially grateful to g. kotliar and p. sun for pressing us to clarify the issue of the magnetic phase diagram .
this work has been supported in part by nsf grant no .
dmr-0424125 and the robert a. welch foundation ( qs ) , and the us doe ( jxz ) .
the edmft approach shares similarity to , but is more general than , the dynamical equations developed in the spin - glass context : j. ye , s. sachdev , and n. read , phys .
lett . * 70 * , 4011 ( 1993 ) ; a. m. sengupta and a. georges , phys .
b * 52 * , 10295 ( 1995 ) ; a. j. bray and m. a. moore , j. phys .
c*13 * , l655 ( 1980 ) . | the identification of magnetic quantum critical points in heavy fermion metals has provided an ideal setting for experimentally studying quantum criticality .
motivated by these experiments , considerable theoretical efforts have recently been devoted to reexamine the interplay between kondo screening and magnetic interactions in kondo lattice systems .
a local quantum critical picture has emerged , in which magnetic interactions suppress kondo screening precisely at the magnetic quantum critical point ( qcp ) .
the fermi surface undergoes a large reconstruction across the qcp and the coherence scale of the kondo lattice vanishes at the qcp .
the dynamical spin susceptibility exhibits @xmath0 scaling and non - trivial exponents describe the temperature and frequency dependence of various physical quantities .
these properties are to be contrasted with the conventional spin - density - wave ( sdw ) picture , in which the kondo screening is not suppressed at the qcp and the fermi surface evolves smoothly across the phase transition . in this article we discuss recent microscopic studies of kondo lattices within an extended dynamical mean field theory ( edmft ) .
we summarize the earlier work based on an analytical @xmath1-expansion renormalization group method , and expand on the more recent numerical results .
we also discuss the issues that have been raised concerning the magnetic phase diagram .
we show that the zero - temperature magnetic transition is second order when double counting of the rkky interactions is avoided in edmft . |
the quadruply lensed quasar sdss j1004 + 4112 , first identified as a gravitational lens candidate from optical imaging and spectroscopy from the sloan digital sky survey ( sdss ; * ? ? ?
* ) , has an exceptionally large image separation of @xmath4 @xcite .
the @xmath5 quasar is multiply imaged by a cluster of galaxies ( rather than a galaxy ) at @xmath6 .
this is the only known quasar strongly lensed by a central part of a massive cluster .
the quasar appears to be radio - quiet since it is not detected in radio surveys such as the first survey @xcite .
subsequent observations revealed additional intriguing aspects of the system . @xcite
discovered time variability in the blue wings of several optical broad emission lines , which can be plausiblly interpreted in terms of microlesning .
_ hubble space telescope _
images detected the fifth image @xcite and multiply imaged background galaxies @xcite .
sdss j1004 + 4112 appears to be an ideal laboratory for exploration of the structure of a large cluster of galaxies and a quasar .
the uniqueness of sdss j1004 + 4112 affords advantages that strongly argue for multiwavelength observations , of which x - ray observations constitute an essential part .
measurement of the x - ray properties of the lensing cluster offers an opportunity to improve lens models of sdss j1004 + 4112 .
moreover , the lensing cluster is the first example of a _ strong - lens selected _ cluster ; it is interesting to determine if the cluster follows the empirical scaling relations between luminosities , temperatures , and masses .
second , we can investigate the magnification ratios of the quasar images at x - ray wavelengths , which may aid the interpretation of the microlensing event that has been observed at optical wavelengths @xcite .
indeed , x - ray flux ratios of lensed quasars are frequently different from optical flux ratios ; this phenomenon has been attributed to microlensing @xcite . in this paper , we report our analysis of an observation of sdss j1004 + 4112 with the _ chandra x - ray observatory_. a key feature of _ chandra _ images is the high angular resolution ( on the order of @xmath7 ) ; this resolution is required to allow reliable separation of the x - ray flux of the lensing cluster from that of the quadruply imaged quasar . in this paper
we will adopt a cosmological model with the matter density @xmath8 , the cosmological constant @xmath9 , and the hubble constant @xmath10 @xcite . at the redshift of the cluster ( @xmath6 )
, @xmath7 corresponds to 7.16 kpc . unless otherwise specified , quoted errors indicate the 90% confidence range
sdss j1004 + 4112 was observed for 80 ks with the _ chandra _ advanced ccd imaging spectrometer ( acis ; * ? ? ?
* ) on 2005 january 1 and 2 .
the data were obtained with the acis s3 ccd operating in vfaint mode .
this ccd has a @xmath11 pixel format with an image scale of @xmath12 pixel@xmath13 .
the target was offset from the nominal aim point with a y - offset of @xmath14 , however , this has little effect on the spatial resolution .
the ccd temperature during the observations was @xmath15 .
the data were processed using the standard software packages ciao 3.2.1 and caldb 3.0.3 . in the analysis of diffuse emission from the cluster ( [ sec : cluster ] ) , periods of high background rates ( defined by @xmath16% higher than the quiescent rates in the 2.57 kev band ) were removed ; the net exposure time is 62042 s. on the other hand , the data without lightcurve - filtering ( 79987 s ) are used in the analysis of quasar components ( [ sec : quasar ] ) since the background counting rate is negligibly small for point sources . the full band ( 0.57 kev ) acis image is shown in figure [ fig : image ] .
the four images of sdss j1004 + 4112 are clearly detected and resolved .
we found astrometry offsets for the _ chandra _ data , ( @xmath17ra , @xmath17dec)=(@xmath18 , @xmath19 ) , from a comparison of three sources that lie within @xmath20 from the aimpoint with those identified by the sdss survey . after correcting for the mean offsets , the measured coordinates of the four x - ray quasar images are consistent with those obtained with _ subaru _
@xcite within @xmath21 .
in addition , extended emission , roughly centered on the quasar images and extending out to @xmath22 , is seen in the x - ray data .
this component should arise from hot gas in the lensing cluster .
the relative offsets of the coordinates of image a and the peak of the diffuse emission are ( @xmath17ra , @xmath17dec)@xmath23(@xmath24 , @xmath25 ) ; the location of the peak of the extended emission is consistent with the position of the brightest cluster galaxy g1 . in the acis - s3 field ,
39 point - like sources including the four quasar images were detected with the wavdetect algorithm with a significance threshold parameter of @xmath26 .
the bright source at ( 10:04:34.262 , + 41:12:20.33 ) with an x - ray flux of @xmath27 ( 0.57 kev ) has been observed serendipitously as part of the follow - up program to spectroscopically monitor components a and b at apo ( richards et al .
2004 ) ; a low s / n spectrum indicates a redshift of @xmath28 , which is marginally consistent with the second most likely photometric redshift ( @xmath29 with 26% confidence ) .
the power - law slope , @xmath30 , connecting the rest frame 2500 and 2 kev flux densities , is @xmath31 , which is roughly consistent with the @xmath30-uv luminosity relation of agns @xcite .
the source at ( 10:04:33.653 , + 41:13:07.24 ) has an x - ray counting rate of @xmath32 , and it is not a counter lens image of the southern bright source as the separation angle is too large .
we derive the spectrum of the cluster component by extracting the data from the circular region within a radius of @xmath33 that is centered on the position of the galaxy g1 .
the fluxes from the four quasar images and several additional point sources were removed from the spectral integration area by excluding all regions within a radius 10 times the size of the point spread function ( psf ) at the source positions . here the size of psf is defined as the 40% encircled - energy radius at 1.5 kev at the source position .
the background was estimated from the data in a surrounding annulus ( @xmath34 ) .
the source and background counts within @xmath35 are @xmath36 and @xmath37 respectively .
the 0.57 kev extracted spectrum is fitted with the mekal thin - thermal plasma model @xcite utilizing the xspec version 11 . in this exercise ,
the galactic hydrogen column density is fixed to @xmath38 @xcite .
the x - ray temperature is constrained to be @xmath39 kev and the metal abundance to be @xmath40 ( the 90% upper limit is @xmath41 ) .
the galactic absorption - corrected , 0.57 kev flux is @xmath42 ( @xmath35 ) .
the bolometric x - ray luminosity within @xmath43 is estimated as @xmath44 , where @xmath43 is defined as the radius within which the average matter density is equal to @xmath45 times the critical density of the universe at the cluster redshift .
the @xmath46-model analysis in
[ subsec : xray_sb ] yields @xmath47 mpc .
this luminosity is lower than the mean value expected from the luminosity - temperature relation of distant clusters , @xmath48 @xcite , but our estimated value is within the observed scatter of the data .
the reduced chi - square of the best - fit model is @xmath49 . to investigate the radial temperature profile , we further analyzed spectra integrated from the inner @xmath50 region and the outer @xmath51 region of the lensing cluster .
the fitted temperatures are @xmath52 kev and @xmath53 kev , respectively .
these results suggest that there is not a strong temperature drop towards the cluster center , indicating radiative cooling is not important in this cluster .
the one - dimensional radial surface - brightness profile of the extended x - ray emission was constructed by adopting a center at g1 and azimuthally averaging the 0.55 kev image , which was corrected for the telescope s vignetting and the detector responses and rebinned by a factor of two ( i.e. , 1 processed image pixel is @xmath54 ) . as shown in figure [ sb ] , the radial profile was fit with the following two models : ( 1 ) the conventional @xmath46-model @xmath55^{-3\beta+1/2}$ ] , and ( 2 ) the profile derived from the universal mass profile proposed by @xcite plus isothermality of the cluster ( * ? ? ?
* hereafter nfw - ssm ) .
the background was set to a constant in the fitting .
we find that both models can fit the observed radial profile reasonably well : the reduced chi - square is @xmath56 and @xmath57 for the @xmath46-model and nfw - ssm model , respectively .
the fitted values of the @xmath46-model parameters are @xmath58 and @xmath59 kpc ) , while the nfw - ssm model results in @xmath60 and @xmath61 kpc ) .
the profile of the intra - cluster medium ( icm ) has a compact core , typical of observed surface brightness profiles of relaxed clusters .
the fit shows excess flux in the measurements compared to the models within @xmath62 kpc ( figure [ sb ] ) .
this is often seen in central regions of relaxed clusters ( e.g. , * ? ? ?
the central emission is significantly extended compared with the psf .
the 3-@xmath63 upper limit on a point - source luminosity is @xmath64 ( 0.55 kev ) assuming a power - law spectrum with index @xmath65 and a source redshift of @xmath6 .
the excess luminosity within @xmath66 is estimated as @xmath67 ( 0.38 kev ) , which is higher than for a typical elliptical galaxy and rather comparable to that of a cd galaxy in nearby clusters ( e.g. , * ? ? ?
* ; * ? ? ?
note that we subtracted the best - fit @xmath46-model component from the total emission assuming the mekal model with @xmath68 kev and @xmath40 for the cluster spectrum .
the @xmath69-band luminosity of the galaxy g1 is @xmath70 from the @xmath71 magnitude of 22.11 @xcite and the color - transformation law for elliptical galaxies @xcite .
the above values may follow the @xmath72 relation for elliptical galaxies ( e.g. , * ? ? ?
* ) and the excess flux may be attributed to hot ism emission from the central galaxy
. however , the x - ray temperature inferred from the hardness ratio is @xmath73 kev assuming a @xmath74 mekal model .
this is marginally high for a typical elliptical galaxy .
thus we might be looking at a superposition of the galaxy component and central emission from the cluster . given the present photon statistics , however , it is not possible to constrain further the origin of the emission .
note that the total x - ray luminosity of unresolved low - mass x - ray binaries is expected to be of the order @xmath75 ( e.g. , * ? ? ?
* ) and contributes only @xmath76% to the excess luminosity .
x - ray emission from the central fifth image @xcite is unlikely : the contribution is estimated to be only @xmath77 counts ( i.e. , about @xmath78% of the central @xmath66 emission ) assuming that the x - ray emission from the fifth image has similar intensity ratios as those measured in the hst acs image .
the observed profile may be better fitted by introducing the two - component @xmath46-models or increasing the inner slope parameter @xmath3 in the nfw - ssm model ; however , we will not pursue this process further since the above two models already provided acceptable fits to the data .
a considerably deeper observation is needed to constrain further the gas profile in the innermost region . from the image analysis ,
the extent of the x - ray emission above the 3-@xmath63 background level is found to be @xmath79 ( figure [ sb ] ) , which is close to an overdensity radius of @xmath80 mpc . under the assumption of hydrostatic equilibrium
, we can infer the mass distribution of the lensing cluster from the x - ray temperature and surface - brightness profiles . because there is not a significant radial dependence of the x - ray temperature in the observations
, we assume isothermality of the gas in the mass estimation . figure [ fig : mx ] shows the cylindrical cluster mass projected within a radius @xmath81 derived from each surface mass distribution profile .
we find that both models yield consistent mass profiles within the measurement errors .
it is interesting to compare the mass derived from x - ray analysis with that from gravitational lensing .
@xcite derived the cylindrical mass to be @xmath82 , and @xcite estimated @xmath83 . from figure
[ fig : mx ] , it is clear that both mass estimations are in excellent agreement with the mass profile from the x - ray observation .
for instance , we find the cylindrical masses within 100 kpc for the @xmath46-model and nfw - ssm model are @xmath84 and @xmath85 , respectively .
the discrepancy between x - ray and lensing masses has been reported in many lensing clusters ( e.g. , * ? ? ?
* ) , and is often ascribed to the projection of the extra matter along line - of - sight , the elongation of the lensing cluster along the line - of - sight direction , or a departure from equilibrium . the excellent agreement between x - ray and lensing masses
implies that none of these effects is significant for sdss j1004 + 4112 . for the nfw - ssm model and @xmath86 @xcite ,
the virial mass and the concentration parameter are constrained to be @xmath87 and @xmath88 .
the value of the concentration parameter is slightly larger than the theoretically expected median value for this virial mass and redshift , @xmath89 @xcite , but is within 2-@xmath63 scatter among different clusters .
the observed source counts for the four images , a d , are 1237 , 1580 , 1312 , and 763 , in the 0.57 kev band , respectively .
light curves were produced with time resolutions of 2048 or 4096 sec .
we did not find any clear sign of time variability within the statistical uncertainties .
note that each component was extracted using a circular region with a radius of @xmath90 in the analysis presented in this section .
the sum of the background and contamination from the cluster emission are estimated to be only @xmath91% of the source counts for a c , and @xmath92% for d , and are negligibly small compared to the statistical errors of the source spectra .
we measure the energy fluxes of lensed images in the _ chandra _ data by fitting each spectrum with a power - law plus gaussian line model ( see the next subsection for details ) .
the results are summarized in table [ tab : flux ] . the x - ray flux ratios differ significantly from those measured in the optical @xcite . since for fold lenses such as sdss j1004 + 4112 , the two images near the critical curve are usually brighter than the other two images and since the optical flux ratios were reproduced with simple mass models @xcite , we believe that the x - ray , rather than the optical , ratios are `` anomalous '' . the optical / x - ray flux ratios of images c and d are almost the same , therefore the most natural interpretation is that the image a is demagnified by a factor of @xmath93 in x - rays .
this interpretation is consistent with the observation that image a appears to be a saddle - point image @xcite which is more likely to be demagnified by perturbations than the other images @xcite . while the optical data compared in table [ tab : flux ] were taken on 2004 april 28 @xcite , the demagnification of image a in x - rays is further supported by optical observations on 2004 november 27 and 2005 april 26 with the magnum telescope @xcite . by linearly interpolating these two @xmath94-band imaging observations ,
we infer optical flux ratios on 2005 january 1 to be b / a=0.69 , c / a=0.40 , and d / a=0.29 , which are not very different from those listed in table [ tab : flux ] . in table [ tab : flux ] , we also show the values of @xmath30 to check the consistency of the x - ray and optical fluxes .
the values are somewhat higher than the @xmath95 that is expected from the uv luminosity of sdss j1004 + 4112 , @xmath96}$ ] ( assuming a magnification factor of @xmath97 for image a ; * ? ? ?
* ) and the best - fit @xmath30-uv luminosity relation @xcite . from table 5 of @xcite , @xmath30 is @xmath98 and @xmath99 for @xmath100 and @xmath101 , respectively .
thus the derived @xmath30 is within the 1-@xmath63 ( 2-@xmath63 ) scatter for a ( b d ) .
the spectrum of the lensed quasar component is displayed in figure [ fig : qso ] . to enhance the signal - to - noise ratio , the total spectrum ( sum of the images a
, b , c , and d ) is plotted .
we fit the spectrum with a power - law model .
again , the galactic hydrogen column density is fixed to @xmath102 since we did not find significant intrinsic absorption .
this model is rejected at the 90% confidence level ( @xmath103 ) .
we plot the residual of the data against the best - fit model in figure [ sb]a , which shows @xmath104 excesses in adjacent bins around 2.3 kev in our frame .
then we mask these bins and those to either side ( 9 bins from 2.262.66 kev ) and fit a power - law model .
the chi - square becomes significantly smaller ( @xmath105 ) .
the above results suggest the presence of an emission line component at @xmath106 kev .
we then fit the spectrum with a power - law plus gaussian line profile model ( see figure [ fig : qso]b ) . in comparison to just a power - law model ,
the fit was improved at the @xmath107% confidence level according to the @xmath108-test ( @xmath109 for three additional parameters ) .
the fitting yields the power - law index @xmath110 , the centroid energy of the line @xmath111 kev , and the gaussian width of the line @xmath112 ev .
the line centroid is consistent with the neutral iron k@xmath3 line at 6.4 kev in the quasar rest frame and the derived @xmath63 suggests that the line is intrinsically broad . from the result , we can compute the equivalent width @xmath113 ev in the quasar rest frame .
the luminosity is @xmath114 in the 210 kev band . in the case of @xmath97 for the image a , though the lens models that are consistent with the data allow a wide range of @xmath115 @xcite , the intrinsic quasar luminosity is estimated to be @xmath116 ( 210 kev ) .
this is within the luminosity range of seyfert galaxies ( e.g. , * ? ? ?
the reduced chi - square of the best fit model is @xmath117 .
there is a well - known negative correlation between the strength of the iron line and the x - ray luminosity in quasars ( e.g. , * ? ? ?
the derived equivalent width is larger by a factor of @xmath93 than that expected from the relation of @xcite and a factor of @xmath118 than that from recent _ xmm - newton _ observations @xcite .
combined with the result in [ sec : xrayflux ] , we speculate that microlensing demagnifies the x - ray continuum of image a , while keeping ( or amplifying ) the fe k@xmath3 line .
moreover , microlensing of emission lines in the optical band has been detected for this lens system @xcite , providing additional evidence for the microlensing hypothesis .
the relative enhancement of fe k@xmath3 line has often been observed in the x - ray emission of lensed quasars @xcite , and interpreted in terms of microlensing .
indeed , the relative enhancement of fe k@xmath3 line can be reproduced if the emission is separated in two regions ( e.g. , * ? ? ?
the origin of the strong fe k@xmath3 line is further explored by analyzing the spectrum of each image , rather than the sum of all images . to measure the line intensities ,
we fit them with the power - law plus gaussian model . the centroid energy and the width of the line as well as the power - law index
are allowed to vary in the fit .
the best - fit parameters and the 90% errors are listed in table [ tab : specfit ] .
we found that the spectral parameters for the four images agree with each other at the 95% confidence level , although the parameters , particularly the line width , show large uncertainties .
the fitting results are also statistically consistent with those derived for the total spectrum .
we then performed spectral fitting with @xmath119 , @xmath63 and @xmath120 fixed to the best - fit values for the total spectrum ( figure [ fig:4spec ] ) and calculated the equivalent widths ( table [ tab : flux ] ) .
although the errors are very large , the fe k@xmath3 line is detected most significantly for image a , supporting the interpretation above .
we also find that the hardness ratios of the images are marginally consistent with each other ( see table [ tab : flux ] ) .
@xcite recently proposed that spectral differences between lensed quasar image components are due to small line - of - sight differences through quasar disk wind outflows .
in particular , the author suggested that sdss j1004 + 4112 components b - d suffer from absorption in a disk - wind outflow and that the blue enhancements in sdss j1004 + 4112 a @xcite could be due to the alleviation of absorption along that sightline .
indeed , the a - b diffference spectrum from keck / lris spectroscopy of c iv region @xcite resembles that of the troughs seen in a broad absorption line ( bal ) quasar .
however , the observed line profiles and transitions involved ( e.g. , he ii ) mean that these features more plausibly result from excess _ emission _ in component a rather than absorption in component b. in addition , quasars with intrinsic uv absorption ( e.g. , bal troughs ) are generally x - ray faint due to a large intrinsic x - ray absorption ( @xmath121;e.g .
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , however , intrinsic absorption was not significantly detected in the x - ray spectra of the four images ; the 90% upper limit on the column density is obtained to be @xmath122 , and @xmath123 for a d , respectively .
here we assumed neutral absorbing material at @xmath5 . in the case
that the absorber is highly ionized and has high velocity flow , as proposed by @xcite , x - ray spectra with considerably better statistics are required to test the model .
moreover , if the differential absorption is responsible for the different x - ray spectra of a and b , we expect ( and @xcite predicts ) that the image a should be less absorbed in the x - ray . instead , however , component a has the smallest x - ray to optical flux ratio and the x - ray flux ratio is @xmath124 ( the 1-@xmath63 error ; see table [ tab : flux ] ) .
thus , given multiple independent indications , we conclude that the microlensing phenomenon is the most natural interpretation for the present data .
we have presented results from _ chandra _ observations of sdss j1004 + 4112 , a large - separation gravitational lens system created by a cluster of galaxies .
we have detected x - ray emission from the lensing cluster as well as four lensed quasar images . from the cluster x - ray emission
, we have constrained the bolometric luminosity and the temperature to be @xmath125 and @xmath39 kev , consistent with the luminosity - temperature relation of distant clusters .
we have reconstructed the mass profile of the lensing cluster assuming isothermality and hydrostatic equilibrium , and found that the mass within 100 kpc excellently agrees with that expected from strong lensing .
x - ray emission from the lensed quasar images displays two anomalies : the presence of a strong fe k@xmath3 line and significant differences in the flux ratios from those found in the optical band . both of these features are suggestive of microlensing .
the idea is supported by the fact that image a , which appears most anomalous , is a saddle - point image .
we are grateful to a. yonehara for useful discussions .
n. o. acknowledges support from the special postdoctral researchers program of riken .
n. i. is supported by jsps through jsps research fellowship for young scientists .
steffen , a. t. , strateva , i. , brandt , w. n. , alexander , d. m. , koekemoer , a. m. , lehmer , b. d. , schneider , d. p. , & vignali , c. 2006 , , accepted strateva , i. v. , brandt , w. n. , schneider , d. p. , vanden berk , d. g. , & vignali , c. 2005 , , 130 , 387 ccccccc a & @xmath126 & @xmath127 & 1.00 & 1.00 & @xmath128 & @xmath129 + b & @xmath130 & @xmath131 & 0.73 & 0.58 & @xmath132 & @xmath133 + c & @xmath134 & @xmath135 & 0.35 & 0.33 & @xmath136 & @xmath137 + d & @xmath138 & @xmath139 & 0.21 & 0.35 & @xmath140 & @xmath141 + ccccccc a & @xmath142 & @xmath143 & @xmath144 & @xmath145 & @xmath146 & 18.5 + b & @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & 9.6 + c & @xmath152 & @xmath153 & @xmath154 & @xmath155 & @xmath156 & 14.9 + d & @xmath157 & @xmath158 & @xmath159 & @xmath160 & @xmath161 & 7.3 + a+b+c+d & @xmath162 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & 35.9 + | we present results from _ chandra _ observations of sdss j1004 + 4112 , a strongly lensed quasar system with a maximum image separation of @xmath0 .
all four bright images of the quasar , as well as resolved x - ray emission originating from the lensing cluster , are clearly detected .
the emission from the lensing cluster extends out to approximately @xmath1 .
we measure the bolometric x - ray luminosity and temperature of the lensing cluster to be @xmath2 and 6.4 kev , consistent with the luminosity - temperature relation for distant clusters .
the mass estimated from the x - ray observation shows excellent agreement with the mass derived from gravitational lensing .
the x - ray flux ratios of the quasar images differ markedly from the optical flux ratios , and the combined x - ray spectrum of the images possesses an unusually strong fe k@xmath3 emission line , both of which are indicative of microlensing . |
throughout this paper , @xmath0 is a set of @xmath6 points in general and convex position in the plane .
the _ convex segment disjointness graph _ , denoted by @xmath2 , is the graph whose vertex set is the set of all line segments with endpoints in @xmath0 , where two vertices are adjacent if the corresponding segments are disjoint .
obviously @xmath2 does not depend on the choice of @xmath0 .
this graph and other related graphs , were introduced by araujo , dumitrescu , hurtado , noy and urrutia @xcite , who proved the following bounds on the chromatic number of @xmath2 : @xmath7 both bounds were improved by dujmovi and wood @xcite to @xmath8 in this paper we improve the lower bound to conclude near - tight bounds on @xmath5 .
[ thm : main ] @xmath9 the proof of theorem [ thm : main ] is based on the observation that eah colour class in a colouring of @xmath2 is a convex thrackle .
we then prove that two maximal convex thrackles must share an edge in common . from this
we prove a tight upper bound on the number of edges in the union of @xmath10 maximal convex thrackles .
theorem [ thm : main ] quickly follows .
a _ convex thrackle _ on @xmath0 is a geometric graph with vertex set @xmath0 such that every pair of edges intersect ; that is , they have a common endpoint or they cross .
observe that a geometric graph @xmath11 on @xmath0 is a convex thrackle if and only if @xmath12 forms an independent set in @xmath2 .
a convex thrackle is _ maximal _ if it is edge - maximal .
as illustrated in figure [ fig : thrackle](a ) , it is well known and easily proved that every maximal convex thrackle @xmath13 consists of an odd cycle @xmath14 together with some degree @xmath15 vertices adjacent to vertices of @xmath14 ; see @xcite . in particular
, @xmath13 has @xmath1 edges . for each vertex @xmath16 in @xmath14
, let @xmath17 be the convex wedge with apex @xmath16 , such that the boundary rays of @xmath17 contain the neighbours of @xmath16 in @xmath14 .
every degree-1 vertex @xmath18 of @xmath13 lies in a unique wedge and the apex of this wedge is the only neighbour of @xmath18 in @xmath13 . ]
a _ @xmath19-action _ on the unit circle @xmath20 is a homeomorphism @xmath21 such that @xmath22 for all @xmath23 .
we say that @xmath24 is _ free _ if @xmath25 for all @xmath23 .
[ lem : agreepoint ] if @xmath24 and @xmath26 are free @xmath19-actions of @xmath20 , then @xmath27 for some point @xmath28 . for points
@xmath29 , let @xmath30 be the clockwise arc from @xmath31 to @xmath32 in @xmath20 .
let @xmath33 . if @xmath34 then we are done
. now assume that @xmath35 . without loss of generality ,
@xmath36 appear in this clockwise order around @xmath20 .
paramaterise @xmath37 with a continuous injective function @xmath38\rightarrow \overrightarrow{x_0g(x_0)}$ ] , such that @xmath39 and @xmath40 .
assume that @xmath41 for all @xmath42 $ ] , otherwise we are done .
since @xmath26 is free , @xmath43 for all @xmath42 $ ] .
thus @xmath44))=\overrightarrow{g(p(0))g(p(1))}=\overrightarrow{g(x_0)x_0}$ ] .
also @xmath45))=\overrightarrow{f(x_0)f(p(1))}$ ] , as otherwise @xmath46 for some @xmath42 $ ] .
this implies that @xmath47 appear in this clockwise order around @xmath20 . in particular , with @xmath48
, we have @xmath49 . thus @xmath50 .
hence @xmath51 for some @xmath42 $ ] . since @xmath24 is a @xmath19-action , @xmath52 .
this is a contradiction since @xmath53 but @xmath54 .
assume that @xmath0 lies on @xmath20 .
let @xmath13 be a maximal convex thrackle on @xmath0 .
as illustrated in figure [ fig : thrackle](b ) , for each vertex @xmath18 in @xmath14 , let @xmath55 be a pair of closed intervals of @xmath20 defined as follows .
interval @xmath56 contains @xmath18 and bounded by the points of @xmath20 that are @xmath57 of the way towards the first points of @xmath0 in the clockwise and anticlockwise direction from @xmath18 .
let @xmath16 and @xmath58 be the neighbours of @xmath18 in @xmath14 , so that @xmath16 is before @xmath58 in the clockwise direction from @xmath18 .
let @xmath59 be the endpoint of @xmath60 in the clockwise direction from @xmath16 .
let @xmath61 be the endpoint of @xmath62 in the anticlockwise direction from @xmath58 .
then @xmath63 is the interval bounded by @xmath59 and @xmath61 and not containing @xmath18 .
define @xmath64 as follows . for each @xmath65 ,
map the anticlockwise endpoint of @xmath60 to the anticlockwise endpoint of @xmath66 , map the clockwise endpoint of @xmath60 to the clockwise endpoint of @xmath66 , and extend @xmath67 linearly for the interior points of @xmath60 and @xmath66 , such that @xmath68 and @xmath69 .
since the intervals @xmath60 and @xmath66 are disjoint , @xmath67 is a free @xmath19-action of @xmath20 . [
lem : common_edge ] let @xmath70 and @xmath71 be maximal convex thrackles on @xmath0 , such that @xmath72 . then there is an edge in @xmath73 , with one endpoint in @xmath74 and one endpoint in @xmath75 .
by lemma [ lem : agreepoint ] , there exists @xmath28 such that @xmath76 .
let @xmath77 and @xmath78 so that @xmath79 and @xmath80 , where @xmath55 and @xmath81 are defined with respect to @xmath70 and @xmath71 respectively . since @xmath72 , we have @xmath82 and @xmath83 .
thus @xmath84 . if @xmath85 then @xmath86 , implying @xmath87 .
thus @xmath88 .
hence @xmath89 .
without loss of generality , @xmath90 . thus @xmath91 .
if @xmath92 then @xmath31 is an endpoint of both @xmath56 and @xmath66 , implying @xmath93 , which is a contradiction .
thus @xmath94 contains points other than @xmath31 .
it follows that @xmath95 and @xmath96 .
therefore the edge @xmath97 is in both @xmath70 and @xmath71 .
moreover one endpoint of @xmath97 is in @xmath74 and one endpoint is in @xmath75 .
let @xmath11 be the directed multigraph with vertex set @xmath98 , where there is a _
arc @xmath97 in @xmath11 if @xmath18 is in @xmath99 and there is a _ red _ arc @xmath97 in @xmath11 if @xmath18 is in @xmath100 . since @xmath72 , every vertex of @xmath11 has outdegree @xmath15 .
therefore @xmath101 and there is a cycle @xmath102 in the undirected multigraph underlying @xmath11 .
in fact , since every vertex has outdegree 1 , @xmath102 is a directed cycle . by construction , vertices in @xmath11
are not incident to an incoming and an outgoing edge of the same color .
thus @xmath102 alternates between blue and red arcs .
the red edges of @xmath102 form a matching as well as the blue edges , both of which are thrackles .
however , there is only one matching thrackle on a set of points in convex position .
therefore @xmath102 is a 2-cycle and the result follows .
[ thm : overcount ] for every set @xmath0 of @xmath1 points in convex and general position , the union of @xmath10 maximal convex thrackles on @xmath0 has at most @xmath103 edges .
suppose that @xmath107 .
thus @xmath108 for all distinct @xmath109 . by lemma [ lem : common_edge ] , @xmath110 and @xmath111
have an edge in common , with one endpoint in @xmath112 and one endpoint in @xmath113 .
hence distinct pairs of thrackles have distinct edges in common . since every maximal convex thrackle has @xmath1 edges and we overcount at least one edge for every pair , the total number of edges is at most @xmath114 .
now assume that @xmath115 .
thus there is a vertex @xmath16 and a pair of thrackles @xmath110 and @xmath111 , such that @xmath116 .
as illustrated in figure [ fig : split ] , replace @xmath16 by two consecutive vertices @xmath117 and @xmath118 on @xmath0 , where @xmath117 replaces @xmath16 in every thrackle except @xmath111 , and @xmath118 replaces @xmath16 in @xmath111 .
add one edge to each thrackle so that it is maximal .
let @xmath119 be the resulting set of thrackles .
observe that @xmath120 , and the number of edges in @xmath121 equals the number of edges in @xmath104 plus @xmath10 . by induction , @xmath119 has at most @xmath122 edges , implying @xmath104 has at most @xmath123 edges .
we now show that theorem [ thm : overcount ] is best possible for all @xmath124 .
let @xmath125 be a set of @xmath10 vertices in @xmath0 with no two consecutive vertices in @xmath125 .
if @xmath126 and @xmath127 are consecutive in this order in @xmath0 , then @xmath128 is a maximal convex thrackle , and @xmath129 has exactly @xmath103 edges in total .
if @xmath130 then , there are @xmath10 convex thrackles whose union is the complete geometric graph on @xmath0 . possibly add edges to obtain @xmath10 maximal convex thrackles with @xmath131 edges in total . by theorem [ thm : overcount ] , @xmath132 .
the quadratic formula implies the result . | let @xmath0 be a set of @xmath1 points in general and convex position in the plane .
let @xmath2 be the graph whose vertex set is the set of all line segments with endpoints in @xmath0 , where disjoint segments are adjacent .
the chromatic number of this graph was first studied by araujo et al . [ _ cgta _ , 2005 ] .
the previous best bounds are @xmath3 ( ignoring lower order terms ) . in this paper
we improve the lower bound to @xmath4 , to conclude a near - tight bound on @xmath5 . |
the _ cellular automaton interpretation _ of quantum mechanics ( qm ) has recently been outlined by g. thooft @xcite .
this presents an attempt to redesign the foundations of quantum theory in accordance with essentially classical concepts , such as determinism and existence of ontological states of reality . which entails a surprising and intuitive explanation of the born rule and the apparent collapse of quantum mechanical states in measurement processes .
such an approach may be founded on the observation that quantum mechanical features arise in a large variety of deterministic and , loosely speaking , `` classical '' models .
so far , however , practically all of these models have been exceptional in that they can not easily be generalized to cover real phenomena , incorporating interactions and relativity .
yet cellular automata ( ca ) may provide the necessary versatility , as we shall presently continue to discuss @xcite .
for a large variety of earlier attempts in this field , see , for example , refs.@xcite and further references there . to begin with , let us recall the linearity of quantum mechanics ( qm ) as a crucial aspect of the unitary dynamics embodied in the schrdinger equation .
this linearity does not depend on the particular object under study , provided it is sufficiently isolated from anything else .
it is reflected in the superposition principle and implies the possibilities of interference effects and of non - factorizable states of composite objects , _
i.e. _ entanglement in multipartite systems .
the linearity of qm has been questioned and nonlinear modifications have been proposed before
notwithstanding various ` no - go ' arguments especially in order to test experimentally the robustness of qm against such nonlinear deformations . until now
no deviations from the predictions of qm have been observed , in particular no indications of nonlinearity .
we have studied a seemingly unrelated _ discrete _ dynamical theory , _
i.e. _ , which appears to deviate substantially from quantum theory .
nevertheless , it has been shown with the help of sampling theory that the deterministic mechanics of the class of discrete _ hamiltonian ca _ can be mapped one - to - one to continuum models pertaining to nonrelativistic qm , which are modified by the presence of a fundamental time scale @xcite .
perhaps surprisingly , this construction of a linear relation between ca and qm with a nonzero discreteness scale , compatible with the consistency of the action principle underlying the discrete dynamics on one side and the locality of the continuum description on the other , requires that both theories are linear @xcite .
this result has led us to consider composite objects formed from ca subsystems @xcite .
clearly , qm is special in that it is characterized not only by interference effects , like any classical wave theory would be , but also by the tensor product structures of states and observables applying for composite systems , which allow for entanglement .
it is not obvious that ca can constitute composites which conform with qm in this aspect , in the limit of negligible discreteness scale .
a main obstruction could have been the fact that the state space of hamiltonian ca is not a complex projective space , since the norm of the analogue of state vectors is not conserved by the dynamics ; instead there is a conserved two - time correlation function , as we shall recall , which becomes the familiar norm only in the continuum limit . in section2 . , we will review the earlier results concerning individual ca and briefly indicate the construction of multipartite ca in such a way that the superposition principle is respected and the composition rules of ca are compatible with those of qm .
this ` outlandish ' perspective based on ca may lead to additional insight in regard to interference and entanglement and eventually to an understanding of nonrelativistic qm in accordance with ref.@xcite .
concluding remarks are presented in section3 .
we describe classical hamiltonian ca with countably many degrees of freedom presently in terms of _ complex integer - valued _ state variables @xmath2 ( also known as _ gaussian integers _ ) , where @xmath3 denote different degrees of freedom and @xmath4 different states labelled by this discrete _ clock variable_. various equivalent forms of the action for such ca exist , as indicated earlier @xcite .
we will employ a particularly compact form here , which is useful for the following construction of composite ca in analogy with multipartite qm systems .
let @xmath5 denote a self - adjoint matrix of gaussian integers that will play the role of the hamilton operator .
furthermore , we introduce the suggestive notation @xmath6 , for any quantity @xmath7 depending on the clock variable @xmath8 . the summation convention for greek indices , @xmath9 ,
will often allow us to simplify notation further by suppressing them altogether , for example , writing @xmath10 .
then , with @xmath2 and @xmath11 as independent variables , the ca action @xmath12 is defined by : @xmath13\ ; : = \ ; \sum_n\big [ \frac{1}{2i}(\psi_n^*\dot\psi_n-\dot\psi_n^*\psi_n)+\psi_n^*\hat h\psi_n\big ] \;\equiv\;\psi^*\hat{\cal s}\psi \;\;,\ ] ] where the operator @xmath14 is a useful abbreviation , _ cf .
_ below . in order to set up the variational principle
, we introduce _ integer - valued _
variations @xmath15 to be applied to a polynomial @xmath16 as follows : @xmath17/2\delta f \;\ ; , \ ] ] and @xmath18 , if @xmath19 .
note that variations of terms that are _ constant , linear , or quadratic _ in integer - valued variables yield analogous results as standard infinitesimal variations of corresponding expressions in the continuum .
making use of these ingredients , we postulate the variational principle : 0.15 cm [ _ ca action principle _ ] 0.15 cm the discrete evolution of a ca is determined by stationarity of the action under arbitrary integer - valued variations of all dynamical variables , @xmath20.@xmath21 0.15 cm let us point out several characteristics of this _ ca action principle _ : + * i ) * while infinitesimal variations do not conform with integer valuedness , there is _ a priori _ no restriction of integer variations . hence _
arbitrary _ integer - valued variations must be admitted . + * ii ) * one could imagine contributions to the action ( [ action ] ) which are of higher than second order in @xmath22 or @xmath23 .
however , in view of arbitrary variations @xmath24 and @xmath25 , such additional contributions to the action must be absent for consistency .
otherwise the number of equations of motion generated by variation of the action , according to eq.([variation ] ) , would exceed the number of variables .
( a limited number of such remainder terms , which are nonzero only for some fixed values of @xmath8 , could serve to encode the _ initial conditions _ for the evolution . )
these features of the _ ca action principle _ are essential in constructing a map between hamiltonian ca and equivalent quantum mechanical continuum models @xcite .
for curiosity , generalizations of the variations defined in eq.([variation ] ) have also been considered , which allow higher than second order polynomial terms in the action . while leading to consistent discrete equations of motion , however , these nonlinear equations generally are beset with undesirable nonlocal features in the corresponding continuum description @xcite .
the equations of motion are obtained by applying the _ ca action principle _ to the action @xmath12 of eq.([action ] ) with the variations as defined in eq.([variation ] ) .
thus , variations @xmath26 and @xmath27 , respectively , yield discrete analogues of the schrdinger equation and its adjoint : @xmath28 \label{delpsi } \dot\psi_n^*&=&-\frac{1}{i}(\hat h\psi_n)^ * \;\ ; , \end{aligned}\ ] ] recalling that @xmath29 and @xmath30 , _ etc .
_ we remark that the action @xmath12 vanishes when evaluated for solutions of these finite difference equations . note that by setting @xmath31 , with real integer - valued variables @xmath32 and @xmath33 , and suitably separating real and imaginary parts of eqs.([delpsistar])([delpsi ] ) , the discrete equations assume a form that resembles hamilton s equations for a network of coupled classical oscillators @xcite : @xmath34 where we also split the self - adjoint matrix @xmath35 into real integer - valued symmetric and antisymmetric parts , respectively , @xmath36 .
these equations led us to the name _
hamiltonian ca _ ; this is furthermore justified by the fact that analogues of poisson brackets and classical like observables can be introduced here @xcite .
the time - reversal invariant equations of motion of section2.1 . give rise to conservation laws which are in _ one - to - one correspondence _ with those of the related schrdinger equation in the continuum .
it is straightforward to demonstrate the following theorem .
0.15 cm [ _ theorema _ ] 0.15 cm for any matrix @xmath37 that commutes with @xmath35 , @xmath38=0 $ ] , there is a _ discrete conservation law _ : @xmath39 for self - adjoint @xmath37 , defined by gaussian integers , this relation is about real integer quantities.@xmath21 a rearrangement of eq.([gconserv ] ) yields the corresponding conserved quantity @xmath40 ( using matrix notation , as before ) : @xmath41 _ i.e. _ a real integer - valued two - point correlation function which is invariant under a shift @xmath42 , @xmath43 . in particular , for @xmath44 , the corresponding conservation law amounts to a constraint on the state variables : @xmath45 this can be anticipated to play a similar role for discrete ca as the familiar normalization of state vectors in continuum qm .
we may also define the following symmetrized quantity : @xmath46 which , by eq.([normal ] ) , is conserved as well .
we have shown earlier how to construct an one - to - one invertible map between the dynamics of discrete hamiltonian ca and continuum qm in presence of a fundamental time scale @xcite . due to this discreteness scale @xmath0
, continuous time wave functions are _ bandlimited _ , _
i.e. _ , their fourier transforms have only finite support in frequency space , @xmath47 $ ] .
hence , we can apply sampling theory , in order to reconstruct continuous time signals , wave functions @xmath48 , from their discrete samples , the ca state variables @xmath2 , and _
vice versa _ @xcite .
let us represent here the resulting mapping rules obtained through the reconstruction formula provided by _
shannon s theorem _
@xcite : @xmath49 \label{npm1 } \psi_{n\pm 1}^\alpha & \longmapsto & \;\exp\big [ \mp l\frac{\mbox{d}}{\mbox{d}t}\big ] \psi^\alpha ( t ) = \psi^\alpha ( t\mp
l ) \;\ ; , \\ [ 1ex ] \label{samp } \psi^\alpha ( nl)&\longmapsto & \;\psi_n^\alpha \;\ ; , \end{aligned}\ ] ] keeping in mind that the continuum wave function is bandlimited .
these results allow to map ca equations of motion , in particular eqs.([delpsistar])([delpsi ] ) to appropriate continuum versions .
corresponding to eqs.([gconserv])([q ] ) , there exist analogous conservation laws and conserved quantities , which are found by applying the mapping rules separately to all wave function factors that appear .
thus , for example , we obtain from eq.([q ] ) the conserved quantity : @xmath50\psi ( t ) \\ [ 1ex ] \label{qcont2 } & \;&= \psi^{*\alpha}(t)\psi^\alpha ( t ) + \frac{l^2}{2}\mbox{re}\;\psi^{*\alpha}(t)\frac{\mbox{d}^2}{\mbox{d}t^2}\psi^\alpha ( t ) + \mbox{o}(l^4 ) \;\ ; , \end{aligned}\ ] ] which shows @xmath0-dependent corrections to the continuum limit , which here amounts to the usual conserved normalization @xmath51 .
similarly , the schrdinger equation with finite-@xmath0 correction terms are obtained @xcite . up to this point , our considerations dealt with individual hamiltonian ca .
based on these results , we will briefly indicate in the following the construction of multipartite systems .
can discrete ca combine to form composite multipartite systems ? a positive answer to this question is crucial , if we would like the quantum features of ca that we have found , so far , to also include the full extent of the superposition principle .
it is responsible for interference effects that we can find in single ca .
however , a more dramatic consequence in qm is the possibility of entanglement in states or observables . when scrutinizing possible constructions of multipartite ca , several requirements appear naturally .
one may wonder whether not only the _ linearity _ of the evolution law but also the _ tensor product structure _ of composite wave functions finds its analogue here .
these are the fundamental ingredients of the usual continuum theory reflected in interference and entanglement . which should be recovered in the continuum limit ( @xmath1 ) of the ca picture , at least .
furthermore , when the discreteness scale @xmath0 is finite , the dynamics of composites of ca which do not interact among each other should lead to _ no spurious correlations _ among them .
this principle of `` no correlations without interactions '' holds in all of known physics .
however , we are facing some serious obstacles which seem to prevent satisfying these requirements , when trying to form composites of hamiltonian ca .
the want - to - be discrete time derivative introduced before , @xmath6 , for any quantity @xmath7 depending on the clock variable @xmath8 , which appears all over in the ca equations of motion and conservation laws , does not obey the _
leibniz rule _ : @xmath52}= \dot a_n\textstyle{\frac{b_{n+1}+b_{n-1}}{2}}+ \textstyle{\frac{a_{n+1}+a_{n-1}}{2}}\dot b_n \neq \dot a_nb_n + a_n\dot b_n \;\;.\ ] ] similar observations can be expected for other definitions one might come up with . ignoring this for a moment ,
consider a trial multi - ca equation of motion analogous to the single - ca eq.([delpsistar ] ) : @xmath53 where @xmath54 may describe a block - diagonal hamiltonian in the absence of interactions among the ca .
then , through eq.([leibniz ] ) , the required _ factorization _ of eq.([psieq ] ) is hindered on the left - hand side , since unphysical correlations will be produced among the components of a factorized wave function , such as @xmath55 or for superpositions of such factorized terms .
for a bipartite system , for example , we obtain immediately : @xmath56 . furthermore , applying the mapping rules of section2.3 , before taking the limit @xmath1 , we find that the bilinear terms here do not converge to the appropriate qm expression . which should be @xmath57 , in order to allow the decoupling of two subsystems that do not interact .
this latter problem is a general one of nonlinear terms in the equations of motion of discrete ca , which we discussed before @xcite : _ the linear map provided by shannon s theorem does not commute with the multiplication implied by the nonlinearities .
_ this follows from the explicit reconstruction formula ( or any variant thereof that is linear ) @xcite . also on the right - hand side of eq.([psieq ] ) we find this obstruction , when trying to map the equation with a factorized wave function to its continuous time description .
it can be observed that the difficulties we pointed out arise from the implicit assumption that the components of a multipartite ca are _ synchronized _ to the extent that they share a common clock variable @xmath8 . in ref.@xcite , we have considered a radical way out of the impasse encountered by resorting to a _ many - time _ formalism .
this means giving up synchronization among parts of the composite ca by introducing a set of clock variables , @xmath58 , one for each one out of @xmath59 components
. it may be surprising to encounter this in the present nonrelativistic context , since the many - time formalism has been introduced by dirac , tomonaga , and schwinger in their respective formulations of relativistically covariant many - particle qm or quantum field theory , where a global synchronization can not be maintained @xcite .
we replace here the single - ca action of eq.([action ] ) by an integer - valued multipartite - ca action : @xmath60 : = \psi^*\big ( \sum_{k=1}^m\hat{\cal s}_{(k)}\;+\;\hat{\cal i}\big ) \psi \;\;,\ ] ] with @xmath61 and , correspondingly , @xmath62 as independent _ gaussian integer _ variables ; the self - adjoint operator @xmath63 incorporates interactions between different ca ; whereas @xmath64 is as introduced in eq.([action ] ) , with the subscript @xmath65 indicating that it acts _ exclusively _ on the pair of indices pertaining to the @xmath66-th single - ca subsystem : @xmath67 \;\;,\ ] ] with summation over _ all _ clock variables ( summation over twice appearing greek indices remains understood ) ; the @xmath68-operation , however , acts only with respect to the explicitly indicated @xmath69 , @xmath70 , while the single - ca hamiltonian , @xmath71 , requires a matrix multiplication , as before .
obviously , we can apply the _ ca action principle _ also to the present situation with the generalized action of eq.([maction ] ) .
this results in the following discrete equations of motion : @xmath72 together with the adjoint equations ; here the interaction @xmath63 , like @xmath71 , is assumed to be independent of the clock variables and the @xmath68-operation acts only with respect to @xmath69 in the @xmath66-th term on the left - hand side .
we have verified in ref.@xcite that the many - time formulation avoids the problems of a single - time multi - ca equation , such as eq.([psieq ] ) , which we discussed .
in particular , in the absence of interactions with each other , between ca subsystems , @xmath73 , _ no unphysical correlations _ are introduced among independent ca subsystems . furthermore , _ continuous multi - time equations _ corresponding to eqs.([meom ] ) are obtained by applying the mapping rules given in section2.3 . to the discrete equations .
we find no problem of incompatibility between multiplication according to nonlinear terms _ vs. _ linear mapping according to _
shannon s theorem _ , since a separate mapping is applied for each one of the clock variables .
this effectively replaces @xmath74 , where @xmath75 is a continuous real time variable . in this way , a _ modified multi - time schrdinger equation _
is obtained : @xmath76 \psi_{\dots t_k\dots}^{\dots\alpha_k\dots } \;=\;\frac{1}{i}\big ( \sum_{k=1}^mh_{(k)}^{\alpha_k\beta_k } \psi_{\dots t_k\dots}^{\dots\beta_k\dots } \;+\;{\cal i}^{\dots\alpha_k\dots\;\beta_1\dots\beta_m } \psi_{\dots t_k\dots}^{\beta_1\dots\beta_m } \big ) \;\;,\ ] ] where an overall factor of two from the left - hand side has been absorbed into the matrices on the right . by construction , here @xmath77
is bandlimited with respect to each variable @xmath75 . performing the continuum limit , @xmath1
, we arrive at the multi - time schrdinger equation ( one power of @xmath78 providing the physical dimension of @xmath71 and @xmath79 ) considered by dirac and tomonaga @xcite . however , when @xmath0 is fixed and finite , modifications in the form of powers of @xmath80 arise on its left - hand side , similarly as in single ca case before . in the present nonrelativistic context
, it may be appropriate to identify @xmath81 , in which case the operator on the left - hand side of eq.([mschroed ] ) , for @xmath1 , can be simply replaced by @xmath82 .
this results in the usual ( single - time ) _ many - body schrdinger equation_. finally , the study of the _ conservation laws _ of the multipartite ca equations of motion can be performed along the lines of section2.2 . and analogous results have been obtained @xcite
. the equivalent discrete or continuous many - time equations ( [ meom ] ) and ( [ mschroed ] ) are both linear in the ca wave function @xmath77 .
therefore , superpositions of solutions of these equations also present solutions and the _ superposition principle _ does indeed hold for multipartite hamiltonian ca . as in the case of single ca , this entails the fact that these discrete systems with all variables , parameters , _ etc . _ presented by gaussian integers
can produce _ interference _ effects as in quantum mechanics .
even more interesting , their composites can also show _ entanglement _ , which is deemed an essential feature of qm .
this follows from the form of the equations of motion , which allow for superpositions of factorized states . for example , in the bipartite case ( @xmath83 ) , assuming that the individual ca are characterized by two degrees of freedom ( @xmath84 ) , a time dependent analogue of the totally antisymmetric _
bell state _ is given by : @xmath85 which may be a solution of appropriate discrete equations of motion .
we conclude with a word of warning concerning the freely used expressions here that we borrowed from qm , such as `` wave functions '' and `` states '' , in particular .
they usually invoke the notion of vectors in a _
hilbert space _ , which turns into a complex projective space upon normalization of the vectors .
however , as has become obvious in section2.2 .
, see eqs.([normal])([q ] ) , and which can be seen similarly in the multipartite ca case @xcite , as long as the ca are truly discrete ( @xmath86 ) , the normalization ( squared ) of vectors is not among the conserved quantities , hence not applicable , but is replaced by a conserved two - time correlation function instead . furthermore
, contrary to first impression , perhaps , the envisaged space of states is _ not _ a hilbert space , since it fails in two respects : the vector - space and completeness properties are missing . first of all , the relevant _ gaussian integers _ are not complete . hence the completeness property of the space of states is lacking , which is built with these integers as underlying scalars .
secondly , the integers _ only _ featuring in all aspects of the ca do not form a field but a commutative ring ( for the multiplication of vectors by such scalars there is no multiplicative inverse , such as exists , _ e.g. _ , for rational , real , or complex numbers ) .
therefore , we do not have a vector space over a field , as usual in qm .
it has to be replaced by a more general structure .
this is known as a module over a ring , presently a module over the commutative ring of gaussian integers .
this allows the construction of a linear space endowed with an integer - valued scalar product , _
i.e. _ a _ unitary space_. taking its incompleteness into account , then , the space of states in the presented ca theory can be classified as a _ pre - hilbert module over the commutative ring of gaussian integers_. in any case , we conclude that superpositions of states , interference effects , and entanglement , as in qm , all can be found already on the `` primitive '' level of the presently considered natural hamiltonian ca , discrete single or multipartite systems which are characterized by ( complex ) integer - valued variables and couplings .
this presents a brief review of earlier work which has demonstrated surprising quantum features arising in integer - valued , hence `` natural '' , _ hamiltonian cellular automata _ @xcite .
the study of this particular class of ca is motivated by thooft s _ cellular automaton interpretation of qm _ @xcite and various recent attempts to construct models which may eventually lead to demonstrating that the essential features of qm can all be understood to emerge from pre - quantum deterministic dynamics .
the single ca we have considered allow practically for the first time to reconstruct quantum mechanical models with nontrivial hamiltonians in terms of such systems with a finite discreteness scale .
furthermore , we have recently extended this study by describing _ multipartite systems _ , analogous to many - body qm . not only is this useful for the construction of more complex models _ per se _ ( especially with a richer structure of energy spectra ) , but it is also necessary , in order to extend the _ superposition principle _ of qm to a description at the ca level .
we find that it can be introduced already there to the fullest extent , compatible with a tensor product structure of multipartite states , which entails the possibilities of their _ interference _ and _
entanglement_. surprisingly , we have been forced in our approach employing sampling theory for the map between ca and an equivalent continuum picture to introduce a many - time formulation , which only appeared in relativistic quantum mechanics before , as introduced by dirac , tomonaga , and schwinger @xcite .
this points towards a crucial further step in these developments , which is still missing , namely a ca model of _ interacting quantum fields_. without the possibility of interacting multipartite ca with quantumlike features , as described here , it is hard to envisage a ca picture of dynamical fields spread out in spacetime .
last not least , a detailed model is still lacking in which the born rule can be related quantitatively to the `` ordinary '' statistical mechanics of deterministic dynamical degrees of freedom .
it is a pleasure to thank g.grssing and j.walleczek for the kind invitation to the conference `` emergent quantum mechanics '' ( vienna , october 2015 ) and h.batelaan , a.caticha and g.t hooft for discussions .
support by the fetzer - franklin fund is gratefully acknowledged .
99 t hooft g 2014 the cellular automaton interpretation of quantum mechanics .
a view on the quantum nature of our universe , compulsory or impossible ? _ preprint _ arxiv:1405.1548 | cellular automata can show well known features of quantum mechanics , such as a linear rule according to which they evolve and which resembles a discretized version of the schrdinger equation .
this includes corresponding conservation laws .
the class of `` natural '' hamiltonian cellular automata is based exclusively on integer - valued variables and couplings and their dynamics derives from an action principle .
they can be mapped reversibly to continuum models by applying sampling theory .
thus , `` deformed '' quantum mechanical models with a finite discreteness scale @xmath0 are obtained , which for @xmath1 reproduce familiar continuum results .
we have recently demonstrated that such automata can form `` multipartite '' systems consistently with the tensor product structures of nonrelativistic many - body quantum mechanics , while interacting and maintaining the linear evolution .
consequently , the superposition principle fully applies for such primitive discrete deterministic automata and their composites and can produce the essential quantum effects of interference and entanglement . |
in reinforcement learning , an agent learns to maximize its discounted future rewards @xcite .
the structure of the environment is initially unknown , so the agent must both learn the rewards associated with various action - sequence pairs and optimize its policy .
a natural approach is to tackle the subproblems separately via a critic and an actor @xcite , where the critic estimates the value of different actions and the actor maximizes rewards by following the policy gradient @xcite .
policy gradient methods have proven useful in settings with high - dimensional continuous action spaces , especially when task - relevant _ policy representations _ are at hand @xcite .
we tackle the problem of learning actor ( policy ) and critic representations . in the supervised setting , representation or deep learning algorithms
have recently demonstrated remarkable performance on a range of benchmark problems .
however , the problem of learning features for reinforcement learning remains comparatively underdeveloped .
the most dramatic recent success uses @xmath1-learning over finite action spaces , and essentially build a neural network critic @xcite . here
, we consider _ continuous _ action spaces , and develop an algorithm that simultaneously learns the value function and its gradient , which it then uses to find the optimal policy .
this paper presents value - gradient backpropagation ( @xmath0 ) , a deep actor - critic algorithm for continuous action spaces with compatible function approximation .
our starting point is the deterministic policy gradient and associated compatibility conditions derived in @xcite . roughly speaking , the compatibility conditions are that 1
. the critic approximate the gradient of the value - function and 2 .
the approximation is closely related to the gradient of the policy .
see theorem [ t : compat ] for details .
we identify and solve two problems with prior work on policy gradients relating to the two compatibility conditions : 1 .
_ temporal difference methods do not directly estimate the gradient of the value function . _ + instead , temporal difference methods are applied to learn an approximation of the form @xmath2 , where @xmath3 estimates the value of a state , given the current policy , and @xmath4 estimates the _ advantage _ from deviating from the current policy @xcite .
although the advantage is related to the gradient of the value function , it is not the same thing .
2 . _ the representations used for compatible approximation scale badly on neural networks . _
+ the second problem is that prior work has restricted to advantage functions constructed from a particular state - action representation , @xmath5 , that depends on the gradient of the policy .
the representation is easy to handle for linear policies .
however , if the policy is a neural network , then the standard state - action representation ties the critic too closely to the actor and depends on the internal structure of the actor , example [ eg : deep_advantage ] . as a result
, weight updates can not be performed by backpropagation , see section [ sec : problem ] .
the paper makes three novel contributions .
the first two contributions relate directly to problems p1 and p2 .
the third is a new task designed to test the accuracy of gradient estimates .
[ [ method - to - directly - learn - the - gradient - of - the - value - function . ] ] method to directly learn the gradient of the value function .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the first contribution is to modify temporal difference learning so that it directly estimates the gradient of the value - function .
the _ gradient perturbation trick _ , lemma [ lem : gradient ] , provides a way to simultaneously estimate both the value of a function at a point and its gradient , by perturbing the function s input with uncorrelated gaussian noise .
plugging in a neural network instead of a linear estimator extends the trick to the problem of learning a function and its gradient over the entire state - action space .
moreover , the trick combines naturally with temporal difference methods , theorem [ thm : extension ] , and is therefore well - suited to applications in reinforcement learning . [ [ deviator - actor - critic - dac - model - with - compatible - function - approximation . ] ] deviator - actor - critic ( dac ) model with compatible function approximation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the second contribution is to propose the deviator - actor - critic ( dac ) model , definition [ def : beh_crit ] , consisting in three coupled neural networks and value - gradient backpropagation ( @xmath0 ) , algorithm [ alg : qprop ] , which backpropagates three different signals to train the three networks .
the main result , theorem [ thm : main ] , is that @xmath0 has compatible function approximation when implemented on the dac model when the neural network consists in linear and rectilinear units .
the proof relies on decomposing the actor - network into individual units that are considered as actors in their own right , based on ideas in @xcite .
it also suggests interesting connections to work on structural credit assignment in multiagent reinforcement learning @xcite .
[ [ contextual - bandit - task - to - probe - the - accuracy - of - gradient - estimates . ] ] contextual bandit task to probe the accuracy of gradient estimates .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + a third contribution , that may be of independent interest , is a new contextual bandit setting designed to probe the ability of reinforcement learning algorithms to estimate gradients .
a supervised - to - contextual bandit transform was proposed in @xcite as a method for turning classification datasets into @xmath6-armed contextual bandit datasets .
we are interested in the _ continuous _ setting in this paper .
we therefore adapt their transform with a twist .
the sarcos and barrett datasets from robotics have features corresponding to the positions , velocities and accelerations of seven joints and labels corresponding to their torques .
there are 7 joints in both cases , so the feature and label spaces are 21 and 7 dimensional respectively .
the datasets are traditionally used as regression benchmarks labeled sarcos1 through sarcos7 where the task is to predict the torque of a single joint and similarly for barrett .
we convert the two datasets into two continuous contextual bandit tasks where the reward signal is the negative distance to the correct label 7-dimensional .
the algorithm is thus `` told '' that the label lies on a sphere in a 7-dimensional space .
the missing information required to pin down the label s position is precisely the gradient . for an algorithm to make predictions that are competitive with fully supervised methods ,
it is necessary to find extremely accurate gradient estimates .
[ [ experiments . ] ] experiments .
+ + + + + + + + + + + + section [ sec : experiments ] evaluates the performance of @xmath0 on the contextual bandit problems described above and on the challenging octopus arm task @xcite .
we show that @xmath0 is able to simultaneously solve seven nonparametric regression problems without observing any labels instead using the distance between its actions and the correct labels .
it turns out that @xmath0 is competitive with recent _ fully supervised _ learning algorithms on the task .
finally , we evaluate @xmath0 on the octopus arm benchmark , where it achieves the best performance reported to date .
an early reinforcement learning algorithm for neural networks is @xmath7 @xcite .
a disadvantage of @xmath7 is that the entire network is trained with a single scalar signal .
our proposal builds on ideas introduced with deep @xmath1-learning @xcite , such as replay .
however , deep @xmath1-learning is restricted to finite action spaces , whereas we are concerned with _
continuous _ action spaces .
policy gradients were introduced in @xcite and have been used extensively @xcite .
the deterministic policy gradient was introduced in @xcite , which also proposed the algorithm @xmath8 .
the relationship between @xmath0 and @xmath8 is discussed in detail in section [ sec : problem ] .
an alternate approach , based on the idea of backpropagating the gradient of the value function , is developed in @xcite .
unfortunately , these algorithms do not have compatible function approximation in general , so there are no guarantees on actor - critic interactions .
see section [ sec : problem ] for further discussion .
the analysis used to prove compatible function approximation relies on decomposing the actor neural network into a collection of agents corresponding to the units in the network .
the relation between @xmath0 and the difference - based objective proposed for multiagent learning @xcite is discussed in section [ sec : local_actors ] .
we use boldface to denote vectors , subscripts for time , and superscripts for individual units in a network .
sets of parameters are capitalized ( @xmath9 , @xmath10 , @xmath11 ) when they refer to matrices or to the parameters of neural networks .
this section recalls previous work on policy gradients . the basic idea is to simultaneously train an actor and a critic .
the critic learns an estimate of the value of different policies ; the actor then follows the gradient of the value - function to find an optimal ( or locally optimal ) policy in terms of expected rewards .
the environment is modeled as a markov decision process consisting of state space @xmath12 , action space @xmath13 , initial distribution @xmath14 on states , stationary transition distribution @xmath15 and reward function @xmath16 .
policy _ is a function @xmath17 from states to actions .
we will often add noise to policies , causing them to be stochastic . in this case , the policy is a function @xmath18 , where @xmath19 is the set of probability distributions on actions .
let @xmath20 denote the distribution on states @xmath21 at time @xmath22 given policy @xmath23 and initial state @xmath24 at @xmath25 and let @xmath26 .
let @xmath27 be the discounted future reward . define the @xmath28\quad\text{and } \label{e : q } \\
\text{value of a policy : } \qquad & j({{\ensuremath{\boldsymbol\mu}}_\theta } ) = \operatorname*{\mathbb e}_{{\mathbf{s}}\sim \rho^{{\ensuremath{\boldsymbol\mu } } } , { \mathbf{a}}\sim { \ensuremath{\boldsymbol\mu}}_\theta}[q^{{\ensuremath{\boldsymbol\mu}}_\theta}({\mathbf{s}},{\mathbf{a } } ) ] .
\label{eq : j}\end{aligned}\ ] ] the aim is to find the policy @xmath29 with maximal value .
a natural approach is to follow the gradient @xcite , which in the deterministic case can be computed explicitly as [ t : dpg]@xmath30 + under reasonable assumptions on the regularity of the markov decision process the policy gradient can be computed as @xmath31 .
\end{aligned}\ ] ] see @xcite . since the agent does not have direct access to the value function @xmath32 , it must instead learn an estimate @xmath33 . a sufficient condition for when plugging an estimate @xmath4 into the policy gradient
@xmath34 $ ] yields an unbiased estimator was first proposed in @xcite .
a sufficient condition in the deterministic setting is : [ t : compat]@xmath30 + the value - estimate @xmath35 satisfies is compatible with the policy gradient , that is @xmath36\ ] ] if the following conditions hold : 1 .
* @xmath37 approximates the value gradient : * + the weights learned by the approximate value function must satisfy @xmath38 , where @xmath39 \label{eq : val_estimate}\ ] ] is the mean - square difference between the gradient of the true value function @xmath32 and the approximation @xmath40 .
* @xmath37 is policy - compatible : * + the gradients of the value - function and the policy must satisfy @xmath41 see @xcite .
having stated the compatibility condition , it is worth revisiting the problems that we propose to tackle in the paper .
the first problem is to directly estimate the gradient of the value function , as required by eq . in condition _
c1_. the standard approach used in the literature is to estimate the value function , or the closely related advantage function , using temporal difference learning , and then compute the derivative of the estimate .
the next section shows how the gradient can be estimated directly .
the second problem relates to the compatibility condition on policy and value gradients required by eq . in condition _
c2_. the only function approximation satisfying _
c2 _ that has been proposed is [ eg : advantage]@xmath30 + let @xmath42 be an @xmath43-dimensional feature representation on states and set @xmath44
. then the value function approximation @xmath45 satisfies condition _ c2 _ of theorem [ t : compat ] .
the approximation in example [ eg : advantage ] encounters serious problems when applied to _ deep _ policies , see discussion in section [ sec : problem ] .
in this section , we tackle the first problem by modifying temporal - difference ( td ) learning so that it directly estimates the gradient of the value function .
first , we developed a new approach to estimating the gradient of a black - box function at a point , based on perturbing the function with gaussian noise .
it turns out that the approach extends easily to learning the gradient of a black - box function across its entire domain .
moreover , it is easy to combine with neural networks and temporal difference learning .
gradient estimates have been intensively studied in bandit problems , where rewards ( or losses ) are observed but labels are not .
thus , in contrast to supervised learning where it is possible to compute the gradient of the loss , in bandit problems the gradient must be estimated .
more formally , consider the following setup .
[ def : bb]@xmath30 + a function @xmath46 is a * zeroth - order black - box * if it can only be queried for _ zeroth - order _ information .
that is , user can request the value @xmath47 of @xmath48 at any point @xmath49 , but can not request the gradient of the function .
we use the shorthand _ black - box _ in what follows . the black - box model for optimization
was introduced in @xcite , see @xcite for a recent exposition . in those papers ,
a black - box consists in a _ first - order oracle _ that can provide both zeroth - order information ( the value of the function ) and first - order information ( the gradient or subgradient of the function ) .
@xmath30 + the reward function @xmath50 is a black box since nature does not provide gradient information .
the value function @xmath51 $ ] is _ not even _ a black - box : it can not be queried directly since it is defined as the expected discounted _ future _ reward .
it is for this reason the gradient perturbation trick must be combined with temporal difference learning , see section [ sec : tdgl ] .
an important insight is that the gradient of an unknown function at a specific point can be estimated by perturbing its input @xcite .
for example , for small @xmath52 the gradient of @xmath46 is approximately @xmath53 $ ] where the expectation is over vectors sampled uniformly from the unit sphere .
the following lemma provides a simple method for estimating the gradient of a function _ at a point _ based on gaussian perturbations : [ lem : gradient]@xmath30 +
the gradient of differentiable @xmath46 at @xmath54 is @xmath55\right\}.\ ] ] by taking sufficiently small variance , we can assume that @xmath48 is locally linear .
setting @xmath56 yields a line through the origin .
it therefore suffices to consider the special case @xmath57 . setting @xmath58,\ ] ]
we are required to show that @xmath59 .
the problem is convex , so setting the gradient to zero requires to solve @xmath60 $ ] , which reduces to solving the set of linear equations @xmath61 = ( w^j - v^j)\operatorname*{\mathbb e}[(\epsilon^j)^2]=(w^j - v^j)\cdot \sigma^2=0\qquad \text { for all $ j$}.\ ] ] the first equality holds since @xmath62=0 $ ] .
it follows immediately that @xmath59 .
the solution to the optimization problem in eq .
is the gradient @xmath63 of @xmath48 at a particular @xmath54 .
the next step is to learn a function @xmath64 that approximates the gradient across a range of values . more precisely , given a sample @xmath65 of points , we aim to find @xmath66 .
\label{eq : grad_optimal}\ ] ] the next lemma considers the case where @xmath67 and @xmath68 are linear estimates , of the form @xmath69 and @xmath70 for fixed representations @xmath71 and @xmath72 .
[ lem : gradl]@xmath30 + let @xmath46 be a differentiable function .
suppose that @xmath73 and @xmath72 are representations such that there exists an @xmath43-vector @xmath74 and a @xmath75-matrix @xmath76 satisfying @xmath77 and @xmath78 for all @xmath79 in the sample .
if we define loss function @xmath80.\ ] ] then @xmath81.\ ] ] follows from lemma [ lem : gradient ] .
in short , the lemma reduces gradient estimation to a simple optimization problem _ given a good enough representation_. jumping ahead slightly to section [ sec : dpg ] , we ensure that our model has good enough representations by constructing two neural networks to learn them .
the first neural network , @xmath82 , learns an approximation to @xmath83 that plays the role of the baseline @xmath84 .
the second neural network , @xmath85 learns an approximation to the gradient .
recall that @xmath86 is the expected value of a state - action pair given policy @xmath23 .
it is never observed directly , since it is computed by discounting over future rewards .
td - learning is a popular approach to estimating @xmath32 through dynamic programming @xcite .
we quickly review td - learning .
let @xmath87 be a fixed representation .
the goal is to find a value - estimate @xmath88 where @xmath89 is an @xmath43-dimensional vector , that is as close as possible to the true value function .
if the value - function were known , we could simply minimize the mean - square error with respect to @xmath89 : @xmath90.\ ] ] unfortunately , it is impossible to minimize the mean - square error directly , since the value - function is the expected discounted future reward , rather than the reward .
that is , the value function is not provided explicitly by the environment not even as a black - box .
the bellman error is therefore used a substitute for the mean - square error : @xmath91\ ] ] where @xmath21 is the state subsequent to @xmath24 .
let @xmath92 be the td - error .
td - learning updates @xmath89 according to @xmath93 where @xmath94 is a sequence of learning rates .
the convergence properties of td - learning and related algorithms have been studied extensively , see @xcite . finally , we apply temporal difference methods to estimate the _ _ gradient _ _ of the value function , as required by condition _
c1 _ of theorem [ t : compat ] .
we are interested in gradient approximations of the form @xmath95 where @xmath96 and @xmath10 is a @xmath75-dimensional matrix .
the goal is to find @xmath76 such that @xmath97 for all sampled state - action pairs .
it is convenient to introduce notation @xmath98 and shorthand @xmath99 .
then , analogously to the mean - square , define the perturbed gradient error : @xmath100,\ ] ] given a good enough representation , lemma [ lem : gradl ] guarantees that minimizing the perturbed gradient error yields the gradient of the value function .
unfortunately , as discussed above , the value function can not be queried directly .
we therefore introduce the bellman gradient error as a proxy @xmath101.\ ] ] set the tdg - error as @xmath102 and , analogously to eq .
, define the tdg - updates @xmath103 where @xmath104 is the @xmath75 matrix given by the outer product .
we refer to @xmath105 as the * perturbed tdg - error*. the following _ extension theorem _ allows us to import guarantees from temporal - difference learning to temporal - difference gradient learning .
[ thm : extension]@xmath30 + guarantees on td - learning extend to tdg - learning .
the idea is to reformulate tdg - learning as td - learning , with a slightly different reward function and function approximation . since the function approximation is still linear , any guarantees on convergence for td - learning transfered automatically to tdg - learning .
first , we incorporate @xmath106 into the state - action pair .
define @xmath107 and @xmath108 second , we define a dot product on matrices of equal size by flattening them down to vectors .
more precisely , given two matrices @xmath109 and @xmath110 of the same dimension @xmath111 , define the dot - product @xmath112 .
it is easy to see that @xmath113 the tdg - error can then be rewritten as @xmath114 where @xmath115 is a linear function approximation .
if we are in a setting where td - learning is guaranteed to converge to the value - function , it follows that tdg - learning is also guaranteed to converge since it is simply a different linear approximation .
thus , @xmath116 and the result follows by lemma [ lem : gradl ] .
this section presents our model , which consists of three coupled neural networks that learn to estimate the value function , its gradient , and the optimal policy respectively .
[ def : beh_crit]@xmath30 + the * deviator - actor - critic ( dac ) * model consists in three neural networks : * * actor - network * with policy @xmath117 ; * * critic - network * , @xmath118 , that estimates the value function ; and * * deviator - network * , @xmath119 , that estimates the gradient of the value function .
gaussian noise is added to the policy during training resulting in actions @xmath120 where @xmath121 . the outputs of the critic and deviator are combined as @xmath122 the gaussian noise plays two roles .
firstly , it controls the explore / exploit tradeoff by controlling the extent to which actor deviates from its current optimal policy . secondly , it controls the `` resolution '' at which deviator estimates the gradient .
the three networks are trained by backpropagating three different signals .
critic , deviator and actor backpropagate the tdg - error , the perturbed tdg - error , and deviator s gradient estimate respectively ; see algorithm [ alg : qprop ] .
an explicit description of the weight updates of individual units is provided in appendix [ sec : explicit ] .
deviator estimates the gradient of the value - function with respect to _
deviations @xmath106 from the current policy_. backpropagating the gradient through actor allows to estimate the influence of actor - parameters on the value function as a function of their effect on the policy .
critic and deviator learn representations suited to estimating the value function and its gradient respectively .
note that even though the gradient is a linear function _ at a point _ , it can be a highly nonlinear function in general .
similarly , actor learns a _ policy _ representation .
we set the learning rates of critic and deviator to be equal @xmath123 in the experiments in section [ sec : experiments ] .
however , the perturbation @xmath106 has the effect of slowing down and stabilizing deviator updates : [ rem : stability]@xmath30 + the magnitude of deviator s weight updates depend on @xmath121 since they are computed by backpropagating the perturbed tdg - error @xmath124 .
thus as @xmath125 , deviator s learning rate essentially tends to zero . in general
, deviator learns more slowly than critic .
this has a stabilizing effect on the policy since actor is insulated from critic its weight updates only depend ( directly ) on the output of deviator .
our main result is that the deviator s value gradient is compatible with the policy gradient of each unit in the actor - network considered as an actor in its own right : [ thm : main]@xmath30 + suppose that all units are rectilinear or linear .
then for each actor - unit in the actor - network there exists a reparametrization of the value - gradient approximator , @xmath68 , that satisfies the compatibility conditions in theorem [ t : compat ] .
the actor - network is thus a collection of interdependent agents that individually follow the correct policy gradients .
the experiments below show that they also collectively converge on useful behaviors .
[ [ overview - of - the - proof . ] ] overview of the proof .
+ + + + + + + + + + + + + + + + + + + + + + the next few subsections prove theorem [ thm : main ] .
we provide a brief overview before diving into the details .
guarantees for temporal difference learning and policy gradients are typically based on the assumption that the value - function approximation is a _ linear _ function of the learned parameters . however , we are interested in the case where actor , critic and deviator are all neural networks , and are therefore highly nonlinear functions of their parameters .
the goal is thus to relate the representations learned by neural networks to the prior work on linear function approximations .
to do so , we build on the following observation , implicit in @xcite : [ rem : active]@xmath30 + a neural network of @xmath126 linear and rectilinear units can be considered as a set of @xmath127 submodels , corresponding to different subsets of units .
the active submodel at time @xmath22 consists in the active units ( that is , the linear units and the rectifiers that do not output 0 ) . the active submodel has two important properties :
* it is a _ linear _ function from inputs to outputs , since rectifiers are linear when active , and * at each time step , learning only occurs over the active submodels , since only active units update their weights .
the feedforward sweep of a rectifier network can thus be disentangled into two steps @xcite .
the first step , which is highly nonlinear , applies a gating operation that selects the active submodel by rendering various units inactive .
the second step computes the output of the neural network via matrix multiplication .
it is important to emphasize that although the active submodel is a linear function from inputs to outputs , it is not a linear function of the weights .
the strategy of the proof is to decompose the actor - network in an interacting collection of agents , referred to as actor - units .
that is , we model each unit in the actor - network as an actor in its own right that . on each time step
that an actor - unit is active , it interacts with the deviator - submodel corresponding to the current active submodel of the deviator - network .
the proof shows that each actor - unit has compatible function approximation .
first , we recall some basic facts about backpropagation in the case of _ rectilinear _ units .
recent work has shown that replacing sigmoid functions with rectifiers @xmath128 improves the performance of neural networks @xcite .
let us establish some notation .
the output of a rectifier with weight vector @xmath129 is @xmath130 the rectifier is * active * if @xmath131 .
we use rectifiers because they perform well in practice and have the nice property that units are _ linear _ when they are active .
the rectifier subgradient is the indicator function @xmath132 consider a neural network of @xmath126 units , each equipped with a weight vector @xmath133 .
hidden units are rectifiers ; output units are linear .
there are @xmath126 units in total .
it is convenient to combine all the weight vectors into a single object ; let @xmath134 where @xmath135 .
the network is a function @xmath136 .
the network has error function @xmath137 with gradient @xmath138 .
let @xmath139 denote the output of unit @xmath140 and @xmath141 denote its input , so that @xmath142 .
note that @xmath143 depends on @xmath10 ( specifically , the weights of lower units ) but this is supressed from the notation .
@xmath30 + the * _ influence _ * of unit @xmath140 on unit @xmath144 at time @xmath22 is @xmath145 @xcite .
the influence of unit @xmath140 on the output layer is the vector @xmath146 .
the following lemma summarizes an analysis of the feedforward and feedback sweep of neural nets .
[ lem : structure]@xmath30 + the following properties hold a. * influence . * + a path is * _ active _ * at time @xmath22 if all units on the path are firing .
the influence of @xmath140 on @xmath144 is the sum of products of weights over all active paths from @xmath140 to @xmath144 : @xmath147 where @xmath148 refer to units along the path from @xmath140 to @xmath144 . b. * output decomposition .
* + the output of a neural network decomposes , relative to the output of unit @xmath140 , as @xmath149 where @xmath150 is the @xmath151-matrix whose @xmath152 entry is the sum over all active paths from input unit @xmath153 to output unit @xmath144 that do not intersect unit @xmath140 . c. * output gradient . *
+ fix an input @xmath154 and consider the network as a function from parameters to outputs @xmath155 whose gradient is an @xmath156-matrix .
the @xmath157-entry of the gradient is the input to the unit times its influence : @xmath158 d. * backpropagated error . *
+ fix @xmath154 and consider the function @xmath159 . let @xmath160 .
+ the gradient of the error function is @xmath161 where the backpropagated error signal @xmath162 received by unit @xmath140 decomposes as @xmath163 .
direct computation .
the lemma holds generically for networks of rectifier and linear units .
we apply it to actor , critic and deviator networks below .
this subsection proves condition _
c1 _ of compatible function approximation for a minimal , linear deviator - actor - critic model .
the next subsection shows how the minimal model arises at the level of actor - units .
@xmath30 + the * minimal model * of a deviator - actor - critic consists in an actor with linear policy @xmath164 , where @xmath165 is an @xmath43-vector and @xmath166 is a noisy scalar .
the critic and deviator together output : @xmath167 where @xmath89 is an @xmath43-vector , @xmath168 is a scalar , and @xmath169 is simply scalar multiplication .
the critic in the minimal model is standard .
however , the deviator has been reduced to almost nothing : it learns a single scalar parameter , @xmath168 , that is used to train the actor .
the minimal model is thus too simple to be much use as a standalone algorithm .
[ lem : loc_compat]@xmath30 + there exists a reparametrization of the gradient estimate of the minimal model @xmath170 such that compatibility condition _
c1 _ in theorem [ t : compat ] is satisifed : @xmath171 let @xmath172 and construct @xmath173 .
clearly , @xmath174 observe that @xmath175 and that , similarly , @xmath176 as required .
the proof proceeds by showing that the compatibility conditions in theorem [ t : compat ] hold for each actor - unit .
the key step is to relate the actor - units to the minimal model introduced above .
[ lem : reduction]@xmath30 + actor - units in a dac neural network are equivalent to minimal model actors .
let @xmath177 denote the influence of unit @xmath140 on the output layer of the actor - network at time @xmath22 .
when unit @xmath140 is active , lemma [ lem : structure]ab implies we can write @xmath178 , where @xmath179 is the sum over all active paths from the input to the output of the actor - network that do not intersect unit @xmath140 .
following remark [ rem : active ] , the active subnetwork of the deviator - network at time @xmath22 is a linear transform which , by abuse of notation , we denote by @xmath180 .
combine the last two points to obtain @xmath181 observe that @xmath182 is a @xmath183-vector .
we have therefore reduced actor - unit @xmath140 s interaction with the deviator - network to @xmath183 copies of the minimal model .
theorem [ thm : main ] follows from combining the above lemmas . compatibility condition _
c1 _ follows from lemmas [ lem : loc_compat ] and [ lem : reduction ] .
compatibility condition _
c2 _ holds since the critic and deviator minimize the bellman gradient error with respect to @xmath10 and @xmath11 which also , implicitly , minimizes the bellman gradient error with respect to the corresponding reparametrized @xmath184 s for each actor - unit .
theorem [ thm : main ] shows that each actor - unit satisfies the conditions for compatible function approximation and so follows the correct gradient when performing weight updates .
it is interesting to relate our approach to the literature on multiagent reinforcement learning @xcite . in particular , @xcite consider the _
structural credit assignment problem _ within populations of interacting agents : how to reward individual agents in a population for rewards based on their collective behavior ? they propose to train agents within populations with a _ difference - based objective _ of the form @xmath185 where @xmath1 is the objective function to be maximized ; @xmath186 and @xmath187 are the system variables that are and are not under the control of agent @xmath140 respective , and @xmath188 is a fixed counterfactual action . in our setting
, the gradient used by actor - unit @xmath140 to update its weights can be described explicitly : @xmath30 + actor - unit @xmath140 follows policy gradient @xmath189 = \operatorname*{\mathbb e}\left[\operatorname*{\nabla}_{\theta^j } { \ensuremath{\boldsymbol\mu}}_{\theta^j}(s)\cdot \big\langle { \ensuremath{\boldsymbol\pi}}^j,{\mathbf{g}}^{\mathbf{w}}(\tilde{{\mathbf{s}}})\big\rangle\right],\ ] ] where @xmath190 is deviator s estimate of the directional derivative of the value function in the direction of actor - unit @xmath140 s influence
. follows from lemma [ lem : structure]b .
notice that @xmath191 in eq . .
it follows that training the actor - network via @xmath0 causes the actor - units to optimize the difference - based objective without requiring to compute the difference explicitly .
although the topic is beyond the scope of the current paper , it is worth exploring how suitably adapted variants of backpropagation can be applied to the reinforcement learning problems in the multiagent setting . [ [ comparison - with - textttcopdac - mathttq . ] ] comparison with @xmath8 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + extending the standard value function approximation in example [ eg : advantage ] to the setting where actor is a neural network yields the following representation , which is used in @xcite when applying @xmath8 to the octopus arm task : [ eg : deep_advantage]@xmath30 + let @xmath192 and @xmath193 be an actor and critic neural network respectively .
suppose the actor - network has @xmath194 parameters ( i.e. the total number of entries in @xmath9 ) .
it follows that the jacobian @xmath195 is an @xmath156-matrix .
the value function approximation is then @xmath196 where @xmath129 is an @xmath194-vector .
weight updates under @xmath8 , with the function approximation above , are therefore as described in algorithm [ alg : copdac ] .
let us compare @xmath0 with @xmath8 , considering the three updates in turn : * _ actor updates . _ + under @xmath0 , the actor backpropagates the value - gradient estimate .
in contrast under @xmath8 the actor performs a complicated update that combines the policy gradient @xmath197 with the advantage function s weights and differs substantively from backprop . * _ deviator / advantage - function updates . _
+ under @xmath0 , the deviator backpropagates the perturbed tdg - error .
in contrast , @xmath8 uses the gradient of the _ actor _ to update the weight vector @xmath129 of the advantage function . + by lemma [ lem : structure]d
, backprop takes the form @xmath198 where @xmath199 is a @xmath183-vector .
in contrast , the advantage function requires computing @xmath200 , where @xmath129 is an @xmath194-vector .
although the two formulae appear similarly superficially , they carry very different computational costs .
+ the first consequence is that the parameters of @xmath129 must exactly line up with those of the policy .
the second consequence is that , by lemma [ lem : structure]c , the advantage function requires access to @xmath201 where @xmath202 is the input from unit @xmath153 to unit @xmath140 .
thus , the advantage function requires access to the input @xmath203 and the influence @xmath204 of every unit in the actor - network . * _ critic updates . _
+ the critic updates for the two algorithms are essentially identical , with the td - error replaced with the tdg - error . in short ,
the approximation in example [ eg : deep_advantage ] that is used by @xmath8 is thus not well - adapted to deep learning .
the main reason is that learning the advantage function requires coupling the vector @xmath129 with the parameters @xmath9 of the actor .
[ [ comparison - with - computing - the - gradient - of - the - value - function - approximation . ] ] comparison with computing the gradient of the value - function approximation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + perhaps the most natural approach to estimating the gradient is to simply estimate the value function , and then use its gradient as an estimate of the derivative @xcite .
the main problem with this approach is that , to date , it has not been show that the resulting updates of the critic and the actor are compatible .
there are also no guarantees that the gradient of the critic will be a good approximation to the gradient of the value function
although it is intuitively plausible .
the problem becomes particularly severe when the value - function is estimated via a neural network that uses activation functions that are _ not smooth _ such as rectifers .
rectifiers are becoming increasingly popular due to their superior empirical performance @xcite .
we evaluate @xmath0 on three tasks : two highly nonlinear contextual bandit tasks constructed from benchmark datasets for nonparametric regression , and the octopus arm .
we do not evaluate @xmath0 on other standard reinforcement learning benchmarks such as mountain car , pendulum or puddle world , since these can already be handled by _ linear _ actor - critic algorithms .
the contribution of @xmath0 is the ability to learn representations suited to nonlinear problems . [
[ cloning - and - replay . ] ] cloning and replay .
+ + + + + + + + + + + + + + + + + + + temporal difference learning can be unstable when run over a neural network . a recent innovation introduced in @xcite that _ stabilizes _ td - learning is to clone a separate network @xmath205 to compute the targets @xmath206 .
the parameters of the cloned network are updated periodically .
we implement a similar modification of the tdg - error in algorithm [ alg : qprop ] .
we also use experience replay @xcite .
@xmath0 is well - suited to replay , since the critic and deviator can learn values and gradients over the full range of previously observed state - action pairs offline .
cloning and replay were also applied to @xmath8 .
both algorithms were implemented in theano @xcite .
the goal of the contextual bandit tasks is to probe the ability of reinforcement learning algorithms to accurately estimate gradients .
the experimental setting may thus be of independent interest .
[ [ description . ] ] description .
+ + + + + + + + + + + + we converted two robotics datasets , sarcos and barrett wam , into contextual bandit problems via the supervised - to - contextual - bandit transform in @xcite .
the datasets have 44,484 and 12,000 training points respectively , both with 21 features corresponding to the positions , velocities and accelerations of seven joints .
labels are 7-dimensional vectors corresponding to the torques of the 7 joints . in the contextual bandit task , the agent samples 21-dimensional state vectors i.i.d . from either the sarcos or barrett training data and
executes 7-dimensional actions .
the reward @xmath207 is the negative mean - square distance from the action to the label .
note that the reward is a scalar , whereas the correct label is a 7-dimensional vector .
the gradient of the reward @xmath208 is the direction from the action to the correct label . in the supervised setting ,
the gradient can be computed . in the bandit
setting , the reward is a zeroth - order black box .
the agent thus receives far less information in the bandit setting than in the fully supervised setting .
intuitively , the negative distance @xmath50 `` tells '' the algorithm that the correct label lies on the surface of a sphere in the 7-dimensional action space that is centred on the most recent action .
by contrast , in the supervised setting , the algorithm is given the position of the label in the action space . in the bandit
setting , the algorithm must estimate the position of the label on the surface of the sphere .
equivalently , the algorithm must estimate the label s direction relative to the center of the sphere which is given by the _ gradient _ of the value function .
the goal of the contextual bandit task is thus to _ simultaneously solve seven nonparametric regression problems when observing distances - to - labels instead of directly observing labels_. the value function is relatively easy to learn in contextual bandit setting since the task is not sequential .
however , both the value function and its gradient are highly nonlinear , and it is precisely the gradient that specifies where labels lie on the spheres . [ [ network - architectures . ] ] network architectures .
+ + + + + + + + + + + + + + + + + + + + + + @xmath0 and @xmath8 were implemented on an actor and deviator network of two layers ( 300 and 100 rectifiers ) each and a critic with a hidden layers of 100 and 10 rectifiers .
updates were computed via rmsprop with momentum .
the variance of the gaussian noise @xmath209 was set to decrease linearly from @xmath210 until reaching @xmath211 at which point it remained fixed . [ [ performance . ] ] performance .
+ + + + + + + + + + + + figure [ f : robo ] compares the test - set performance of policies learned by @xmath0 against @xmath8 .
the final policies trained by @xmath0 achieved average mean - square test error of 0.013 and 0.014 on the seven sarcos and barrett benchmarks respectively . remarkably , @xmath0 is competitive with fully - supervised nonparametric regression algorithms on the sarcos and barrett datasets , see figure 2bc in @xcite and the results in @xcite .
it is important to note that the results reported in those papers are for algorithms that are given the labels and that solve _ one regression problem at a time_. to the best of our knowledge , there are no prior examples of a bandit or reinforcement learning algorithm that is competitive with fully supervised methods on regression datasets . for comparison , we implemented backprop on the actor - network under full - supervision .
backprop converged to .006 and .005 on sarcos and barrett , compared to 0.013 and 0.014 for @xmath0 .
note that backprop is trained on 7-dim labels whereas @xmath0 receives 1-dim rewards .
[ [ accuracy - of - gradient - estimates . ] ] accuracy of gradient - estimates .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the true value - gradients can be computed and compared with the algorithm s estimates on the contextual bandit task .
[ f : robo_grad ] shows the performance of the two algorithms .
gradient - error converges to @xmath212 on both tasks .
s gradient estimate , implicit in the advantage function , converges to 0.03 ( sarcos ) and 0.07 ( barrett ) .
this confirms that @xmath0 yields significantly better gradient estimates .
@xmath8 s estimates are significantly worse for barrett compared to sarcos , in line with the worse performance of @xmath8 on barrett in fig .
[ f : robo ] .
it is unclear why @xmath8 s gradient estimate gets worse on barrett for some period of time . on the other hand ,
since there are no guarantees on @xmath8 s estimates , it follows that its erratic behavior is perhaps not surprising .
[ [ comparison - with - bandit - task - in- . ] ] comparison with bandit task in @xcite .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + note that although the contextual bandit problems investigated here are lower - dimensional ( with 21-dimensional state spaces and 7-dimensional action spaces ) than the bandit problem in @xcite ( with no state space and 10 , 25 and 50-dimensional action spaces ) , they are nevertheless much harder .
the optimal action in the bandit problem , in all cases , is the constant vector @xmath213 $ ] consisting of only 4s .
in contrast , sarcos and barrett are nontrivial benchmarks even when fully supervised .
the octopus arm task is a challenging environment that is high - dimensional , sequential and highly nonlinear .
[ [ desciption . ] ] desciption .
+ + + + + + + + + + + the objective is to learn to hit a target with a simulated octopus arm @xcite .
settings are taken from @xcite .
importantly , the action - space is _ not _ simplified using `` macro - actions '' .
the arm has @xmath214 compartments attached to a rotating base .
there are @xmath215 state variables ( @xmath216 , @xmath217 position / velocity of nodes along the upper / lower side of the arm ; angular position / velocity of the base ) and @xmath218 action variables controlling the clockwise and counter - clockwise rotation of the base and three muscles per compartment . after each step , the agent receives a reward of @xmath219 , where @xmath220 is the change in distance between the arm and the target .
the final reward is @xmath221 if the agent hits the target .
an episode ends when the target is hit or after 300 steps .
the arm initializes at eight positions relative to the target : @xmath222 .
see appendix [ sec : octopus_details ] for more details .
[ [ network - architectures.-1 ] ] network architectures .
+ + + + + + + + + + + + + + + + + + + + + + we applied @xmath0 to an actor - network with @xmath223 hidden rectifiers and linear output units clipped to lie in @xmath224 $ ] ; and critic and deviator networks both with two hidden layers of @xmath223 and @xmath225 rectifiers , and linear output units .
updates were computed via rmsprop with step rate of @xmath226 , moving average decay , with nesterov momentum @xcite penalty of @xmath227 and @xmath227 respectively , and discount rate @xmath228 of @xmath229 .
the variance of the gaussian noise was initialized to @xmath210 .
an explore / exploit tradeoff was implemented as follows .
when the arm hit the target in more than 300 steps , we set @xmath230 ; otherwise @xmath231 .
a hard lower bound was fixed at @xmath232 .
we implemented copdac - q on a variety of architectures ; the best results are shown ( also please see figure 3 in @xcite ) .
they were obtained using a similar architecture to @xmath0 , with sigmoidal hidden units and sigmoidal output units for the actor .
linear , rectilinear and clipped - linear output units were also tried . as for @xmath0 , cloning and experience replay
were used to increase stability .
[ [ performance.-1 ] ] performance .
+ + + + + + + + + + + + figure [ f : oct ] shows the steps - to - target and average - reward - per - step on ten training runs .
@xmath0 converges rapidly and reliably ( within @xmath233 steps ) to a stable policy that uses less than 50 steps to hit the target on average ( see supplementary video for examples of the final policy in action ) .
@xmath0 converges quicker , and to a better solution , than @xmath8 .
the reader is strongly encouraged to compare our results with those reported in @xcite . to the best of our knowledge
, @xmath0 achieves the best performance to date on the octopus arm task .
[ [ stability . ] ] stability .
+ + + + + + + + + + it is clear from the variability displayed in the figures that both the policy and the gradients learned by @xmath0 are more stable than @xmath8 .
note that the higher variability exhibited by @xmath0 in the right - hand panel of fig .
[ f : oct ] ( rewards - per - step ) is misleading .
it arises because dividing by the number of steps which is lower for @xmath0 since it hits the target more quickly after training inflates @xmath0 s apparent variability .
value - gradient backpropagation ( @xmath234 is the first deep reinforcement learning algorithm with compatible function approximation for continuous policies .
it builds on the deterministic actor - critic , @xmath8 , developed in @xcite with two decisive modifications .
first , we incorporate an explicit estimate of the value gradient into the algorithm .
second , we construct a model that decouples the internal structure of the actor , critic , and deviator so that all three can be trained via backpropagation .
@xmath0 achieves state - of - the - art performance on two contextual bandit problems where it simultaneously solves seven regression problems without observing labels .
note that @xmath0 is competitive with recent _ fully supervised _
methods that solve a _
single _ regression problem at a time .
further , @xmath0 outperforms the prior state - of - the - art on the octopus arm task , quickly converging onto policies that rapidly and fluidly hit the target .
[ [ acknowledgements . ] ] acknowledgements .
+ + + + + + + + + + + + + + + + + we thank nicolas heess for sharing the settings of the octopus arm experiments in @xcite .
45 [ 1]#1 [ 1]`#1 ` urlstyle [ 1]doi : # 1 adrian k agogino and kagan tumer .
unifying temporal and structural credit assignment problems . in _ aamas _ , 2004 .
adrian k agogino and kagan tumer . analyzing and visualizing multiagent rewards in dynamic and stochastic environments .
_ journal of autonomous agents and multi - agent systems _ , 170 ( 2):0 320338 , 2008 .
l c baird .
residual algorithms : reinforcement learning with function approximation . in _ icml _
, 1995 .
david balduzzi . .
in _ arxiv:1509.01851 _ , 2015 .
david balduzzi , hastagiri vanchinathan , and joachim buhmann .
kickback cuts backprop s red - tape : biologically plausible credit assignment in neural networks . in _ aaai _ , 2015 .
andrew g barto , richard s sutton , and charles w anderson .
neuronlike adapative elements that can solve difficult learning control problems .
_ ieee trans .
systems , man , cyb _ , 130 ( 5):0 834846 , 1983 .
f bastien , p lamblin , r pascanu , j bergstra , i goodfellow , a bergeron , n bouchard , and y bengio . .
in _ nips workshop : deep learning and unsupervised feature learning _ , 2012 .
j bergstra , o breuleux , f bastien , p lamblin , r pascanu , g desjardins , j turian , d warde - farley , and yoshua bengio .
theano : a cpu and gpu math expression compiler . in _ proc .
python for scientific comp .
( scipy ) _ , 2010 .
george e dahl , tara n sainath , and geoffrey hinton .
improving deep neural networks for lvcsr using rectified linear units and dropout . in _
ieee int conf on acoustics , speech and signal processing ( icassp ) _ , 2013 .
christoph dann , gerhard neumann , and jan peters .
policy evaluation with temporal differences : a survey and comparison .
_ jmlr _ , 15:0 809883 , 2014 .
marc peter deisenroth , gerhard neumann , and jan peters . .
_ foundations and trends in machine learning _ , 20 ( 1 - 2):0 1142 , 2011 . miroslav dudk , dumitru erhan , john langford , and lihong li . .
_ statistical science _ , 290 ( 4):0 485511 , 2014
. y engel , p szab , and d volkinshtein . learning to control an octopus arm with gaussian process temporal difference methods . in _ nips _ , 2005 .
michael fairbank and eduardo alonso . .
ieee world conference on computational intelligence ( wcci ) _ , 2012 .
michael fairbank , eduardo alonso , and daniel v prokhorov . .
_ ieee trans .
_ , 240 ( 12):0 20882100 , 2013 .
abraham flaxman , adam kalai , and h brendan mcmahan .
online convex optimization in the bandit setting : gradient descent without a gradient . in _ soda _ , 2005 .
xavier glorot , antoine bordes , and yoshua bengio . eep sparse rectifier neural networks . in _ proc .
14th int conference on artificial intelligence and statistics ( aistats ) _ , 2011 .
carlos guestrin , michail lagoudakis , and ronald parr .
coordinated reinforcement learning . in _
icml _ , 2002 .
roland hafner and martin riedmiller .
reinforcement learning in feedback control : challenges and benchmarks from technical process control .
_ machine learning _ , 84:0 137169 , 2011 .
g hinton , nitish srivastava , and kevin swersky .
lecture 6a : overview of minibatch gradient descent .
chris holmesparker , adrian k agogino , and kagan tumer .
combining reward shaping and hierarchies for scaling to large multiagent systems . _ the knowledge engineering review _
, 2014 .
michael i jordan and r a jacobs . .
in _ nips _ , 1990 .
sham kakade .
a natural policy gradient . in _ nips _ , 2001
. vijay r konda and john n tsitsiklis .
actor - critic algorithms . in _ nips _ , 2000 .
samory kpotufe and abdeslam boularias .
gradient weights help nonparametric regressors . in _ advances in neural information processing systems ( nips ) _ , 2013 .
sergey levine , chelsea finn , trevor darrell , and pieter abbeel . .
_ arxiv:1504.00702 _ , 2015 .
volodymyr mnih , koray kavukcuoglu , david silver , andrei a. rusu , joel veness , marc g. bellemare , alex graves , martin riedmiller , andreas k. fidjeland , georg ostrovski , stig petersen , charles beattie , amir sadik , ioannis antonoglou , helen king , dharshan kumaran , daan wierstra , shane legg , and demis hassabis .
human - level control through deep reinforcement learning .
_ nature _ , 5180 ( 7540):0 529533 , 02 2015 .
vinod nair and geoffrey hinton .
ectified linear units improve restricted boltzmann machines . in _
icml _ , 2010
. a s nemirovski and d b yudin .
_ problem complexity and method efficiency in optimization_. wiley - interscience , 1983 .
duy nguyen - tuong , jan peters , and matthias seeger . .
in _ nips _ , 2008 .
jan peters and stefan schaal .
policy gradient methods for robotics . in _ proc .
ieee / rsj int .
robots syst .
. daniel v prokhorov and donald c wunsch . .
_ ieee trans .
_ , 80 ( 5):0 9971007 , 1997 . maxim raginsky and alexander rakhlin . .
_ ieee trans .
inf . theory _ , 570 ( 10):0 70367056 , 2011 . david silver , guy lever , nicolas heess , thomas degris , daan wierstra , and martin riedmiller .
deterministic policy gradient algorithms . in _ icml _ , 2014 .
nitish srivastava , geoffrey hinton , alex krizhevsky , ilya sutskever , and ruslan salakhutdinov .
dropout : a simple way to prevent neural networks from overfitting .
_ jmlr _ , 15:0 19291958 , 2014 .
r s sutton and a g barto .
_ reinforcement learning : an introduction_. mit press , 1998 .
richard sutton , david mcallester , satinder singh , and yishay mansour .
policy gradient methods for reinforcement learning with function approximation . in _ nips _ , 1999 .
richard sutton , hamid reza maei , doina precup , shalabh bhatnagar , david silver , csaba szepesvri , and eric wiewiora .
fast gradient - descent methods for temporal - difference learning with linear function approximation . in _ icml _ , 2009 .
richard sutton , csaba szepesvri , and hamid reza maei . a convergent @xmath235 algorithm for off - policy temporal - difference learning with linear function approximation . in _ adv in neural information processing systems ( nips )
_ , 2009 .
shubhendu trivedi , jialei wang , samory kpotufe , and gregory shakhnarovich . .
in _ uai _ , 2014 .
john tsitsiklis and benjamin van roy .
an analysis of temporal - difference learning with function approximation .
_ ieee trans .
, 420 ( 5):0 674690 , 1997 .
niklas wahlstrm , thomas b. schn , and marc peter deisenroth . .
_ arxiv:1502.02251 _ , 2015 .
y wang and j si . .
_ ieee trans .
_ , 120 ( 2):0 264276 , 2001 .
ronald j williams .
simple statistical gradient - following algorithms for connectionist reinforcement learning .
_ machine learning _ , 8:0 229256 , 1992 .
m d zeiler , m ranzato , r monga , m mao , k yang , q v le , p nguyen , a senior , v vanhoucke , j dean , and g hinton . on rectified linear units for speech processing . in _
icassp _ , 2013 .
* appendices *
it is instructive to describe the weight updates under @xmath0 more explicitly .
let @xmath236 , @xmath237 and @xmath238 denote the weight vector of unit @xmath140 , according to whether it belongs to the actor , deviator or critic network .
similarly , in each case @xmath239 or @xmath240 denotes the influence of unit @xmath140 on the network s output layer , where the influence is vector - valued for actor and deviator networks and scalar - valued for the critic network .
weight updates in the deviator - actor - critic model , where all three networks consist of rectifier units performing stochastic gradient descent , are then per algorithm [ alg : pb ] .
units that are not active on a round do not update their weights that round .
listing 1 summarizes technical information with respect to the physical description and task setting used in the octopus arm simulator in xml format .
< constants > < frictiontangential>0.4</frictiontangential > < frictionperpendicular>1</frictionperpendicular > <
pressure>10</pressure > < gravity>0.01</gravity > < surfacelevel>5</surfacelevel > < buoyancy>0.08</buoyancy >
< muscleactive>0.1</muscleactive > < musclepassive>0.04</musclepassive >
< musclenormalizedminlength>0.1</musclenormalizedminlength > < muscledamping>-1</muscledamping > < repulsionconstant>.01</repulsionconstant > < repulsionpower>1</repulsionpower > < repulsionthreshold>0.7</repulsionthreshold >
< torquecoefficient>0.025</torquecoefficient > < /constants > | this paper proposes @xmath0 , a deep reinforcement learning algorithm for continuous policies with compatible function approximation .
the algorithm is based on two innovations .
firstly , we present a temporal - difference based method for learning the _ gradient _ of the value - function . secondly , we present the deviator - actor - critic ( dac ) model , which comprises three neural networks that estimate the value function , its gradient , and determine the actor s policy respectively .
we evaluate @xmath0 on two challenging tasks : a contextual bandit problem constructed from nonparametric regression datasets that is designed to probe the ability of reinforcement learning algorithms to accurately estimate gradients ; and the octopus arm , a challenging reinforcement learning benchmark .
@xmath0 is competitive with _ fully supervised methods _ on the bandit task and achieves the best performance to date on the octopus arm .
policy gradient , reinforcement learning , deep learning , gradient estimation , temporal difference learning |
long - distance electron transfer ( et ) plays an essential role in biological energy conversion @xcite .
representative are those in photosynthetic reaction centers in which photon energy is converted to electrochemical energy via series of ets through redox centers embedded in transmembrane protein .
a simple but fundamental question open to microscopic investigation is how the protein environment is involved in the et : the protein structure could be involved passively by simply holding the redox centers at appropriate spatial configuration , or actively by providing intermediate virtual states for superexchange et mechanism @xcite . to address this and related questions ,
quantum mechanical investigation based on realistic molecular structure is essential .
however , first - principles treatment of electrons in large biomolecular systems is still a formidable task . in this
regard , fragment - based approaches , such as the fragment molecular orbital ( fmo ) @xcite , divide - and - conquer @xcite , and many others @xcite appear promising . in a series of papers , we have reported calculations of et coupling matrix element and et pathways @xcite from the linear - combination of fmos ( fmo - lcmo ) @xcite with the two - body correction ( fmo2 ) .
the method was found to give accurate et couplings over four orders of magnitude along the et distance when the bond detached atoms ( bda ) were not involved or when the minimal atomic basis set was used @xcite .
nevertheless , the problem of degraded mo energies caused by bdas with atomic basis sets larger than minimal set , that had been already pointed out in ref .
@xcite , carried over to the et analysis .
recently , however , kobori et al . have found notable remedy of mo energies with the three - body correction ( fmo3 ) @xcite . following this
, we study in this work how the fmo3 correction affects the et coupling energy and et pathway analysis .
we also examine selection of the fmo space . in ref .
@xcite , in order to remove linear dependence of basis functions associated with bdas , a canonical transformation of hamiltonian matrix , @xmath0 where the matrix @xmath1 diagonalizes overlap matrix , was employed .
however , this transformation often mixes the fmos in unwanted ways for the et pathway analysis , particularly for studying inter - fragment tunneling current .
we found that the problem can be evaded by a projection to restricted fmo space instead of the canonical transformation of eq .
( [ eq : uhu ] ) .
for instance , with fmo - lc(vc)mo scheme , which restricts to the minimal - valence ( v ) plus core ( c ) mo space , the smallest eigenvalue of overlap matrix for systems studied in this work was 0.225 , which is large enough to regard the fmo space linearly independent @xcite . for numerical demonstration , we examine hole transfer between two triptophane ( trp ) residues bridged by helical polyproline oligopeptide , which serves a good model of long - distance ets observed in metal - derivatized oligoprolines @xcite .
previous theoretical works @xcite have employed the same model systems to study the effects of solvent and bridge - conformation dynamics on the electronic coupling .
section [ sec : theory ] outlines the theoretical framework .
section [ sec : comput ] describes the computational details .
applications to hole transfer between two trp molecules bridged by polyproline linker systems are discussed in sec .
[ sec : results ] .
section [ sec : concl ] concludes .
here we outline the fmo - lcmo method @xcite to explain the present projection scheme to the restricted fmo space .
the fmo - lcmo hamiltonian up to the three - body fmo3 correction is described as @xmath2 in which @xmath3 consists of monomer fock matrices , and @xmath4 and @xmath5 are the two- and three - body corrections .
the intra - fragment block of @xmath4 , between fmos @xmath6 and @xmath7 in the same fragment @xmath8 , is given by @xmath9 whereas the inter - fragment block is @xmath10 the terms in the right - hand - side of eqs .
( [ eq : fmo2ii ] ) and ( [ eq : fmo2ij ] ) will be defined in eqs .
( [ eq : hii])([eq : hijx ] ) . the fmo3 correction to the intra- and
inter - fragment blocks are @xmath11 and @xmath12 as noted in introduction , the canonical transformation of eq .
( [ eq : uhu ] ) is not desired for the et pathway analysis .
we thus simply limit the number of fmos and project the hamiltonian matrix to this set .
we denote this restricted fmo ( rfmo ) space . therefore , the terms in the right - hand - side of eqs .
( [ eq : fmo2ii])([eq : fmo3ij ] ) in the selected rfmo space @xmath13 are defined as follows . with the fragment monomer , dimer , and trimer represented by @xmath14 , and @xmath15 ,
the intra - fragment blocks are defined by @xmath16 and @xmath17 whereas the inter - fragment blocks are @xmath18 in the summation over @xmath19 in the right - hand - side , all the dimer and trimer fmos are taken , except the spurious ones stemming from the bdas . to calculate the et coupling matrix element @xmath20 , we employ two methods ; generalized mulliken - hush ( gmh ) @xcite and bridge green function ( bgf ) @xcite .
the gmh method scales the donor - acceptor mo energy splitting @xmath21 by a formula @xmath22 in which @xmath23 , @xmath24 and @xmath25 are the diagonal and off - diagonal dipole matrix elements .
it assumes that the hamiltonian and dipole matrix elements scale similarly for states involved in ets . despite its simplicity
, the gmh formula ( [ eq : gmh ] ) has been successfully applied to a number of et reactions .
the bgf method has been also demonstrated to give accurate and robust results with @xmath26 in which the sums over @xmath27 and @xmath28 exclude donor and acceptor mos , @xmath29 and @xmath30 .
the first term in the right - hand - side is the direct coupling between @xmath29 and @xmath30 .
@xmath31 is the overlap matrix .
@xmath32 is the bridge green function defined as @xmath33 in which @xmath34 is the projection operator to the mo space external to the donor - acceptor mos .
the electron tunneling energy @xmath35 is naturally defined as the average of donor - acceptor orbital energies , @xmath36 . in the tunneling current analysis @xcite
, the et coupling @xmath20 is expressed as a sum of tunneling current @xmath37 between basis fmos @xmath13 @xcite , @xmath38 @xmath39 in which the summation over @xmath27 and @xmath28 are over the fmos within fragments @xmath8 and @xmath40 , and @xmath41 denotes the spatial region assigned to the donor molecule .
the inter - orbital current @xmath37 is computed from the electronic hamiltonian and overlap matrices and the coefficients of fmo - lcmo , @xmath42 and @xmath43 , that represent the mixing of bridge fmos to the donor and acceptor fmos , @xmath29 and @xmath30 , in the initial ( ) and final ( ) states , @xmath44 and @xmath45 , @xmath46 @xmath47 @xmath48 these are thus computed straightforwardly from the fmo - lcmo method .
the normalized inter - fragment tunneling current is defined by @xmath49 which satisfies @xmath50
for the purpose of benchmarking the fmo3-lcmo calculations , we consider proline - trimer bridged systems , trp-(pro)@xmath51-trp , with two types of helix structure of polyproline ( pp ) linker , one with cis - configurations ( cispp ) and the other with trans - configurations ( transpp ) of peptide bonds .
the former and latter proline trimers have the backbone dihedral angles @xmath52 of approximately @xmath53 and @xmath54 , respectively .
the cispp and transpp structures are schematically drawn in fig .
1 . apparently the transpp is more stretched .
starting from typical conformations of cispp and transpp , local strains were removed by geometry optimization at the b3lyp/6 - 31g(d)-d3 level , with the d3 version of grimme s dispersion correction with the original d3 damping function @xcite .
we used gaussian09 program @xcite for the geometry optimization .
the resultant molecular structures are displayed in fig .
s1 and table s1 of the supplementary material . in the fmo calculation , the trp-(pro)@xmath51-trp chain was divided into six fragments as designated in fig .
2 , in which p1-p3 are the three prolines and mw denotes the main - chain of trp ( w ) .
the @xmath55-carbon atoms were treated as bdas .
the homos of trp fragments were taken as the donor and acceptor mos of hole transfer .
the electronic coupling and tunneling current from the fmo2-lcmo and fmo3-lcmo methods were compared to a reference calculation with the rhf hamiltonian of the entire system of six fragments projected to the fmo space .
we denote this last scheme fmo6-lcmo with the hamiltonian matrix @xmath56 where @xmath57 are mos in the rfmo space and @xmath58 and @xmath59 are the mos and mo energies of the entire system .
we consider a minimal - valence plus core rfmo space that includes the same number of mos as the minimal basis set such as sto-3 g . we denote this lc(vc)mo space .
other choices of rfmo space , lumo , lumo+2 , lumo+6 , lumo+10 , to check the dependence on the size of rfmo space , are obtained by augmenting the lower unoccupied mos of each fragment to the occupied space .
the 6 - 31g(d ) basis set was used .
the total number of basis aos for the entire system is 774 .
the fmo2 and fmo3 calculations include additional 50 aos from the bdas .
the number of fmos in the occupied , lc(vc)mo , and full spaces for each fragment are summarized in table i. for basis sets larger than the double - zeta basis , the lc(vc)mo scheme gives significant reduction .
moreover , it removed the problem of linear - dependence observed previously @xcite , as the smallest eigenvalue of the overlap matrix in the lc(vc)mo space was 0.225 .
we used the program gamess @xcite for the conventional fmo calculation @xcite . to estimate all inter - fragment tunneling currents including the long - distance ones
, we did not employ the electrostatic dimer ( es - dim ) approximation @xcite that avoids self - consistent field calculations of the far separated dimers in fmo calculations . | the linear - combination of fragment molecular orbitals with three - body correction ( fmo3-lcmo ) is examined for electron transfer ( et ) coupling matrix elements and et pathway analysis , with application to hole transfer between two triptophanes bridged by cis- and trans - polyproline linker conformations . a projection to the minimal - valence - plus - core fmo space
was found to give sufficient accuracy with significant reduction of computational cost while avoiding the problem of linear dependence of fmos stemming from involvement of bond detached atoms . |
study of the parameters of galactic components via decomposition of the rotation curves is often ambiguous in the case of spiral galaxies .
the same rotation curve often may be explained by models with very different parameters of the components .
the dark - to - light mass ratio may vary several times between the models explaining well the same rotation curve , especially if the latter is smooth and featureless ( see for example * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
incorporating information about the stellar velocity dispersion from observations helps to significantly constrain the masses of galactic components @xcite and relieve some ambiguities that appear in the modeling of rotation curves , but observing the distributions of the velocity dispersion is a rather difficult task .
the stellar disk thickness can be used as an alternative to the stellar velocity dispersion @xcite .
studies of the vertical structure of stellar disks alone allow the constraint of some dark halo parameters @xcite .
note that not all galaxies are suitable for such studies : inferring the structural parameters for a spiral galaxy with a significant bulge from photometric data alone is dangerous and should be supplemented with concerns of different mass - to - light ratio for different galactic subsystems ( see discussion by * ? ? ?
* regarding difference between photometrically and dynamically identified subsystems ) .
infrared observations are crucial for studies of the structure of edge - on galaxies .
the dust attenuation makes the extraction of the structural parameters difficult from the optical data . in the nir bands the galaxies are much more transparent , which facilitates studies of the structure of galaxies
the all - sky 2mass survey @xcite offers a good opportunity to increase the number of edge - on galaxies available for studies in the near - infrared bands .
although faint parts of the galaxies are not seen in the 2mass images , the thin disks of the galaxies are obtained with sufficient signal - to - noise ratio ( s / n hereafter ) . in this paper
we focus on the determination of stellar disk structural parameters corrected for the internal extinction .
one approach to the problem is to perform the modeling of the radiation transfer in 3-d stellar - gaseous disks ( like in * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
such studies require rather deep images , although there are attempts to apply this kind of analysis to larger sample of moderate quality @xcite . in the present paper
we develop another approach to study the extinction - corrected structure of the stellar galactic disks comparing the parameters of the disks obtained from different nir bands . in section 2
we discuss the sample selection and evaluation of the stellar disk s structural parameters , consider reliability of the parameters , and study the radial variation of the disk scaleheight . in sections 3 and 4 we show how the comparison of the parameters obtained in the three nir bands can be used to estimate the extinction - corrected scaleheight in the galactic disks .
section 5 considers the connection between the galactic disk thickness with parameters of the dark halos around the spiral galaxies in the frames of a simplified model .
section 6 summarizes results of the paper .
as a pre - defined sample of edge - on galaxies , the revised flat galaxies catalog ( rfgc hereafter , * ? ? ?
* ) is chosen .
the catalog contains visually - classified extragalactic objects with major - to - minor axes ratio a / b @xmath5 7 . to secure enough pixels for studying the vertical structure , we select objects with the major axis greater than 1 arcmin .
this size is estimated directly from the 2mass images at the level of s / n @xmath6 3 .
note that this size is significantly less than commonly used diameter d@xmath7 . as a result
, we select 139 spiral edge - on galaxies that have images in all three 2mass bands : h , j , and k@xmath0 .
we apply the technique described by @xcite to obtain the structural parameters of stellar disks of the galaxies disks .
this method is based on the analysis of photometric profiles drawn parallel to the major and minor axes of galaxy at a one - pixel interval .
the volume brightness in the stellar disks is assumed to change in the radial @xmath8 or vertical @xmath9 directions as follows : @xmath10 where @xmath1 and @xmath2 are the scalelength and scaleheight of the disk , respectively , and @xmath11 is the central surface luminosity of the face - on disk , which corresponds to the central surface brightness @xmath3 . the model photometric profiles are obtained by the integration of equation ( [ vol_disk ] ) along the line - of - sight and then by convolution with the instrumental profile .
the best - fit parameters @xmath1 , @xmath2 , and @xmath12 are estimated for each radial or vertical profile , and then their median values are taken as the resulting scales and central surface brightness of the stellar disks , whereas the standard deviations of @xmath1 , @xmath2 , and @xmath12 estimated from different profiles are taken as uncertainty of the parameters .
the central regions of the galaxies in which bulges may exist are excluded from our analysis .
the parameters @xmath1 , @xmath2 , and @xmath3 are obtained in the j , h , and k@xmath0 bands for selected 139 galaxies .
the @xmath13-based parameters are shown in table 1 .
figure 1 shows histograms of distribution of the scale heights ( top panel ) and scale lengths ( bottom panel ) estimated for our program galaxies . in both panels , the j- and h - band scales are normalized by those in the k@xmath0 band . it can be noticed that the scaleheights look systematically larger , on average , in the j band than those in the k@xmath0 for the same galaxy .
we will discuss this effect below : in [ three_band ] using our observational data , and in
[ im_sim ] with the help of artificial data . as in @xcite
, we investigate how the scaleheight @xmath2 of the stellar disks in our galaxies change with the distance to the center @xmath8 .
our approach to the estimation of the @xmath2 is independent of the apriori assumptions about behavior of the scaleheight with radius , in a contrast with 3-d modeling of the radiation transfer in the stellar and dust disks . since we analyze the vertical profiles separately from each other , this approach is insensitive to possible bending of the galactic disks .
note that if the resulting scaleheight @xmath14 demonstrates a systematic growth with distance to the center , it suggests that the real radial gradient of the disk scaleheight may be a few times greater due to the projection effects .
we fit @xmath14 with a linear function of radius @xmath8 and find the radial gradients of @xmath2 ( expressed in units of @xmath2 ) per scalelength , see figure 2 .
different lines in figure 2 show histograms of distribution of the gradient in the j , h , and k@xmath0 bands in our whole sample .
the median gradient @xmath15 in the @xmath13 is essentially zero ( -0.01 ) , as well as in the h ( 0 . ) and j ( 0.01 ) bands .
note that some galaxies in histogram in figure 2 indeed show noticeable radial gradients of the scaleheight . in combination with projection effects
, it suggests significant gradients @xmath15 in a small fraction of our galaxies . the dust disks embedded into the stellar components of the galaxies may significantly distort the structural parameters of the stellar disks estimated even from the nir images .
we check reliability of the estimated parameters via the following simulations .
a set of artificial images is created using equation ( [ vol_disk ] ) and typical parameters @xmath1 and @xmath2 .
the surface brightness @xmath3 is replaced by the luminosity surface density @xmath12 in our simulations .
each stellar disk contains an embedded dust disk inside with scalelength of 1.5@xmath1 and scaleheight of 1/3 @xmath2 ( according to * ? ? ?
. the face - on opacity of the dust disk corresponds to @xmath16 = 1 mag .
this value should be close to the upper limit for the face - on extinction in spiral galaxies @xcite , finally , the model disks are randomly inclined by 85 - 90 degrees and their images ( i.e. 2-d projections to the plane of sky ) are obtained via integration along the line - of - sight .
the images are then convolved with the instrumental profile typical for the observations .
then the poisson and gaussian noise patterns are added to create sets of typically thirty artificial images with the same parameters but with different random noise pattern .
the images are analyzed by the same way as the real galaxies ; the mean values of @xmath1 , @xmath2 , and @xmath12 and their r.m.s . are estimated and then compared with the input parameters .
figures 3 and 4 give examples how well the structural parameters can be extracted from the artificial edge - on disks made with stars and dust should they be observed in the j and k@xmath0 nir bands .
although the scalelength and the central surface brightness can be recovered with a fairly good accuracy from both j and k@xmath0 bands , the scaleheight in the j band looks systematically thicker due to the dust absorption effects only .
the k@xmath0 scaleheight in this case is found to be about 15% greater than the model input scaleheight .
we check how the chosen values of the model parameters affect our conclusions .
the amount of the dust opacity is responsible for increasing the estimated scale height and makes it thicker by 10 - 15% in the k@xmath0 if @xmath16 is of the order of 1 mag .
much higher values of the central dust opacity , more than 3 mag , make stellar disks look about twice as thick even in the k@xmath0 band . at the same time
, the disk scalelength shows relatively small variations in this case .
variations of the dust scalelength between @xmath1 and 2@xmath1 do not change the estimated structural parameters significantly .
the dust disk thickness affect the estimated stellar disk thickness in the most power when it is about a half of @xmath2 .
once the dust disk is getting comparable in thickness with the stellar one , it affects mostly the stellar disk surface brightness , and not @xmath2 . the inclination of the stellar disk to the line - of - sight is an important parameter and it increases the stellar disk thickness estimated in our modeling by more than 15% , if the inclination is getting to be greater than 5 degrees . on the other hand ,
the relatively small inclination has almost no effect on the recovered stellar disk scalelength and central surface brightness .
more than 5 degrees inclination in galactic disks would create problem with estimation of @xmath2 ( see also * ? ? ?
* ) , but such galaxies should fail the rfgc catalog criteria and we do nt expect to have many of highly inclined disks in our sample . if the size of the modeling disk is so small that its true scaleheight is much less than the seeing , the resulting @xmath2 gets independent of the original disk scaleheight .
our selection of objects for the analysis by size of their major axis prevents such situation .
given all other parameters fixed , the recovering scalelength , scaleheight , and surface brightness are independent of the numerical value of @xmath1 , @xmath2 , and @xmath12 unless @xmath2 is getting smaller than a pixel in the model images .
in general , our modeling suggests that we can estimate the structural parameters with of order of ten percent precision using our method . at the same time , the sensitivity to the variations of the scaleheight within the same galaxy with different amount of extinction ( like in the case of observations in different nir bands ) should be by the order of magnitude better .
as we can see from our observational results shown in figure 1 , the stellar disks look thicker in the j band and thinner in k@xmath0 for the same object , in agreement with our modeling from
[ im_sim ] .
two factors may be responsible for such a difference in real galaxies : the vertical gradient of stellar population properties and the dust attenuation .
let s consider the effects of the radial and vertical gradients of stellar population in galactic disks .
the radial gradients of metallicity in the thin disks are of the order of -0.07 dex / kpc for the late - type galaxies @xcite , and even less in our galaxy @xcite .
according to the population synthesis models by @xcite , this creates gradients in ( j - k@xmath0 ) color of the order of -0.015 mag / kpc ( i.e. the periphery is bluer ) .
hence , the scalelength in j should appear 4 % longer than that in k for a typical scalelength values of 3 kpc .
a radial age gradient in combination with the metallicity change may rise this scalelength change to 5% . as we see , these numbers are within our accuracy of estimation of the scalelength .
the vertical gradients of metallicity and age , and hence contribution of the stellar population effects to the vertical color variations within the thin stellar disk , are usually small @xcite .
even if we assume the large values for the vertical gradient of [ fe / h ] as in @xcite , -0.2 dex / kpc , this would create less than 1% longer scaleheights in the j than that in the k@xmath0 due to the stellar population gradients .
the amplitude of the changes of scalelengths and scaleheights between the h and k@xmath0 bands is roughly a half of that between the j and k@xmath0 .
we conclude that the scaleheights of the pure thin stellar disk ( i.e. without dust ) estimated from the j , h , and k@xmath0 images should appear the same .
all significant scaleheight variations have to be addressed to the reddening effects .
this is in agreement with conclusions by @xcite .
note that in some cases of extraordinary objects like a superthin galaxy ugc 7321 @xcite , the reddening alone can not explain the observed optical color ( b - r ) , ( r - h ) variations in the vertical direction and vertical gradients of metallicity and age are required for the stellar population . at the same time , @xcite note that the superthin s h - k color gradients might be explained by the wavelength - dependent dust absorption alone .
following @xcite , we consider a vertical luminosity profile of an edge - on image of a stellar disk with the volume brightness @xmath17 .
embedded co - planar dust disk with its scaleheight @xmath18 absorbs and scatters the stellar light in dependence of the dust extinction coefficient @xmath19 . for simplicity , we neglect the dust scatter appealing to @xcite conclusion that the structural parameters of stellar disks in 3-d modeling of the light distribution with or without scattering are not significantly different . as in @xcite ,
the brightness distribution along the minor photometric axis ( z ) in such a disk is @xmath20 , where @xmath21 is the brightness profile in the case of zero extinction , and @xmath22 is the optical depth across the whole edge - on disk , which also depends on the vertical distance over the galactic plane @xmath9 . for small @xmath22 , @xmath23 .
the observed scaleheight @xmath24 of the stellar disk with added extinction is connected with its projected brightness along the selected profile as @xmath25 .
given @xmath26 , we find that @xmath27 . therefore , estimating the thickness of the stellar disk with dust from its edge - on profiles we expect to find @xmath28 where @xmath29 is the @xmath24 estimated from the disk s brightness profile at a certain photometric band , and the optical depth in the plane of the disk in certain photometric band is @xmath30 .
we assume that @xmath31 equals to 0.276 , 0.176 , and 0.112 in the j , h , and k@xmath0 bands , respectively @xcite . since the v - band optical depth @xmath32 and @xmath33 are fixed for the same disk , we use linear regression and fit a line to three data points ( for j , h , and k@xmath0 values ) at @xmath31 - @xmath29 diagram to find @xmath33 . for the convenience of comparison of the vertical scales ,
we normalize the j and h scaleheights in each galaxy by their k@xmath0-band scaleheight .
the upper panel in figure 5 compares the normalized scaleheights .
each line in the panel connects three thicknesses for one galaxy .
the tendency to observe the stellar disks thicker in the j than in the k@xmath0 is well seen .
the histogram in the lower panel in figure 5 shows the distributions of all normalized @xmath2 in three nir colors .
the vertical thick line designates all scaleheights in the @xmath13 because of the normalization .
the mean ratio of @xmath2 in the j and h to that in k@xmath0 is 1.16 and 1.08 , respectively .
we calculate the scaleheight @xmath33 corrected for the extinction ( or extinction - free hereafter ) using equation ( [ z_e ] ) , which gives @xmath34 .
the distribution of @xmath33 normalized by the @xmath2(k@xmath0 ) is also shown in the lower panel in figure 5 .
the median @xmath33/@xmath2(k@xmath0 ) is about 0.89 , i.e. the thin stellar disks in our galaxies are 11% thinner than it can be estimated from the analysis of the vertical profiles from the k@xmath0-band images . the extinction - free scaleheights are given in table 1 .
we observe no systematic variations in the scalelength between the nir bands , as it can be seen in figure 6 , which is the same as figure 5 but drawn for the scalelengths .
this is not a surprise because our simulations shown in figures 3 and 4 suggest rather small systematic variations of the radial scalelength among the nir bands , and proximity of the @xmath1 estimated from the j , h , and k@xmath0 bands to its extinction - corrected value . using the scalelength @xmath1 from the k@xmath0 band and extinction corrected scaleheights
we find that the mean radial - to - vertical scale ratio in the thin extinction - free stellar disks of galaxies in our sample is 5.6 .
the thickness of the stellar disk is sensitive to the gravitational potential of the disk + bulge + halo system in the vertical direction ( e.g. * ? ? ?
therefore , the relative thickness of the stellar disk @xmath4 may provide some information about the spherical - to - disk mass ratio . as a first step
, we consider a very simplified model of a galaxy that consists of a stellar exponential disk embedded into a spherical dark halo .
the total mass of the exponential stellar disk with the central surface density @xmath35 is @xmath36 the total mass of all components of the galaxy is @xmath37 where @xmath38 is the disk s rotational velocity . following @xcite and @xcite , we assume that the disk thickness depends on the local surface density @xmath39 and vertical velocity dispersion @xmath40 : @xmath41 and the ratio of the vertical - to - radial velocity dispersions is @xmath42 .
this is a rather coarse but fair approximation , see @xcite .
we incorporate an assumption of the marginal stability of the stellar disk , i.e. @xmath43 where the toomre parameter @xmath44 @xcite . here
@xmath45 designates the epicyclic frequency , and @xmath46 for the flat rotation curves ( see * ? ? ?
* ) . substituting @xmath40 and @xmath39 from equations ( [ z0h_1 ] ) , ( [ z0h_2 ] ) , and ( [ z0h_3 ] ) into ( [ z02 ] )
, we find @xmath47 @xmath48 @xmath49 .
therefore , @xmath50 numerical n - body simulations @xcite in which the spherical subsystem was introduced as a fixed potential confirm the equation ( [ zh ] ) .
figure 7 is adopted from @xcite with additions from later simulations by khoperskov ( private communication ) .
figure 7 shows that the stellar disk thickness @xmath4 is sensitive to the relative mass of the spherical component @xmath51 normalized by the @xmath52 .
both @xmath51 and @xmath52 are determined within the limits of the stellar disk ( four scalelengths in the simulations ) .
circles in figure 7 correspond to different numerical models with various parameters of the disk and dark halo subsystems , as well as with different bulges .
the upper curve corresponds to the bulgeless models , whereas the lower one is for the models with some bulge contribution .
contemporary and more realistic n - body simulations include `` live '' halos that respond to the gravitational potential of disks .
the relation @xmath4 versus @xmath53 should look very similar in those simulations as in figure 7 for the galaxies in which the spherical subsystems dominate by mass ( i.e. for large @xmath54 ) . for the spherical subsystems with masses @xmath55 , figure 7 illustrates an extreme case , and its top left part ( large @xmath4 and small @xmath54 ) demonstrates the lower limit for mass of the spherical subsystem given disk thickness @xmath4 .
preliminary results from n - body simulations ( khoperskov , private communication ) with live halo suggest not essential corrections to @xmath54 inferred from @xmath4 according to figure 7 ( of the order of 15% ) .
we encourage reader to assume our further estimations of halo - to - disk mass ratio made from the disk thickness as lower limits for the real @xmath54 .
following figure 7 , we calculate the distribution of the dark halo mass @xmath56 expressed in the units of the stellar disk mass for our galaxies .
the rfgc members are `` flat '' galaxies that are not expected to harbor large bulges , so we assume here that @xmath57 . we find the following relation for bulgeless galaxies from figure 7 : @xmath58 . according to this formula , the median value of @xmath59 is about 1.3 within our sample .
the histogram of distribution of @xmath59 is shown in figure 8 .
no any correlations between @xmath59 and absolute magnitude , maximum of rotational velocity , or radial gradient of the scaleheight is found for our galaxies . a tendency to observe higher @xmath59 in the objects with larger scalelength @xmath1
is found , although this relation shows large uncertainty .
following the same basic assumptions , we can include the central surface density @xmath35 into consideration . as it follows from our definition of the disk and total galaxy mass , @xmath60 .
we perform a linear fit between @xmath61 and @xmath62 for our galaxies with known @xmath38 and found that @xmath63 .
it seems more correct to utilize non edge - on galaxies to study such a correlation , and @xcite and @xcite suggest @xmath64 and @xmath65 , respectively , for arbitrary inclined disks .
if the latter relation is accepted , it gives @xmath66 , and in combination with equation ( [ zh ] ) we come to the following relation between the disk thickness and its central surface density @xmath67 the upper panel in figure 9 shows the relation between @xmath68 ( i.e. extinction - corrected inverse thickness of the disks ) and the central surface brightness in the @xmath13 band @xmath69 .
here @xmath69 was corrected for the extinction according to formula ( [ z_e ] ) .
the correction is rather small , and its median value for our sample of galaxies is about 0.1 mag / arcsec@xmath70 .
the lower panel in figure 9 demonstrates the plot of @xmath68 versus the central surface density estimated from observations as @xmath71 .
the numerical coefficient in the formula corresponds to the k@xmath0-band solar absolute magnitude 3.33 mag .
we assume the constant ( m / l@xmath72 ) = 1 here .
the trend in figure 9 is in agreement with formula ( [ zh_s0 ] ) .
the very big scatter of points in figure 9 can be attributed to the relatively low accuracy in the @xmath3 , @xmath1 , and @xmath33 , and even more uncertainty is due to the non - constant value of the mass - to - light ratio ( m / l ) .
the scatter of this relation can be lowered dramatically if only large ( @xmath73 kpc ) galaxies with the scaleheight gradient in the @xmath13 @xmath74 are considered , see figure 10 . in this case
the inverse disk thickness ( @xmath68 ) depends on @xmath75 as @xmath76 where @xmath77 = @xmath78 .
the solid curve in figure 10 represents this equation .
apparently , the large spiral galaxies whose stellar disks are not flared can be better described by the toy - model developed above .
establishing the correlation between the @xmath3 and @xmath68 may be useful for estimating the disk thickness and @xmath59 in disky galaxies with arbitrary inclination .
note that our model is oversimplified , and attempts to estimate the @xmath59 ratio only .
should we need to know the values of @xmath56 or @xmath52 taken apart , or structural parameters of the dark halo , the rotation curve modeling has to be incorporated .
such additional parameter as the disk thickness helps to decrease the ambiguity during the rotation curves decomposition .
a special attention should be given to correspondence between the photometrically and dynamically inferred parameters of galactic subsystems .
an example of n - body simulations of an early - type disk galaxy @xcite demonstrates that photometry may fail to trace the dynamically distinctive subsystems in the central parts of bulge - harboring galaxies . in our consideration
we avoid central parts of galaxies , and using the rfgc objects prevents us from considering bulge - dominated stellar systems .
the weakness of the simplified assumptions in the modeling described above encourages us to apply a more realistic simulations of the stellar disk vertical structure .
since it requires extensive numerical simulations and better quality infrared data , and will gain from the incorporation of rotation curves decomposition , we defer such more complex consideration to a further paper .
we study how the dust absorption affects the basic structural parameters ( disk central surface brightness , vertical and radial scales ) of stellar disks in spiral galaxies estimated from the nir j , h , and k@xmath0 images . using 2mass data ,
we compare the structural parameters estimated from different nir bands focusing on the scaleheight of thin stellar disks .
the stellar disks look thinner in the 2mass k@xmath0 band in the comparison with the h and j. we employ this fact to figure out the extinction - corrected scaleheight @xmath2 of the thin stellar disk . using 139 relatively large galaxies selected from the 2mass catalog , we find that the mean vertical scaleheight ( @xmath2 ) ratios are 1.16:1.08:1.00:0.89 in the j : h : k@xmath0:extinction - free bands , respectively .
this means that the radial scalelength estimated from the @xmath13 images is very close to the extinction - corrected one , whereas the scaleheight is overestimated by 11% on average .
the mean extinction correction for the face - on central surface brightness is only about 0.1 mag / arcsec@xmath70 in the k@xmath0 .
the median radial - to - vertical scale ratio for our sample is about 6 . using a relation between the stellar disk thickness and the halo - to - disk mass ratio obtained from a simplified model
, we estimate the dark - to - luminous mass ratio in our galaxies within the limits of their optical disks .
its mean value is about 1.3 .
a relation between the stellar disk thickness and its central surface brightness is observed for the galaxies from our sample , although it shows significant scatter and is affected by uncertainty in additional parameters .
this relation is more prominent for large ( @xmath73 kpc ) galaxies from our sample and can be utilized for coarse estimating the disk thickness and relative mass of the dark halo in large spiral galaxies arbitrarily inclined to the line - of - sight .
this project was partly supported by grant rfbr-07 - 02 - 00792 .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
computational equipment for db was provided by aas under a small research grant .
we appreciate recommendations from anonymous referee that improved the paper .
we acknowledge the usage of the hyperleda database ( http://leda.univ-lyon1.fr ) .
lcccccccccc name & @xmath1(k ) & d@xmath1 & z0(k ) & d@xmath2 & @xmath3(k)&d@xmath3&@xmath2(cor.)&@xmath2(cor)/@xmath1&@xmath3(k , cor)&@xmath79 + & kpc & kpc & kpc & kpc&mag/@xmath80&mag/@xmath80&kpc & & mag/@xmath80 &
+ rfgc0095 & 0.72 & 0.06 & 0.31 & 0.08 & 17.69 & 0.26 & 0.27 & 0.375 & 17.61 & 0.06 + rfgc0099 & 4.35 & 0.06 & 0.95 & 0.17 & 17.51 & 0.18 & 0.83 & 0.191 & 17.41 & 1.02 + rfgc0102 & 1.76 & 0.03 & 0.48 & 0.11 & 17.49 & 0.20 & 0.46 & 0.260 & 17.46 & 0.49 + rfgc0124 & 8.15 & 0.24 & 1.36 & 0.19 & 17.94 & 0.18 & 1.12 & 0.137 & 17.81 & 1.79 + rfgc0152 & 5.33 & 0.06 & 0.88 & 0.14 & 17.73 & 0.21 & 0.76 & 0.143 & 17.66 & 1.67 + + + .
the solid , dashed , and dash - dotted lines designate the @xmath13 , h , and j gradients , respectively .
the three short vertical lines above the histogram indicate the median values of the gradients in the @xmath13 , h , and j bands ( from left to right ) , that are essentially equal to zero .
[ fig4 ] ] normalized by that in the k@xmath0 .
each a galaxy is shown by a dashed line , which connects the @xmath2 values in the j , h , and k@xmath0 ( the latter equals to unit for all objects ) .
the solid line shows the case of equality of all scaleheights .
bottom panel : histograms of distribution of the scaleheights normalized by those in the k@xmath0 band . dashed line corresponds to the j band , dotted - to the h band .
the solid line designates the extinction - corrected scaleheights .
the numbers under the letters show the mean values of the scaleheights in the corresponding bands .
the thick vertical line designates formally the scaleheights in the @xmath13 that are all equal to unity due to the normalization .
[ fig3 ] ] band .
dashed line corresponds to the j band , dotted - to the h band .
the thick vertical line designates formally the scalelengths in the @xmath13 that are all equal to unity due to the normalization .
[ fig3 ] ] is shown against the spherical - to - disk mass ratio @xmath54 . here
@xmath51 is the mass of the spherical component ( halo and bulge , in a general case ) , and @xmath52 is the mass of the disk component .
the @xmath51 and @xmath52 are estimated within the limits of the stellar disk ( 4@xmath1 ) .
this figure comes from simplified n - body simulations @xcite and each symbol corresponds to certain numerical model .
the models were run with a variety of parameters for the disk , dark halo , and bulge .
the upper curve corresponds to the bulgeless models , whereas the lower one is for the models with some bulge contribution .
[ fig5 ] ] corrected for the internal extinction versus the central surface brightness in the k@xmath0 band .
bottom panel : the same relation where @xmath3 is replaced with the central surface density calculated as @xmath81 , see text .
[ fig7 ] ] corrected for the extinction versus the central surface density of the stellar disks ( see text ) for the sample of large ( @xmath73 kpc ) and non - flared disk galaxies .
the solid line shows the best - fit to the data ( see text ) . | we present results of an analysis of the j , h , and k@xmath0 2mass images of 139 spiral edge - on galaxies selected from the revised flat galaxies catalog .
the basic structural parameters scalelength ( @xmath1 ) , scaleheight ( @xmath2 ) , and central surface brightness of the stellar disks ( @xmath3 ) are determined for all selected galaxies in the nir bands .
the mean relative ratios of the scaleheights of the thin stellar disks in the j : h : k@xmath0 bands are 1.16:1.08:1.00 , respectively . comparing the scaleheights obtained from the nir bands for the same objects
, we estimate the scaleheights of the thin stellar disks corrected for the internal extinction .
we find that the extinction - corrected scaleheight is , on average , 11% smaller than that in the k - band .
using the extinction - corrected structural parameters , we find that the dark - to - luminous mass ratio is , on average , 1.3 for the galaxies in our sample within the framework of a simplified galactic model .
the relative thicknesses of the stellar disks @xmath4 correlates with their face - on central surface brightnesses obtained from the 2mass images .
we also find that the scaleheight of the stellar disks shows no systematic growth with radius in most of our galaxies . |
the physics of hyperfine structure ( hfs ) is driven by magnetic interactions .
this physics has a short - range nature , and is more complicated and challenging than `` softer '' regimes in atomic physics .
this is especially true of the nuclear contribution to hyperfine structure , because the nuclear current density is more complicated than the nuclear charge density and less well understood@xcite . until very recently hyperfine splittings in light hydrogenic atoms were by far the most precisely measured atomic transitions .
many of these very accurate experiments date back nearly half a century .
theoretical predictions are far less accurate , but have improved considerably in recent years .
non - recoil and non - nuclear contributions@xcite are known through order @xmath0 , where @xmath1 is the fermi hyperfine energy ( viz . , the leading - order contribution ) and @xmath2 is the fine - structure constant . because the hadronic scales for recoil and certain types of nuclear corrections are the same ,
recoil corrections are treated on the same footing as nuclear corrections@xcite , and we will call both types `` nuclear corrections . ''
it is very likely that the uncalculated qed terms of order @xmath3 in light atoms contribute less than 0.1 ppm .
although much larger than the experimental errors , this is still significantly smaller than the nuclear corrections .
we restrict ourselves to hydrogenic s - states in this work , because these states maximize nuclear effects . .
state & h & @xmath4h & @xmath5h & @xmath5he@xmath6 ' '' '' + 1s & @xmath733 & 138 & @xmath738 & @xmath7212 + 2s & @xmath733 & 137 & & @xmath7211 + table i is an updated version of the corresponding table in ref.@xcite .
because hyperfine splittings are dominated by short - range physics , we expect the splittings in the @xmath8 s - state to be proportional to @xmath9 , where @xmath10 is the non - relativistic wave function of the electron .
forming the fractional differences ( in parts per million ) between @xmath11 and @xmath12 leads to the tabulated results .
these large differences reflect neither experimental errors nor uncertainties in the qed calculations ; they directly reflect large nuclear contributions .
one complication in performing the nuclear calculation is obtaining a final result that is tractable for numerical calculations .
a framework fortunately exists for performing systematic expansions@xcite of nuclear matrix elements in powers of @xmath13 , where @xmath14 is a typical nuclear momentum scale that can be taken to be roughly the pion mass ( @xmath15 140 mev ) , and @xmath16 is the large - mass qcd scale ( @xmath17 1 gev ) typical of qcd bound states such as the nucleon , heavy mesons , nucleon resonances , etc .
this framework , called power counting , also extends to nuclei , where @xmath18 specifies a typical correlation length ( and a reasonable nearest - neighbor distance ) in light nuclei ( @xmath17 1.4 fm ) .
this expansion in powers of @xmath19 should converge moderately well .
for the purposes of this initial work , we restrict ourselves to leading - order terms in the nuclear corrections .
this restriction eliminates nuclear corrections of relativistic order , which we will briefly discuss later .
there have been relatively few relativistic calculations in light nuclei because of the complexity of the nuclear force . the calculations that exist
are known to generate rather small corrections@xcite , and they are especially small in the deuteron case because of its weak binding@xcite .
few calculations exist for the much more complicated three - nucleon systems@xcite .
most nuclear physics knowledge and lore in light nuclei is non - relativistic in nature . in processes that involve virtual excitation of intermediate nuclear states ( each state @xmath20 with its own energy , @xmath21 , relative to the ground - state energy , @xmath22 )
the excitation energy @xmath23 is of order @xmath24 and typically is a correction to the leading order@xcite .
consistency therefore demands that we drop such terms .
the nuclear recoil energy @xmath25 , where @xmath26 is the nuclear mass , has the same scale and can also be dropped .
our goal is to evaluate the deuteron contribution in leading order as carefully as possible , and to use the impulse approximation to evaluate the @xmath5he and @xmath5h results .
this restriction should be accurate to within the 10 - 15% uncertainty of our leading - order approximation and adequate for a first calculation of the trinucleon sector .
this level of accuracy does require the inclusion of the intrinsic hyperfine structure of the nucleons .
we will see that the calculated results are good for deuterium and tritium and fairly good for @xmath5he@xmath6 .
our goal is to present a simple and compelling picture of the nuclear hyperfine structure in @xmath27 , @xmath5h , and @xmath5he@xmath6 , using a unified approach for all . to accomplish this
we invoke a simple and intuitive model of trinucleon structure based on su(4 ) symmetry that is sufficiently accurate to explain the patterns in table i.
the hyperfine interactions that interest us are simple ( effective ) couplings of the electron spin to the nuclear ( ground - state ) spin : @xmath28 where @xmath29 is the electron ( pauli ) spin operator and @xmath30 is the nuclear spin ( i.e. , the total angular momentum@xcite ) operator .
other couplings of the electron spin are possible and either generate no hyperfine splitting , none in s - states , or higher - order ( in @xmath2 ) contributions .
coupling of the electron spin to the electron angular momentum ( i.e. , the electron spin - orbit interaction , which vanishes in s - states ) is one example .
we begin with a sketch of the fermi hyperfine splitting in order to establish our notation and conventions .
the electron charge operator ( @xmath31 ) is a space scalar , and by itself does not generate a term of the desired form because nuclear - spin information can not be transmitted via the exchange of the appropriate component of the virtual photon s propagator .
the leading - order term in the hyperfine energy shift consequently has the ( usual ) form of a current - current interaction . @xmath32
where @xmath33 is the current density of the ( dirac ) electron and @xmath34 is the nuclear ( ground - state ) current density . ignoring higher - order ( in @xmath2 ) terms
this expression can be manipulated into the form @xmath35 where @xmath36 is the electron mass and @xmath10 is the usual non - relativistic atomic @xmath37 s - state wave function .
the restriction to atomic s - states limits the current to the magnetic - dipole part , which can be written in terms of the magnetic - dipole operator , @xmath38 , as @xmath39 and @xmath38 can be written in terms of the nuclear ground - state magnetization density @xmath40 as @xmath41 where @xmath42 and @xmath43 defines the nuclear magnetic moment .
we finally obtain the fermi hyperfine energy in the limit @xmath44 : @xmath45 where the factor of @xmath46 leads to a hyperfine splitting of @xmath47 .
all additional contributions will be measured as a fraction of this energy .
we note that using the point - like @xmath48 and eqn .
( 3a ) in eqn .
( 1 ) leads to an atomic matrix element @xmath49 ( in the standard dirac notation ) that can be evaluated analytically for higher - order coulomb corrections@xcite to the leading contribution of @xmath50 , where @xmath51 is the nuclear charge .
note that @xmath1 is independent of nuclear structure . naively calculating higher - order ( in @xmath2 ) corrections from the first part of eqn .
( 4 ) fails because only that part of @xmath10 inside the nucleus ( i.e. , within the magnetic `` size '' , @xmath52 ) contributes to the integral , and that is the ( only ) part of the electron s wave function significantly modified by the nuclear charge distribution , @xmath53 , whose radius , @xmath54 , is nearly the same as @xmath52 . in other words
a proper calculation@xcite must take into account modifications of @xmath55 by @xmath56 , and that necessarily involves one order in @xmath2 higher than @xmath1 .
thus we need to perform a consistent second - order ( in @xmath2 ) calculation of the electron - nucleus interaction , depicted in fig .
these graphs are constructed from the direct , crossed , and nuclear seagull contributions to the nuclear compton amplitude .
only the forward - scattering part of this amplitude is required for the @xmath57 corrections to @xmath1 , and this generates a short - range atomic operator that samples the ( upper - component ) s - state wave functions only near the origin .
the resulting energy shift is then given by @xmath58 where @xmath59 is the lepton compton amplitude and @xmath60 is the corresponding nuclear compton tensor , both of which are required to be gauge invariant .
the lepton tensor can be decomposed into an irreducible spinor basis ; we can ignore odd matrices and spin - independent components , since they do not contribute to the hyperfine structure at @xmath61 .
we also ignore ( with one exception treated later ) terms that couple two currents together .
it is easy to show that since the nuclear current scales as @xmath62 ( the conventional components of the current have explicit factors of @xmath63 ) , two of them should scale as @xmath64 and generate higher - order ( in @xmath62 ) terms .
this leaves a single term representing a charge - current correlation @xmath65 } { ( q^2 + i \epsilon)^2 ( q^2 - 2 m_e q_0 + i \epsilon)}\ , .
\eqno(6)\ ] ] the seagull terms @xmath66 and @xmath67 are of relativistic order@xcite @xmath68 and can be dropped .
although the term @xmath69 is of non - relativistic order , it does not contribute in conventional approaches because of crossing symmetry .
the explicit form for @xmath70 ( suppressing the nuclear ground - state expectation value , but including all intermediate states , @xmath71 ) is @xmath72 which greatly simplifies eqn .
( 6 ) in the limit @xmath73 and @xmath74 , leading finally in this limit to a very simple form when closure is used @xmath75 which is infrared divergent . using @xmath76 and @xmath77 together with @xmath78 , and a lower - limit ( infrared ) @xmath79cutoff , @xmath80 ,
we find @xmath81 where there is an implied ( nuclear ) expectation value . the constant term does not contribute because of the derivative , the @xmath82-term contributes a term proportional to @xmath83 , and the last term is the term we are seeking .
the remaining singular ( second ) term must be treated more carefully .
the part proportional to @xmath84 leads ( because of the integral over @xmath85 ) to a contribution proportional to @xmath51 , the total nuclear charge , and @xmath86 , the nuclear magnetic moment .
since this contribution is already part of @xmath87 , keeping this term would amount to double counting , and we therefore ignore it .
the @xmath85-term on the other hand generates ( unretarded ) dipole transitions , and the singularity @xmath88 arises from neglecting @xmath89 and @xmath36 terms .
theorem@xcite for unretarded electric dipole transitions ( @xmath90 ) generates an additional factor of @xmath91 ( i.e. , @xmath80 ) so this singular term is actually of the form @xmath92 and requires a careful calculation . a rather tedious evaluation of this term in eqn .
( 6 ) leads to @xmath93 @xmath94 where @xmath95 in intermediate states .
a similar term arises in the product of two currents mentioned above eqn .
( 6 ) and leads to @xmath96 @xmath97 .
these contributions are completely unimportant , as we now demonstrate .
the constant terms ( including @xmath98 ) are proportional to @xmath99 , which vanishes for non - relativistic dipole operators because they commute .
replacing @xmath91 by a constant ( viz . , the closure approximation ) similarly vanishes .
it is straightforward to show that the nuclear matrix element also vanishes in zero - range approximation , where one neglects the deuteron d - state and the potential in intermediate states ( the structure of the dipole operator weights the tails of the wave functions , and this minimizes the effect of the intermediate - state nuclear potential ) . in perturbation
theory it is possible to show that only the spin - orbit combination of potentials in intermediate states contributes ( i.e. , central and tensor terms cancel ) , and this small potential is of relativistic order @xmath100 , which we have previously agreed to ignore .
one can also show that the non - vanishing deuteron contribution is proportional to @xmath101 , the square of the d- to s - wave asymptotic normalization constant @xmath102 , which is extremely small .
a recent brute - force numerical calculation confirms these estimates@xcite .
the terms @xmath103 and @xmath104 are therefore numerically negligible and can be ignored .
our final result in leading order is a relatively simple expression originally developed in a limiting case by low@xcite for the deuteron , as sketched by bohr@xcite for the same system : @xmath105 where both an atomic and nuclear expectation value is implied , but has been ignored in eqn .
( 11a ) and subsequent equations for reasons of simplicity .
a more convenient representation of this result is obtained by dividing both sides by the expression for the fermi hyperfine energy given by eqn .
since the wigner - eckart theorem guarantees that the resulting form of eqn .
( 11a ) must be proportional to @xmath106 ( which cancels in the ratio ) , we arrive at a simple but powerful expression for the leading - order contribution : @xmath107 where @xmath108 and an expectation value is required of the z ( or `` 3 '' ) component of the vector @xmath109 in the nuclear state with maximum azimuthal spin ( i.e. @xmath110 ) .
the intrinsic size of the nuclear corrections is given by ( @xmath111 ) = @xmath738 ppm @xmath112fm ] , where @xmath112fm ] is the value of the low moment in eqn .
( 11c ) in units of fm .
the results of table i therefore suggest ( correctly ) that low moments are on the order of a few fermi in light nuclei , which is quite sensible .
we predicate our discussion with the deuteron in mind .
other nuclei can and will be treated _
mutatis mutandis_. the isospin of the deuteron ( @xmath113 ) makes it a useful first case .
we note that the nuclear physics in eqn . ( 11 ) involves the correlation between the nuclear charge operator , @xmath114 , and the nuclear current operator , @xmath115 .
if one inserts a complete set of states between these operators , there will be both elastic contributions ( i.e. , ground - state expectation values ) that are called zemach corrections@xcite , and inelastic contributions ( called nuclear polarization corrections ) .
although we will calculate ( or estimate ) both types , it is much easier to calculate the sum of the two .
the nuclear charge operator contains both isoscalar and isovector pieces , and is non - relativistic in leading order .
we ignore relativistic corrections , as we discussed earlier .
we therefore write the charge operator in the form @xmath116 where @xmath117 and @xmath118 and @xmath119 this decomposes the i@xmath120 nucleon s charge operator into proton plus neutron parts .
the densities @xmath121 and @xmath122 are the intrinsic charge densities of the proton and neutron , respectively , while @xmath123 and @xmath124 are the proton and neutron isotopic projection operators , respectively .
the coordinate @xmath125 is the distance of the i@xmath120 nucleon from the nuclear center of mass . we expect that the neutron charge density should play a very minor role , and we will find ( later ) that its contribution is only a few percent of that of the proton .
rather sophisticated models exist for the fourier transform ( i.e. , the form factor ) of @xmath121@xcite .
the nuclear current operator is more complicated , even if we ignore relativistic corrections ( which we will ) .
the problem is the mechanism underlying the nuclear force ( viz .
, the exchange of charged mesons ) , which can also contribute to the nuclear current in the form of meson - exchange currents ( mec ) . these currents largely vanish for isoscalar transitions ( such as the deuteron ground state ) because there is no net flow of charge , but they can be sizable ( 10% - 20% ) for isovector transitions .
one can show that their contribution to the deuteron in eqn . (
11 ) almost entirely vanishes , and we will henceforth ignore these currents below eqn .
we formally expand the current into convection , spin - magnetization , and meson - exchange parts @xmath126 where @xmath127 is the nuclear convection current , @xmath128 is the i@xmath120 nucleon s momentum in the nuclear center - of - mass frame , @xmath129 is the impulse - approximation magnetic - moment operator , @xmath130 is the nuclear meson - exchange current , and @xmath131 is the nucleon magnetization density for the i nucleon expressed in terms of protons and neutrons separately .
the quantities @xmath132 and @xmath133 are the ( total ) proton and neutron magnetic moments , while @xmath134 and @xmath135 are the intrinsic proton and neutron magnetization densities ( normalized to 1 ) . using eqns .
( 12 ) and ( 13 ) the energy shift in eqns . ( 11 ) can be evaluated . rather than split eqn .
( 11c ) into zemach terms ( by inserting intermediate ground states between @xmath136 and @xmath137 ) and polarization terms ( by inserting intermediate excited nuclear states between @xmath136 and @xmath137 ) , we will use the fact that our charge and current operators are each given by a sum over single - nucleon operators .
thus their product can be decomposed into single - nucleon plus two - nucleon operators .
these forms are particularly convenient to evaluate .
we first write in an obvious notation that @xmath138 note that the quantity @xmath139 was not part of low s original work , nor was there any evidence at that time that such a term might be significant .
we next use eqns .
( 13 ) to manipulate the magnetization part of the current in eqn .
( 11c ) into the form ( recall that @xmath140 ) @xmath141 while the convection current can be reduced to @xmath142 the one - body @xmath143 part of the convection - current contribution vanishes upon integration , as does the second ( tensor ) term in the magnetization contribution .
shifting the variables @xmath85 and @xmath84 by @xmath125 leads to @xmath144 where @xmath145 and @xmath146 determine the proton and neutron parts of the one - body current .
note that the quantities @xmath147 and @xmath148 are the usual proton and neutron zemach terms , and we have listed in eqn .
( 18a ) the value of the proton zemach moment recently determined directly from the electron - scattering data for the proton@xcite ( the neutron has not been evaluated ) . in numerical work
described below we will use simple forms for the neutron and proton form factors : a dipole form for the proton charge and magnetic form factors and the neutron magnetic form factor @xmath149 and a modified galster@xcite form for the neutron charge form factor @xmath150 . to incorporate into our calculations the numerical value given by eqn .
( 18a ) we use @xmath151 @xmath152 , which reproduces this value for the dipole case ( see appendix a for moments and correlation functions determined by this choice of form factors ) .
the r.m.s . radius determined by this @xmath153 is 0.86 fm , which is slightly smaller than the proton charge radius@xcite but slightly larger than its magnetic radius@xcite and thus represents an average value .
the much smaller neutron moment , @xmath154 ( see appendix a ) , can be adequately represented using this value of @xmath153 and @xmath155 fm@xmath4 , which determines the neutron charge radius@xcite .
these numbers lead to @xmath157 fm , which we will use below .
because this value is such a small fraction of the proton result , the uncertainty in the neutron value plays no significant role . equation ( 17 ) is still a nuclear operator , and its expectation value depends on the nucleus .
we begin with the deuteron ( which has @xmath158 ) and this eliminates the @xmath159 terms in @xmath123 and @xmath124 .
the spin terms then sum to @xmath160 , which is not the total angular momentum ( it lacks @xmath161 , the orbital angular momentum contribution ) .
the expectation value of @xmath160 is @xmath162 , where @xmath163 is the z - component of the deuteron total angular momentum operator and @xmath164 is the amount of @xmath165-wave in the deuteron wave function ( typically , slightly in excess of 5.6% ) . in the state of maximum @xmath163
this leads to @xmath166 which is completely dominated by the proton .
note that the @xmath165-wave prefers to anti - align with the spin , which leads to the reduction in eqn .
( 19 ) . a similar ( though more complicated ) analysis is possible for @xmath5he and @xmath5h ( see appendix b ) .
the traditional ( and very useful ) decomposition of the trinucleon wave function uses representations of spin - isospin symmetry ( viz .
, su(4 ) ) .
in addition to the somewhat larger d - state component @xmath167 , the significant @xmath168-wave component comes in two distinct types : the dominant s - state ( @xmath169 ) with a completely antisymmetric spin - isospin wave function and completely symmetric space wave function , and the mixed - symmetry s@xmath170-state @xmath171 .
the representations of su(4 ) were used long ago to decompose contributions to the trinucleon magnetic moments , and this leads to@xcite @xmath172 \left ( \frac{1+\tau_3}{2 } \right ) + \left [ \begin{array}{c } \xi \\ \gamma \end{array } \right ] \left ( \frac{1-\tau_3}{2 } \right ) \right\ } \
, , & ( 20b ) \cr}\ ] ] where @xmath173 specify the two sign cases in eqns .
( 20a ) and ( 20b ) .
note that @xmath159 is the third component of total isospin and @xmath30 is the ( total ) nuclear - angular - momentum operator .
the first and second terms in eqn .
( 20b ) determine @xmath5he and @xmath5h , respectively , while upper and lower components refer to @xmath174 in eqn .
finally we obtain @xmath175 and @xmath176 where the point - neutron , su(4 ) symmetry limit ( i.e. , the s - state only , which is indicated by the arrow ) is discussed below .
this result is very easy to interpret . in the dominant s - state ( corresponding to @xmath177 and @xmath178 )
the two `` like '' nucleons ( e.g. , the protons in @xmath5he ) have opposite and cancelling spins , while the `` unlike '' nucleon ( e.g. , the neutron in @xmath5he ) carries all of the spin and determines both the magnetic moment and the single - nucleon contribution to the hyperfine structure ( if we ignore meson - exchange currents and the convection current ) . ignoring those currents leads to @xmath179 and @xmath180 , where the arrow indicates the su(4 ) limit .
the @xmath5he single - nucleon contribution to hfs in this limit comes solely from the tiny neutron contribution , while the corresponding @xmath5h contribution is simply @xmath147 and becomes identical to the free proton zemach moment .
the two - nucleon contributions are more complicated and are determined by correlation functions . in the deuteron these correlations must be between a neutron and a proton and are of the types : @xmath181 , @xmath182 , and @xmath183 ( convection current only ) . in the trinucleons
there are additional types : @xmath184 , @xmath185 , and @xmath186 ( convection current only ) . as we noted earlier
the correlations involving @xmath187 will be very small , and we have ignored the tiny @xmath188 convection - current correlation in @xmath5h .
the two - nucleon contributions contained in eqns .
( 15 ) and ( 16 ) can be manipulated into simpler forms by shifting @xmath85 by @xmath125 and @xmath84 by @xmath189 , leading to @xmath190 where @xmath191 and the three correlation functions are defined by @xmath192 where in eqns .
( 24b , d , f ) we have decomposed the charge and magnetic distributions in terms of isospin projectors and radial functions @xmath193 and @xmath194 .
explicit forms for these functions are given in appendix a. note that the quantity in parenthesis in eqns .
( 24b , d ) is the ( dominant ) spin part of the magnetic moment operator , and that we ignore the contribution of two neutron charge distributions in eqn .
in the limit of no finite size each of the three non - vanishing radial functions ( @xmath195 ) simply equals @xmath196 . the special case of the deuteron is easily dealt with . with our assumptions about the nucleon form factors
there are only two distinct types of products contained in @xmath197 and @xmath198 ; these are the dipole - dipole form of @xmath199 correlation @xmath200 and the dipole - galster form of @xmath201 and @xmath202 correlation @xmath203 .
the conventional form of the deuteron wave function ( suppressing the spin and isospin wave functions ) is @xmath204 which leads to the useful relations @xmath205 using @xmath206 for the deuteron , while @xmath207 this leads immediately to @xmath208 where we have removed a factor of @xmath209 from the magnetic moments ( i.e. , they are now given in units of nuclear magnetons ) . in the limit of vanishing neutron charge distribution and point protons , this expression becomes @xmath210 which is low s expression@xcite for the complete deuteron finite - size effect ( in leading order ) . in the next section
we will refer to the integrals of radial functions such as @xmath211 , ( i.e. , including the numerical factors , but without the magnetic moments in eqn .
( 27 ) ) as `` low moments . ''
the two low moments for point - like nucleons are , therefore , @xmath212 and @xmath213 , as given in eqn .
the proton hfs has been discussed in detail recently@xcite and we have nothing more to add .
the recently evaluated proton zemach moment is listed in table ii , and it leads to a @xmath758.2(6 ) khz contribution to the hydrogen hfs , which equals @xmath741.0(5 ) ppm . when added to the usual qed and
recoil corrections@xcite there is a 3.2(5 ) ppm discrepancy with experiment , which can be attributed to hadronic polarization and ( possible ) additional recoil corrections .
* table ii*. & + proton & neutron & & proton & neutron & total & + 1.086(12 ) & @xmath70.042 & fm & @xmath740.0 & @xmath71.1 & @xmath741.1 & khz + we begin our discussion of the deuteron with the single - nucleon contribution given by eqn .
table ii lists the proton zemach moment and the neutron moment determined by our choice of form factors .
the neutron result is only 4% of the proton value in magnitude , and the opposite sign reflects the fact that the ( overall neutral ) neutron has negative charge at large distances that balances positive charge at short distances . using the value of @xmath214 ( corresponding to the av18 potential model@xcite ) in eqn .
( 19 ) leads to the nucleon - moment deuterium hfs contributions listed on the right side of table ii .
the proton result differs from that in hydrogen by the factor of @xmath215 and the statistical factors for the deuterium and hydrogen hfs .
* table iii*. @xmath216 & @xmath217 & @xmath218 & @xmath219 & @xmath220 & @xmath221 & @xmath222 & + 3.081&@xmath70.115&3.271&@xmath70.015&@xmath70.126&0.001&@xmath70.003&fm + the deuterium low moments are listed in table iii , and the resulting hfs is listed in table iv , both for point - like nucleons ( only the proton charge contributes ) and for nucleons with finite size . the moments themselves and the resulting hfs are defined in eqn .
( 28 ) for point - like nucleons and in eqn .
( 27 ) for finite nucleons .
it is obvious that d - waves and the neutron s charge distribution play a minor role .
the proton charge distribution and the neutron magnetic distribution have a somewhat larger effect ; @xmath223 is larger than @xmath216 by about 6% .
* table iv*. @xmath216 & @xmath217 & @xmath218 & @xmath219 & @xmath220 & @xmath221 & @xmath222 & + 84.9 & @xmath73.2 & 90.2 & 0.6 & @xmath73.5 & 0.0 & 0.0 & khz + one can also compute the zemach moment of the entire deuteron by constructing the charge ( @xmath224 ) and magnetic ( @xmath225 ) form factors and using the equivalent momentum - space version of the zemach moment formula : @xmath226 various contributions and limits are listed in table v. results for point - like nucleons are listed to the left and include the contributions from the s - wave spin - magnetization current , the d - wave spin - magnetization current , and the orbital ( convection ) current , followed by the total contribution . including identical dipole nucleon form factors for the proton s charge and the neutron s magnetization densities ( which multiplies both @xmath227 and @xmath228 by @xmath229 - see eqn .
( a3 ) ) leads to the rightmost result and corresponds to an increase of about 10% .
the experimental result of 2.593(16 ) fm was obtained directly from the electron - scattering data@xcite , and is approximately 2% smaller ( 4 standard deviations ) than our non - relativistic calculation .
this difference is the expected size of relativistic corrections from mec . * table v*. & & + @xmath230 & @xmath231 & orb & zemach & zemach & + 2.324 & @xmath70.035 & 0.094 & 2.383 & 2.656 & 2.593(16 ) & fm
+ * table vi*. & + nucleon & low & total & nucleon & low & total & + 0.0 & 81.8 & 81.8 & @xmath741.1 & 87.3 & 46.2 & khz + the one - body ( nucleon zemach ) and two - body ( low ) contributions to the total deuterium hfs are listed in table vi . because there is then no point - nucleon contribution to the one - body part ,
the low term is the sole contribution and leads to a very large result .
the finite - nucleon case has considerable cancellation between the two , and totals only about half the size of the point - nucleon limit .
one can also break the total result down into deuteron zemach ( elastic ) terms plus polarization ( inelastic ) terms .
this is indicated in table vii .
the polarizability term is more than twice the elastic ( zemach ) term , and reflects how easily a weakly bound system can be excited compared to a system like the nucleon , which is difficult to excite . in the latter case
the polarization term is only about 10% of the zemach term .
* table vii*. & + zemach & polar & total & zemach & polar & total & + @xmath729.5 & 111.2 & 81.8 & @xmath732.8 & 79.1 & 46.2 & khz + the physics of the nuclear correction to the deuterium hfs is straightforward and completely dominated by the proton zemach moment and the @xmath181 low contribution .
nuclear structure plays little role except to fix the size of the ( radial ) low moment .
the signs were fixed by the sign of the proton magnetic moment ( for the one - body term ) and the neutron magnetic moment ( for the two - body term ) , and are opposite . if we incorporate the additional minus sign in eqn .
( 11b ) , the naively expected sign of the one - body terms should be @xmath7 , while that of the low contribution should be @xmath232 , as we found in the deuterium case .
as we will see , however , nuclear structure can play an exceptional role in the trinucleon , and these expectations are not fulfilled in two cases .
the required @xmath5h and @xmath5he matrix elements were calculated using wave functions obtained from a faddeev calculation@xcite .
the ( second - generation ) av18@xcite potential was used , together with an additional tm@xmath170 three - nucleon force@xcite whose short - range cutoff parameter had been adjusted for each case to provide the correct binding energies .
individual one - body ( labelled `` nucleon '' ) and two - body ( labelled `` low '' ) terms are tabulated together with their total in table viii .
note that the @xmath5he case ( which has proton number @xmath233 ) is uniformly enhanced ( compared to the h cases ) by a factor of @xmath234 contained in @xmath235 in eqn .
( 11a ) . for the same reason the two protons in @xmath5he effectively double that low moment .
taking those factors into account the @xmath5he low term becomes comparable in size to that of the deuteron .
the ( approximate ) su(4 ) symmetry that dominates light nuclei@xcite provides an explanation for the relative sizes of the entries in this table , as well as the unexpected signs ( see above ) of the @xmath5h one - body term and the @xmath5he two - body term .
the two protons in @xmath5he have their spins anti - aligned in the su(4 ) limit , and this cancellation leads to the small net result and unexpected sign for the one - body part , which is determined by small components of the wave function .
the protons in @xmath236h and @xmath5h make comparable one - body contributions , since the proton in @xmath5h carries the entire spin in the su(4 ) limit .
for the same reason the low term in @xmath5h is very small because the two neutron @xmath237 terms largely cancel , since the proton carries all of the spin in the su(4 ) limit .
the neutron zemach moment plays only a very small role in the one - body terms ( as we found for the deuteron ) except for @xmath5he , which has a greatly suppressed proton contribution .
the tensor term also is quite small .
the convection current terms are negligible for @xmath5h , but the @xmath186 contribution in @xmath5he is approximately 5% of the total . *
table viii*. & + nucleon & low & total & nucleon & low & total & + @xmath750.6 & @xmath79.6 & @xmath760.1 & 14 & 1428 & 1442 & khz + our final results are listed in table ix .
the first line of the table is the same as that in table i , showing the fractional difference of experiment and qed theory in ppm .
that fractional difference is recomputed in the second line when the nuclear corrections are added to the theoretical result . in the proton case
the zemach and recoil corrections slightly over - correct , but the overall result is consistent with the expectation that the polarization corrections are positive and must be less than 4 ppm@xcite . for the nuclear cases the quality of our results must be considered quite good , given the size of our hadronic expansion parameter .
the deuterium case is particularly close to experiment , and this is likely due to the small binding energy , which tends to minimize relativistic corrections@xcite .
the quality of our trinucleon results range from very good in the @xmath5h case ( @xmath238 residue ) to adequate in the @xmath5he case ( @xmath239 residue ) .
the large disparity in the two cases is undoubtedly due to missing mec , particularly the isovector ones .
even this amount of missing strength is only slightly larger than our expansion parameter . .
theory & h & @xmath4h & @xmath5h & @xmath5he@xmath6 ' '' '' + qed only & @xmath733 & 138 & @xmath738 & @xmath7212 + qed + hadronic & 3.2(5 ) & @xmath73.1 & 1.2 & @xmath746 + previous work on this topic is quite old@xcite , except for the deuterium@xcite case .
the older work relied on the breit approximation for the electron physics , which is sufficient only for the leading - order corrections .
it used an adiabatic treatment of the nuclear physics based on the bohr picture of the nuclear hyperfine anomaly , which is far more complex than the treatment that we have presented .
uncalculated qed corrections and poorly known fundamental constants ( such as @xmath2 ) led to estimates of nuclear effects that were many tens of ppm in error .
although the nuclear physics at that time was not adequate to perform more than qualitative treatments of the trinucleons , the su(4 ) mechanism was known and this allowed a qualitative understanding .
the only previous attempt to treat the three nuclei simultaneously was in ref .
they found nuclear corrections of about 200 ppm for deuterium , 20 ppm for @xmath5h , and @xmath7175 ppm for @xmath5he@xmath6 . except for the deuterium case ( which involves significant cancellations )
this has to regarded as quite successful , given the knowledge available at that time .
we have performed a calculation of the nuclear part of the hfs for @xmath4h , @xmath5h , and @xmath5he@xmath6 , based on an expansion parameter adopted from @xmath240pt , a unified nuclear model , and modern second - generation nuclear forces .
this is the first such calculation , and the results are quite good . details of the results can be understood in terms of the approximate su(4 ) symmetry that dominates the structure of light nuclei .
the work of jlf was performed under the auspices of the u. s. dept . of energy , while the work of glp was supported in part by the doe .
the correlation functions that we require are built from the charge and magnetic form factors for protons and neutrons .
the generic correlation function has the form @xmath241 where the charge density @xmath242 and magnetization density @xmath243 are normalized to one .
although a variety of functional forms have been proposed for these densities , few published forms have high accuracy over the low - momentum - transfer region that is important for zemach moments .
fortunately the proton s zemach moment was recently determined to high accuracy , and the neutron s is sufficiently small that any credible model should suffice . for the neutron charge form factor
we will assume a modified galster form@xcite @xmath244 with @xmath151 @xmath152 ( determined below ) and @xmath245 = 0.0190 fm@xmath4 .
this is accurate enough@xcite for our purposes , both at low values of @xmath246 and at moderate values of @xmath246 . for the proton charge form factor and the proton and neutron magnetic form factors we choose the tractable and venerable dipole form @xmath247 which is a reasonable ( but only moderately accurate ) approximation .
these two forms inserted into eqn .
( a1 ) generate the two correlation functions that we require : @xmath248 and @xmath249 : @xmath250 and @xmath251 the first moment of these functions is the linear zemach moment @xmath252 which we identify with the recently determined proton moment : 1.086(12 ) fm .
this restricts @xmath153 to be 4.029(45 ) @xmath152 , which we also use for the neutron .
the rms radius for a dipole with this value of @xmath153 is 0.86 fm , slightly smaller than the proton s charge radius , but slightly larger than the magnetic radius by a few percent , and this represents our level of accuracy ( except for the measured proton zemach moment ) .
we can use these functions to determine the appropriate correlation functions @xmath253 where the limiting form holds only for small ( @xmath254 ) , and similarly @xmath255 the remaining functions that we require are determined by @xmath256 and @xmath257 from the former we obtain @xmath258 and @xmath259 and from the latter @xmath260 and @xmath261 note that these functions have been normalized so that @xmath262 , @xmath263 , and @xmath264 become @xmath196 in the limit of large @xmath153 .
we resort to the simple zero - range approximation in order to make a rough estimate of the effect of nucleon finite size on the dominant low moment of the deuteron .
this approximation is most accurate for asymptotic ( long - range ) quantities , but will substantially overestimate short - range effects .
we ignore the d - state and assume everywhere the asymptotic s - wave function @xmath265 , where @xmath266 @xmath267 .
this leads to an expansion parameter @xmath268 0.115 .
the matrix element of @xmath262 ( relative to the point - nucleon value ) is approximately @xmath269 , which produces an increase @xmath270 from nucleon finite size .
this is too large by a factor of two compared to detailed calculations , but shows that the finite - size effect is enhanced by the large numerical coefficient of the logarithmic term beyond what is expected from a @xmath271 correction term .
many features of the trinucleon systems , @xmath5he and @xmath5h , can be determined in a semi - quantitative fashion ( accurate at the @xmath1710% level ) by simplifying the wave functions to the dominant component alone .
wave function components have traditionally been classified according to their combined spin - isospin symmetry , determined by the generators of su(4 ) . in this scheme the su(4 ) generators for a system of @xmath272 nucleons
are determined by the intrinsic spins and isospins of the individual nucleons : @xmath273 @xmath274 and @xmath275 all wave function components are labelled by the ( combined ) intrinsic spin of the three nucleons ( @xmath276 = 1/2 or 3/2 ) and the total isospin ( @xmath277 = 1/2 or 3/2 ) .
wave function spin and isospin components are then determined by : ( 1 ) the @xmath278 $ ] or antisymmetric state ( @xmath276 = 1/2 , @xmath277=1/2 ) , which combines with a completely symmetric space wave function to form the dominant s - state ; ( 2 ) the mixed - symmetry state , which can be separated into ( @xmath276 = 1/2 , @xmath277 = 1/2 ) , @xmath279 = 3/2 , @xmath277=1/2 ) , and @xmath279 = 1/2 , @xmath277 = 3/2 ) components .
the first term contributes to the small s@xmath170-state , while the second generates the d - state(s ) , and the third contributes only to tiny isospin impurities .
the remaining spin - isospin representation has a tiny @xmath279 = 1/2 , @xmath277 = 1/2 ) symmetric component ( called the s@xmath280-state , with a totally antisymmetric space wave function ) , and a @xmath281 = 3/2 , @xmath277 = 3/2 ) d - wave isospin impurity .
the number of components in the order discussed is ( 4 ) + ( 4 + 8 + 8) + ( 4 + 16 ) = 64 , as expected . ignoring the tiny s@xmath282-state , very small isospin impurities , and the negligible p - states ,
we can therefore write in an obvious but schematic notation for the trinucleon wave functions @xmath283 it was shown many years ago@xcite that expectation values of the su(4 ) generators for the trinucleons have very simple forms . @xmath284 and @xmath285 where a spin - isospin expectation value of the nuclear spin and isospin operators is still required .
as expected , the spatially symmetric s - state dominates the trinucleon ground states @xmath286 because it minimizes the kinetic energy .
the mixed - symmetry s@xmath170-state is much smaller @xmath287 , while the very strong nuclear tensor force generates a relatively large d - state component @xmath167 .
if one ignores the s@xmath170- , s@xmath282- , p- , and d - state components , the remaining s - state wave function factorizes into a completely symmetric space wave function and a completely antisymmetric spin - isospin wave function , which greatly facilitates calculating matrix elements .
the mixed spin - isospin generator @xmath288 has the very useful and simple property for the s - state @xmath289 which follows ( except for the factor of @xmath290 ) from the wigner - eckart theorem and the properties of the @xmath291 $ ] state .
999 h. arenhvel , _ czech .
j. phys . _
* 43 * , 259 ( 1993 ) .
j. l. friar , _ few - body systems _ * 22 * , 161 ( 1997 ) . m. i. eides , h. grotch , and v. a. shelyuto , _ phys
. rep . _
* 63 * , 342 ( 2001 ) .
s. g. karshenboim and v. g. ivanov , _ eur .
phys . j. d _ * 19 * , 13 ( 2002 ) ; _ phys
. lett .
b _ * 524 * , 259 ( 2002 ) .
m. j. zuilhof and j. a. tjon , phys .
c * 22 * , 2369 ( 1980 ) .
d. r. phillips , g. rupak , and m. j. savage , phys . lett .
b * 473 * , 209 ( 2000 ) .
f. gross , in _ modern topics in electron scattering _ , ed . by b.
frois and i. sick ( world scientific , singapore , 1991 ) , p. 219 .
this lovely review thoroughly and clearly discusses relativistic effects in nuclear electromagnetic interactions .
a. r. edmonds , _ angular momentum in quantum mechanics _ , ( princeton , 1960 ) .
although nuclear physicists almost universally denote the total nuclear angular momentum by @xmath137 , we have chosen to use @xmath30 instead .
this more closely conforms to conventional usage in atomic physics .
nuclear physicists should be careful not to confuse our @xmath30 with the simple sum of individual nucleon spins .
a. bohr and v. f. weisskopf , _ phys . rev . _ * 77 * , 94 ( 1950 ) .
this was the first detailed calculation of the combined effect of the charge and magnetic nuclear densities on the hyperfine structure in heavy atoms ; it used a model for the nuclear densities .
zemach@xcite investigated light atoms and used perturbation theory without assumptions about the form of the nuclear densities . c. zemach , _ phys .
rev . _ * 104 * , 1771 ( 1956 ) .
j. l. friar , _ phys . rev .
c _ * 16 * , 1540 ( 1977 ) . j. l. friar and g. l. payne , _ phys . rev .
c _ ( submitted ) .
f. low , _ phys .
rev . _ * 77 * , 361 ( 1950 ) ; f. e. low and e. e. salpeter , _ phys .
_ * 83 * , 478 ( 1951 ) .
this work calculated the correlation - function part of the nuclear hfs for deuterium , as originally sketched by bohr@xcite .
a. bohr , _ phys .
_ * 73 * , 1109 ( 1948 ) .
j. l. friar and i. sick , _ phys .
b _ * 579 * , 285 ( 2004 ) .
s. galster , _ et al . _ , _ nucl .
_ * b32 * , 221 ( 1971 ) .
i. sick , _ phys .
b _ * 576 * , 62 ( 2003 ) .
i. sick , ( private communication ) .
j. l. friar , j. martorell , and d. w. l. sprung , _ phys .
a _ * 56 * , 4579 ( 1997 ) .
this work contains a discussion of the neutron s size .
j. l. friar , b. f. gibson , and g. l. payne , _ ann .
nucl . part .
_ * 34 * , 403 ( 1984 ) .
this review discusses the usefulness of those results .
e. l. tomusiak , m. kimura , j. l. friar , b. f. gibson , g. l. payne , and j. dubach , _ phys .
c _ * 32 * , 2075 ( 1985 ) .
this reference tested an impulse approximation formula ( which is quite accurate , though not exact ) for the trinucleon magnetic moments that is based on an su(4 ) decomposition : @xmath292 -\haf ( 2 t_3 ) ( { \mu^{\phantom{|}}_p}- { \mu^{\phantom{|}}_n } ) [ 1 - \frac{4}{3 } p_{s^\prime } -\frac{2}{3}p_d ] + \haf p_d [ 1+\frac{1}{3 } ( 2 t_3)]$ ] , where the @xmath293 states have been ignored , and @xmath294 and @xmath164 are the probabilities of the s@xmath295 and d states , respectively .
the quantity @xmath296 is + 1 for @xmath5he and 1 for @xmath5h .
j. l. friar , g. l. payne , v. g. j. stoks , and j. j. de swart , _ phys .
* b 311 * , 4 ( 1993 ) .
r. b. wiringa , v. g. j. stoks , r. schiavilla , _ phys .
c _ * 51 * , 38 ( 1995 ) .
s. a. coon , m. d. scadron , p. c. mcnamee , b. r. barrett , d. w. e. blatt , and b. h. j. mckellar , _ nucl .
phys . _ * a317 * , 242 ( 1979 ) ; j. l. friar , d. hber , and u. van kolck , _ phys .
c _ * 59 * , 53 ( 1999 ) .
the latter paper includes a @xmath240pt derivation of an improved tucson - melbourne three - nucleon potential ( originally derived in the former paper ) , which is usually denoted tm@xmath170 .
the latter can be viewed as a second - generation three - nucleon force .
r. g. sachs and j. schwinger , _ phys .
rev . _ * 70 * , 41 ( 1946 ) .
r. n. faustov and a. p. martynenko , _ eur .
c _ * 24 * , 281 ( 2002 ) . v. w. hughes and j. kuti , _ ann .
nucl . part .
sci . _ * 33 * , 611 ( 1983 ) .
r. avery and r. g. sachs , _ phys .
_ * 74 * , 1320 ( 1948 ) .
e. n. adams ii , _ phys .
rev . _ * 81 * , 1 ( 1951 ) .
a. m. sessler and h. m. foley , _ phys .
rev . _ * 98 * , 6 ( 1955 ) .
d. a. greenberg and h. m. foley , _ phys .
_ * 120 * , 1684 ( 1960 ) . i. b. khriplovich and a. i. milstein , _ zh . eksp
. teor .
_ * 125 * , 205 ( 2004 ) [ _ jetp _ * 98 * , 181 ( 2004 ) ] ; a. i. milshtein , i. b. khriplovich , and s. s. petrosyan , _ j. exp .
phys . _ * 82 * , 616 ( 1996 ) ; _ phys .
b _ * 366 * , 13 ( 1996 ) .
these papers are noteworthy because they calculate the sub - leading - order correction to the deuterium hfs in zero - range approximation .
we have calculated only the leading order here , but have done so in the spirit of traditional nuclear calculations , while also calculating tritium and @xmath5he@xmath6 . | hyperfine intervals in light hydrogenic atoms and ions are among the most accurately measured quantities in physics .
the theory of qed corrections has recently advanced to the point that uncalculated terms for hydrogenic atoms and ions are probably smaller than 0.1 parts per million ( ppm ) , and the experiments are even more accurate .
the difference of the experiments and qed theory is interpreted as the effect on the hyperfine interaction of the ( finite ) nuclear charge and magnetization distributions , and this difference varies from tens to hundreds of ppm .
we have calculated the dominant component of the 1s hyperfine interval for deuterium , tritium and singly ionized helium , using modern second - generation potentials to compute the nuclear component of the hyperfine splitting for the deuteron and the trinucleon systems .
the calculated nuclear corrections are within 3% of the experimental values for deuterium and tritium , but are about 20% discrepant for singly ionized helium .
the nuclear corrections for the trinucleon systems can be qualitatively understood by invoking su(4 ) symmetry .
= 0.25 in = 0.25 in = .10 in = 1ex = 2ex = 6ex = 3ex plus 1000pt minus 1000pt # 1 # 1= to + + _ theoretical division , los alamos national laboratory + los alamos , nm 87545 _ + and + + _ dept . of physics and astronomy + univ . of iowa + iowa city , ia 52242 _ |
heavy quarks are an ideal probe for the qgp .
they are produced in the primordial hard collisions of the nuclear reaction and therefore probe the created medium during its entire evolution process .
when the system cools down they hadronize , and their decay products can finally be detected .
therefore , heavy - quark observables provide new insights into the interaction processes within the hot and dense medium .
two of the most interesting observables are the elliptic flow , @xmath0 , and the nuclear modification factor , @xmath1 , of open - heavy - flavor mesons and their decay products like `` non - photonic '' single electrons .
the measured large elliptic flow , @xmath0 , of open - heavy - flavor mesons and the `` non - photonic single electrons or muons '' from their decay underline that heavy quarks take part in the collective motion of the bulk medium , consisting of light quarks and gluons .
the nuclear modification factor shows a large suppression of the open - heavy flavor particles spectra at high transverse momenta ( @xmath3 ) compared to the findings in pp collisions .
this also supports a high degree of thermalization of the heavy quarks with the bulk medium . in this letter
we explore the medium modification of heavy - flavor @xmath3 spectra , using a hybrid model , consisting of the ultra - relativistic quantum molecular dynamics ( urqmd ) model @xcite and a full ( 3 + 1)-dimensional ideal hydrodynamical model @xcite to simulate the bulk medium . the heavy - quark propagation in the medium
is described by a relativistic langevin approach @xcite .
similar studies have recently been performed in a thermal fireball model with a combined coalescence - fragmentation approach @xcite , in an ideal hydrodynamics model with a lattice - qcd eos @xcite , in a model from kolb and heinz @xcite , in the bamps model @xcite , the martini model @xcite as well as in further studies and model comparisons @xcite .
the urqmd hybrid model has been developed to combine the advantages of transport theory and ( ideal ) fluid dynamics @xcite .
it uses initial conditions , generated by the urqmd model @xcite , for a full ( 3 + 1 ) dimensional ideal fluid dynamical evolution , including the explicit propagation of the baryon current . after a cooper - frye transition back to the transport description
, the freeze out of the system is treated dynamically within the urqmd approach .
the hybrid model has been successfully applied to describe particle yields and transverse dynamics from ags to lhc energies @xcite and is therefore a reliable model for the flowing background medium .
the diffusion of a `` heavy particles '' in a medium consisting of `` light particles '' can be described with help of a fokker - planck equation @xcite as an approximation of the collision term of the corresponding boltzmann equation .
it can be mapped into an equivalent stochastic langevin equation , suitable for numerical simulations .
the drag and diffusion coefficients for the heavy - quark propagation within this framework are taken from a resonance approach @xcite , where the existence of d - mesons and b - mesons in the qgp phase is assumed , as well as a @xmath4-matrix approach @xcite in which quark - antiquark potentials are used for the calculation of the coefficients in the qgp .
the initial production of charm quarks in our approach is based on a glauber approach . for the realization of the initial collision dynamics we use the urqmd model .
we perform a first urqmd run excluding interactions between the colliding nuclei and save the nucleon - nucleon collision space - time coordinates .
these coordinates are used in a second , full urqmd run as possible production coordinates for the charm quarks . as momentum distribution for the initially produced charm quarks at @xmath5 we use @xmath6 with @xmath7 , @xmath8 and @xmath9 and for bottom quarks
@xmath10 with @xmath11 and @xmath12 .
these distributions are taken from @xcite .
the @xmath3 distribution for charm quarks at @xmath13tev is obtained from a fit to pythia calculations . @xmath14 with the coefficients @xmath15 , @xmath16 and @xmath17 . starting with these distributions as initial conditions we propagate the heavy quarks at each hydro - timestep .
we use the urqmd / hydro s cell velocities , the cell temperature , the size of the time - step , and the @xmath18-factor for the calculation of the momentum transfer , propagating all quarks independently .
our approach provides us only with the heavy - quark distributions . since
heavy quarks can not be measured directly in experiments we include a hadronization mechanism for d - mesons and b - mesons , via the use of a quark - coalescence mechanism . to implement this coalescence
we perform our langevin calculation until the decoupling temperature is reached .
subsequently we add the momenta of light quarks to those of the heavy quarks .
first we performed our calculations in au+au collisions at @xmath5 in a centrality range of 20%-40% . to compare our results to the single - electron spectra measured by phenix we use pythia for the decay of the heavy quarks to heavy flavor electrons and apply a rapidity cut of @xmath19 .
[ rhic ] ( left ) shows our results for the elliptic flow @xmath0 . for a decoupling temperature of @xmath20
we obtain a reasonable agreement with the experimental data except for low @xmath3 bins . here
a depletion effect can be seen .
this effect is due to the radial velocity of the medium , which is in case of a developed elliptic flow larger in @xmath21 than in @xmath22 direction .
consequently there is a depletion of particles with high @xmath23 in the low @xmath3 region and smaller elliptic flow .
this effect is more important for heavier particles and a larger radial flow @xcite . in fig .
[ rhic ] ( right ) the nuclear modification factor @xmath1 for non - photonic single electrons is depicted .
( left ) and nuclear modification factor @xmath1 ( right ) of electrons from heavy quark decays in au+au collisions at @xmath24 using a coalescence mechanism .
we use a rapidity cut of @xmath19 . for a decoupling temperature of @xmath20
we get a reasonable agreement to data @xcite , scaledwidth=100.0% ] ( left ) and nuclear modification factor @xmath1 ( right ) of electrons from heavy quark decays in au+au collisions at @xmath24 using a coalescence mechanism .
we use a rapidity cut of @xmath19 . for a decoupling temperature of @xmath20
we get a reasonable agreement to data @xcite , scaledwidth=100.0% ] also here we obtain a good agreement with the data , especially in case of using the t - matrix coefficients or a low decoupling temperature .
now we performed the same calculations , but in pb+pb collisions at @xmath25 in a centrality range of 30%-50% .
the analysis is done in a rapidity cut of @xmath19 in line with the alice data .
here we made use of the coalescence mechanism with a decoupling temperature of @xmath26 only since we achieved the best results using this configuration at rhic energies . in the alice experiment d - mesons are measured .
therefore we do not need to perform the decay to electrons this time .
[ lhc ] ( left ) depicts our results for the elliptic flow compared to alice measurements . of d - mesons in pb+pb collisions at @xmath27tev compared to data from the alice experiment .
( talk by z. conesa del valle at qm 2012 , data not published yet . ) a rapidity cut of @xmath19 is employed .
right : @xmath1 of d - mesons in pb+pb collisions at @xmath28 compared to experimental data from alice @xcite .
a rapidity cut of @xmath19 is employed.,scaledwidth=100.0% ] of d - mesons in pb+pb collisions at @xmath27tev compared to data from the alice experiment .
( talk by z. conesa del valle at qm 2012 , data not published yet . ) a rapidity cut of @xmath19 is employed .
right : @xmath1 of d - mesons in pb+pb collisions at @xmath28 compared to experimental data from alice @xcite .
a rapidity cut of @xmath19 is employed.,scaledwidth=100.0% ] additionally , apart form the calculation using the coalescence mechanism , also a calculation using a fragmentation mechanism is shown , since fragmentation might get more important at higher @xmath3 bins , as measured at lhc . as fragmentation mechanism we used the peterson fragmentation @xcite .
@xmath29 ^ 2},\ ] ] here @xmath30 is a normalization constant , @xmath31 the relative - momentum fraction obtained in the fragmentation of the charm quarks and @xmath32 .
both @xmath0 calculations are in agreement with the alice data set . using the fragmentation function a sharper rise of the elliptic flow at low @xmath3
is reached , while at medium @xmath3 the flow using the coalescence approach is stronger . at high @xmath3
both hadronization mechanisms lead to similar results . a complementary view on the drag and diffusion coefficients is provided by the nuclear suppression factor @xmath1 .
figure [ lhc ] ( right ) shows the calculated nuclear modification factor @xmath1 of d - mesons at lhc .
here we compare to two data sets available , for @xmath33 and @xmath34 mesons . in line with the experimental data
the simulation is done for a more central bin of @xmath35-@xmath36 . in case of the coalescence approach
we find a maximum of the @xmath1 at about @xmath37 followed by a sharp decline to an @xmath1 of about @xmath38 at high @xmath3 .
the fragmentation approach leads to a different result at low @xmath3 . a very sharp @xmath1 drop - off from low to high
@xmath3 is seen . at high @xmath3
the two approaches nearly converge .
concerning the difference of the results using the fragmentation and coalescence mechanism new @xmath0 and @xmath1 measurements , especially at low @xmath3 , would be very helpful to draw conclusions on the hadronization mechanism at lhc .
to summarize , we presented in this letter our results on the medium modification of heavy quarks at rhic and lhc energies using the nuclear modification factor @xmath1 and the elliptic flow @xmath0 as observables . at rhic energies we compared different sets for drag and diffusion coefficients and obtained the best agreement to experimental measurements if using a resonance model with a decoupling temperature of @xmath26 . at lhc
we compared a coalescence approach and a fragmentation approach as hadronization mechanism using the resonance model at a decoupling temperature of @xmath26 .
both approaches describe the ellitpic flow @xmath0 in pretty good agreement with the experimental data while for the @xmath1 a major disagreement between our models at low @xmath3 can be seen that needs to be resolved by new measurements .
we are grateful to the center for scientific computing ( csc ) at frankfurt for providing computing resources .
t. lang gratefully acknowledges support from the helmholtz research school on quark matter studies .
this work is supported by the hessian loewe initiative through the helmholtz international center for fair ( hic for fair ) .
j. s. acknowledges a feodor lynen fellowship of the alexander von humboldt foundation .
this work is supported by the office of nuclear physics in the us department of energy s office of science under contract no .
de - ac02 - 05ch11231 .
99 s. a. bass _ et al .
_ , prog . part . nucl
. phys .
* 41 * ( 1998 ) 255 [ prog . part .
* 41 * ( 1998 ) 225 ] [ arxiv : nucl - th/9803035 ] .
m. bleicher _ et al .
_ , j. phys .
* g25 * ( 1999 ) 1859 [ arxiv : hep - ph/9909407 ] .
d. h. rischke , s. bernard and j. a. maruhn , nucl .
a * 595 * ( 1995 ) 346 [ arxiv : nucl - th/9504018 ] .
d. h. rischke , y. pursun and j. a. maruhn , nucl .
a * 595 * ( 1995 ) 383 [ erratum - ibid .
a * 596 * ( 1996 ) 717 ] [ arxiv : nucl - th/9504021 ] .
r. rapp and h. van hees , arxiv:0903.1096 [ hep - ph ] . h. van hees , m. mannarelli , v. greco and r. rapp , phys .
* 100 * ( 2008 ) 192301 [ arxiv:0709.2884 [ hep - ph ] ] . h. van hees , v. greco and r. rapp , arxiv:0706.4456 [ hep - ph ] .
v. greco , h. van hees and r. rapp , arxiv:0709.4452 [ hep - ph ] .
h. van hees , m. mannarelli , v. greco and r. rapp , eur .
j. c * 61 * ( 2009 ) 799 [ arxiv:0808.3710 [ hep - ph ] ] .
r. rapp , d. cabrera , v. greco , m. mannarelli and h. van hees , arxiv:0806.3341 [ hep - ph ] .
r. rapp and h. van hees , arxiv:0803.0901 [ hep - ph ] .
m. he , r. j. fries and r. rapp , arxiv:1204.4442 [ nucl - th ] .
m. he , r. j. fries and r. rapp , arxiv:1208.0256 [ nucl - th ] .
j. aichelin , p. b. gossiaux and t. gousset , acta phys .
b * 43 * ( 2012 ) 655 [ arxiv:1201.4192 [ nucl - th ] ] .
j. uphoff , o. fochler , z. xu and c. greiner , phys .
c * 84 * ( 2011 ) 024908 [ arxiv:1104.2295 [ hep - ph ] ] .
j. uphoff , o. fochler , z. xu and c. greiner , phys .
b * 717 * ( 2012 ) 430 [ arxiv:1205.4945 [ hep - ph ] ] . c. young , b. schenke , s. jeon and c. gale , phys .
c * 86 * ( 2012 ) 034905 [ arxiv:1111.0647 [ nucl - th ] ] .
g. d. moore and d. teaney , phys .
c * 71 * ( 2005 ) 064904 [ arxiv : hep - ph/0412346 ] .
i. vitev , a. adil and h. van hees , j. phys .
* g34 * ( 2007 ) s769 [ arxiv : hep - ph/0701188 ] .
b. gossiaux , j. aichelin , t. gousset and v. guiho , j. phys .
* g37 * ( 2010 ) 094019 [ arxiv:1001.4166 [ hep - ph ] ] .
b. gossiaux , s. vogel , h. van hees , j. aichelin , r. rapp , m. he and m. bluhm , arxiv:1102.1114 [ hep - ph ] .
p. gossiaux , j. aichelin and t. gousset , prog .
* 193 * ( 2012 ) 110 [ arxiv:1201.4038 [ hep - ph ] ] . h. petersen , j. steinheimer , g. burau , m. bleicher and h. stocker , phys .
c * 78 * ( 2008 ) 044901 [ arxiv:0806.1695 [ nucl - th ] ] .
s. a. bass , a. dumitru , m. bleicher , l. bravina , e. zabrodin , h. stoecker and w. greiner , phys .
c * 60 * ( 1999 ) 021902 [ arxiv : nucl - th/9902062 ] .
a. dumitru , s. a. bass , m. bleicher , h. stoecker and w. greiner , phys . lett .
b * 460 * ( 1999 ) 411 [ arxiv : nucl - th/9901046 ] .
j. steinheimer , m. bleicher , h. petersen , s. schramm , h. stocker and d. zschiesche , phys .
c * 77 * ( 2008 ) 034901 [ arxiv:0710.0332 [ nucl - th ] ] .
j. steinheimer , v. dexheimer , h. petersen , m. bleicher , s. schramm and h. stoecker , phys . rev .
c * 81 * ( 2010 ) 044913 [ arxiv:0905.3099 [ hep - ph ] ] .
h. petersen , g. y. qin , s. a. bass and b. muller , phys .
c * 82 * ( 2010 ) 041901 [ arxiv:1008.0625 [ nucl - th ] ] .
h. petersen , phys .
c * 84 * ( 2011 ) 034912 [ arxiv:1105.1766 [ nucl - th ] ] .
j. steinheimer , s. schramm and h. stocker , phys .
c * 84 * ( 2011 ) 045208 [ arxiv:1108.2596 [ hep - ph ] ] .
b. svetitsky , phys . rev .
d * 37 * ( 1988 ) 2484 .
m. golam mustafa , d. pal and d. kumar srivastava , phys .
c * 57 * ( 1998 ) 889 [ erratum - ibid .
c * 57 * ( 1998 ) 3499 ] [ nucl - th/9706001 ] .
h. van hees and r. rapp , phys .
c * 71 * ( 2005 ) 034907 [ nucl - th/0412015 ] .
h. van hees , v. greco and r. rapp , phys .
c * 73 * ( 2006 ) 034913 [ nucl - th/0508055 ] .
p. b. gossiaux and j. aichelin , phys .
c * 78 * ( 2008 ) 014904 [ arxiv:0802.2525 [ hep - ph ] ] . m. he , r. j. fries and r. rapp , phys .
b * 701 * ( 2011 ) 445 [ arxiv:1103.6279 [ nucl - th ] ] . h. van hees , v. greco and r. rapp , phys .
c * 73 * ( 2006 ) 034913 [ arxiv : nucl - th/0508055 ] .
p. huovinen , p. f. kolb , u. w. heinz , p. v. ruuskanen and s. a. voloshin , phys .
b * 503 * ( 2001 ) 58 [ arxiv : hep - ph/0101136 ] .
d. krieg and m. bleicher , eur .
j. a * 39 * ( 2009 ) 1 [ arxiv:0806.0736 [ nucl - th ] ] .
a. adare _ et al .
_ [ phenix collaboration ] , phys .
c * 84 * ( 2011 ) 044905 [ arxiv:1005.1627 [ nucl - ex ] ] .
[ alice collaboration ] , jhep * 1209 * ( 2012 ) 112 [ arxiv:1203.2160 [ nucl - ex ] ] . c. peterson , d. schlatter , i. schmitt and p. m. zerwas , phys .
d * 27 * ( 1983 ) 105 . | we calculate the heavy quark evolution in heavy ion collisions and show results for the elliptic flow @xmath0 as well as the nuclear modification factor @xmath1 at rhic and lhc energies . for the calculation
we implement a langevin approach for the transport of heavy quarks in the urqmd ( hydrodynamics + boltzmann ) hybrid model . as drag and diffusion coefficients
we use a resonance approach for elastic heavy - quark scattering and assume a decoupling temperature of the charm quarks from the hot medium of @xmath2 . at rhic energies we use a coalescence approach at the decoupling temperature for the hadronization of the heavy quarks to d - mesons and b - mesons and a sub - following decay to heavy flavor electrons using pythia . at lhc
we use an additional fragmentation mechanism to account for the higher transverse momenta reached at higher collision energies . |
bilayer quantum hall systems at total filling factor @xmath0 exhibit one of the most interesting many - body correlation effects : spontaneous interlayer phase coherence , which is solely caused by the coulomb interaction in the limit of zero interlayer tunneling .
recently this phenomenon has received drastic experimental support by spielman _
@xcite , who discovered a strong enhancement in the zero - bias interlayer tunneling conductance for interlayer distance @xmath1 in the regime of very little tunneling , where @xmath2 is the magnetic length .
these experiments strongly indicated a presence of spontaneous interlayer coherence at small interlayer distances .
the ground state in the limit of zero interlayer distance can be shown to be spontaneously interlayer phase coherent due to layer symmetry .
one of the most convenient way to see this is to use the pseudospin representation in which the top ( bottom ) layer is denoted as pseudospin @xmath3 ( @xmath4 ) . in the limit of zero interlayer separation ,
there is exact pseudospin su(2 ) symmetry so that our bilayer system at @xmath5 with pseudospin can be exactly mapped onto a single layer system @xmath6 with real spin .
then , it is straightforward to see that the ground state should be a pseudospin ferromagnet in order to reduce the coulomb energy cost , i.e. hund s rule . on the other hand
, it has been known that , in the limit of large interlayer separation , the ground state is composed of two split composite fermion fermi seas @xcite .
therefore , it is clear that there should be a quantum phase transition at a critical interlayer distance @xmath7 since the ground states at small and large interlayer separations are not adiabatically connected .
however , it is not at all clear what happens near @xmath7 . in spite of intense theoretical efforts in the past
, the true nature of phase transition near @xmath7 has been elusive .
analytical approaches based on hatree - fock theories @xcite were not able to reliably treat strong quantum fluctuations near the phase transition , which led to the erroneous prediction of broken translational symmetry .
on the other hand , analytic approaches utilizing effective field theories @xcite were not directly based on the microscopic hamiltonian of the many - body coulomb interaction .
so its validity should be justified by other means such as exact diagonalization . in particular ,
mapping of the original many - body coulomb hamiltonian to a spin model hamiltonian is not expected to be valid near the critical interlayer separation .
previous numerical studies based on exact diagonalization @xcite , however , did not take into account fundamental fluctuations of interlayer number difference due to intrinsic , quantum - mechanical uncertainty in layer indices .
consequently those studies could not provide any direct information related to spontaneous interlayer phase coherence .
it is crucial at this point to distinguish between pseudospin ferromagnetic order and interlayer phase coherence .
there is no fundamental reason to believe that they are identical because the interlayer phase coherence is a necessary condition , but not a sufficient condition for pseudospin ferromagnetism .
we will show in this article that , at intermediate interlayer distances , there is another sort of long - range interlayer phase coherence due to pseudospin spiral ordering .
it should be emphasized that this pseudospin spiral order is spontaneous in that it exists solely because of the coulomb interaction without any external parallel magnetic field .
the formation of pseudospin spiral order has been considered previously in the presence of parallel magnetic field @xcite .
this article is organized as follows .
we will begin by defining the hamiltonian of bilayer quantum hall systems in sec .
[ sec_hamil ] .
the pseudospin magnetization is computed as a function of interlayer distance @xmath8 in the limit of zero interlayer tunneling in sec .
[ sec_sx ] where the intrinsic , quantum - mechanical fluctuation of interlayer number difference is fully taken into account , based on the analogy with superconductivity . defined as the point where the pseudospin ferromagnetic order is destroyed , the critical interlayer distance @xmath7 is reliably determined . in sec .
[ sec_spiral ] we focus on the nature of the instability which destroys pseudospin ferromagnetic order at @xmath7 .
we will show that the instability is due to the formation of pseudospin spiral order which is manifested in the lowest energy excitations . by computing a new order parameter for pseudospin spiral
, we will show in sec .
[ op ] that the ground state itself has the pseudospin spiral order .
finally , it is argued in section [ conclusions ] that a signature of the phase transition from pseudospin ferromagnetic order to our predicted pseudospin spiral order may have been already observed experimentally .
let us begin by writing the hamiltonian for the bilayer quantum hall systems : @xmath9 where @xmath10 is the lowest landau level ( lll ) projection operator .
@xmath11 represents the usual coulomb interaction between electrons : @xmath12 where @xmath13 is the magnetic length , @xmath8 is the interlayer distance , and @xmath14 is the lateral separation between the _ i_-th and _ j_-th electrons . in the above
we have used a pseudospin representation to distinguish the top ( @xmath3 ) and the bottom ( @xmath4 ) layers . in general
, we define the pseudospin operator as follows : @xmath15 where @xmath16 is the usual pauli matrix , and @xmath17 denotes the lll orbital index .
note that @xmath18 is half the electron number difference between the two layers , and @xmath19 is associated with interlayer tunneling .
we take the real spin to be fully polarized .
the tunneling hamiltonian @xmath20 in eq.([hamiltonian ] ) can be written as : @xmath21 where @xmath22 is the single particle interlayer tunneling gap .
although eq.([h_t ] ) is valid for general @xmath22 , we are interested only in the limit of zero interlayer tunneling , i.e. @xmath23 , which is appropriate when considering spontaneous interlayer coherence ( note that the @xmath24 limit is not the same as the @xmath25 situation ) .
we analyze the hamiltonian in eq.([hamiltonian ] ) by using exact diagonalization ( via a modified lanczos method ) in the torus geometry @xcite .
as mentioned in the beginning , it will be shown that there is a new long - range interlayer phase coherence due to pseudospin spiral order at sufficiently large @xmath8 .
but , first , we should determine the critical interlayer distance , @xmath7 , at which pseudospin ferromagnetic order terminates . of course , the most natural order parameter for the pseudospin ferromagnetic order is the pseudospin magnetization in the @xmath26-direction : @xmath27 .
it is tempting to apply exact diagonalization techniques to finite systems with fixed number of electrons in each layer .
but , @xmath19 changes the interlayer number difference by unity ( @xmath28 ) so that the ground state expectation value is precisely zero when computed naively without any explicit interlayer tunneling .
this problem has been addressed in an _ ad hoc _ manner by introducing explicit interlayer tunneling @xcite , which , however , severely obscures the effect of spontaneous phase coherence and may produce misleading results .
the real solution is to realize that there is intrinsic , quantum - mechanical uncertainty in the layer index of electrons at small interlayer distance ( even in the limit of zero tunneling ) so that the true ground state @xmath29 is a linear combination of various states with different @xmath18 : @xmath30 where @xmath31 is the lowest energy state with @xmath32 , @xmath33 is a sharply peaked function of @xmath34 with width @xmath35 , and @xmath36 is the number of electrons .
it is important to remember that even the halperin ( 1,1,1 ) state , @xmath37 , has @xmath38 , which can be proved by analyzing the following , explicit wavefunction including both the orbital and pseudospin part @xcite : @xmath39 |0\rangle .
\label{111}\ ] ] @xmath37 is the exact ground state at @xmath40 .
because it can be exactly mapped onto the bcs wavefunction , it is fruitful to consider an analogy with the bcs theory for general @xmath8 . following the bcs theory
, one can compute the pseudospin magnetization @xmath19 as follows : @xmath41 where the last two step are justified in the thermodynamic limit since @xmath33 is sharply peaked at @xmath42 .
although @xmath27 should be exactly @xmath43 at @xmath40 in the thermodynamic limit due to the exact pseudospin su(2 ) symmety , there are some finite - system size corrections .
after some algebra on the ( 1,1,1 ) state , eq.([bcs ] ) shows that @xmath44 is @xmath45 for n odd ( @xmath46 ) , and is @xmath47 for n even ( @xmath48 ) .
therefore , the pseudospin magnetization should be scaled as @xmath49 ( @xmath50 ) for @xmath36 odd ( even ) .
fig.[fig1 ] shows the pseudospin magnetization per particle ( defined in eq.([bcs ] ) ) as a function of @xmath51 for various numbers of electrons .
as conventional in finite system analysis , the critical interlayer distance can be obtained from the inflection point of @xmath27 : @xmath52 .
although @xmath7 was estimated to be around @xmath53 previously in various approaches , it should be emphasized that the pseudospin magnetization @xmath27 is computed for the first time in this article without any interlayer tunneling . and therefore @xmath7 is reliably determined without any ambiguity . ) as a function of interlayer distance @xmath51 for various numbers of electrons @xmath36 .
[ fig1],width=220 ]
we now address the nature of instability causing the destruction of pseudospin ferromagnetic order .
this question is , in turn , directly related to the nature of the new ground state at @xmath54 . to be specific
, we consider the lowest energy excitations which are responsible for the ground state instability .
fig.[fig2 ] shows the dispersion curves of the lowest energy excitations at @xmath55 and @xmath56 for a finite system with the total number of electrons @xmath57 and the number of pseudospin - up electrons @xmath58 ( or @xmath59 ) .
there are several points to be emphasized .
first , there is clearly a linearly dispersing goldstone mode at small interlayer distances , such as @xmath55 .
however , @xmath60regarding the discreteness of finite - system wavevectors , the goldstone mode seems to vanish for @xmath61 , which coincides with the destruction of pseudospin ferromagnetic order @xcite .
consistent with the conclusion from @xmath27 , this shows that the ground state undergoes a phase transition around @xmath62 .
second , contrary to the prediction of the random phase approximation ( rpa ) theory @xcite , there is no softening of the dispersion curve ( roton ) at any finite momentum for any interlayer distance .
remember that the rpa theory predicts a roton around @xmath63 , which was believed to cause an instability toward charge density wave order .
this already shows that previous theories have some serious flaws in describing the phase transition at @xmath62 .
we will show in the next paragraph that there are completely different low - energy excitations at large @xmath51 .
third , several previous numerical studies @xcite were unfortunately based on the exact diagonalization of @xmath64 system , which , under careful investigation @xcite , can be shown to predict a wigner crystal at large @xmath8 rather than a quantum hall liquid , which is a finite - system artifact of boundary condition and special number of electrons .
fourth , these lowest - excitation energies are exactly zero in the @xmath65 limit where there is no interaction between electrons with different pseudospins . and @xmath56 for a finite system with the total number of electrons @xmath57 and the number of pseudospin - up electrons @xmath58 ( or @xmath59 ) .
it will be shown that the lowest energy state in the momentum channel @xmath66 ( denoted by arrows ) is the pseudospin spiral state .
[ fig2],width=240 ] however , the most crucial aspect of the dispersion curve is that there are completely new low - energy excitations due to the pseudospin spiral order , which are denoted by arrows in fig.[fig2 ] .
the lowest energy state in the momentum channel @xmath66 ( denoted by arrows in fig.[fig2 ] ) begins to be ripped out of the well - defined dispersion curve at @xmath67 , and it becomes completely separated from all the other excitations at @xmath68 so that it becomes really the lowest energy excitation among all momentum channels .
the same behavior is found in all studied systems with different values of @xmath36 and @xmath69 , with exception of the @xmath64 system due to a finite - size artifact as mentioned in the above .
note that there are three other degenerate excitations in the momentum channels @xmath70 , @xmath71 , and @xmath72 due to reflection symmetries and periodic boundary conditions of torus geometry .
these peculiar excitations are completely different from usual rotons because ( i ) they form isolated points rather than a part of conventional dispersion curve , and ( ii ) their momentum increases as @xmath73 , which diverges in the thermodynamic limit .
we will show later that these peculiar properties are precisely the properties of pseudospin spiral state . in order to understand why these excitations are the pseudospin spiral states
, it would be best if we first examine the following pseudospin spiral wavefunction : @xmath74 |0\rangle \label{spiral}\ ] ] where @xmath17 denotes the linear momentum in the landau gauge , i.e. , @xmath75 and @xmath76 is the linear system size ( note that @xmath77 at @xmath0 ) .
we will show later in fig.[fig3 ] by explicitly computing the overlap between @xmath78 and the lowest - energy excitations ( denoted by arrows in fig.[fig2 ] ) that the lowest - energy excitations are the pseudospin spiral states .
but , first , we would like to emphasize that the pseudospin spiral state @xmath79 is the exact ground state in the presence of parallel magnetic field @xcite .
a surprising result of this article is that , even without parallel magnetic field , spiral states are the lowest energy excitations which cause the instability of pseudospin ferromagnetic order . .
[ fig3],width=220 ] to elucidate the physical meaning of pseudospin spiral state , examine eq.([spiral ] ) which reveals that @xmath80 contains diagonal interlayer correlations between the states with @xmath81 and @xmath82 .
remember that the interlayer correlation between @xmath83 and @xmath84 is identical to the diagonal interlayer correlation between electrons with different pseudospins which are separated laterally in the @xmath85-direction by @xmath86 . in turn , this diagonal interlayer correlation is physically equivalent to the pseudospin spiral order with spiraling period of @xmath87 because the pseudospin part of wavefunction is essentially given by @xmath88 pseudospin spiral excitations ( and therefore the diagonal interlayer correlation ) are schematically depicted in fig.[fig4 ] . , where the ground state is conjectured to be a bound state of two pseudospin spiral states with opposite winding direction , i.e. a bound state of ( b ) and ( c ) .
[ fig4],width=249 ] now we compute the momentum of pseudospin spiral states by directly applying the magnetic translation operator onto @xmath89 .
the explicit algebra shows that the momentum of @xmath90 is @xmath91 which is exactly the momentum of the lowest - energy excitation seen in fig.[fig2 ] if @xmath92 .
note that the momentum of this pseudospin spiral state is actually proportional to its spiraling period .
that is the reason why the momentum of pseudospin spiral state with long spiraling period diverges in the thermodynamic limit .
similarly , we can explain why the pseudospin spiral state occurs in an isolated momentum channel .
consider two pseudospin spiral states with long period , say @xmath92 and @xmath93 , whose momenta differ by a factor proportional to @xmath94 .
therefore , the pseudospin spiral states with two different long spiraling periods do not occur in adjacent momenta .
it is important to note that @xmath78 is the pseudospin spiral excitation in the limit of long spiraling period ( @xmath95 ) , which initiates the destruction of pseudospin ferromagnetic order .
therefore , it may be indicative of a second order phase transition @xcite .
as mentioned earlier , fig.[fig3 ] shows the overlap between the pseudospin spiral state @xmath78 in eq.([spiral ] ) and the lowest - energy excited state with @xmath96 @xcite .
as one can see , the overlap is high overall at small @xmath51 , and is very close to unity ( @xmath97 ) especially around @xmath98 which coincides with the point where @xmath27 begins to deviate from the fully saturated value of @xmath99 .
this shows that the instability of pseudospin ferromagnetic order is indeed caused by pseudospin spiral order .
it should be noted that , though the overlap is lowered at larger @xmath51 , the low energy excitation is in general adiabatically connected to the pseudospin spiral state . in fact , the reason for lower overlaps at larger @xmath51 is that our trial wavefunction in eq.([spiral ] ) is not very accurate anymore for larger @xmath51 since it is based on the assumption that the ground state is the ( 1,1,1 ) state .
of course , the ground state itself is affected by the pseudospin spiral instability which , after all , destroys the pseudospin ferromagnetic order at the critical interlayer distance .
therefore , it would be satisfactory if , in addition to the evidence due to the low energy excitation , one can prove directly that the ground state has the pseudospin spiral order at @xmath100 .
the next section will be devoted to the pseudospin spiral order in the ground state . with @xmath92 ) measuring the diagonal interlayer correlation .
note that the diagonal interlayer correlation is physically equivalent to the pseudospin spiral order .
[ fig5],width=249 ]
it is conjectured that the ground state at large @xmath8 is in fact a bound state of two pseudospin spiral states with opposite winding direction , i.e. a bound state of ( b ) and ( c ) in fig.[fig4 ] , which requires a complicated interaction between the two spiral states ( note that this conjecture is consistent with the fact that the ground state itself does not break translational symmetry ) .
so instead of constructing a trial wavefunction we take a more transparent approach by defining a new order parameter which measures the pseudospin spiral order ( or equivalently the diagonal interlayer correlation ) : @xmath101 the following properties of @xmath102 are worth mentioning .
( i ) @xmath102 is a generalization of pseudospin magnetization @xmath19 which measures the vertical interlayer correlation : @xmath103 .
so the relationship between @xmath102 and @xmath90 is analogous to that between @xmath19 and @xmath37 .
\(ii ) when acting upon momentum eigenstates , @xmath102 alters momentum of the eigenstates , though changes in momentum are negligible in the thermodynamic limit .
so , in order to avoid any confusion , we compute @xmath104 instead .
\(iii ) in order to be meaningful in the thermodynamic limit , @xmath104 should be scaled as @xmath105 .
\(iv ) since the ( 1,1,1 ) state does not have any pseudospin spiral order , @xmath106 should be zero in the thermodynamic limit . however , there are some finite - system size corrections : @xmath107 .
we subtract these finite system size corrections from the expectation value of @xmath108 .
fig.[fig5 ] plots the ground state expectation value of @xmath109 as a function of interlayer distance @xmath51 .
as one can see , @xmath110 is peaked around @xmath111 which is consistent with the critical interlayer distance @xmath7 .
so it is shown that indeed the ground state contains the pseudospin spiral order at @xmath100 .
in fact , fig.[fig5 ] implies that the pseudospin ferromagnetic order and the pseudospin spiral order may coexist for a range of @xmath8 since the expectation value of pseudospin spiral order parameter is not zero even in interlayer distances smaller than @xmath7 .
of course , in the thermodynamic limit , @xmath110 will be zero at @xmath112 since there is no interlayer correlation in the two split fermi seas of composite fermions . finally , we would like to discuss the physical origin of pseudospin spiral order .
first , let us remind ourselves that the pseudospin ferromagnetic state can be viewed as a condensate of neutral pairs of electrons and correlation holes which are bound directly across the interlayer separation .
now for sufficiently large @xmath8 the ground state has the pseudospin spiral order , or equivalently diagonal interlayer correlation , which can be also viewed as a condensate of neutral electron - hole pairs with extended size and higher symmetry such as @xmath113-wave , as compared to the @xmath114-wave - like pair in pseudospin ferromagnet .
now , the origin of spiral order can be identified with the origin of extended pairs which is due to weak pairing .
of course , the pairing between electrons and correlation holes is weakened as d increases .
eventually the binding is completely lost when the split composite fermion fermi seas are formed .
we conclude by mentioning some experimental implications . recently ,
kellogg _ et al . _ observed an enhanced longitudinal drag resistance at @xmath51 as large as @xmath115 where the enhanced tunneling conductance is absent @xcite .
theoretically , tunneling is renormalized from the bare single particle tunneling gap @xmath22 to the greatly enhanced renormalized tunneling @xmath116 .
therefore , it is natural to assume that the enhanced tunneling conductance becomes absent as soon as the pseudospin ferromagnetic order is destroyed at @xmath62 . on the other hand , we have shown that there is a new interlayer phase coherence due to the pseudospin spiral order for @xmath117 which , we speculate , causes the longitudinal drag anomaly because of remaining interlayer correlation . regarding the hall resistance ,
our pseudospin spiral state is expected to be incompressible , as shown in the dispersion curve .
however , the lowest energy gap is estimated to be very small so that the hall plateau will be difficult to be observed in current temperature ranges and impurity concentrations .
the author is very grateful to s. das sarma for his careful reading of manuscript and his constant support throughout this work .
the author is also indebted to e. demler , s. m. girvin , a. kaminski and v. w. scarola for their insightful comments .
this work was supported by arda . as shown later
, the ground state at large @xmath8 acquires a pseudospin spiral order which is associated with diagonal interlayer correlation . since this diagonal interlayer correlation creates a _ nonlocal _ correlation between laterally separated electrons ( in top layer ) and correlation holes ( in bottom layer ) for sufficiently large @xmath8 , it is no longer valid to map the original long - range coulomb hamiltonian to a usual spin model which is based on the existence of _ local _ spin operator
. therefore , there is no fundamental reason for the goldstone mode at sufficiently large @xmath8 .
we have proved that for @xmath64 the energy dispersion in the first brillouin zone has a pattern associated with rectangular - lattice wigner crystal which is commensurate with rectangular unit cell .
also , this pattern persists even in the @xmath65 limit . in terms of the viewpoint that the interlayer coherent state is a macroscopic condensate of neutral electron - hole pairs ,
the pseusospin spiral instability with long spiral period can be interpreted as a gradual change in the nature of neutral pairs from tightly bound @xmath114-wave - like pairing to a pairing with extended size and higher symmetry . and
this continuous change in the nature of pairing may indicate a second order phase transition which is consistent with the viewpoint from the pseusospin spiral instability with long spiral period .
note that in order to compute the overlap one should project @xmath80 into the hilbert space of fixed @xmath18 since @xmath80 is a linear combination of various @xmath18 eigenstates similar to the ( 1,1,1 ) state . | using exact diagonalization of bilayer quantum hall systems at total filling factor @xmath0 in the torus geometry , we show that there is a new long - range interlayer phase coherence due to spontaneous pseudospin spiral order at interlayer distances larger than the critical value at which the pseudospin ferromagnetic order is destroyed .
we emphasize the distinction between the interlayer phase coherence and the pseudospin ferromagnetic order . |
it is generally accepted that the massive black - holes ( mbhs ) found in the centres of most galaxies grow via accretion and via mergers following mergers between their host galaxies . however
, little is known about the seeds from which mbhs grow @xcite .
the seeds could be _ massive _
( i.e. , @xmath4 ) , in which case the first epoch of mergers between galactic black holes at high redshift will generate gravitational waves ( gws ) that could be detected by the future space - borne detector lisa @xcite
. however , these seeds could also be _ light _
( i.e. , @xmath5 ) and , in that case , the gws generated during these first mergers will be at frequencies between the sensitive bands of lisa and the ground - based instruments currently in operation ligo , virgo and geo 600 @xcite , or their upgraded advanced versions . in this paper
we show that third - generation ground - based interferometers and second - generation space - based detectors may be able to fill this gap and directly probe the first mergers between light mbh seeds .
light seed black holes might be produced as the remnants of pop iii stars @xcite .
models including light seeds predict dozens of mbh binary ( mbhb ) coalescences per year in the mass range @xmath6 @xcite . here and elsewhere
we will stretch the definition of mbh to mean any black hole in the centre of a dark matter halo .
these events are mostly at high redshift , @xmath7 , and therefore only events with total mass @xmath8 will be accessible to lisa . to observe systems in the @xmath9 range will require a gw detector sensitive to frequencies in the @xmath10hz range .
this might be achieved in space by second - generation instruments such as the big - bang - observer @xcite , decigo @xcite or alia @xcite , or on the ground by third - generation interferometers . in this paper
we will use decigo as an example of a future space - based detector , and the einstein gravitational - wave telescope ( et ) @xcite as an example of a future ground - based detector .
this choice was motivated by the fact that et is currently undergoing a design study within europe , and a technology demonstration mission for decigo is currently under development in japan .
decigo is a very ambitious mission requiring new technology and so the timescale is rather uncertain , but at present the launch target is @xmath11 .
the timetable for et is also rather uncertain , but construction will start no earlier than @xmath12 .
we will demonstrate that these instruments will have the capability to detect seed black - hole binaries , and we will discuss whether these observations could tell us definitively that we are observing light seeds produced by pop iii stars .
we have already estimated the seed black - hole event rates for et @xcite as input for the design study , but in this paper we compare and contrast these results with those for decigo , and provide additional details regarding the distribution of the detectable sources and parameter - estimation accuracy .
this paper is organised as follows . in section
[ methods ] we describe our calculations , including the astrophysical model used to construct the event populations , the waveform models used and the sensitivity curves of the various instruments . in section [ rates ]
we provide details of the detectable seed black - hole merger events , including the number of events and the mass and redshift distributions of events . in section [ parest ]
we discuss the parameter estimation accuracy that might be achieved for these systems by a network of third - generation ground - based detectors , before summarizing our conclusions in section [ discuss ] .
to generate event populations , we trace the galaxy merger hierarchy using monte - carlo merger - tree realisations built on the extended press - schechter formalism @xcite , using simulations described in detail in @xcite . in these simulations , dark matter halos are populated with @xmath5 seed black holes at redshift @xmath13 , and it is assumed that these black holes merge as their host halos merge , and accrete mass efficiently . in the present work , we assumed a standard lcdm cosmology with wmap 1-year parameters @xcite and considered two variants of the pop iii seed model . in both cases
, we used the same merger history for the dark - matter halos and assumed that the seeds accrete at the eddington limit during each merger episode , but the models differed in the initial mass - distribution of the seeds : ( i ) the _ vhm , ems _ ( volonteri - haardt - madau ( vhm ) with equal - mass seeds ) model assumes equal @xmath14 seeds ; ( ii ) the _ vhm , smd _ model ( vhm with seed - mass distribution ) takes the seed - mass distribution to be uniform in log - mass for masses in the range @xmath15 .
we have also studied two further scenarios that differ in the model of accretion onto the seeds , using prescriptions in @xcite .
results from all four scenarios were considered for the et study in @xcite , but we now restrict to the two representative models _
vhm , ems _ and _ vhm , smd _ to avoid overloading the figures in this paper .
these models produce event populations that are consistent with observational constraints at @xmath16 from the x - ray and optical quasar luminosity function and from the observed faint x - ray counts of agns @xcite .
the models predict @xmath17 mbhb coalesences per year in the universe when summed over all black hole masses , but uncertainties in the assumptions of the model could change this by a factor of a few either way .
we are using these models to study black holes at @xmath18 , but they have been tuned to reproduce observations at lower redshift , @xmath19 . indeed ,
accretion onto light black holes may be very inefficient @xcite , which adds further uncertainties to the picture .
however , we are using these specific models only to indicate the potential of future gw observations as a probe of seed black holes .
our results are not designed to be robust predictions of the event rates given our present limited understanding of the details of the physical processes involved .
we compute signal - to - noise ratios ( snrs ) , @xmath20 , for these sources using the usual expression , @xmath21 , where @xmath22 is the noise - weighted inner - product @xmath23 in which @xmath24 is the fourier transform of the waveform , and @xmath25 is the one - sided power spectral density of the noise in the detector .
to assess parameter - estimation accuracies , we compute the inverse of the fisher information matrix ( fim ) @xmath26 where @xmath27 denotes the model parameters . when considering multiple detectors in a network , the above expressions are still valid if the inner product is replaced by the sum of the inner products over the individual detectors .
to model the gravitational waves generated by mbhb mergers , we use the phenomenological waveform model ( imr ) described in @xcite .
this waveform model includes the inspiral , merger and ringdown radiation in a consistent way using a prescription of the form @xmath28 expressions for @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath34 and @xmath35 are given in eqs .
( 4.14)-(4.19 ) and tables i - ii of @xcite . for these sources ,
much of the inspiral is outside of the range of ground - based detectors like et , so it is necessary to include the contributions from merger and ringdown in order to accumulate a significant snr .
the waveforms are functions of the redshifted total mass of the source , @xmath36 , and the symmetric mass ratio , @xmath37 . in the present calculation
we have used the coefficients from the original paper @xcite , although updated coefficients are available @xcite .
we have verified that the snrs computed from the two sets of coefficients are nearly identical . as a check of our results
, we also computed the snr for events using the effective - one - body - numerical - relativity ( eobnr ) waveform family introduced by buonanno et al .
a comparison of the et snrs is shown in fig .
[ f : eobvimr ] .
for comparable mass ratios , @xmath38 , the snrs predicted by the eobnr waveforms are somewhat higher than the imr snrs ( by up to @xmath39 when the merger and ringdown phases dominate the snr ) , while for asymmetric mass ratios , @xmath40 , they tend to be somewhat lower .
however , neither waveform family has been shown to be valid for @xmath41 .
these comparisons give us confidence that the detection - rate estimates quoted here are reasonably trustworthy . and for two different choices of mass ratio @xmath42 and @xmath43 .
events are at a fixed luminosity distance of @xmath44gpc , corresponding to @xmath45 , and we are computing the snr for one 10 km right - angle detector with the et sensitivity.,scaledwidth=60.0% ] we consider snrs and event rates for lisa @xcite , decigo @xcite , advanced ligo @xcite and et . in addition , we present results for two different et configurations a `` broad - band '' configuration as described in @xcite which we refer to as `` et '' and which was used to obtain the results shown in fig .
[ f : eobvimr ] , and a `` xylophone '' configuration which trades off sensitivity at higher frequency for improved sensitivity near @xmath2hz and was described in @xcite . for the snr calculations , we compute sky- and orientation - averaged snrs , which are obtained by computing the snr for an optimally oriented source using eq .
( [ snr ] ) and dividing by @xmath46 .
the specific detector therefore enters only through the specification of the noise spectral density .
the noise spectral densities we assumed for each detector are illustrated in fig .
[ fig : noisecurve ] . for lisa and decigo the @xmath47 illustrated is for an equivalent right - angle interferometer , obtained by multiplying the usual @xmath47 for a 60@xmath48 interferometer by @xmath49 .
this ensures the sky - averaging for these detectors is done in exactly the same way as the others .
the decigo noise curve is for a single channel of one decigo - like interferometer .
the target decigo design calls for four interferometers , which would enhance the snr by a factor of 2 compared to that in a single interferometer .
the einstein telescope will be a 10 km scale laser - interferometer .
a key target in the et design is the capability to measure polarisation at a single site .
this is achievable by having two coplanar detectors at the site , offset by 45@xmath48 .
the currently favoured design is for a triangular facility containing three 10 km 60@xmath48 detectors , as this has lower facility costs , and we will refer to this as a `` single et '' .
the spectral density shown for both et configurations in fig .
[ fig : noisecurve ] , and which we use in our snr calculations , is for a single right - angle 10 km detector @xcite .
the sensitivities of two right - angle 10 km detectors and a `` single et '' are factors of @xmath50 and @xmath51 higher , respectively , than that of one right - angle 10 km detector . in section [ parest ] , we will present estimates of the accuracy with which a network of et - like detectors can measure the parameters of seed black hole mergers . we consider four third generation network " configurations i ) one et at the geographic location of virgo , plus a second right - angle 10 km detector at the location of ligo hanford or perth ( australia ) ; ii ) as configuration ( i ) plus a third 10 km detector at the location of ligo livingston ; iii ) as configuration ( i ) but with the hanford / perth 10 km detector replaced by a second et ; and iv ) three ets , one at each of the sites . so that they represent equivalent right - angle interferometers.,scaledwidth=60.0% ]
in fig . [ fig : snrvm ] we show how the snr in each of the detectors varies as a function of the redshifted total mass for sources at fixed luminosity distance and for two different values of mass ratio , @xmath52 .
advanced ligo is unsurprisingly beaten at all masses by et , but et and lisa are complementary in that et is more sensitive for @xmath53 , while lisa is better above that mass .
this is advantageous , as these two detectors may well be operating concurrently .
decigo beats every other detector for all masses .
however , this relies on extrapolation of the decigo noise curve over many decades of frequency , and it is not yet clear how valid that extrapolation will be . a less ambitious second generation space - based detector like alia @xcite would be most sensitive for @xmath54 , but would be beaten by et / lisa below / above that range .
the xylophone configuration significantly improves et s sensitivity for black hole mergers with @xmath55 , as the instrumental noise is significantly reduced at precisely the frequency where most of the snr is accumulated for these systems .
this is very important for the seed black hole event rate .
we note that the snrs of events at redshift @xmath45 will be @xmath56 for et , @xmath57 for lisa and @xmath58 for decigo . .
all sources are at a fixed luminosity distance of @xmath44gpc.,scaledwidth=60.0% ] to estimate the number of events that each of the detectors would see , we used 1000 merger - tree realisations for each of the two seed black - hole initial mass - distributions described in section [ methods ] and constructed a catalogue of the merger events that occurred over three years in each case . for each event , we computed the matched filtering snr that would be obtained for that event by each detector , using eq .
( [ snr ] ) . specifying the snr threshold that will be required in the detector for an event to be resolvable then determines the number of events that will be observed .
the event rate is shown as a function of snr threshold in fig .
[ fig : nvthresh ] . for the et configurations the snr threshold is the snr in a single right - angle interferometer with the target sensitivity , and for lisa / decigo it is the snr in a single michelson channel .
the snr threshold that is likely to be required is @xmath59 for the network of detectors , which corresponds to snrs in a single interferometer of 5.3 for a single et , or snrs of 4.8 , 3.9 , 3.8 and 3.1 for the et network configurations ( i)(iv ) described in section [ detmod ] , or an snr of 5.7 in a single michelson channel of lisa / decigo ( assuming that these instruments can be thought to consist of two independent data channels only ) .
the snr threshold of @xmath60 may be optimistic when other considerations are taken into account , in particular foreground confusion from other gravitational wave sources .
this figure illustrates the sensitivity of the event rate to the threshold that is ultimately required .
advanced ligo is not included in fig .
[ fig : nvthresh ] because it will not detect any events in these astrophysical scenarios .
we see from fig .
[ fig : nvthresh ] that an et network will detect between a few and a few tens of events , which will be mostly early mergers between light seeds ; lisa will detect several tens of events , which will be mostly mergers between heavier black holes that have already undergone several merger events ; and even a single decigo - like interferometer will provide a complete survey of these events , which is why the decigo curves are flat for all values of the snr threshold shown in the plot .
we note that the et xylophone configuration would detect significantly more seed black hole events at relatively high snr than the broadband et configuration
. this should be an important consideration when final design decisions are made .
we emphasise that these results should not be considered to be robust predictions for the event rates , due to the uncertainties in the astrophysical models , but are designed to illustrate the potential science that these future detectors could do .
the event rate alone does not provide a proper comparison between the different detectors , since it suppresses information about the characteristics of the events . in fig .
[ fig : dndm ] we show the mass and redshift distributions of events detected by each instrument for the _ vhm , ems _ and _ vhm , smd _ scenarios .
each curve is normalised so that the integral under the curve is equal to 1 , and the mass distribution is for the intrinsic mass , @xmath61 , not the redshifted mass .
these plots reflect the complementarity of the detectors noted earlier .
lisa and et are almost completely complementary in mass et / lisa detects events only below / above @xmath62 but probe a comparable range in redshift .
the distribution of decigo events follows the true astrophysical distribution of mergers most mergers are between @xmath63 black holes in the _ vhm , ems _ scenario , but the mass distribution is quite flat in the _ vhm , smd _ case , in the range @xmath64 . in both cases
the redshift distribution of events is quite flat and extends up to @xmath65 .
if decigo was replaced by alia , then it would not be able to detect the tail of @xmath66 black holes found in the _ vhm , smd _ model , leaving only et able to probe those systems .
et events are predominantly of low mass @xmath67@xmath68 in both configurations and would have redshift @xmath69 if it was operated in the broad band mode ; if et was operated in the xylophone configuration , it would be able to detect mergers out to @xmath70 and indeed would be dominated by events with @xmath71 .
lisa events are predominantly of high mass @xmath72@xmath73 and redshift @xmath69 .
the coalescences detected by lisa are between black holes that have already undergone several merger events , while et probes the earliest mergers between seed black holes .
thus , lisa and et together will probe the full merger history of galaxies in these scenarios .
decigo or alia by themselves would have the same capability , which is an important scientific motivation for building these ambitious instruments . [
cols="^,^ " , ]
in this paper we have explored the capability of future gravitational - wave detectors to probe the hierarchical assembly of structure in the universe starting from light seeds of massive black holes , born during supernovae of population iii stars . in the astrophysical scenarios that we considered ,
lisa would detect several tens of events , but can not probe black hole masses below @xmath74 .
while these observations will have some power to distinguish between light - seed and heavy - seed scenarios ( see , e.g. , @xcite ) , lisa can not probe the first generation of mergers .
the einstein telescope will complement lisa by detecting a few to a few tens of black hole mergers per year in the @xmath2@xmath75 range , detecting the first epoch of mergers and thus directly probing the mass - distribution of the seeds .
decigo would provide a complete survey of mbhb mergers occurring over its lifetime . the launch date for this instrument
is rather uncertain , but if it achieves the target of 2024 it may operate at the same time as et , allowing simultaneous measurements
. more likely , decigo will be later than this , in which case the detections it makes will confirm earlier results from lisa and et .
these future observations will provide detailed information on the assembly of structure and stringent constraints on viable models of galaxy formation and growth .
the detection of at least one light seed will itself be of huge significance in constraining the heavy - seed model .
of the two alternative designs for et that we have considered , it is quite clear that the xylophone configuration is to be preferred for the detection of these sources .
it not only enhances the event rate , but it allows et to detect mergers at higher redshift , which is important for having confidence that the events are seed mergers .
this should be born in mind when the configuration for et is finalised .
we have also discussed parameter estimation for these merger events .
measuring the parameters of the short - lived events with @xmath76 will require a network of concurrently operating detectors .
we have shown that a ground - based network of et detectors should be able to measure the redshifted mass and the luminosity distance of the majority of light - seed merger events to accuracies of @xmath77 and @xmath78 respectively .
this should be sufficient for us to say with confidence that we are observing mergers involving light seeds of massive black holes at high redshift .
our results indicate that advanced ligo will not detect any events from these sources .
this is in contrast to a prediction made in @xcite .
however , that previous calculation was based on a semi - analytic prescription for the black hole merger history which overpredicts the number of events involving low - mass black holes at low redshift as compared to our merger - tree calculations . in the model of @xcite
, et would detect several hundred events rather than several , which our calculations suggest is very optimistic .
the current work should not be taken as a robust calculation of the properties of the events that will be observed by future gravitational - wave detectors , but as an indication of the potential science that these detectors will do .
more work is needed to properly quantify exactly what gravitational - wave observations will tell us about the growth of structure .
nonetheless , we expect that the calculations described here will be of use to researchers interested in quantifying such questions in the future .
we would like to thank m. volonteri for useful discussions and for providing the monte - carlo realisations of the halo and mbh merger hierarchy .
we also thank a. buonanno , p. ajith , y. pan and e. ochsner for very helpful discussions of the imr and eobnr waveforms .
we used code from the lsc algorithm library for the eobnr waveform family .
jg s work is supported by the royal society .
i m is partially supported by nasa atp grant nnx07ah22 g to northwestern university . as acknowledges support from national science foundation grants no .
phy 06 - 53462 and no .
phy 05 - 55615 , and nasa grant no .
nng05gf71 g , awarded to the pennsylvania state university .
av is partially supported by the uk science and technology facilities council .
10 sesana a. , volonteri m. & haardt f. , 2007 , mnras , 377 , 1711 bender p.
et al . , 1998 _ lisa pre - phase a report ; second edition _
, mpq 233 sigg d. et al . , 2008 , class .
quantum grav .
25 , 114041 acernese f. et al . , 2008 , class .
quantum grav .
25 , 114045 grote h. et al . , 2008 , class .
quantum grav .
25 , 114043 madau p. & rees m. j. , 2001 , apj , 551 , l27 sesana a. , haardt f. , madau p. & volonteri m. , 2004 , apj , 611 , 623 phinney e. s. et al . , 2003 , the big bang observer , nasa mission concept study kawamura s. et al . , 2006 ,
quantum grav . , 23 , s125 bender p. l. , armitage , p. j. , begelman , m. c. & pema , r. , 2005 , _ massive black hole formation and growth _
, white paper submitted to the nasa seu roadmap committee .
hild s. , chelkowski s. & freise a. , 2008 , arxiv:0810.0604 freise a. , chelkowski s. , hild s. , del pozzo w. , perreca a. & vecchio a. , 2009 , class .
quantum grav . , accepted , arxiv0804.1036 sesana a. , gair j. r. , mandel , i. & vecchio a. , 2009 , preprint arxiv:0903.4177 press w. h. & schechter p. , 1974
, apj , 187 , 425 volonteri m. , haardt f. & madau p. , 2003 ,
apj , 582 , 599 volonteri m. , salvaterra r. & haardt f. , 2006 , mnras , 373 , 121 spergel d. n. et al , 2003 , apjs , 148 , 175 shankar f. , salucci p. , granato g. l. , de zotti g. , danese l. , 2004 , mnras , 354 , 1020 hopkins p. , hernquist l. , cox t. , di matteo t. , robertson b. , springel v. , 2005 , apj , 632 , 81 alvarez m. a. , wise j. h. & abel t. , 2008 , arxiv:0811.0820 milosavljevic m. , couch s. m. & bromm v. , 2008 , arxiv:0812.2516 ajith p. et al . , 2008 , phys.rev.d , 77 , 104017 ajith p. , 2008
quantum grav .
25 , 114033 buonanno a. et al . , 2007 , phys.rev.d 76 , 104049 fritschel p. , 2003
, in cruise m. & saulson p. ( eds . )
spie conference series , 4856 , 282 hild s. , chelkowski s. , freise a. , franc j. , morgado n. , flaminio r. & desalvo r. , 2009 , arxiv:0906.2655 arun k. g. et al . , 2009 , class .
quantum grav .
26 , 094027 van der sluys m. , mandel i. , raymond v. , kalogera v. , roever c. & christensen n. , 2009 , arxiv:0905.1323 vallisneri m. , 2008 , phys .
d 77 , 042001 plowman j. e. , jacobs d. c. , hellings , r. w. , larson s. l. & tsuruta s. , 2009 , preprint arxiv:0903.2059 wyithe j. s. b. & loeb , a. , 2004 , apj , 612 , 597 | identifying the properties of the first generation of seeds of massive black holes is key to understanding the merger history and growth of galaxies .
mergers between @xmath0 seed black holes generate gravitational waves in the @xmath1@xmath2hz band that lies between the sensitivity bands of existing ground - based detectors and the planned space - based gravitational wave detector , the laser interferometer space antenna ( lisa ) . however , there are proposals for more advanced detectors that will bridge this gap , including the third generation ground - based einstein telescope and the space - based detector decigo . in this paper
we demonstrate that such future detectors should be able to detect gravitational waves produced by the coalescence of the first generation of light seed black - hole binaries and provide information on the evolution of structure in that era .
these observations will be complementary to those that lisa will make of subsequent mergers between more massive black holes .
we compute the sensitivity of various future detectors to seed black - hole mergers , and use this to explore the number and properties of the events that each detector might see in three years of observation . for this calculation , we make use of galaxy merger trees and two different seed black hole mass distributions in order to construct the astrophysical population of events .
we also consider the accuracy with which networks of future ground - based detectors will be able to measure the parameters of seed black hole mergers , in particular the luminosity distance to the source .
we show that distance precisions of @xmath3 are achievable , which should be sufficient for us to say with confidence that the sources are at high redshift . |
one of the most striking differences between crystalline and amorphous solids is the difficulty of defining analogs of lattice defects such as vacancies , interstitials and dislocations in the latter @xcite .
the lack of a unique reference configuration in an amorphous solid precludes the use of such geometrical devices such as ideal lattice sites or burgers circuits .
recent studies of the mechanical deformation of glasses do offer a clue , however @xcite .
it is now accepted that one may identify regions in the amorphous solid which undergo _ non - affine _
distortions @xcite , i.e. displacements that can not be represented as a uniform strain or rotation acting on a reference volume , @xmath0 , in the presence of external stresses .
the dynamics of such regions has been used to describe irreversible plastic deformation of amorphous solids , in ways roughly analogous to that of dislocations in crystals @xcite .
continuing along this line of thinking , one may ask whether configurations involving topological defects in crystalline solids may be singled out using only measures of the local non - affine displacements @xcite of atoms away from their ideal reference positions , _ without _ explicit reference to a burgers circuit ? in ref .
@xcite an affirmative answer to this question was given for the case of dislocation dipoles in the ideal two - dimensional ( 2d ) triangular lattice .
it was shown @xmath1 that specific localised vibrational modes contribute to non - affine displacements in crystals , @xmath2 the largest contribution to thermally excited non - affine fluctuations arises from particle displacements that tend to introduce a dislocation dipole , i.e. localised _ defect precursor _ modes and @xmath3 this contribution is separated by a large gap from the next largest contribution , which represents other , more complicated deformations of @xmath0 . here
, we continue the program initiated in @xcite for a lattice structure that is the geometric dual of the triangular lattice , namely , the planar honeycomb ( phc ) crystal .
while emphasising that our work is not necessarily specific to any particular material , we must mention that there are , indeed , compelling reasons to study the phc lattice .
this decade s two most ground - breaking materials , with tremendous technological applications are graphene @xcite and hexagonal boron nitride ( h - bn ) films @xcite .
they are known to be flat mono - layers of atoms arranged in the form of a two - dimensional ( 2d ) honeycomb lattice .
apart from these , the phc lattice appears in many other atomic and macro - molecular systems such as confined water @xcite , assemblies of patchy colloids @xcite and nano - particles @xcite .
mechanical properties of lattice defects in these materials play a major role in deciding their applicability in various facets of the industry .
defects not only determine strength but also influence the topological characteristics of 2d sheets by coupling to the local curvature , leading to rolling up of 2d sheets into tubes and spheres or the creation of ripples on the surface @xcite .
obtaining greater insight into the physics of such defects in the phc lattice is therefore crucial to our understanding of these features exhibited by graphene and similar materials . as an example , consider the stone - wales @xcite ( sw ) topological defect observed in @xmath4-bonded carbon materials as well as in boron nitride nanotubes and nanosheets @xcite .
a sw defect is formed due to an in - plane rotation of a carbon - carbon bond by 90with respect to the bond center , resulting in bond reconstruction and eventual transformation of four hexagons into two heptagons and two pentagons ( see fig .
[ fig : sw_transformation ] ) .
the sw defects play a significant role in the formation of various carbon nanostructures , through local amorphization of graphene sheets @xcite .
they are also known to influence the electronic @xcite as well as transport properties @xcite of graphene , which can be exploited in the development of carbon - based electronic devices . in this paper
we focus on lattice deformations of the ideal phc crystal with the specific objective of identifying non - affine localised modes which act as precursors to the formation of topological defects
. a description of lattice defects in terms of local non - affine displacements may be useful for many reasons .
firstly , there is some evidence that non - affine displacements are associated with defect nucleation @xcite and may act as a predictor for its appearance .
secondly , this is a natural choice for describing processes such as amorphization where a solid breaks down by the proliferation of lattice defects @xcite .
lastly , this may prove to be a common language for describing mechanical deformation in both crystalline and amorphous matter .
we begin by generalising the coarse - graining procedure introduced in @xcite to derive exact statistics of the non - affinity parameter @xmath5 quantifying non - affine displacement fluctuations in the phc lattice . examining the nature of the localised vibrational modes responsible for producing @xmath5 , we discover that one of the largest contributions to @xmath5 arises from a mode that appears as a precursor to the sw topological defect .
as expected , non affine fluctuations in general and the sw precursor mode in particular increase in amplitude as the solid is made softer .
we also show that in the same limit , correlations between sw precursor fluctuations become stronger : the correlation length increases and so does the lifetime of these fluctuations . spatial correlations between sw precursor modes are also strongly anisotropic and therefore encourage clustering , producing cascades of sw distortions of the kind described in @xcite .
our results may be explicitly confirmed by direct use of video microscopy @xcite on patchy colloids .
these form phc lattices , and the interaction strength may be varied using surface modifications @xcite .
some of our results may also have implications for the mechanical behaviour and amorphization of graphene as discussed later .
the rest of the paper is organised as follows . in the next section
we describe in detail the process by which we identify non - affine fluctuations of a coarse graining region in the phc lattice . in order to obtain explicit and analytic results and to make our calculations generic
, we use a simple parametrisation of the particle interactions in terms of harmonic nearest neighbour ( nn ) and next near neighbour ( nnn ) harmonic bonds .
this is motivated by the fact that for small fluctuations any realistic interaction may be mapped onto a harmonic solid . in section 3
, we describe our results for the single point and two point statistics of defect precursors .
we end the paper with a summary of our main results and conclusions .
at any non - zero temperature @xmath6 , atomic displacements fluctuate from their reference , ideal lattice positions with amplitudes determined by their interactions . given any sub - volume @xmath0 within the solid , these fluctuations may be either _ affine _ or _ non - affine_. the former case represents fluctuations that simply deform the entire sub - volume uniformly . in two dimensions ,
there are only four such affine transformations possible : a uniform change in volume , a uniaxial strain , a shear or a uniform rotation .
if the number of particles within the sub - volume is larger than three , then there is a possibility of deformation modes which are non - affine , i.e. can not be described by a linear combination of the four simple affine transformations of @xmath0 .
the statistics of these non - affine displacements have previously been obtained for the one dimensional chain and the two dimensional triangular lattice @xcite .
it has also been shown that dominant non - affine fluctuations in the triangular lattice correspond to specific atomic displacements that tend to produce lattice defects @xcite .
we now use the ideas introduced in @xcite and @xcite to obtain the spatio - temporal statistics of non - affine fluctuations in the planar honeycomb ( phc ) structure . of the phc lattice , marked in green .
the central basis atoms ( orange , labelled @xmath7 and @xmath8 ) possess 12 neighbors ( blue ) within this domain .
the nearest neighbor bonds are indicated by bold lines while the next nearest neighbor bonds for one of the central basis particles are shown by dotted lines .
( b ) phonon dispersion curves @xmath9 for @xmath10 and @xmath11 , respectively .
the inset defines the high symmetry directions @xmath12 , @xmath13 and @xmath14 in the brillouin zone .
note that _ all _ the modes soften as @xmath15 decreases . as @xmath16 , the shear modulus vanishes and the transverse acoustic branch becomes unstable ( @xmath17 ) at small @xmath18 .
, width=491 ] [ fig : coarse_graining_volume ] we start by defining a simple set of interactions that can stabilise a phc lattice . these can not be standard central force potentials as these usually lead to closed packed solids such as the triangular lattice in two dimensions @xcite .
open lattices need strongly orientation dependent potentials . in atomic and molecular systems
this is due to covalent or hydrogen bonds .
for example , modelling of properties of carbon based materials is generally performed using bond - order , short - ranged empirical potentials such as the brenner @xcite and tersoff @xcite potentials .
similar potentials are also used to model water @xcite . in colloidal solids ,
one requires patchy interactions , i.e. heterogeneities on the surface of the colloid particles that produce angle dependent forces @xcite . here , in order to obtain results that are sufficiently generic , we use a phc network of particles connected by harmonic bonds , noting that _ all _ interactions are harmonic in the limit of small fluctuations .
accordingly , we assume the following general hamiltonian , @xmath19 where @xmath20 and @xmath21 are the momentum and displacement ( from the ideal reference ) of particle @xmath22 , @xmath23 is the particle mass and @xmath24 is the stiffness of the harmonic bond between particles @xmath22 and @xmath25 . it is well known that a phc lattice with only nn interactions is unstable @xcite . to make the lattice mechanically stable , we need to add either nnn bonds between the particles or include an energy for bond bending . in this work , we choose the former path although we have checked that the two methods give qualitatively the same results .
we therefore set @xmath26 if @xmath27 form a nn pair and @xmath28 if they are next nearest neighbours .
all other @xmath24 are set to zero .
if we call @xmath29 the lattice parameter , i.e. the length of a nn bond in the reference configuration , then @xmath29 , @xmath30 and @xmath31 set the units for length , energy and time .
we therefore set them to unity in the following .
for the coarse - graining analysis it will be useful to have the interaction energy written in a more generic form that makes it easy to switch to a representation in terms of fourier modes .
looking at fig .
[ fig : coarse_graining_volume](a ) , which shows a portion of the phc lattice , the structure may be constructed by inserting a basis of two atoms separated by a distance @xmath29 on a triangular primitive lattice . from now on we therefore label each particle by the pair @xmath32 , where @xmath22 refers to the specific basis or unit cell , and @xmath33 identifies the particle within the two - atom basis for the lattice .
we call the particle displacements @xmath34 and their cartesian components @xmath35 with @xmath36 . the equilibrium reference positions of the lattice points in @xmath0 , are denoted by 2d constant vectors @xmath37 , where the index @xmath38 refers to the basis , @xmath39 being the index of the central pair , and @xmath40 refers to individual atoms in the basis . the instantaneous positions are similarly denoted by the vectors @xmath41 .
indicating particle displacements as , @xmath42 @xmath43 we define @xmath44 as the @xmath45 column matrix @xmath46 $ ] . in terms of these variables ,
the configurational part of the hamiltonian takes the generic form @xmath47 here @xmath48 is the overall , system - wide dynamical matrix .
now define the fourier transform of the particle displacements , @xmath49 such that the real - space displacements are @xmath50 .
here and below the @xmath18-integral runs over the brillouin zone , and the normalising factor @xmath51 is its volume .
the dynamical matrix in @xmath18-space , @xmath52 , for the phc lattice with nn and nnn interactions then has the following structure : @xmath53 with @xmath54 the row and column entries are arranged in the order @xmath55 , @xmath56 , @xmath57 , @xmath58 here .
the eigenvalues @xmath59 of the above dynamical matrix give the phonon dispersion curves for each of the @xmath60 phonon branches .
the dispersion curves for @xmath10 and @xmath61 are plotted in fig .
[ fig : coarse_graining_volume](b ) . as expected
there are two acoustic branches where @xmath62 as @xmath63 and two optical branches .
as @xmath16 the lattice becomes unstable @xcite to shear fluctuations and @xmath9 for transverse acoustic phonons is imaginary at small @xmath18 . from the dynamical matrix one
can now determine the statistics of all displacement fluctuations .
the variances of the displacements in fourier space are given by @xmath64 where the angular brackets indicate a thermal average and @xmath65 is the eigenvector of @xmath66 corresponding to the @xmath67-th phonon branch . here and in similar formulas below we omit an overall factor of @xmath68 that simply scales all displacement variances . utilizing the above fourier space variances one can then express the covariances of the real - space displacements as @xmath69 we now move on to the definition of coarse - grained elastic strain and non - affine displacements .
the green shading in fig . [ fig : coarse_graining_volume](a ) identifies the coarse - graining region @xmath0 we will use .
it consists of a central basis ( shown in orange ) surrounded by a nearest neighbour shell of @xmath70 atoms ( shown in blue ) .
note that @xmath0 is the smallest non - trivial neighbourhood consistent with phc symmetry that includes all the interactions necessary for obtaining a stable solid . to define non - affine displacements
, we consider relative displacements with respect to either of the two atoms in the central basis . labelling the central basis as @xmath71 and the others @xmath72
, we denote these relative displacements by @xmath73 = @xmath74 .
the non - affinity then measures @xcite how well these relative displacements are approximated by the best - fit affine deformation : @xmath75 ^ 2 \bigg ) \
\end{split } \label{eq : falk_langer}\ ] ] where the minimisation is over all four elements of the two - dimensional deformation @xmath76 .
we include in the sum over triples @xmath77 all relative displacements across nn and nnn bonds involving either one of the particles in the central pair .
the choice of nn and nnn bonds is made as these bonds carry the physical interactions .
each atom in the central pair has @xmath78 such bonds , so the total number of terms in the sum in ( [ eq : falk_langer ] ) is @xmath79 , the @xmath80 arising from the double counting of the central nn bond @xmath81 .
expression ( [ eq : falk_langer ] ) , can be concisely written @xcite as @xmath82 where @xmath83 is the column vector @xmath84 formed out of the elements of the deformation tensor , @xmath85 is a column vector collecting the relevant @xmath86 , and @xmath87 is a @xmath88 \times ( 2\times 2 ) = 34 \times 4 $ ] matrix with entries @xmath89 . on carrying out the minimization in ( [ eq : min_e ] ) one obtains @xmath90 with @xmath91 .
the corresponding minimum value of the non - affinity parameter is given by @xmath92 here @xmath93 turns out to be a projection matrix that projects onto those directions of @xmath94 that can not be modelled by an affine strain .
one easily checks that @xmath95 is idempotent and hence has eigenvalues of @xmath96 and @xmath7 , as required for a projection . .
note that the eigenvalues spectrum bunches together into band - like structures .
the curves corresponding to the largest @xmath97 eigenvalues are labelled by numbers . on the right
, we show the eigen - displacements ( red points and lines ) corresponding to these eigenvalues .
the reference lattice positions are shown in blue .
a stone - wales defect will result from the precursor mode @xmath8 if anharmonic interactions ( neglected in this work ) makes the nnn bonds a - b and c - d strong while breaking the nn bonds a - d and b - c as a consequence of this deformation.,width=491 ] we now proceed to obtain the statistics of @xmath5 . for the particle displacements in @xmath0 , the covariances @xmath98 are given by @xmath99\,\frac{1 } { \omega^2_s({\bf q})}\end{aligned}\ ] ] this hermitian matrix with these elements , denoted @xmath100 below , determines the effective interactions between the coarse grained variables @xmath94 such that ensemble averages of any observable @xmath101 defined within @xmath0 are given by @xmath102 in the next section , we shall use this formula to obtain coarse grained statistics of the non - affine displacements and of the non - affinity parameter @xmath5 .
we note two technical details : in the matrix @xmath100 we use the same convention as already in @xmath95 , @xmath103 , @xmath87 and @xmath85 , i.e. we include only the triples @xmath77 that are used in the definition ( [ eq : falk_langer ] ) .
this makes the matrix @xmath100 of size 34 @xmath104 34 . on the other hand , with @xmath105 particles in @xmath0
there are only @xmath106 independent relative displacement components , so @xmath100 must have 8 zero eigenvalues . in expressions involving @xmath107 , the inverse should then always be understood as regularised appropriately , e.g. by the addition of a small constant along the diagonal of @xmath100 . and @xmath108 for @xmath10 ( blue ) and @xmath11 ( red ) .
predictions from exact calculation ( line ) are compared with results obtained from md simulations ( points ) of a @xmath109 phc lattice .
the system was allowed to equilibrate for @xmath110 md steps with a time step of @xmath111 , after which configurations were collected over a maximum of @xmath112 md steps at intervals of @xmath113 time steps.,width=377 ]
the non - trivial eigenvalues @xmath114 of @xmath115 correspond to _ non - affine _ displacements .
accordingly we have from ( [ eq : chi ] ) for the mean non - affinity @xmath116 and for its variance , @xmath117 to obtain the statistics of @xmath5 , we need to evaluate the brillouin zone integrals that define @xmath100 numerically .
we use a @xmath118 point gauss - legendre quadrature routine to perform all such numerical integrations over the brillouin zone in this paper .
the computed @xmath115 is a @xmath119 matrix with @xmath120 eigenvalues . as explained above this
has 8 eigenvalues that are automatically zero , and a further 4 that represent the four affine deformations in 2d , namely volumetric , axial , shear and rotation ; recall that @xmath95 projects into the space orthogonal to these deformations .
the remaining 22 non - trivial eigenvalues relate to distinct non - affine modes and are plotted in fig .
[ fig : eigenvalue_k2_k1 ] as a function of @xmath15 .
all our numerical results are given for a dimensionless temperature of @xmath121 , where a harmonic approximation should be reliable provided that anharmonic interaction coefficients are of order unity in our dimensionless units .
as @xmath122 , most of the eigenvalues diverge as @xmath123 .
this behavior is expected since the @xmath18-integral defining @xmath100 involves the inverse of @xmath124 , which vanishes for the soft phonon branch as @xmath125 .
remarkably , the eigenvalue spectrum contains large _ gaps _ with the three largest eigenvalues being separated from the others by almost an order of magnitude unless @xmath15 is close to unity .
the eigenvectors corresponding to these three eigenvalues are also shown in fig .
[ fig : eigenvalue_k2_k1 ] . out of these
, two deformations correspond to breathing modes , where the central basis is shifted relative to the others in @xmath0 in the horizontal and vertical directions .
the eigenmode corresponding to @xmath126 is interesting : it represents a rotation of the central basis relative to the reference positions .
a rotation of @xmath127 together with a reconnection of the bonds would nucleate a sw defect .
this is , then , one of the non - affine precursor fluctuations that must precede nucleation of sw defects .
our calculations show that such precursor fluctuations are the most prominent ( and _ softest _ ) non - affine distortions that tend to change the lattice topology . and @xmath128 for @xmath129 and @xmath11 .
note that while @xmath130 is isotropic , @xmath128 shows prominent quadrupolar anisotropy .
both correlations increase with decreasing @xmath15.,width=510 ] to obtain the probability distribution of @xmath5 , @xmath131 , we begin with the characteristic function @xcite , @xmath132 here , @xmath133 is the variable conjugate to @xmath5 . by carrying out the gaussian integrals
, we obtain @xmath134 the @xmath115 matrix above is hermitian and can be diagonalized by a unitary transformation .
hence , writing the determinant in terms of products of its eigenvalues we can also write @xmath135 a numerical inverse fourier transform of @xmath136 then gives @xmath131 . at a more intuitive level , one observes from the form of ( [ eq : phi ] ) that @xmath131 is the distribution of the sum of the squares of uncorrelated gaussian random variables with variances @xmath137 . to determine the probability distribution of the sw mode we proceed similarly .
let @xmath138 be the projection of any @xmath94 in @xmath0 along the sw eigenmode , i.e. @xmath139 where @xmath140 is the eigenvector of @xmath115 corresponding to the sw eigenvalue .
defining the corresponding characteristic function with @xmath141 being the variable conjugate to @xmath138 , @xmath142 on performing completion of squares of the argument of the numerator of the above expression , we get , @xmath143 finally , on carrying out the above gaussian integrals followed by inverse fourier transforming the result , yields , as @xmath144 has a zero mean gaussian distribution with covariance matrix @xmath100 , @xmath138 is also gaussian with distribution @xmath145 the probability distributions for @xmath5 and @xmath138 are shown in fig . [ fig : projection](a ) and ( b ) , where we have compared the analytical results , obtained by fourier transforming the corresponding characteristic functions , with data obtained from molecular dynamics ( md ) simulations using the parallelized md package lammps @xcite .
note that both @xmath146 and @xmath147 increases as @xmath15 .
it is straightforward to show that @xmath148 , i.e. @xmath140 belongs to the non - trivial eigenspace of @xmath95 , implying immediately that @xmath149 .
the leading behaviour of @xmath146 as @xmath16 is the same . and
( b ) @xmath150 for @xmath10 ( blue ) and @xmath11 ( red ) .
points are from simulations of a @xmath151 phc lattice while the lines are from our exact results .
the simulation data was obtained from @xmath152 md configurations and further averaged over shifts of the origin of the time axis.,width=377 ] in order to calculate the spatio - temporal correlations of the non - affinity @xmath5 and the sw projection @xmath138 , we need to consider simultaneously displacement differences in two neighborhoods @xmath0 and @xmath153 centered at @xmath154 and * @xmath155 * , at times @xmath156 and @xmath157 respectively.the non - affinity and sw projection are defined as @xmath158 and @xmath159 for the coarse graining volume @xmath0 , with particle displacements within this region at a time @xmath156 being represented by the vector @xmath160.for the region @xmath161 at a time @xmath157 , we correspondingly have the quantities @xmath162 and @xmath163 . the displacement correlations in this case are defined as @xmath164 where the matrix @xmath165 has elements given by the expression @xmath166\frac{\cos[\omega_s({\bf q})(t'-t)]}{\omega^2_s({\bf q})}\end{aligned}\ ] ] the correlation between @xmath167 and @xmath168 can be calculated using wick s theorem as @xcite , @xmath169 where the @xmath170 denote the eigenvalues of the matrix @xmath171 .
similarly , the correlation of @xmath172 and @xmath173 is given by , @xmath174 in a homogeneous solid in equilibrium , these correlation functions are functions only of the relative coordinates @xmath175 and times @xmath176 . from now on we will simply denote these by @xmath177 and @xmath156 respectively .
we first focus on the spatial correlations at equal time .
these are depicted in fig .
[ rho - s ] , where we plot @xmath178 and @xmath179 for two values of @xmath15 .
the @xmath5 correlation are short ranged , isotropic functions qualitatively similar to those for the triangular lattice @xcite and increase in magnitude as @xmath15 decreases .
the sw precursor correlations are also short ranged but show prominent quadrupolar anisotropy .
the range and the anisotropy increases grows as the phc lattice softens with decreasing @xmath15 .
non - affine distortions are localised in space being composed of many incommensurate lattice distortions .
the sw precursor mode , can condense into sw defects when an - harmonic terms in the interactions become prominent in the small @xmath15 regime . in fig .
[ fig : rho_time_correlation ] we turn to the local correlations as a function of time .
the figure shows the normalized correlations , @xmath180 and @xmath181 .
our results are compared with those obtained from molecular dynamics simulations of a @xmath182 site harmonic honeycomb net .
as is evident from the figure , the correlations decay with time showing that they are transient .
the decay occurs in an oscillatory fashion due to the presence of a large number of mutually incommensurate frequencies .
the mean frequency of these oscillations decreases as @xmath16 , indicating larger relaxation times of the induced deformation .
calculated at different times @xmath183 and @xmath184 , for the harmonic honeycomb lattice with @xmath185 .
note the development of strong spatial anisotropy with time.,width=491 ] finally , to demonstrate the behavior of the correlations over both space and time , we plot @xmath186 in fig.[rho - s - t ] , for @xmath185 , at four specific values of @xmath156 .
interestingly , the spatial anisotropy of the correlations increases with time even as their amplitude decays .
the behaviour of @xmath187 is similar to that observed in the triangular lattice @xcite .
in this paper we have carried out a detailed study of the displacement fluctuations of a coarse - graining volume , @xmath0 , within the phc crystal . at non - zero temperatures , thermal fluctuations
produce random affine local deformations ( such as scaling , shear or rotation ) , but also distortions of @xmath0 that are impossible to represent as an affine transformation .
we have obtained the full single point and two - point statistics of these non - affine fluctuations in the harmonic limit .
the approach to the limit where the lattice is unstable causes a divergence of the mean non - affinity , and of the spatio - temporal scales of the correlation functions .
we have shown that the vibrational modes that contribute to non - affine fluctuations are given by the eigenvectors of the matrix @xmath115 .
this is constructed from displacement correlations @xmath100 and the operator @xmath95 that projects all atomic displacements onto the non - affine subspace .
the eigenvalues of @xmath115 are seen to group together into three well defined clusters with large gaps between them .
as the lattice softens , most eigenvalues diverge but the clusters remain distinct . we have shown that one of the three largest eigenvalues involves a rotation of the basis unit relative to the rest of the atoms in @xmath0 and therefore would tend to nucleate a sw defect in a system with a more realistic interaction potential .
we have obtained the statistics and correlations of these localised sw precursor modes in the phc lattice .
most work on lattice defects in the phc crystal have centered on graphene for obvious reasons and have involved mainly the calculation of defect formation energies and energy barriers at zero temperature . within our model
, we estimate that a value of @xmath188 reproduces the phonon spectrum of graphene reasonably well . comparing the velocities for acoustic phonons for @xmath61 with the actual values in graphene
, we also estimate that our dimensionless temperature corresponds to a real temperature of about @xmath189 .
our predictions for the spatial and temporal correlations for @xmath5 and the sw precursors should be accurate for these parameters .
sw defects in graphene typically have a very high activation energy ( @xmath190 ev ) as determined by ab - initio calculations @xcite .
this can be lowered by extrinsic mechanisms such as hydrogen adsorption , but only to about 2.54 ev @xcite .
this would make sw defects unlikely in pristine graphene at temperatures which are relevant .
our calculations suggest that this picture may need to be modified .
entropic contributions to defect stability , arising especially from lattice vibrations that have , to the best of our knowledge , received relatively less attention , appear to be at least equally important .
this is not particularly surprising , in retrospect , because in 2d the loosely packed and open phc lattice is expected to be strongly influenced by thermal fluctuations @xcite .
our calculations also show that sw defects are highly correlated in space and time due to elastic interactions . interpreting the space - time correlation functions @xmath191 as a linear response function @xcite
, one can argue that an isolated sw defect at any point @xmath177 would induce other sw defects at neighbouring lattice points .
the spatial distribution of these induced sw defects would be anisotropic and follow @xmath186 as shown in fig .
[ rho - s ] . similarly , fig .
[ rho - s - t ] , shows that the anisotropy in the sw precursor correlations increases with time .
this may be interpreted as implying that sw defects are nucleated in cascades @xcite , which enables these defects to proliferate during amorphization @xcite at which point the harmonic theory breaks down .
how does external stress @xmath192 influence @xmath5 and the sw precursors ? to answer this question one only needs to include @xcite the term @xmath193 in the hamiltonian ( [ harm ] ) , where @xmath194 is the best - fit _ local _ strain , @xmath195 , at particle @xmath22 . to lowest order in @xmath192 , the variation of @xmath5 is then @xmath196 , where the dependence on @xmath22 enters in @xmath165 via the relative displacement between the location where stress is applied , @xmath197 , and the location where we are measuring @xmath5 , say @xmath198 .
the probability @xmath108 remains gaussian with the same variance but now the @xmath199 symmetry of @xmath138 is broken and @xmath108 is shifted with a mean @xmath200 .
stress , @xmath192 , also does not affect the space - time correlation functions of @xmath138 .
the density of sw defects will be modified , though , by curvature terms @xcite arising from out of plane fluctuations that have not been included in our calculation .
curvature then acts as an external field coupling to @xmath5 and @xmath138 .
the effect of external fields on @xmath5 has been studied in the context of the triangular lattice @xcite and should lead to similar results here . since microscopic and instantaneous atomic coordinates are indeed difficult to obtain for systems such as graphene and h - bn , we believe that our predictions may be tested in future using video microscopic data @xcite on patchy colloids @xcite that form a phc lattice .
similar experiments have been successfully carried out in simpler colloids @xcite so it would be interesting to see results for the more complex phc structure . in future
we would like to investigate the three dimensional fcc lattice where a similar association of non - affine displacements with defects has been observed @xcite . in three dimensional close packed solids ,
the elementary topological defect is a small dislocation loop which may be difficult to identify using a burgers circuit for a fluctuating crystal at @xmath201 .
however , local non - affinity should remain an easily recognised marker .
analysis of non - affine modes in progressively disordered solids should also elucidate the connection , if any , between topological defects in crystals and non - affine droplet fluctuations or shear transformation zones in amorphous matter @xcite .
am thanks tifr for a junior research fellowship and sg thanks csir , india for a senior research fellowship .
useful discussions with t. saha - dasgupta are acknowledged .
r. w. baluffi , _ introduction to the elasticity theory for crystal defects _ , ( cambridge university press , cambridge , 2012 ) .
p. chaikin and t. lubensky , _ principles of condensed matter physics _ , ( cambridge press , cambridge , 1995 )
. m. l. falk and j. s. langer , annu . rev . condens .
matter phys .
* 2 * , 353 ( 2010).a . s. argon .
acta . met .
, * 27 * , 47 , ( 1979 ) .
m. l. falk , j. s. langer , phys .
e , * 57 * , 7192 ( 1998 ) . c. e. maloney and a. lema " itre , , 195501 ( 2004 ) .
s. s. schoenholz , a. j. liu , r. a. riggleman and j. rottler , phys .
x * 4 * , 031014 ( 2014 ) .
s. ganguly , s. sengupta , p. sollich and m. rao , phys .
e * 87 * , 042801 ( 2013 ) .
s. ganguly , s. sengupta , p. sollich arxiv:1502.00042 j. h. warner , f. schffel , a. bachmatiuk , m. h. rmmeli , eds .
_ graphene : fundamentals and emergent applications _
( elsevier , amsterdam , 2013 ) j. j. pouch and s.a .
alterovitz eds .
_ synthesis and properties of boron nitride _ ( trans tech , brookfield , 1990 )
koga , k. zeng , x. c. tanaka , h. , 5262 , ( 1997 ) f. romano , f. sciortino , nat . mater . * 10 * , 171 ( 2011 ) c. mondal , a. h. khan , b. das , s. acharya and s. sengupta , sci . rep . * 3 * , 2612 , ( 2013 ) .
m. bowick and l. giomi , adv .
* 58 * , 449 ( 2009 ) .
a. stone and d. wales , chem .
phys . lett . *
128 * , 501 ( 1986 ) .
y. miyamoto _
et al . _ ,
phys . rev .
b * 69 * , 121413 ( 2004 ) .
h. bettinger , t. dumitric , g. scuseria , and b. yakobson , phys .
b * 65 * , 041406 ( 2002 ) .
j. kotakoski , a. krasheninnikov , u. kaiser , and j. meyer , , 105505 ( 2011 ) .
j. carlsson and m. scheffler , phys .
rev . lett . * 96 * , 046806 ( 2006 ) .
s. ihnatsenka and g. kirczenow , phys .
b * 88 * , 125430 ( 2013 ) .
t. das , s. ganguly , s. sengupta , m. rao , arxiv:1312.3746 o. ori , f. cataldo , and m. v. putz , int . j. mol .
* 12 * , 7934 ( 2011 ) .
k. zahn , a. wille , g. maret , s. sengupta , and p. nielaba , phys .
lett . * 90 * , 155506 ( 2003 ) .
s. glotzer and m. solomon , nat .
* 6 * , 557 , ( 2007 ) .
d. w. brenner , phys .
b * 42 * , 9458 ( 1990 ) .
j. tersoff , phys .
b * 37 * , 6991 ( 1988 ) .
j. cserti and g. tichy , eur .
* 25 * , 723 ( 2004 ) .
plimpton , s. j. comp .
* 117 * , 1 ( 1995 ) .
s. letardi , m. celino , f. cleri , and v. rosato , surf .
sc . * 496 * , 33 ( 2002 ) .
m. kabir , s. mukherjee , and t. saha - dasgupta , phys .
b * 84 * , 205404 ( 2011 ) .
y. yamayose , y. kinoshita , y. doi , a. nakatani and t. kitamura , eur .
80 * , 40008 , ( 2007 ) .
r. zwanzig , _ nonequilibrium statistical mechanics _
( oxford university press , oxford , 2001 ) k. franzrahe , p. keim , g. maret , p. nielaba , s. sengupta , phys .
e , * 78 * , 026106 ( 2008 ) .
p. keim , g. maret , u. herz , and h. h. von grnberg , phys .
lett . * 92 * , 215504 ( 2004 ) . | certain localised displacement fluctuations in the planar honeycomb lattice may be identified as precursors to topological defects .
we show that these fluctuations are among the most pronounced _ non - affine _ distortions of an elemental coarse graining volume of the honeycomb structure at non zero temperatures .
we obtain the statistics of these precursor modes in the canonical ensemble , evaluating exactly their single point and two - point spatio - temporal distributions , for a lattice with harmonic nearest neighbour and next near neighbour bonds . as the solid
is destabilised by tuning interactions , the precursor fluctuations diverge and correlations become long - lived and long - ranged . |
quantum information processing based on superconducting qubits has made considerable progress in recent years ( see for example the reviews @xcite ) .
several platforms have been realised and the first quantum protocols have been already implemented experimentally @xcite . in this respect
superconducting nano - circuits are by now considered among the most promising implementations of solid - state quantum processors . over the years
the judicious choice of new designs and the improvement in materials and fabrication has increased the coherence properties of superconducting nano circuits by several orders of magnitude . however , beyond standard relaxation and decoherence studies , the dissipation and heat transport properties of these systems are only little investigated and poorly understood , albeit important from the practical point of view .
the newly born field of quantum thermodynamics @xcite has highlighted that quantum coherence and quantum correlations can play prominent role in the design of efficient thermal processes .
several theoretical proposals have already appeared discussing heat engines based on the dynamics of nano - circuits and qubits @xcite , some with a direct implementation with superconductors @xcite . on a related topic ,
quantum limited heat transport by phonons , photons , and electrons has been investigated in solid state circuits both theoretically @xcite and experimentally @xcite .
this large body of activity quests for the research on new designs with enhanced functionalities relying on the quantum properties of the constituent circuits . here
, we present a new device of this family .
our focus is to introduce the coupling between a pair of qubits as elements to control heat transport in quantum regime .
we present a quantum heat switch connected to two reservoirs . using a second order quantum master equation
, we quantitatively obtain the expression for power mediated by this two - qubit system . as a result of the coherent coupling , when power is enhanced by up to one order of magnitude the bandwidth at maximum power
is boosted as compared to the case in which the two qubits are independent . and @xmath1 operated by control parameter @xmath2 .
( b ) a quantum heat switch discussed in this work .
the principle of a basic heat switch is shown in fig . [ fig1]a . here ,
when the switch is `` on '' , heat flows from hot bath to the cold one . in quantum circuits a heat switch can be realized by coupling the two baths by an intermediate element with tunable parameters .
for instance this element can act as an energy filter thanks to its level structure .
a basic element in this respect is either a classical or quantum @xmath0 circuit admitting heat only at or around its resonance frequency @xcite . in steady state with unequal temperatures of the two baths ( cold @xmath3 and hot @xmath4 )
a qubit can act as such a switch @xcite .
we show that with our suggested design ( fig .
[ fig1]b ) composed of two coupled qubits , this operation can be boosted . in fig .
[ fig1]b , the superconducting qubits in the middle are mutually coupled @xmath5 and each consists of a loop interrupted by josephson junctions .
they are further coupled to two resonators with baths @xmath6 via mutual inductances @xmath7 for @xmath8 and each resonator is a series @xmath9 circuit .
resistors @xmath1 and @xmath10 , with temperatures @xmath11 and @xmath12 act as cold and hot reservoirs , respectively .
we write the total hamiltonian of the system with its baths shown in fig .
[ fig1]b as @xmath13 where @xmath14 are the hamiltonians of the two qubits , @xmath15 is the coupling between them , @xmath16 are the hamiltonians of the hot and cold reservoirs , and @xmath17 and @xmath18 the couplings of qubits 1 and 2 to the two reservoirs
. the various components of the hamiltonian of each qubit can be written as @xmath19 for @xmath20 with @xmath21 the overall energy scale of each qubit , @xmath22 the dimensionless energy splitting at @xmath23 , and @xmath24 the pauli matrices . the flux @xmath2 is applied to the qubits using control parameter @xmath25 . here
@xmath26 and @xmath27 is the superconducting flux quantum .
@xmath28 on the coupling parameter @xmath29 for different values of dimensionless flux @xmath30 , ( a ) @xmath31 , ( b ) @xmath32 . in both panels
] for the noise , we assume linear coupling as @xmath34 , and similarly for other couplings . here , @xmath35 is the coupling of the reservoir @xmath36 to each qubit , and @xmath37 is the noise current of the cold / hot environment .
the interaction hamiltonian @xmath15 between two qubits which is the energy corresponding to the standard inductive coupling of two current loops , is given by @xmath38 where @xmath39 are the current operators of the two qubits .
therefore , with @xmath40 , we have @xmath41 and the total hamiltonian of the two coupled qubits , @xmath42 , reads @xmath43 for quantitative analysis , we use the bell basis of maximally entangled states @xmath44 . considering the fully symmetric system , @xmath45 , and @xmath46 , and the relevant matrices in this basis ( details are provided in the appendix )
the hamiltonian is given by @xmath47 where @xmath48 .
the eigenenergies @xmath49 ( normalized by @xmath50 ) of the hamiltonian ( [ hamilton1 ] ) are given by @xmath51=0.\ ] ] out of the eigenenergies , @xmath52 represents a `` protected state '' as will become evident in what follows , and the rest of the energies @xmath53 form a three - state system .
figure [ fig2]a , b displays these eigenenergies as a function of @xmath29 for two different values of @xmath30 .
it is vivid that coupling influences all energy levels and splits the initially degenerate levels @xmath54 and @xmath55 .
introducing flux to the system ( varying @xmath30 ) , modifies the three level structure .
it is then straightforward to find the corresponding normalized eigenstates by solving a set of three linear equations , yielding @xmath56 and @xmath57^{\rm t}.$ ] the transition rates between eigenstates @xmath58 and @xmath59 having an energy spacing @xmath60 of the coupled qubits are sensitive to current noise produced by two baths at the frequency @xmath61 and can be calculated from fermi s golden rule .
these rates are given by @xmath62 where @xmath63 , @xmath64 , @xmath65 and @xmath66^{-1 } \frac{2\hbar\omega_{ij}}{r_{\rm b}(1-e^{-\hbar\omega_{ij}/k_bt_{\rm b}})}$ ] is the unsymmetrized noise spectrum . here ,
@xmath67 and @xmath68 are the bare resonance angular frequency and the quality factor of each @xmath0-circuit .
now we indeed see that @xmath69 is a protected in this set - up state since @xmath70 , and hence @xmath71 for all @xmath58 and @xmath72 .
( black lines ) and @xmath73 ( red lines ) as a function of drive for five different values of coupling .
dot - dashed line corresponds to dimensionless resonance frequency @xmath74 of the @xmath0 resonators .
( right ) flux @xmath30 dependence of the transmitted dimensionless power @xmath75 for different values of coupling .
the colors of the labels from left panels correspond to the related ones in the right panel .
the parameters @xmath76 , @xmath77 , @xmath78 , @xmath79 , and @xmath33 were used for all panels . ] in steady state , the master equation ( see appendix ) yields the populations @xmath80 of the levels as @xmath81 where @xmath82 is the total transition rate @xmath83 due to both the baths from eigenstate @xmath58 to @xmath59 .
the indices assume values @xmath84 , and their cyclic permutations .
equation ( [ diagrho ] ) applies , when the system is initialized in the subspace @xmath85 .
the expression of power to bath @xmath36 can be written in the form @xmath86 it is easy to show that @xmath87 , where @xmath88 and @xmath89 .
so , we can simplify the expression of power in a two - qubit system to @xmath90 ) between three energy levels 1 , 3 , and 4 as a function of @xmath30 .
the dashed lines illustrate transition rates in the case of decoupled qubits @xmath91 while the solid lines are for coupled qubits @xmath92 corresponding to the maximum power transfer .
the parameters are @xmath93 , @xmath94 , @xmath78 , @xmath79 , @xmath33 .
in left panels of fig . [ fig3]a , we plot the energy separation , @xmath95 ( black lines ) and @xmath73 ( red lines ) , for five different values of coupling @xmath29 as a function of flux bias @xmath30 .
dot - dashed line represents the dimensionless resonance frequency @xmath96 .
the dependence of transmitted power on flux @xmath30 for these values of coupling is shown in the right panel of fig .
[ fig3 ] . with no coupling ( @xmath97 ) the two level separations are equal : @xmath98 . by increasing coupling @xmath29 from @xmath99 to @xmath100 ,
the power initially decreases because of increasing difference between energy levels @xmath101 and @xmath102 . further increasing @xmath29 the power increases abruptly obtaining the maximum at @xmath103 .
this is because , here the @xmath73 separation meets the resonance frequency at @xmath31 . by further increasing coupling ,
@xmath73 will cross @xmath104 at two values of @xmath30 and leads to two peaks in power , but the maxima decrease because of larger @xmath95 .
it is clear that the @xmath30 dependence of @xmath73 becomes stronger when @xmath29 increases from 0 to 0.2 .
this is the feature that determines the bandwidth ( in @xmath30 ) of the filter .
the mechanism of increased power with @xmath29 is shown schematically in fig .
[ fig4 ] ( left ) : the total transition rates ( relaxation and excitation ) increase dramatically as shown in fig .
[ fig4 ] ( right ) . there the relevant transition rates between levels for decoupled and coupled @xmath92 qubits are depicted . numerical results of the power @xmath105 as a function of coupling at different bath temperatures are displayed in the left panel of fig .
here we have fixed the temperature of the cold bath and vary that of the hot one . based on this figure we see the obvious result : the higher the difference between the temperatures , the higher the power transferred to the cold bath .
the precise evolution of power as a function of coupling @xmath29 is a result of the interplay of the level structure of the coupled qubits and the parameters of the baths .
the quality factors do not influence the maximum of the transmitted power because the transition rates do not depend on @xmath106 at resonance . yet in our example , high @xmath106 decreases @xmath105 at @xmath97 strongly .
this is because under these conditions , the @xmath0-circuit is not in resonance with the qubits at any value of @xmath30 . as a function of coupling between the qubits for different bath temperatures .
the solid lines display maximum power with respect to @xmath30 for each value of coupling while the dashed lines depict the power at @xmath31 .
( right ) dependence of the transmitted dimensionless power @xmath107 on flux @xmath30 for three different values of coupling : @xmath97 ( red line ) , @xmath103 ( blue line ) , and @xmath108 ( black line ) .
insets : entanglement as functions of @xmath30 ( top ) and coupling ( bottom ) for the same temperatures as in the main frame .
the parameters for all the panels are @xmath94 , @xmath78 , @xmath79 , @xmath33 , @xmath93 ( brown lines ) , @xmath109 ( green lines ) , @xmath110 ( purple lines ) , and @xmath111 ( blue lines ) .
[ fig5 ] ] given the prominent role of the coherent coupling in determining the boost in the emitted power it is relevant to explore if / how quantum correlations are the underlying reason for this effect .
as we will see in the following , entanglement is indeed present and plays some role although it is not in a direct correspondence with the enhancement factor .
entanglement plays an important role in quantum information processing @xcite .
bipartite systems formed of two coupled qubits have attracted lots of attention in this respect @xcite . in order to assess the degree of entanglement of the coupled qubits in our set - up , we adopt the concurrence introduced by wootters @xcite as @xmath112 , where @xmath113 are the eigenvalues arranged in decreasing order of the non - hermitian matrix @xmath114 . here
@xmath115 and @xmath116 denotes the complex conjugation of @xmath117 and @xmath118 and @xmath119 are the relevant pauli matrices .
concurrence varies from @xmath99 for a completely unentangled state to @xmath101 for a maximally entangled one . the entanglement @xmath120 is given by @xmath121 , where @xmath122/2 $ ] .
the dependence of entanglement on flux and coupling is shown in the inset of the left panel of fig .
[ fig5 ] ( top and bottom ) , respectively , for different bath temperatures .
it is clear that at optimal value of coupling @xmath92 , the qubits are highly entangeled up to 30% .
finally it is fair to note that the enhancement of power when coupling the qubits is somewhat case - dependent . for the presented case of @xmath123 this effect
is prominent , since with increasing coupling one eventually meets the resonance condition discussed .
the numerics shows that in the opposite regime , @xmath124 , there is also weak enhancement , due to matching of @xmath125 condition at finite @xmath29 . yet , when the system is at the resonance @xmath126 already at @xmath97 , the transmitted maximum power decreases monotonically as a function of @xmath29 .
in order to assess the bandwidth of the filter @xmath127 , we calculate the curvature of the eigenenergies with respect to @xmath30 , i.e. @xmath128}^{-1}|_{q=0}$ ] . differentiating the expression of eigenenergies of the hamiltonian eq .
( [ ee1 ] ) twice with respect to @xmath30 and putting @xmath129 at @xmath31 , we have @xmath130 this yields for the width @xmath131}^{-1}.\ ] ] for @xmath97 , we have @xmath132}^{-1}|_{q=0}=\frac{\delta}{2}$ ] .
we want to compare this value to that at optimal coupling where @xmath133 at @xmath31 , which occurs according to eqs .
( [ enk0 ] ) at @xmath134 , i.e. at @xmath135 where @xmath136 is the dimensionless resonator frequency . in fig .
[ fig3 ] , @xmath137 , which yields @xmath138 for the optimum where the level separation @xmath73 meets the resonance frequency at @xmath31 ; this is consistent with our numerics .
inserting eq .
( [ kappaopt ] ) into eq .
( [ filter ] ) , we have @xmath139 the expression above allows us to see how much one can improve the bandwidth of filtering by coupling the two qubits . if we divide expression ( [ filter1 ] ) by the corresponding width for decoupled qubits in the example of fig .
3 , we obtain an improvement in the filter bandwidth as @xmath140 i.e. it is narrower by factor 6.2 . to make our argument more general , we compare the bandwidth at maximum power of decoupled and optimally coupled qubits . to start with , we assume a fixed value for @xmath104 . for decoupled qubits
the maximum power is obtained at the resonance condition @xmath141 . in this case
, we have for the decoupled qubits @xmath142}^{-1}|_{q=0 } = \omega_{lc}/4 $ ] .
the ratio @xmath143 of the bandwidth at maximum power for optimally coupled and fully decoupled qubits , is given by @xmath144 it is clear that one benefits of setting @xmath145 , and for @xmath146 we obtain the ultimate improvement of @xmath147 .
( in our numerical example in fig .
[ fig3 ] this improvement would be @xmath148 ) .
we separately show the transmitted power for decoupled @xmath91 and coupled @xmath149 and @xmath150 qubits as a function of flux bias @xmath30 in a smaller range in the right panel of fig .
[ fig5 ] . here based on numerical results for transmitted power ,
the ratio of the full width at half maximum ( fwhm ) of the peaks at coupled @xmath92 and decoupled @xmath91 cases equals 6.8 , which is consistent with our analytical result 6.2 given above for the bandwidth of the filter .
broader and weaker peak of single qubit in comparison with the sharper peak of coupled qubits indicates that coupled qubits provide a more selective heat switch .
we have introduced a design of a quantum heat switch based on two coupled qubits .
we present explicit results of the amount of heat transferred between the hot and the cold baths .
we have shown that at optimal value of coupling , when the energy level difference of the two excited states crosses the resonance frequency of the environment , the corresponding transition rates , and thus the transmitted power increase dramatically , in our example by an order of magnitude . as a result
, we achieve a four times narrower bandwidth at maximum power of the heat switch based on coupled qubits as compared to that of two independent ones .
in general , we write the master equation for the coupled qubit system in the absence of active time - dependent driving as @xmath151 where the rhs arises from coupling to the baths and can be written in the rotating wave approximation and for the case @xmath152 , and @xmath153 as @xmath154 , and we find the rates @xmath155 , yielding @xmath156 @xmath157 is the total transition rate from state @xmath158 to state @xmath59 . for the diagonal elements we then have the master equation in the form @xmath159
the power to the cold reservoir , @xmath160 , can be obtained in the standard way as @xmath161 , where @xmath162 $ ] is the operator of power to the coupled qubits from the cold bath , and @xmath163 .
we then find that @xmath164 we write with standard notations the power transmitted between the hot and cold baths in linear response as @xmath165\rangle,\ ] ] where the subscript @xmath166 refers to the interaction picture .
after a straightforward calculation we obtain @xmath167 by the identity given in the main text we can finally show that this expression is equivalent to @xmath168
the relevant matrices in bell basis are given by @xmath169 @xmath170 @xmath171 another regime , where we find explicit results for eigenenergies and eigenstates is that with finite @xmath29 but at @xmath31 .
we have then the solutions @xmath174 the corresponding eigenstates are @xmath175 | we present a quantum heat switch based on coupled superconducting qubits , connected to two @xmath0 resonators that are terminated by resistors providing two heat baths . to describe the system we use a standard second order master equation with respect to coupling to the baths .
we find that this system can act as an efficient heat switch controlled by the applied magnetic flux .
the flux influences the energy level separations of the system , and under some conditions , the finite coupling of the qubits enhances the transmitted power between the two baths , by an order of magnitude under realistic conditions . at the same time , the bandwidth at maximum power of the switch formed of the coupled qubits is narrowed . |
scientific consensus dictates that the x - ray spectra of black hole x - ray binaries ( xrbs ) and active galactic nuclei ( agns ) are modelled by three spectral components : a multicolour black body component representing the emission from the accretion disc in the soft x - rays , a power law with a cutoff or no cutoff representing the thermal / non - thermal comptonisation of soft photons from a population of hot electrons in the hard x - rays , and a reflection component representing the reprocessed emission of hard x - rays from the accretion disc , i.e. photoelectrically absorbed or compton - scattered emission , in the intermediate energies , that most prominently manifests itself in the form of an iron line .
however , the magnitude of these components remains unknown at a given time .
for example , the magnitude of the soft component in the hard state is usually difficult to determine .
however , knowledge of the accretion disc geometry in the hard state , in particular the location of the inner edge of the accretion disc , is critical for the properties and/or the validity of all accretion disc models .
it is especially important for radiatively inefficient models such as advection dominated accretion flow ( adaf , * ? ? ?
* ; * ? ? ?
the quiescent properties of black hole transients are often cited as evidence for adafs and black hole event horizons @xcite . however , for the adaf model to be relevant , the inner edge of the accretion disc must increase by orders of magnitude between the outburst and quiescent states .
therefore , it is important to determine the magnitude of the accretion disc and reflection components in the hard state a difficult task due to the low flux compared to the comptonisation component .
in fact , even the physical mechanism for the x - ray emission in the hard state is open to question , with possible contributions from direct synchrotron emission from the jet @xcite or synchrotron self - compton radiation from the base of the jet @xcite .
this kind of spectral degeneracy is especially true for the galactic xrbs and a striking example can be seen e.g. in @xcite , where three very different models are fitted equally well to the same data set of cyg x-1 , despite the excellent quality of the data that was obtained by all the x - ray satellites in orbit at the time .
in addition , the magnitudes of different components in the intermediate x - ray states are difficult to determine as all the components are luminous
. extra complications in the interpretation of x - ray spectra are caused by the attenuation of interstellar matter in galactic xrbs and the so - called warm absorbers ( possibly associated with the winds of the accretion discs ) in agns .
thus , modelling the x - ray spectra of accreting black holes often leads to a problem of degeneracy , i.e. multiple distinct models fit the observed data equally well .
even if an apparently good fit is obtained between the data and the model , it does not necessarily imply a match between theory and physical reality . in this paper
we demonstrate the use of linear unsupervised decomposition methods in separating a set of time series of x - ray spectra into subcomponents corresponding to distinct spectral components from the disc and the population of hot electrons . in section [ spectral_decomposition ]
we will compare three different decomposition methods , namely principal component analysis ( pca ) , independent component analysis ( ica ) and non - negative matrix factorisation ( nmf ) , using simulated spectra mimicking the spectral behaviour from a typical xrb .
this analysis provides a better estimate of the x - ray continuum models required to fit the x - ray spectra in xrbs by taking into account also the spectral variability in addition to the fitting of the time - averaged spectra . in section [ discussion ]
we apply the nmf to a set of spectra from the stellar mass black hole xrb gx 3394 . with a sufficiently long set of observations throughout the hardness - intensity diagram ( hid ) from gx 339@xmath04 ,
we study how spectral decomposition reveals the change of spectral components in the hid . in section [ conclusions ]
we present the conclusions of this paper and make suggestions for the future use of the spectral decomposition methods in the context of xrbs .
x - ray spectra can be decomposed via matrix factorisation techniques .
generally , this problem falls into the category of blind source separation ( bss ) , in which a set of source signals is estimated from a set of mixed signals with no information on the source signals or the mixing process . in this case
the data matrix is a collection of discrete x - ray spectra , @xmath1 , consisting of flux values with energy @xmath2 measured at time @xmath3 .
this matrix is taken to consist of a mix of a few , up to @xmath4 , separate source signals @xmath5 weighted over the energy bands by a weight matrix @xmath6 .
essentially , the columns of the weight matrix can be thought to represent a constant spectral component that has a variable amplitude dictated by the rows of the source signal matrix for each @xmath4 . in practice , @xmath4 is much smaller than @xmath3 or @xmath2 , as the intention is to describe the information in @xmath7 concisely . @xmath8 and
@xmath9 are subsequently estimated in such a way that @xmath10 , i.e. @xmath11 the bss problem is very underdetermined since multiple equally valid solutions exists . however , by placing clever requirements on the factorisation one can reduce the number of possible solutions and obtain meaningful results .
pca , for example , assumes that the source signals are minimally correlated to each other and ica assumes that the source signals are maximally independent of each other . a slightly different approach is introduced in nmf which imposes structural constraints , i.e. the non - negativity of the constituent matrices @xmath8 and @xmath9 . in the following we create a simulated dataset that mimics the spectral effects that are present in the usual x - ray data of xrbs and tackle it using the methods outlined above . in order to gauge the appropriateness of a decomposition method ( pca / ica / nmf ) we created a simulated dataset that mimics the spectral effects that are present in the usual x - ray data of xrbs : absorbed ( phabs ) disc black body spectra ( diskbb ) with changing normalisation and temperature values and cutoff power law spectra ( cutoffpl ) with changing normalisation , spectral index and cutoff values . for
` spectral pathways ' we use distinguishable functions such as sine and saw waves .
in addition , we vary the disc temperature and the power law index so as to produce soft and hard x - ray states .
we use isis @xcite to fake 200 spectra with an exposure of 5 ks , varying the black body normalisation sinusoidally from 500 to 2500 , the black body temperature linearly from 0.3 kev to 1.3 kev and back , the power law normalisation as a saw wave from 1.0 to 2.0 , power law index linearly from 1.5 to 2.5 and back , and a power law cutoff sinusoidally from 40 kev to 120 kev ( fig .
[ sim ] ) .
we used the _ rxte _ response function from gx 3394 pointing 70110010400 and then unfolded the spectra using isis to flux units ( kev photons s@xmath12
@xmath13 kev@xmath12 ) .
we would like to note that by using `` flux - corrected '' spectra , response matrix features are introduced in addition to the physical processes , which could have an effect in the low flux regimes .
as pca , ica and nmf are all methods that rely on _ linear _ decomposition , they are suited best for finding the variability of the normalisations of spectral components , and can behave erratically when dealing with more complicated effects such as changing optical depth and electron temperature that varies the cutoff and spectral slope of the x - ray spectra . when presented with data that varies non - linearly , linear methods can , however , be used as the non - linearity
can be approximated as a collection of several linear components .
thus , we do not expect the decomposition to return the variability of the parameters themselves , but rather the fluxes of the spectral components that are collectively produced by the varying parameters .
the quality of the decomposition is then measured by how well they are able to reproduce the disc and power law fluxes . in the following
we review the different spectral decomposition methods and how they are able to reproduce the original fluxes of the simulated spectral components .
pca ( review e.g. * ? ? ?
* ) is one of the standard tools of time series analysis that has been used in astronomy mainly for stellar spectral classification ( e.g. * ? ? ?
* ) , galaxy spectral classification ( e.g. * ? ? ?
* ) , and quasar spectral classification ( e.g. * ? ? ?
. for studying variable x - ray spectra pca has been used in agns e.g. by @xcite and @xcite , and in xrbs by @xcite and @xcite . however ,
one of the drawbacks in using pca for decomposing spectra is its tendency to cancel components as it works with the mean - subtracted spectra .
this results in components that have either too exaggerated or diminished an impact on the x - ray spectra depending on the values of their normalisation .
additionally , if the eigenvectors of the principal components have both positive and negative values , it results in pivoting behaviour of the spectra , i.e. when the normalisation of that component increases it increases the impact of the component in the positive part and decreases the impact in the negative part of the eigenvector .
however , this turns out to be a good proxy for the power law index @xcite .
we have used a singular value decomposition ( svd ) in calculating the components of individual , mean - subtracted spectra that comprise @xmath7 on the left - hand side of eq .
[ eq1 ] . more detailed description of the method can be found in e.g. @xcite , @xcite and @xcite .
-diagram of the simulated dataset .
after @xmath14 the lev settles down to a straight line indicating the start of the noise .
likewise , in the @xmath15-diagram @xmath4 achieves the value where further adding the number of components decreases the reduced @xmath15 value only by a small amount after @xmath14.,scaledwidth=50.0% ] -diagram .
_ left : _ the number of @xmath4 components above the noise level when the data is selected to be below a certain flux threshold .
_ right : _ the number of @xmath4 components above the noise level when the data is selected to span a certain number of observations from the beginning of the simulation.,scaledwidth=50.0% ] in general , we would expect the quality of the factorisation , i.e. its similarity with the original data , to be an increasing function of the degree of the factorisation @xmath4
. a `` good '' value of @xmath4 would not be a local maximum for quality , but instead a point where the response of the quality to @xmath4 changes from being steep to shallow ( i. e. , a `` good '' value of @xmath4 provides a substantially better approximation than nearby smaller values , but only a slightly worse approximation than nearby larger values ) .
one method of choosing the degree of the factorisation in pca is the log - eigenvalue ( lev ) diagram ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in the lev diagram
significant components can be distinguished from the noise by their deviation from the geometrical progression , i.e. straight line in the diagram .
in addition , we devise a method similar to lev diagram to measure the quality of the factorised spectra using a median of reduced @xmath15 -values of the resulting factorisation when compared to the simulated dataset .
this method calculates the reduced @xmath15-values between each individual spectrum from the factorisation @xmath16 , on the right - hand side of eq .
[ eq1 ] , with different degrees of factorisation and the spectra from the simulated dataset with associated errors and takes the median of these values , thus producing a quality measure of how well the factorisation with degree @xmath4 fits in to the data and reducing the number of components to those that vary above the noise level . as mentioned above the chosen degree of factorisation should be a point where the median of reduced @xmath15 -value changes from being steep to shallow .
in addition , the value should be close to 1 so as to portray faithfully the original spectra .
let us call this method the @xmath15-diagram and it can be formulated as follows : @xmath17^{2}}{\mathrm{max}(j)-k } \bigg\}.\ ] ] the @xmath15-diagram will be used for ica and nmf results as well so as to make the comparison between different methods easier .
this is because the lev diagram can not be used for ica and nmf as in these methods the number of components of the factorisation has to be determined before starting the analysis itself .
[ pcalev ] shows the lev diagram and the @xmath15-diagram for the simulated dataset with different degrees of factorisation .
the lev diagram in fig .
[ pcalev ] indicates the start of the noise after @xmath14 , thus indicating that six components is sufficient to explain the simulated x - ray spectra .
likewise , in the @xmath15-diagram after @xmath14 there appears to be no improvement to the @xmath18 value by adding more components , and thus it is sufficient to take six components into account .
this also demonstrates that the @xmath15-diagram leads to the same conclusion as the lev diagram .
based on these methods it suffices to take into account six principal components for explaining the simulated x - ray spectra .
we also studied how the number of @xmath4 above the noise level changes in the @xmath15-diagram on a flux - limited or time - limited sample . in this case
, we use the same simulation setup , models , and parameter ranges , but increase the number of spectra five - fold to increase resolution .
[ clips ] shows how @xmath4 changes if the sample is restricted below some flux threshold ( left panel ) , or if the sample is restricted to span a number of observations from the beginning of the simulation ( right panel ) . to obtain @xmath14
, we see that use of the whole dataset is not necessary , although it is still a substantial amount ( the flux level 0.7 photons s@xmath12 @xmath13 roughly divides the data in two equal sets ) . however , even in smaller samples the number of significant components is only slightly smaller . of the principal components and
the bottom panel shows the weights @xmath6 of the principal components with the inset showing the last five with a zoomed range for the y - axis.,scaledwidth=50.0% ] fig .
[ pca ] shows the six principal components ; the source signals @xmath5 and weights @xmath6 across the energies , derived from the simulated dataset .
it is clear that the first component ( @xmath19 ) corresponds to the variations in the disc component , and the second component ( @xmath20 ) correspond to the variations in the power law component .
however , it is not clear how the remaining components contribute to the disc and power law components and they likely produce random negative and positive corrections to the disc and/or power law components .
a comparison of the first two components with the disc and power law fluxes from the simulation is shown in section [ quality ] .
ica ( for a review see
e.g. * ? ? ?
* ) is a linear solution to bss similar to pca , but instead of uncorrelatedness and orthogonality ica relies on statistical independency of the constituent components . similarly to pca , ica has been used in astronomy mainly for galaxy spectral classification ( e.g. * ? ? ?
* ; * ? ? ?
statistical independency is a more strict rule than uncorrelatedness , as the uncorrelatedness is implied in independency but not vice versa .
ica relies on the central limit theorem which states that the mean of random processes will always approach gaussian distribution .
thus , ica searches for non - gaussian components .
ica suffers from the same problems in spectral decomposition as pca , as it also works with the mean - substracted spectra resulting in positive and negative weights and pivoting source spectra .
we have used a fastica algorithm @xcite in calculating the components of individual , pre - whitened spectra that comprise @xmath7 , on the left - hand side of eq .
[ eq1 ] using negative entropy to search for the non - gaussian components .
ica has the disadvantage to pca that it does not arrange the components via their fraction of variance .
additionally , the factorisation degree has to be determined before running the algorithm . as in pca
, we use the same method to determine the quality of the factorisation using the @xmath21-diagram as explained in section [ degree ] .
[ icachi ] shows the @xmath21-diagram for different degrees of factorisation .
similar to pca results , after @xmath14 there appears to be no improvement to the @xmath18 value by adding more components , and thus it is enough to take six components into account for the ica analysis .
-diagram of the simulated dataset .
after @xmath14 the @xmath15-diagram achieves the value where further adding the number of components decreases the reduced @xmath15 value only by a small amount , thus it is sufficient to take six components into account for the matrix factorisation.,scaledwidth=50.0% ] and the bottom panel shows the weights @xmath6 of the independent components.,scaledwidth=50.0% ] fig .
[ ica ] shows the six independent components , divided into the source signals and weights as previously , derived from the simulated dataset . unlike in pca
, all the components seem to be relevant in either the disc and/or power law component . based on the energy range of the weights and the simulated fluxes in fig .
[ sim ] it appears that the @xmath22 independent components represent the disc and @xmath23 independent components represent the power law spectral components .
however , it can be seen that when the power law flux is low the disc flux leaks to all independent components .
a comparison of the disc and power law components as represented by the respective independent components with the disc and power law fluxes from the simulation is shown in section [ quality ] .
nmf @xcite provides an exciting alternative to traditional dimensional - reduction methods . in nmf
, samples are represented by non - negative combinations of canonical components .
the structure found by nmf methods is thus often very different from , and more intuitive to interpret than , that of more traditional eigenvector - based methods , such as pca . by constructing samples as positive combinations
, nmf also has the potential to disentangle canonical components which often overlap to create particular community samples .
the price of this more intuitive decomposition is that the factorisation is approximate , i.e. not unique as in pca , and components depend on the dimension of the decomposition , requiring greater care in the interpretation . in nmf the matrices @xmath9 and @xmath8 are found by minimising a cost function under the constraint that they must be non - negative . such a cost function can be constructed using some distance measure @xmath24 between @xmath7 and @xmath16 . in this paper
, we use the generalised kullback - leibler ( kl ) divergence , @xmath25 where @xmath26 .
the kl divergence measures the information lost when @xmath27 is used to approximate @xmath7 , and thus minimising it will result in maximising the information of @xmath7 in @xmath27 .
additionally , it has been shown in @xcite that minimising the above cost function is equivalent to maximising the likelihood of generating @xmath7 from @xmath27 when @xmath7 is assumed to be poisson distributed with mean @xmath27 .
therefore , the kl divergence is appropriate cost function to use with x - ray counting data . in this paper
we use the nmf package @xcite that calculates the standard nmf @xcite by picking random starting values for @xmath9 and @xmath8 from a uniform distribution @xmath28 $ ] and then updating iteratively 10000 times to find a local minimum of the cost function with a multiplicative rule from @xcite : @xmath29 @xmath30 to ensure that the algorithm does not get stuck in a local minimum we repeated the minimisation process for 50 different starting points ( 3050 starting points is considered sufficient in @xcite and @xcite to get robust estimate of the factorisation degree @xmath4 ) for each value of @xmath4 and 300 different starting points for the chosen degree of factorisation , and used the factorisation which minimised the cost function in eq .
[ cost ] .
nmf is not a hierarchical method ; each component depends on the choice of degree @xmath4 and thus this choice should be made with care .
similar to pca and ica above , we use the @xmath15-diagram as explained in section [ degree ] .
[ nmfchi ] shows the @xmath21-diagram for different degrees of factorisation .
after @xmath14 there seems to be no improvements to the @xmath18 value by adding more components , and thus it is enough to take six components into account for the nmf analysis .
-diagram of the simulated dataset .
after @xmath14 the @xmath15-diagram achieves the value where further adding the number of components decreases the reduced @xmath15 value only by a small amount.,scaledwidth=50.0% ] of the nmf components and the bottom panel shows the weights @xmath6 of the nmf components.,scaledwidth=50.0% ] fig .
[ nmf ] shows the components derived by nmf from the simulated dataset .
the components are clearly divided to those representing the disc component ( @xmath31 ) and the power law component ( @xmath32 ) .
the two components representing the disc retain even some information of the original parameters , @xmath14 showing the increasing and decreasing disc temperature and @xmath33 sinusoidally varying normalisation , though some mixture of both components can be clearly seen .
the components representing the power law are more mixed with each other , but some similarity can be distinguished of the power law normalisation with @xmath20 and @xmath34 components , power law index with @xmath19 component and the cutoff with @xmath35 component .
one has to bear in mind that one - to - one correspondence with the model parameters is not expected as their effects to the spectra is non - linear , but nmf performs quite well in distinguishing the disc and power law components from each other and tackling the non - linear effects . in the following
we will compare all the decompositions from pca , ica and nmf to the fluxes derived from the disc and power law components of the simulated dataset .
it is interesting to note that all methods converge on the same degree of factorisation however with seemingly different components .
this shows that individual components of the factorisations do not likely represent any meaningful parameters of the spectral model , which is expected as the parameters vary in non - linear fashion .
however , it is possible to compile the disc and power law component from a collection of the factorisation components .
[ quality ] shows the correlation between the model flux values ( see fig .
[ sim ] ) and individual or different combinations of the @xmath5 representing the disc or power law component of the different decomposition methods described above .
the pca can fairly well distinguish the disc flux .
the power law flux is trickier to produce , and the @xmath20 component comes closest to that .
it is not clear how the correcting components ( @xmath36 ) would improve the match if added or subtracted from these `` main '' components .
in addition , it seems that the components do not resemble the variations of the individual model parameters .
ica performs slightly better in distinguishing the flux components .
components @xmath22 add up to match the disc flux and @xmath23 to power law flux .
however , likewise in pca the correlation breaks down for low power law fluxes .
nmf performs the best in distinguishing the two flux components .
the components representing the disc flux ( @xmath31 ) are equally good as in pca and ica .
the components representing the power law flux ( @xmath32 ) distinguish the power law flux of the simulated data the best out of the three methods with smaller spread and continuing the correlation to low power law fluxes as well . )
.,scaledwidth=50.0% ] the unique feature of nmf that allows only positive values for the factorisation , allows the resulting @xmath6 and @xmath5 to be united to form the spectra of individual components .
thus , we select the @xmath4 that corresponds to the disc and power law components mentioned above , and form the disc and power law spectra as @xmath37 and @xmath38 respectively .
the resulting spectra are then fed into isis area and a nominal 1 sec integration time are assumed for fitting purposes .
the units of the ascii spectra should be photons s@xmath12 @xmath13 , thus @xmath39 and @xmath40 are multiplied by the energy bin widths and divided by the bin energy .
it is also important to set minimum_stat_err variable lower than the input uncertainties to keep isis from modifying them . ] and fitted with diskbb or cutoffpl model for @xmath39 and @xmath40 respectively .
the parameters of the fits are then compared ( see fig . [ plotparams ] ) to the original ones in fig .
it is clear that exactly similar parameter tracks are not achieved , but this is mainly driven by low flux values which increase the degeneracy of the model parameters .
gx 3394 is a low - mass black hole x - ray binary discovered in the early 1970s @xcite .
the mass of the black hole is @xmath41 5.8@xmath42 @xcite at a distance of 79 kpc @xcite .
the inclination is uncertain though likely less than 60@xmath43 @xcite .
gx 3394 exhibits recurrent outbursts and is one of the most observed xrbs , thus making it a very suitable object for detecting temporal changes in its x - ray spectra .
in addition , the x - ray spectra of gx 3394 can be usually fit with relatively simple model consisting of an absorbed multicolour disc black body component and/or a power law component with a cutoff , and an iron line ( e.g. * ? ? ? * ) .
gx 3394 is often used as the prime example of a source exhibiting hysteresis in its x - ray spectra , i.e. the change from the power law dominated hard state to the disc dominated soft state occurs at a higher luminosity than the change from the soft state back to the hard state in the same outburst ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the outbursts of gx 3394 also link radio observations to the spectral evolution such that the x - ray flux correlates with the radio flux in the hard state @xcite , the radio flux exhibits strong , transient outbursts before quenching when the source transits to the soft state , and the radio flux is again detected when the source transits back to the hard state but without strong outbursts @xcite .
these characteristics make gx 3394 a good source to probe accretion physics and the disc - jet connection .
other interesting observations include possible detections of the accretion disc in the hard state @xcite which have implications on the accretion geometry and compact jet production .
thus , we selected gx 3394 as the source to test the best unsupervised spectral decomposition method found in this paper , namely the nmf analysis ( see section [ spectral_decomposition ] ) , because of the amount of data available , the ease of fitting the x - ray spectra with simple models , and the clear hysteresis the source exhibits with implications to the accretion physics .
we analysed the archived _ rxte _ observations of gx 3394 from year 2002 to year 2010 that include both _ rxte_/pca and high - energy x - ray timing experiment ( _ rxte_/hexte ) data totalling to 934 pointings .
we also exclude pointings that exhibit large data gaps in the _ rxte_/hexte spectrum .
small data gaps ( @xmath443 energy channels ) are filled by interpolating data from the neighbouring points .
the whole data set includes four outbursts from 2002 , 2004 , 2007 and 2010 .
a comprehensive look into these outbursts can be found e.g. from @xcite and @xcite and references therein .
each pointing is individually reduced by the standard method as described in the _ rxte _ cook book using heasoft 6.13 .
the individual spectra for the analysis is prepared as follows : the _ rxte_/pca spectrum is extracted from the top layer of pcu2 , which has been operational for all the selected pointings .
rxte_/hexte spectrum is extracted from cluster b for data previous to 14 december 2009 when the cluster stopped rocking and it was permanently left as an off - source pointing position . after that we used cluster a for the source data and estimated the background using cluster b ( due to cluster a not rocking and fixed to on - source pointing position )
rxte_/pca and _ rxte_/hexte spectra are then unfolded and the flux values extracted from 325 kev for _ rxte_/pca and from 20200 kev for _ rxte_/hexte . _
rxte_/hexte spectra are then normalised to the _ rxte_/pca such that the overlapping energy range from 2025 kev is the same for both detectors . however , _
rxte_/pca data is used in this region in the final spectra . for the degree of the nmf analysis ( see text for more details ) .
_ bottom : _ the individual @xmath6 of the factorisation for each @xmath4 .
the inset shows the same plot but for smaller range of @xmath6 to better show components at higher energies .
the components for @xmath45 can be attributed to the disc , @xmath46 to the iron line and power law.,scaledwidth=50.0% ] we used the exact same method as described in section [ nmf_sec ] to analyse the spectral data from gx 3394 .
[ plotnmfdata ] shows the degree of factorisation in the upper panel for gx 3394 data .
we choose @xmath34 for the degree as it provides substantially better approximation than smaller degrees , but only a slightly worse approximation than larger degrees as was explained in section [ nmfdegree ] .
the lower panel of fig .
[ plotnmfdata ] shows the individual @xmath6 of the factorisation . from these
we can see that the @xmath47 corresponds to the disc , @xmath46 corresponds to the iron line and the cutoff power law .
the @xmath48 component also shows line - like features in the high energies which arise from the _ rxte_/hexte background lines present in the _ rxte_/hexte spectra after 2010 . as was shown in section [ quality ] , we can form the disc and power law fluxes by adding together the appropriate @xmath5 , in this case @xmath49 and @xmath50 respectively , taking into account the iron line flux in the latter . as was stated in section [ quality ] these correspond fairly well to the actual spectral component fluxes .
thus , _ it is possible to follow the evolution of the disc and power law fluxes without doing any spectral fitting to the data_. in addition , it is possible to construct a disc fraction luminosity diagram ( dfld ) and a power law fraction luminosity diagram ( pfld ) using @xmath51 and @xmath52 .
here we use the pfld defined as the total flux , @xmath53 , as a function of @xmath54 , which equals to 1 when the total flux is composed only of power law flux , and 0 when the total flux is composed only of disk flux . if we take the above results as portraying the actual disc and power law components we can form the disc and power law spectra as @xmath55 and @xmath56 respectively , and feed them into isis for spectral fitting .
we use the original data errors for fitting both @xmath39 and @xmath40 spectra and do not add any systematic error to the spectra .
the @xmath39 spectra were fit with a model phabs@xmath57diskbb using an energy range 315 kev , and @xmath40 with a model phabs@xmath57cutoffpl using an energy range 10140 kev , thus leaving out the iron line region for simplicity and to better concentrate on the power law continuum .
[ fitfig ] shows the results of the modelling plotted on the pfld .
this is done so as to show the evolution of the parameters rather than to provide hard values .
as it is not clear how the data errors would be divided into the @xmath39 and @xmath40 components the resulting parameters have bigger errors than when fitted to the whole spectra .
thus , the parameter values in this presentation are more robust when they are changing gradually , and likely does not represent the actual value when they are in a region where the parameter value changes abruptly across the whole parameter range ( i.e. a mix of colours in fig .
[ fitfig ] ) .
in addition , parameter values that have their error values equal to the minimum or maximum value allowed are removed . from fig .
[ fitfig ] we can see the following : 1 . the power law normalisation increases with the overall luminosity and the power law index increases with the power law fraction , the exception being when the power law fraction is below 0.01 .
the scatter in the lower part of the hard state appears to be due a changing value of the power law index from @xmath5811.5 .
2 . as the power law fraction starts to decrease in the hard state , the disc normalisation starts to increase .
3 . the cutoff energy decreases with the luminosity in the hard state . up to the critical luminosity ( as discussed above )
the cutoff is over 100 kev , but starts to decrease after the critical luminosity down to @xmath583040 kev .
the difference between the transition fluxes when transferring to the soft state seem to occur from a difference in the disc normalisation and power law normalisation ( an indication of a difference could be seen in the cutoff energy as well ) , while the disc temperature and power law index change in similar fashion for each transition .
the disc gets hotter when transitioning to the soft state and there is an indication of heating up again when transitioning back to the hard state as well .
6 . in the soft state
both the disc normalisation and temperature decrease slowly .
these results correspond well to the ones found in previous studies ( e.g. in * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in addition , the nmf analysis is able to track the changes of the disc component in the low disc fraction regime . these correspond to the regions where the disc component is weak and
not required statistically in the spectral fits that have been performed previously . however , it is possible that the low energy tail of the response matrices that arise from escape and fluorescence peaks is mimicking the disc component in the faint spectral states .
but , these peaks are most likely caused by photons that come from the hard spectral component and thus have the same variability producing soft excess to the hard , and not to the soft , factorised component .
this shows the strength of the nmf analysis in situations where the disc component is harder to detect by using only spectral fitting . while the fluxes of the disc and power law components can be distinguished fairly well in the nmf analysis ( fig .
[ quality ] ) , one has to be careful with the values of spectral components in the low flux regimes as these are most likely degenerate ( see fig .
[ plotparams ] and compare to fig .
[ sim ] ) .
as was found in previous studies , at the soft state the disc luminosity is driven by the @xmath59 -relation mimicking the emission from a black body with radius @xmath60 and temperature @xmath61 .
it is usually assumed that the above relation is achieved when the inner disc is located at the innermost stable circular orbit ( isco ) , though multiple factors can affect this scaling ( see a more complete discussion on the difference between these factors in @xcite ) .
[ lumtemp ] shows the luminosity - temperature diagram for a set of selected points with a best - fit line in black , and the whole pfld in the inset with the selected points in red .
the best - fit line follows approximately the relation @xmath62 , which differs from the usual relationship .
however , as the _ rxte _ is sensitive only for energies 3 kev and upwards , the missing softer x - ray flux , as well as the absorption , affect the luminosity - temperature diagram as discussed in @xcite ( where their original correlation was @xmath63 and corrected @xmath64 ) .
here we are concerned with the disc - temperature scaling in the broad sense and assume that when taking into account properly the data below 3 kev and the effect of absorption the scaling should be close to @xmath65 . from fig .
[ lumtemp ] it can be seen that _
gx 3394 exhibits the disc - temperature relation in the disc component right after it starts the transition to the soft state _
, i.e. the inner disc seems to be at the isco already in the beginning of the intermediate state right after the peak of the outburst .
this relation breaks in the soft state when the power law fraction achieves values close to @xmath580.30.4 . , which differs from the expected theoretical relation ( see text for possible reasons).,scaledwidth=50.0% ]
in this paper we have demonstrated that unsupervised linear spectral decomposition methods can be used to follow the evolution of distinct spectral components .
the non - linearities present in the original spectral components can be taken into account by adding multiple linear components together based on their weights across the spectral energies .
these methods can resolve differently varying spectral components given that they present a measurable effect on the fluxes .
of the three methods used to tackle a simulated data set we showed that the non - negative matrix factorisation performs best over principal component analysis and independent component analysis .
the non - negative matrix factorisation also has the additional benefit that the resulting factorisations are always positive , thus the components of these factorisations can be used to fit spectral models separately . tracking the individual spectral parameter variations is not as robust compared to fluxes but they can be tracked reasonably well with good quality spectra and high flux values .
this has the advantage over normal spectral fitting where different models fit the same spectra or a specific spectral component is not required statistically in the fits .
we applied the non - negative matrix factorisation to a set of _ rxte _ spectra of the galactic black hole x - ray binary gx 3394 that includes four outbursts from the source .
we found that five components are required to accurately represent the spectra , and that these can be divided into disc and power law components taking into account the iron line in the latter .
fitting the factorised disc and power law components separately with an appropriate spectral model corroborates the notion that the inner edge of the disc is at the innermost stable circular orbit at the onset of the intermediate state following the peak of the outburst .
however , this analysis should be refined by including data below the _ rxte _ soft x - ray limit . in the future
we envision that unsupervised linear spectral decomposition methods will be used in multiple situations involving the detection of separate spectral components . in order to extend these preliminary studies in detecting the accretion disc in the low disc flux states , as seen for example in the hard state , we would need to have access to both data from multiple sources and detectors more sensitive to the softer x - ray bands . in order to detect the spectral component responsible for the quasi - periodic oscillations in xrbs ,
different timing scales could be tested .
finally , these methods provide an alternative way of detecting the spectral components without performing actual spectral fitting , which might prove to be an invaluable tool when dealing with large datasets .
i thank the the referee for the careful reading of the manuscript and excellent suggestions that improved the paper .
i thank diana hannikainen and thomas maccarone for useful discussions .
i gratefully acknowledge support from emil aaltonen sti . 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 .
allen j.t . ,
hewett p.c . ,
richardson c.t . ,
ferland g.j .
, baldwin j.a . , 2013 , mnras , 430 , 3510 brunet j .-
p . , tamayo p. , golub t.r .
, mesirov j.p . , 2004 , pnas , 101 , 4164 cadolle bel , m. et al . , 2011 , a&a , 534 , a119 connolly a. j. et al . , 1995 ,
aj , 110 , 2655 corbel s. , fender r.p . , tzioumis a.k . , nowak m. , mcintyre v. , durouchoux p. , sood , r. , 2000 , a&a , 359 , 251 corbel s. , coriat m. , brocksopp c. , tzioumis a.k . , fender r.p . , tomsick j.a
, buxton m.m .
, bailyn c.d . , 2013 , mnras , 428 , 2500 cowley a.p . , schmidtke p.c .
, hutchings j.b . , crampton d. , 2002 , aj , 123 , 1741 del santo m. , malzac j. , jourdain e. , belloni t.m . , ubertini p. , 2008
, mnras , 390 , 227 dunn r.j.h . , fender r.p . , krding e.g. , cabanac c. , belloni t. , 2008 , mnras , 387 , 545 falcke h. , biermann p. l. , 1999 , a&a , 342 , 49 fender r.p .
, belloni t.m . ,
gallo e. , 2004 , mnras , 355 , 1105 francis p.j .
, hewett p.c . ,
foltz c.b . ,
chaffee f.h . , 1992 ,
apj , 398 , 476 garcia m.r .
, mcclintock j.e . , narayan r. , callanan p. , barret d. , murray s.s . , 2001 , apj , 553 , l47 gaujoux r. , seoighe c. , 2010 , bmc bioinformatics , 11 , 367 hannikainen d.c .
, hunstead r.w .
, campbell - wilson d. , sood r.k . , 1998 , a&a , 337 , 460 houck j.c . ,
denicola l.a .
, 2000 , in manset n. , veillet c. , crabtree d. , eds .
. ser . 216 , astronomical data analysis software and systems ix , p. 591
hutchins l.n . ,
murphy s.m . ,
singh p. , graber j.h .
, 2008 , bioinformatics , 24(23 ) , 2684 hynes r.i .
, steeghs d. , casares j. , charles p.a .
, obrien k. , 2003 , apj , 583 , l95 hyvrinen a. , 1999 , ieee transactions on neural networks 10(3 ) , 626 hyvrinen a. , karhunen j. , oja e. , 2001 , independent component analysis , 2nd ed . , john wiley & sons ichimaru s. , 1977 , apj , 214 , 840 jolliffe i.t . , 2002 , principal component analysis , 2nd ed .
, springer - verlag , new york koljonen k.i.i .
, mccollough m.l . ,
hannikainen d.c .
, droulans r. , 2013 , mnras , 429 , 1173 lee d.d . ,
seung h.s . , 1999 ,
nat , 401 , 788 lee d.d . ,
seung h.s . , 2001 , in leen t.k . ,
dietterich t.g . , tresp v. , eds . , advances in neural information processing systems 13 : algorithms for non - negative matrix factorization .
mit press , cambridge , ma , p. 556
lu h. , zhou h. , wang j. , wang t. , dong x. , zhuang z. , li c. , 2006 , aj , 131 , 790 malzac j. et al . , 2006 , a&a , 448 , 1125 markert t.h .
, canizares c.r . , clark g.w .
, lewin w.h.g . ,
schnopper h.w . ,
sprott g.f . , 1973 ,
apj , 184 , l67 markoff s. , falcke h. , fender r. , 2001 , a&a , 372 , l25 markoff s. , nowak m.a . , wilms j. , 2005 , apj , 635 , 1203 mcclintock j.e . , narayan r. , garcia m.r . , orosz j.a . ,
remillard r.a . ,
murray s.s . , 2003 ,
apj , 593 , 435 miller j.m . , homan j. , steeghs d. , rupen m. , hunstead r.w . , wijnands r. , charles p.a . , fabian a.c . , 2006 ,
apj , 653 , 525 miyamoto s. , kitamoto s. , hayashida k. , egoshi w. , 1995 , apj , 442 , l13 motta s. , muoz - darias t. , casella p. , belloni t. , homan j. , 2011 , mnras , 418 , 2292 muoz - darias t. , casares j. , martnez - pais i.g . , 2008 , mnras , 385 , 2205 narayan r. , yi i. , 1994 , apj , 428 , l13 narayan r. , garcia m.r . , mcclintock j.e . , 1997 , apj , 478 , l79 nowak m.a . , 1995 ,
pasp , 107 , 1207 nowak m.a . , wilms j. , dove j.b . , 2002 , mnras , 332 , 856 nowak m.a .
et al . , 2011 ,
apj , 728 , 13 paatero p. , tapper u. , 1994 , environmetrics , 5(2 ) , 111 parker m.l .
, marinucci a. , brenneman l. , fabian a.c .
, kara e. , matt g. , walton d.j . , 2014 , mnras , 437 , 721 reis r.c . , fabian a.c . , miller j.m . , 2010 , mnras , 402 , 836 russell d.m . , maitra d. , dunn r.j.h . , markoff s. , 2010 , mnras , 405 , 1759 sajda p. , shuyan d. , parra l.c , 2003 , in unser m.a , aldroubi a. , laine a.f . ,
eds , proc .
spie 5207 , wavelets : applications in signal and image processing x , san diego , p. 321
salvesen g. , miller j.m .
, reis r.c . ,
begelman m.c . , mnras , 431 , 3510 stiele h. , motta s. , muoz - darias t. , belloni t.m .
, 2011 , mnras , 418 , 1746 tomsick j.a .
et al . , 2008 , apj , 680 , 593 vaughan s. , fabian a.c . , 2004 ,
mnras , 348 , 1415 whitney c.a . , 1983 , a&as , 51 , 443 zdziarski a.a . , gierliski m. , mikoajewska j. , wardziski g. , smith d.m . , harmon b.a . , mnras , 351 , 791 | in this paper , we explore unsupervised spectral decomposition methods for distinguishing the effect of different spectral components for a set of consecutive spectra from an x - ray binary .
we use well - established linear methods for the decomposition , namely principal component analysis , independent component analysis and non - negative matrix factorisation ( nmf ) . applying these methods to a simulated dataset consisting of a variable multicolour disc black body and a cutoff power law
, we find that nmf outperforms the other two methods in distinguishing the spectral components .
in addition , due the non - negative nature of nmf , the resulting components may be fitted separately , revealing the evolution of individual parameters . to test the nmf method on a real source , we analyse data from the low - mass x - ray binary gx 3394 and found the results to match those of previous studies .
in addition , we found the inner radius of the accretion disc to be located at the innermost stable circular orbit in the intermediate state right after the outburst peak .
this study shows that using unsupervised spectral decomposition methods results in detecting the separate component fluxes down to low flux levels .
also , these methods provide an alternative way of detecting the spectral components without performing actual spectral fitting , which may prove to be practical when dealing with large datasets .
[ firstpage ] accretion , accretion discs methods : data analysis stars : black holes
x - rays : binaries x - rays : individual : gx 3394 x - rays : stars . |
multiple - quantum ( mq ) coherences in a system of dipolar coupled spins can be detected by mq nmr methods @xcite . in the simplest cases ( one - dimensional spin chains and rings , a system of spin - carrying atoms ( molecules ) in nanopores ) ,
analytical and numerical methods were developed @xcite in order to calculate the intensities of mq nmr coherences . usually spin - lattice relaxation is not taken into account at investigations of mq nmr dynamics because the characteristic times of spin - lattice relaxation are longer than the characteristic time of mq nmr dynamics @xcite . at the same time , mq nmr methods have been also used to measure decoherence for highly correlated spin states @xcite and to study the scaling of the decoherence rate with the number of correlated spins @xcite .
mq nmr investigations showed both experimentally @xcite and with computer simulations for a system of equivalent spins @xcite that the relaxation time decreases with the increase in the mq coherence order and in the number of spins .
it was also shown @xcite that mq nmr allows us to find a relationship between the intensity of the mq coherence of the second order and a measure of the quantum entanglement which is the main resource of quantum computers @xcite .
however , that relationship was found without taking into account effects of spin - lattice relaxation @xcite .
a theoretical interpretation of such results is impossible without taking into account those effects .
a theory of spin - lattice relaxation is also useful for traditional problems of nmr ( paramagnetic impurities , internal molecular motions , etc .
) @xcite . in the present paper
we develop a phenomenological theory of spin - lattice relaxation for a system consisting of two dipolar coupled spins in the mq nmr experiment at low temperatures .
we calculate the intensities of mq coherences of the zeroth and plus / minus second orders and discuss the effect of spin - lattice relaxation on mq nmr dynamics .
we show that the developed approach is also applicable for linear spin chains when the approximation of nearest neighbor interactions is used .
we discuss a connection of quantum entanglement with the intensity of mq nmr coherence of the second order at different spin - lattice relaxation times .
we consider a system consisting of two dipolar coupled spins ( @xmath0 ) in the conditions of the mq nmr experiment @xcite .
initially , the system is in the thermodynamic equilibrium state in a strong external magnetic field @xmath1 the density matrix @xmath2 of this state in the laboratory reference frame is given by @xmath3 where @xmath4 is the larmor frequency,@xmath5 the gyromagnetic ratio , @xmath6 the temperature , @xmath7 the bolzman constant , @xmath8 , and @xmath9 is the operator of the projection of spin @xmath10 onto the axis @xmath11 we can neglect indirect spin spin interactions and chemical shifts in eq .
( [ rho0 ] ) because they are much smaller than dipole
dipole interactions ( ddi ) .
the hamiltonian of the ddi of the two spins , @xmath12 in the strong external magnetic field @xcite is @xmath13 where @xmath14 is the coupling constant and @xmath15 we study mq nmr dynamics of this two spin system . the mq nmr experiment @xcite consists of four periods : preparation , evolution , mixing and detection .
the spin system is irradiated on the preparation period by the multi pulse sequence in which every period consists of eight resonance pulses @xcite .
it is suitable to switch to the rotating ( with the larmor frequency ) reference frame @xcite where the multi pulse sequence consists of the four bloks of the form @xmath16 where @xmath17 and @xmath18 are the time intervals between consecutive pulses and @xmath19 denotes pulses turning spins about axis @xmath20 @xcite .
the ddi are oscillating fast , if the period @xmath21 of the multi
pulse sequence is less than the inverse coupling constant @xmath22 @xcite . as a result
, spin dynamics of the system on the preparation period of the mq nmr experiment is governed by the averaged non - secular hamiltonian @xcite @xmath23 where @xmath24 and @xmath25 ( @xmath26 ) are raising and lowering operators ( @xmath27 ) .
spin dynamics on the preparation period is described by the liouville equation for the density matrix @xmath28 @xcite , @xmath29,\ ] ] and @xmath30 it is worth to emphasize that mq nmr coherences are determined by the density matrix at the end of the preparation period of the mq nmr experiment @xcite . in order to take into account spin - lattice relaxation one
has to introduce , generally speaking , a non linear term with respect to the density matrix @xmath28 in the right hand side of eq .
( [ liouv ] ) . in practice ,
such equation can be solved with the linearization of the right hand side after it is expressed as an expansion in powers of @xmath31 , where @xmath32 is the equilibrium density matrix in the rrf .
this procedure is valid , if the deviation of @xmath28 from @xmath32 is small .
one can expect a small deviation , if the time independent @xmath33 is a solution of the liouville equation .
it is reasonable to suppose that @xcite @xmath34 the temperature t is of an order of the lattice temperature .
its exact value is not significant in our consideration .
we will consider the millikelvin temperature range , where one can suppose @xcite that @xmath35 @xmath36 and @xmath37 is the 4 by 4 unit matrix .
then the liouville equation with a linearized phenomenological term for spin
lattice relaxation reads @xmath38-\frac{\rho(t)-\overline{\rho}}{t_{mq}},\ ] ] where @xmath39 is the spin
lattice relaxation time of mq nmr coherences .
a reorienting spin group can be responsible for relaxation @xcite .
we consider relaxation only as a pertubation of mq nmr dynamics , i.e. we mean that @xmath40 we assume also that the spin
lattice relaxation time @xmath41 in the laboratory reference frame is large @xmath42 and we neglect it .
it is worth to emphasize that the equilibrium density matrix @xmath32 describes the spin system in the rrf , where there is no any external magnetic field , i.e. the zeeman reservoir is absent in the considered case .
thus , relaxation of mq nmr coherences can be described with only one relaxation time , @xmath43 introducing @xmath44 and @xmath45 one can find that eq .
( [ liouv1 ] ) can be written as @xmath46,\ ] ] and @xmath47 at the end of the preparation period @xmath48 of the mq nmr experiment , the solution of eq .
( [ liouv2 ] ) can be written as @xmath49 the first term in the right hand side of eq .
( [ solv1 ] ) was obtained in @xcite , @xmath50 where @xmath51 using eq .
( [ solv1 ] ) one can find that @xmath52 the density matrix @xmath53 after the evolution period @xmath54 of the mq nmr experiment @xcite , during which the decoding field @xmath55 ( in frequency units ) acts , is @xmath56 we did not consider spin - lattice relaxation on the evolution period since we assumed that @xmath57 finally , after the mixing period @xmath58 during which the hamiltonian @xmath59 changes its sign , the density matrix @xmath60 becomes @xmath61 calculating the longitudinal magnetization @xmath62 at the end of the mq nmr experiment @xcite and taking into account that @xmath63 one can find that only mq nmr coherences of zero and plus / minus second orders emerge in the our system and their intensities @xmath64 @xmath65 are : @xmath66 note that all performed calculations are also valid for any equilibrium density matrix @xmath32 which commutes with operators @xmath59 and @xmath67 it is also interesting to note that an analogous approach can be applied to a linear chain of @xmath68 interacting spins in the approximation of nearest neighbour interactions @xcite .
only mq nmr coherences of zero and plus / minus second orders emerge in such system .
using exact expressions for the intensities of these mq nmr coherences @xcite , which were obtained without taking into account spin - lattice relaxation , one can obtain @xmath69 where @xmath70 it is worth to emphasize that we neglect dipole - dipole interactions of next nearest neighbours which are only eight times weaker than the interactions of nearest neighbours .
it means that the developed approach is justified only when @xmath71 in considered systems we have found a simple exponential law for spin
lattice relaxation of mq nmr coherences .
it is well known @xcite that entanglement is the main resource for quantum devices ( in particular , quantum computers ) .
mq nmr is one of the numerous methods of obtaining entangled states @xcite .
it turned out @xcite that intensity of the mq nmr coherence of the second order ( without spin - lattice relaxation ) is intimately connected with the concurrence @xcite which describes quantitatively entanglement of quantum systems . in order to calculate the concurrence @xcite at the end of the preparation period
one should determine the matrix @xmath72 as @xmath73 where @xmath74 is the complex conjugate matrix @xmath75 of eq .
( [ solv2 ] ) in the standard basis @xmath76 and @xmath77 @xmath78 is the pauli matrix .
the concurrence @xmath79 of the two - spin system with the density matrix @xmath75 of eq .
( [ solv2 ] ) equals to @xcite @xmath80 where @xmath81 @xmath82 @xmath83 and @xmath84 are the square roots of the eigenvalues of the matrix product @xmath85 although this product is not hermitian , it has real nonnegative eigenvalues @xcite .
one can find from eqs .
( [ term1]),([solv2]),([flip]),([conc ] ) that the concurrence @xmath86 in our system is @xmath87.\end{gathered}\ ] ] this expression coincides with the one obtained in @xcite in the limit of the infinite spin lattice relaxation time .
the entangled state emerges at temperatures @xmath88 where @xmath89 if one takes @xmath90 the entangled state emerges at the temperature @xmath91 when @xmath92 @xcite . according to eq .
( [ temp ] ) the entangled states emerge at @xmath93 i.e. at @xmath94 the temperature @xmath95 decreases with the decrease in the spin - lattice relaxation time . using eqs .
( [ cohs ] ) and ( [ concdec ] ) one can find a simple relationship between intensities of mq nmr coherences of the second order and the concurrence : @xmath96}-\\\frac{e^{-\tau / t_{mq}}}{2\cosh^2\beta/2}-\frac{1-e^{-\tau / t_{mq}}}{2}.\end{gathered}\ ] ] this formula coincides again with the analogous one obtained in @xcite at @xmath97 in order to obtain the entanglement measure of the two - qubit system we average over the state of the first spin and obtain the reduced density matrix for the second spin @xcite .
the reduced density matrix thus depends on the fluctuations of the first spin .
these fluctuations give rise to the fluctuations of the entanglement measure @xcite .
the rms fluctuations , @xmath98 of the entanglement entropy of the two - qubit system are @xcite @xmath99,\ ] ] where the concurrence @xmath79 is determined by eq .
( [ concdec ] ) .
the dependencies of the sum of the mq nmr coherences of the plus / minus second order , the concurrence @xmath86 and the entanglement fluctuations on the dimensionless spin - lattice relaxation rate @xmath100 at @xmath101 are shown in fig .
[ fig ] . at long spin - lattice relaxation times
the concurrence approximately coincides with the sum of the intensities of the mq nmr coherences of the plus / minus second order as it was obtained in @xcite . at @xmath102 mq nmr dynamics
determines the entanglement and its fluctuations .
lattice relaxation is the dominant factor when @xmath103 we note also that the concurrence decreases with a decrease of the spin - lattice relaxation time and entanglement fluctuations are getting more important .
in conclusion we emphasize that quantum entanglement can be studied with mq nmr methods and the effect of spin - lattice relaxation on mq nmr coherences is important both for traditional magnetic resonance @xcite and quantum information processing @xcite .
we are grateful to d.e .
feldman for many helpful discussions .
this work was supported by the program of the presidium of russian academy of sciences no . 7 ( the program `` development of the methods for obtaining chemical substances and creating new materials '' ) .
( dash line ) , the entanglement fluctuations ( dash - dot line ) versus the dimensionless spin - lattice relaxation rate @xmath104 the spin - spin coupling constant @xmath105 the duration of the preparation period @xmath106 @xmath107 and @xmath108,width=283 ] 99 j. baum , m.munowitz , a.n .
garroway , and a. pines , multiple - quantum dynamics in solid state nmr , j. chem .
83 ( 1985 ) 2015 - 2025 .
g. cho , j.p .
yesinowski , h-1 and f-19 multiple - quantum nmr dynamics in quasi - one - dimensional spin clusters in apatites , j. phys .
( 1996 ) 15716 - 15725 .
feldman , s. lacelle , multiple quantum nuclear magnetic resonance in one - dimensional quantum spin chains , j. chem .
( 1997 ) 7067 - 7084 .
doronin , a.v .
fedorova , e.b .
feldman , et al .
, multiple quantum nmr dynamics of spin-1/2 carrying molecules of a gas in nanopores , j. chem .
131 ( 2009 ) 104109 .
krojanski , d. suter , scaling of decoherence in wide nmr quantum registers , phys .
( 2004 ) 090501 .
h. cho , p. cappellaro , d.g .
cory , et al .
, decay of highly correlated spin states in a dipolar - coupled solid : nmr study of @xmath109 , phys .
b 74 ( 2006 ) 224434 . g.a .
alvarez , d. suter , nmr quantum simulation of localization effects induced by decoherence , phys .
104 ( 2010 ) 230403 .
doronin , e.b .
feldman , a.i .
zenchuk , numerical analysis of relaxation times of multiple quantum coherences in the system with a large number of spins , j. chem .
( 2011 ) 034102 .
feldman , a.n .
pyrkov , evolution of spin entanglement and an entanglement witness in multiple - quantum nmr experiments , jetp letters , 88 ( 2008 ) 398 - 401 .
nielsen , i.l .
chuang , quantum computation and quantum information , university press , cambridge , 2000 .
a. abraham , the principles of nuclear magnetism , clarendon press , oxford , 1961 .
u. haeberlen , j.s .
waugh , coherent averaging effects in magnetic resonance , phys .
175 ( 1968 ) 453 - 467 .
m. goldman , spin temperature and nuclear magnetic resonance in solids , clarendon press , oxford , 1970 .
a. abraham , m. goldman , nuclear magnetism : order and disorder , clarendon press , oxford , 1982 .
moskvich , a.m. polyakov , s.p .
prozorov , v.e .
zobov , a.v .
ponomarenko , the spin lattice relaxation of the moving two spin group in multipulse spin locking , phys
( b ) 116 ( 1983 ) k35-k38 .
feldman , i.i .
maximov , multiple quantum dynamics in linear chains and rings of nuclear spins in solids at low temperatures , j. magn .
reson . 157
( 2002 ) 106 - 113 .
furman , v.m .
meerovich , v.l .
sokolovsky , entanglement and multiple quantum coherence dynamics in spin clusters , quantum information processing 8 ( 2009 ) 379 - 386 .
wootters , entanglement of formation of an arbitrary state of two qubits , phys .
80 ( 1998 ) 2245 - 2248 .
horn , c.r .
johnson , matrix analysis , university press , cambridge , 1985 .
feldman , m.a .
yurishchev , fluctuations of quantum enatnglement , jetp letters 90 ( 2009 ) 70 - 74 . | a phenomenological theory of spin - lattice relaxation of multiple - quantum coherences in systems of two dipolar coupled spins at low temperatures is developed .
intensities of multiple - quantum nmr coherences depending on the spin - lattice relaxation time are obtained .
it is shown that the theory is also applicable to finite spin chains when the approximation of nearest neighbour interaction is used .
an application of this theory to an estimation of the influence of decoherence processes on quantum entanglement and its fluctuations is briefly discussed . |
almost exactly 100 years ago in 1917 willem de sitter published two seminal papers @xcite in which he presented his , now famous , exact solution of einstein s gravitational field equations without matter but with a positive cosmological constant @xmath2 . together with the pioneering work from the same year by einstein himself @xcite
, they mark the very origin of relativistic cosmology .
since then , the vacuum de sitter universe with @xmath4 and its formal antipode with @xmath5 , later nicknamed the anti - de sitter universe , have become truly fundamental spacetimes .
they admit a cornucopia of applications , ranging from purely mathematical studies , theoretical investigations in classical general relativity , quantum field theory and string theory in higher dimensions , to contemporary `` inflationary '' and `` dark matter '' cosmologies , modeling naturally the observed accelerated expansion of the universe , both in its distant past and future .
this remarkable number of applications in various research branches clearly stems from the fact that the ( anti-)de sitter spacetime is highly symmetric and simple , yet it is non - trivial by being curved everywhere . in fact , it is _ maximally symmetric _ and has a _ constant curvature _ , just as the flat minkowski space .
of course , this greatly simplifies all analyses and studies .
one would be inclined to expect that , after a century of thorough investigation of various aspects of these important spacetimes , there is not much left to add or discover .
quite surprisingly , still there are interesting new properties which deserve some attention .
to celebrate the centenary of de sitter space , we take the liberty of presenting here yet another family of coordinate representations of de sitter and anti - de sitter 4-dimensional solutions , in which the metric is _ diagonal _ and typically _ static and axially symmetric_. it naturally arises in the context of @xmath6-metrics @xcite with a cosmological constant @xmath2 , which belong to a wider class of plebaski demiaski non - expanding spacetimes @xcite . as far as we know , this family has not yet been considered and systematically described .
we hope that it could find useful applications in general relativity and high energy physics theories . in particular , in our work we will present and analyze the family of metrics for de sitter and anti de sitter spacetimes which can be written in a unified form @xmath7 with @xmath8 where @xmath9 and @xmath10 are two independent discrete parameters .
the metric has the geometry of a _ warped product _ of two
_ 2-spaces of constant curvature _ , namely @xmath11 ( according to the sign of @xmath12 ) spanned by @xmath13 , and @xmath14 ( according to the sign of @xmath2 ) spanned by @xmath15 .
the warp factor is @xmath16 . also , there are two obvious symmetries corresponding to the killing vectors @xmath17 and @xmath18 .
the metric is clearly static when @xmath19 .
it is also axially symmetric when the axis given by @xmath20 is admitted ( for @xmath21 and for @xmath22 ) , in which case this axis is regular for @xmath23 .
the metric ( [ eq : pddads])([eq : p ] ) is a special subcase of the plebaski
demiaski non - expanding ( and thus kundt ) type d solutions , see metric ( 16.27 ) of @xcite .
it is obtained by setting @xmath24 and @xmath25 ( so that @xmath26 ) with the identification @xmath27 this gives vacuum @xmath6-metrics , as classified by ehlers and kundt @xcite , generalized here to any value of the cosmological constant @xmath2 . when @xmath28 , the curvature singularity at @xmath29 disappears and the metric reduces to the conformally flat ( anti-)de sitter space with ( [ eq : p ] )
. therefore , the family of de sitter and anti de sitter metrics ( [ eq : pddads])([eq : p ] ) can be viewed as a _ natural background for the b - metrics with @xmath2 _ , and will thus play a key role in understanding their geometrical and physical properties .
this is our main motivation .
the structure of the present work is simple . in the next section [ sec2 ] we recall basic properties of ( anti-)de sitter spacetimes , introducing the notation . in section [ sec3 ]
we identify all distinct subclasses of the general metric ( [ eq : pddads])([eq : p ] ) .
these are studied in detail in subsequent section [ sc : ds ] for the de sitter case @xmath4 , and for the anti de sitter case @xmath5 in section [ sc : ads ] .
there are also four appendices in which we present a unified form of the new parametrizations , their mutual relations , further metric forms , and transformations to standard ( anti-)de sitter metrics .
minkowski , de sitter and anti - de sitter spacetimes are the most fundamental and simplest exact solutions of einstein s field equations
. they are maximally symmetric , have constant curvature , and are the only vacuum solutions with a vanishing weyl tensor they are conformally flat and satisfy the field equations @xmath30 , where @xmath2 is a cosmological constant , so that in four spacetime dimensions the curvature is @xmath31 .
minkowski , de sitter and anti - de sitter spacetimes are thus einstein spaces with vanishing , positive , and negative cosmological constant @xmath2 , respectively
. specific properties of these classic spacetimes have been investigated and employed in various contexts for a century .
they were described in many reviews , such as by schrdinger @xcite , penrose @xcite , weinberg @xcite , mller @xcite , hawking and ellis @xcite , birell and davies @xcite , bik @xcite , bik and krtou @xcite , grn and hervik @xcite , or griffiths and podolsk @xcite .
let us recall here only those expressions and properties that will be important for our analysis . since the original works by de sitter @xcite and lanczos @xcite it is known that the de sitter manifold can be conveniently visualized as the hyperboloid @xmath32 embedded in a flat five - dimensional minkowski space @xmath33 this geometrical representation of de sitter spacetime is related to its symmetry structure characterised by the ten - parameter group so(1,4 ) .
the entire hyperboloid ( [ c1 ] ) is covered by coordinates @xmath34 , @xmath35}$ ] , @xmath36}$ ] , @xmath37 such that @xmath38 in which the de sitter metric takes the flrw form @xmath39 the spatial sections at a fixed synchronous time @xmath40 are 3-spheres @xmath41 of constant positive curvature which have radius @xmath42 .
these contract to a minimum size @xmath43 at @xmath44 , and then re - expand .
the de sitter spacetime thus has a natural topology @xmath45 .
this most natural parametrisation of the hyperboloid is illustrated in the left part of figure [ fig : dshyp1 ] .
_ global causal structure _ of the de sitter spacetime can be analyzed by introducing a conformal time @xmath46 and the conformal factor @xmath47 by @xmath48 the metric ( [ c4 ] ) then becomes @xmath49 corresponding to the parametrization @xmath50 the de sitter spacetime is thus conformal to a part of the einstein static universe .
its boundary given by @xmath51 is located at @xmath52 and @xmath53 , which correspond to past and future conformal infinities @xmath54 and @xmath55 , respectively , as illustrated in the right part of figure [ fig : dshyp1 ] .
these infinities have a _
spacelike character_. for more details see , e.g. , @xcite .
analogously , the anti - de sitter manifold can be viewed as the hyperboloid @xmath56 embedded in a flat five - dimensional space @xmath57 which has two temporal dimensions @xmath58 and @xmath59 .
this representation of anti - de sitter space reflects its symmetry structure characterised by the ten - parameter group of isometries so(2,3 ) .
the most natural static coordinates @xmath60 covering the entire hyperboloid are with respect to the parametrization ( 5.3 ) employed in @xcite . ]
@xmath61 see the left part of figure [ fig : adshyp1 ] , in which the anti - de sitter metric reads @xmath62 any spatial section @xmath63 const .
is a 3-space of constant negative curvature spanned by @xmath64 , @xmath36}$ ] , @xmath37 .
the singularities at @xmath65 and @xmath66 are only coordinate singularities . .
the surface drawn is that for @xmath67 ( that is @xmath66 ) . with @xmath68 and @xmath69 reintroduced , each point represents a 2d hemisphere .
right : the global conformal structure of anti - de sitter spacetime ( with @xmath70 ) .
it is conformal to the region @xmath71 of the einstein static universe represented here ( by dashed lines ) as a solid cylinder of radius @xmath72 and infinite length .
the boundary @xmath73 is the anti - de sitter conformal infinity @xmath74 , which has a timelike character and topology @xmath75 .
the penrose diagram , corresponding to 2d shaded section , is obtained by fixing the coordinate @xmath69 . ] the two temporal dimensions @xmath58 and @xmath59 in the 5d flat space are parametrised using ( [ d3 ] ) by a _ single _
time coordinate @xmath76 that is _ periodic _ : values of @xmath76 which differ by a multiple of @xmath77 represent the same points on the hyperboloid .
thus , the anti - de sitter spacetime defined in this way has the topology @xmath78 , and contains closed timelike worldlines .
this periodicity of @xmath76 is not evident in the four - dimensional metric ( [ d4 ] ) , and it is possible to take @xmath79 .
such a range of coordinates corresponds to an infinite number of turns around the hyperboloid .
it is usual to unwrap the circle @xmath80 and extend it to the whole @xmath81 instead , without reference to the parametrisation ( [ d3 ] ) .
one thus obtains a _ universal covering space _ of the anti - de sitter universe with topology @xmath82 without closed timelike curves .
to represent _ global causal structure _ of the anti - de sitter spacetime , a conformal spatial coordinate @xmath83 and the conformal factor @xmath47 may be introduced as @xmath84 writing @xmath85 , the metric ( [ d4 ] ) then takes the form @xmath86 corresponding to the parametrization @xmath87 the whole ( universal ) anti - de sitter spacetime is thus conformal to the region @xmath71 of the einstein static universe , see the right part of figure [ fig : adshyp1 ] . the anti - de sitter conformal infinity @xmath88 , given by @xmath51 ,
is located at the boundary @xmath73 ( corresponding to @xmath89 ) .
in contrast to the de sitter space , the conformal infinity @xmath88 in the anti - de sitter spacetime forms a _ timelike surface _ @xmath73 .
first , we are going to summarize different possible forms of the metric ( [ eq : pddads ] ) with ( [ eq : q ] ) , ( [ eq : p ] ) for the ( anti-)de sitter space , depending on the cosmological constant @xmath4 or @xmath5 , and the two discrete parameters @xmath92 and @xmath93 . to keep the correct signature of the metric , the metric function @xmath94 _ must be positive _ , @xmath95 ( otherwise there would be two additional temporal coordinates @xmath96 and @xmath97 ) .
this puts a constraint on the parameter @xmath90 and the range of @xmath96 . on the other hand ,
the function @xmath98 _ can be both positive and negative _ , depending on @xmath91 , @xmath90 , and the range of @xmath99 . for @xmath19 , the coordinate
@xmath40 is temporal and @xmath99 is spatial , while for @xmath100 , @xmath99 is temporal and @xmath40 is spatial .
the boundary @xmath101 localizes the _ killing horizon _ related to the killing vector field @xmath102 . in the case of de sitter universe , this coincides with the _ cosmological horizon _ separating the static and dynamic regions .
the coordinate singularity at @xmath103 , where the norm of the killing vector field @xmath18 vanishes , localizes the _ axis of symmetry_. this occurs either when @xmath21 or @xmath22 . in both these cases such an axis
is given by @xmath104 , and it is regular for the range @xmath37 . the list of all possible subcases of the general metric ( [ eq : pddads])([eq : p ] ) , with the admitted ranges of @xmath96 and @xmath99 , are summarized in table [ tbl : ds ] for @xmath4 , and in table [ tbl : ads ] for @xmath5 .
.summary of all possible subcases for @xmath105 and different values of @xmath12 , @xmath106 .
the symbol indicates admitted solutions ( their geometry is @xmath107 ) , while those with @xmath108 are forbidden . [ cols="^,^,^,^,^,^,^",options="header " , ] the coordinates @xmath109 only cover _ the part _
@xmath110 of de sitter spacetime , while they cover _ the whole _ anti - de sitter spacetime ( except in the cases @xmath111 for which @xmath112 ) .
* case @xmath4 : * the de sitter metric ( [ eq : ds11ds ] ) for @xmath111 , @xmath113 ( in which the coordinates @xmath114 are relabeled to @xmath115 ) can be explicitly transformed to the metric ( [ eq : ds10ds ] ) for @xmath111 , @xmath116 by applying @xmath117 similarly , the metric ( [ eq : ds11ds ] ) for @xmath118 can be explicitly transformed to the metric ( [ eq : ds1 - 1ds ] ) for @xmath111 , @xmath119 by applying @xmath120 these relations are easily obtained by comparing ( [ eq : ds11para ] ) to ( [ eq : ds10par ] ) , and ( [ eq : ds11para ] ) to ( [ eq : ds1 - 1par ] ) , respectively .
* case @xmath5 : * interestingly , the same transformations ( [ relds1 ] ) and ( [ relds2 ] ) relate the anti - de sitter metric ( [ eq : ads11ds ] ) for @xmath111 , @xmath113 ( with the coordinates relabeled as @xmath115 ) to ( [ eq : ads10ds ] ) for @xmath111 , @xmath116 and ( [ eq : ads1 - 1ds ] ) for @xmath111 , @xmath119 , respectively . comparing the parameterizations ( [ eq : ads10par ] ) and ( [ eq : ads01par ] ) we find a transformation between the anti - de sitter metric ( [ eq : ads10ds ] ) for @xmath111 , @xmath116 and the form ( [ eq : ads01ds ] ) for @xmath121 , @xmath113 ( relabeled to @xmath122 ) : @xmath123 also , the anti - de sitter metric ( [ eq : ads-11ds ] ) for @xmath124 , @xmath113 ( relabeled to @xmath115 )
is transformed to the metric ( [ eq : ads-10ds ] ) for @xmath124 , @xmath116 by @xmath125 while ( [ eq : ads-11ds ] ) for @xmath124 , @xmath113 is transformed to the metric ( [ eq : ads-1 - 1ds ] ) for @xmath124 , @xmath119 by @xmath126
the new metrics analyzed in this work can also be expressed in closely related forms . in these , both (
warped ) parts representing the constant - curvature 2-spaces are more clearly seen because they are all put into the `` canonical '' forms using triginometric and/or hyperbolic functions . in this appendix
we describe all such related metrics . for each subcase , we present the transformation , the resulting metric , and the corresponding paramaterization of the 5d hyperboloid : * subcase * @xmath127 , @xmath128 : @xmath129 transforms ( [ eq : ds11ds ] ) to @xmath130 respectively , corresponding to the parametrizations @xmath131 z_1 = { \displaystyle a\cos { \vartheta}\cos{\psi}\cosh\frac{t}{a}}\ , , \\[8pt ] z_2 = { \displaystyle a|\cos { \vartheta}| \sin{\psi}}\ , , \\[8pt ] z_3 = { \displaystyle a\sin { \vartheta}\cos{\phi}}\ , , \\[8pt ] z_4 = { \displaystyle a\sin { \vartheta}\sin{\phi}}\ , , \end{array } \!\right\ } \hspace{0 mm } \textrm{for}\ |q|<1\ , , \ \left . \ !
\begin{array}{l } z_0 = { \displaystyle a\cos { \vartheta}\sinh\frac{t}{a}\cosh { \psi}}\ , , \\[8pt ] z_1 = { \displaystyle a\cos { \vartheta}\sinh\frac{t}{a}\sinh { \psi}}\ , , \\[6pt ] z_2 = { \displaystyle \pm a|\cos { \vartheta}| \cosh\frac{t}{a } } \\[2pt ] z_3 = { \displaystyle a\sin { \vartheta}\cos{\phi}}\ , , \\[6pt ]
z_4 = { \displaystyle a\sin { \vartheta}\sin{\phi}}\ , , \end{array } \!\right\ } \hspace{0 mm } \textrm{for}\ |q|>1\ , .
\label{eq : ds11apar}\ ] ] * subcase * @xmath127 , @xmath132 : @xmath133 transforms ( [ eq : ds10ds ] ) to @xmath134 corresponding to @xmath135 * subcase * @xmath127 , @xmath136 : @xmath137 transforms ( [ eq : ds1 - 1ds ] ) to @xmath138 corresponding to @xmath139 * subcase * @xmath127 , @xmath128 : @xmath140 transforms ( [ eq : ads11ds ] ) to @xmath141 respectively , corresponding to the parametrizations @xmath142 z_1 = { \displaystyle a\sinh { \vartheta}\cos{\psi}\cosh\frac{t}{a}}\ , , \\[8pt ] z_2 = { \displaystyle a\sinh |{\vartheta}| \sin{\psi}}\ , , \\[8pt ] z_3 = { \displaystyle \pm a\cosh { \vartheta}\sinh{\zeta}}\ , , \\[8pt ] z_4 = { \displaystyle \pm a\cosh { \vartheta}\cosh{\zeta}}\ , , \end{array } \!\right\ } \hspace{0 mm } \textrm{for}\ |q|<1\ , , \ \left . \ ! \begin{array}{l } z_0 = { \displaystyle a\sinh { \vartheta}\sinh\frac{t}{a } \cosh{\psi}}\ , , \\[8pt ] z_1 = { \displaystyle a\sinh { \vartheta}\sinh\frac{t}{a } \sinh{\psi}}\ , , \\[6pt ] z_2 = { \displaystyle \pm a\sinh |{\vartheta}| \cosh\frac{t}{a } } \\[2pt ] z_3 = { \displaystyle \pm a\cosh { \vartheta}\sinh{\zeta}}\ , , \\[6pt ] z_4 = { \displaystyle \pm a\cosh { \vartheta}\cosh{\zeta}}\ , , \end{array } \!\right\ } \hspace{0 mm } \textrm{for}\ |q|>1\ , .
\label{eq : ads11aparpl}\ ] ] * subcase * @xmath127 , @xmath132 : @xmath143 transforms ( [ eq : ads10ds ] ) to @xmath144 corresponding to @xmath145 * subcase * @xmath127 , @xmath136 : @xmath146 transforms ( [ eq : ads1 - 1ds ] ) to @xmath147 corresponding to @xmath148 * subcase * @xmath149 , @xmath150 : the transformation ( [ relads01 ] ) puts ( [ eq : ads01ds ] ) into ( [ eq : adscoflat ] ) , and similarly for ( [ eq : ads0 - 1ds ] ) . * subcase * @xmath151 , @xmath128 : @xmath152 transforms ( [ eq : ads-11ds ] ) to @xmath153 corresponding to @xmath154 * subcase * @xmath151 , @xmath132 : @xmath155 transforms ( [ eq : ads-10ds ] ) to @xmath156 corresponding to @xmath157 * subcase * @xmath151 , @xmath136 : @xmath158 transforms ( [ eq : ads-1 - 1ds ] ) to @xmath159 respectively , corresponding to the parametrizations @xmath160 z_1 = { \displaystyle \pm a\cosh { \vartheta}\cos\frac{t}{a}\sinh { \psi}}\ , , \\[8pt ] z_2 = { \displaystyle a\sinh { \vartheta}\cos{\phi}}\ , , \\[8pt ] z_3 = { \displaystyle a\sinh { \vartheta}\sin{\phi}}\ , , \\[8pt ] z_4 = { \displaystyle a\cosh { \vartheta}\sin\frac{t}{a}}\ , , \end{array } \!\right\ } \hspace{0 mm } \textrm{for}\ |q|<1\ , , \ \left . \ ! \begin{array}{l } z_0 = { \displaystyle \pm a\cosh { \vartheta}\sinh{\psi}\sinh\frac{t}{a}}\ , , \\[8pt ] z_1 = { \displaystyle \pm a\cosh { \vartheta}\sinh{\psi}\cosh\frac{t}{a}}\ , , \\[6pt ] z_2 = { \displaystyle a\sinh { \vartheta}\cos{\phi } } \\[2pt ] z_3 = { \displaystyle a\sinh { \vartheta}\sin{\phi}}\ , , \\[6pt ] z_4 = { \displaystyle \pm a\cosh { \vartheta}\cosh{\psi}}\ , , \end{array } \!\right\ } \hspace{0 mm } \textrm{for}\ |q|>1\ , .
\label{eq : ads-1 - 1aparpl}\ ] ]
* subcase * @xmath127 , @xmath132 : with @xmath163 the metric ( [ eq : ds10ds ] ) is put into the well - known exponentially expanding flat flrw form of the de sitter spacetime ( see ( 4.14 ) in @xcite ) with @xmath164 , @xmath165 : @xmath166 * subcase * @xmath127 , @xmath128 : using @xmath170 the metric ( [ eq : ads11ds ] ) for @xmath118 is transformed into @xmath171 which is simply related via @xmath172 , @xmath173 to standard form of the anti - de sitter spacetime ( see ( 5.16 ) in @xcite ) @xmath174 * subcase * @xmath151 , @xmath132 : by applying @xmath179 the metric ( [ eq : ads-10ds ] ) is put into the axially symmetric form @xmath180 which is related via @xmath181 , @xmath182 to the standard form ( [ eq : adsexp1 ] ) .
* subcase * @xmath151 , @xmath136 : with @xmath183 for @xmath184 , ( [ eq : ads-1 - 1ds ] ) is transformed into the anti - de sitter metric ( [ d4 ] ) in the global static coordinates ( for @xmath118 , the function @xmath185 is replaced by @xmath186 ) .
w. de sitter , _ over de relativiteit der traagheid : beschouingen naar aanleiding van einstein s hypothese _ , koninklijke akademie van wetenschappen te amsterdam * 25 * , 12681276 ( 1917 ) ; proc .
amsterdam * 19 * , 12171225 ( 1918 ) .
j. bik , _ selected solutions of einstein s field equations : their role in general relativity and astrophysics _ , in einstein s field equations and their physical implications , lecture notes in physics * 540 * ( springer , 2000 ) , 1126 . | in this work we present and analyze a new class of coordinate representations of de sitter and anti - de sitter spacetimes for which the metrics are diagonal and ( typically ) static and axially symmetric . contrary to the well - known forms of these fundamental geometries , that usually correspond to a @xmath0 foliation with the 3-space of a constant spatial curvature ,
the new metrics are adapted to a @xmath1 foliation , and are warped products of two 2-spaces of constant curvature . this new class of ( anti-)de sitter metrics depends on the value of cosmological constant @xmath2 and two discrete parameters @xmath3 related to the curvature of the 2-spaces .
the class admits 3 distinct subcases for @xmath4 and 8 subcases for @xmath5 .
we systematically study all these possibilities .
in particular , we explicitly present the corresponding parametrizations of the ( anti-)de sitter hyperboloid , visualize the coordinate lines and surfaces within the global conformal cylinder , investigate their mutual relations , present some closely related forms of the metrics , and give transformations to standard de sitter and anti - de sitter metrics . |
scattering of waves from a group of scatterers is rich in interesting physical phenomena .
examples are electrons or phonons scattering from defects or impurities in crystals and light scattering from conjugated molecules or biological complexes .
multiple scattering effects are nontrivial , especially when scatterers are placed close , i.e. , inside each other s effective radius @xmath4 , where @xmath5 is the total cross section of a single scatterer .
if the scatterers are resonant , their effective radius can be much larger than their physical size or force range .
a classical example is the case of weak , fixed frequency sound incident on two proximate , small identical air bubbles in water @xcite .
quantum mechanically , for two scatterers placed together well within the resonant wave length of scattered particle , extremely narrow proximity resonance can appear @xcite .
three or more proximate scatterers lead to related effects . in this paper
we have formulated a novel general method , eigenmode expasion , for quantum scattering of a system of @xmath1-wave point scatterers of any number and geometric configuration .
the eigenmode expansion is based on green function and is equivalent to solving lippmann - schwinger equation exactly taking all orders of multiple scattering into account .
it is obvious from eigenmode expansion approach that the @xmath2 matrix of such a system is of rank @xmath0 and we are able to construct the @xmath2 matrix explicitly .
many features of the spectrum ( total scattering cross section as a function of energy ) of identical proximate scatterers can be explained simply by this approach .
for example , we can naturally explain the position , width , and superradiant or subradiant nature of resonance peaks , and the unsymmetric fano lineshape of sharp proximity resonance peaks based on the high energy tail of a broad band .
we are also surprised to find off - resonant eigenmodes of _ randomly distributed _ identical proximate scatterers exhibit great symmetry and regularity .
similar systems of @xmath0 interacting resonances have been previously studied , e.g. @xcite .
though our eigenmode expansion approach is of course equivalent to standard scattering theory , it gives new physical insights , as we demonstrate below .
in addition , the quantum scattering of @xmath0 proximate @xmath1-wave scatterers is related to the superradiance problem @xcite .
again , our eigenmode expansion gives new insights to superradiance .
we start by introducing our model for @xmath1-wave point scatterers . consider a single elastic scatterer placed at @xmath6 , the asymptotic form of the total wave function is : @xmath7 the first term on the rhs is the incoming plane wave , and the second term is the scattered wave .
the angular dependent scattering amplitude @xmath8 satisfies the optical theorem : @xmath9 where the total cross section @xmath10 the @xmath2 matrix of the scatterer in the subspace of energy shell @xmath11 , @xmath12 , can be defined as an operator acting on functions of @xmath13 : @xmath14 we assume the force range of the scatterer is small compared to the wavelength of the scattered particle , so it only scatters the @xmath1-wave component ( @xmath15 ) of the incoming wave .
the now angular independent scattering amplitude is related to the phase angle @xmath16 by @xmath17 .
the @xmath2 matrix is of rank 1 .
@xmath12 has only one non-0 eigenvalue @xmath18 , whose corresponding eigenvector is the 0th order sperical harmonic .
a simple scatterer with one internal breit - wigner type resonance can be modeled as : @xmath19 the total cross section has a lorentzian shape ( dashed line in fig .
[ fig1]a ) : @xmath20 where @xmath21 and @xmath22 .
resonance corresponds to the pole of @xmath23 on complex @xmath24 plane , at @xmath25 , where the `` scattered wave '' can exist without the incoming wave . physically @xmath26 and @xmath27 correspond to position and half width of the resonance peak .
we further assume the force range of the scatterer is so small that it can be considered as existing only at one point .
in this point scatterer model , eq .
( [ fundamental ] ) holds not only asymtotically , but over all space .
since any incoming wave @xmath28 of energy @xmath21 can be written as a superposition of plane waves with @xmath11 , the total wave function can be written as : @xmath29 where the free space green function @xmath30 .
the only significant difference is the extra factor @xmath31 , the amplitude of the incoming wave at the point scatterer .
we are now ready to tackle the multiple - scatterer problem @xcite .
the total wave function for a system of @xmath0 scatterers fixed at positions @xmath32 , @xmath33 , @xmath34 , @xmath35 is : @xmath36 where @xmath37 is the scattering amplitude of the @xmath38-th scatterer , @xmath39 is the free space green function from @xmath40 : @xmath41 and for simplicity we have omitted all implicit @xmath42 ( energy ) dependence .
the @xmath31 in eq .
( [ gpoint ] ) is replaced by the amplitude at the @xmath38-th scatterer of the @xmath38-th effective incoming wave @xmath43the sum of the incoming wave and waves scattered by all scatterers except by the @xmath38-th scatterer itself : @xmath44 the above equation provides a system of linear equations for @xmath45 .
we can express them explicitly in matrix form .
define vector for incoming wave @xmath46 and vector for effective incoming waves @xmath47 as : @xmath48 , @xmath49 ( @xmath50 ) .
define the @xmath51 free space green matrix @xmath52 as : @xmath53 and matrix @xmath54 as : @xmath55 substitute @xmath56 with @xmath40 in eq .
( [ nbasic ] ) , we get @xmath57 where @xmath58 the physical implication of eq .
( [ l_s ] ) is clear .
formally @xmath59 each term @xmath60 @xmath61 means the incoming wave is scattered @xmath62 times , since @xmath54 means scattering once by one scatterer , and @xmath52 means free propagation from one scatterer to another . an integral equation similar to eq .
( [ l_s ] ) is known in general scattering theory as the lippmann - schwinger equation @xcite .
resonances occur at poles of @xmath63 on complex @xmath24 plane . the scattered wave @xmath64 depends only on @xmath46 , the amplitudes of the incoming wave at scatterers .
asymptotically @xmath65 if the incoming wave is a plane wave @xmath66 , the sum in parenthesis of eq .
( [ x1 ] ) is nothing else but the scattering amplitude @xmath67 .
for a single scatterer , for incoming wave @xmath15 ( @xmath68 ) , the scattered wave is proportional to @xmath69the outward part of the incoming wave with a phase factor .
is there any incoming wave for multiple scatterers such that the scattered wave is proportional to the outward part of the incoming wave with a phase factor ?
this is equivalent to looking for eigenvectors of the @xmath2 matrix of the whole system . in the spherically symmetric case ,
diagonalization of @xmath2 matrix is essentially spherical harmonic partial wave expansion .
some textbooks have discussed the formal diagonalization of the @xmath2 ( or @xmath70 ) matrix for a general system , and its application to scattering of particles with spin and scattering reactions .
we have found , as will be shown immediately , that for a system of @xmath1-wave point scatterers of any number and geometric configuration , the @xmath2 matrix can be analytically represented under some special basis , and can be diagonalized easily . from eq .
( [ nbasic ] ) , we see the scattered wave of multiple scatterers is a superposition of @xmath39 s , so we try to write the incoming wave as a superposition of @xmath71 s : @xmath72 or @xmath73 , where @xmath74 , and the @xmath75 spherical bessel matrix @xmath76 is defined as : @xmath77 we have @xmath78 , and from eq .
( [ phisc ] ) , we get @xmath79_i = \frac{1}{k } \sum_{i=1}^{n } g_i(\vec{r } ) [ \mathbf{tj}\vec{q_j}]_i,\ ] ] where matrix @xmath80 is defined as @xmath81 if @xmath82 is an eigenvector of @xmath83 : @xmath84 the scattered wave @xmath85 is proportional to the outward part of the incoming wave .
the angular dependence of the incoming and scattered wave is @xmath86 each @xmath82 or @xmath87 can be regarded as an _ eigenmode _ or eigenchannel of the system .
@xmath88 is the relative amplitude of the scattered wave from the @xmath38-th scatterer .
eigenvalue @xmath89 is related to the phase angle of the @xmath90-th eigenmode @xmath91 by @xmath92 eigenmodes have some basic properties that can be proved straight forwardly .
( 1 ) @xmath82 can always be chosen real due to time reversal symmetry : @xmath93 ( 2 ) the eigenmodes obey orthonormal relations upon normalization : @xmath94 ( 3 ) @xmath89 s and @xmath87 s are actually eigenvalues and eigenvectors of the @xmath2 matrix : @xmath95 it is clear now that @xmath87 s here play the same role as spherical harmonics in spherically symmetric case .
spherical harmonics are eigenchannels under special conditions , namely , when the system is spherically symmetric .
any incoming wave can be written as : @xmath96 , where @xmath97 is orthogonal to each @xmath98 .
immediately we see the @xmath97 part does not scatter at all , because each scatterer only scatters @xmath1-wave and @xmath97 is orthogonal to each @xmath98 . for a single scatterer , the @xmath2
matrix is of rank @xmath99 . for a system of @xmath0 scatterers , the @xmath2
matrix is of rank @xmath0 , since the system only scatters incoming waves belonging to the @xmath0-dimensional function space the sum in eq .
( [ x5 ] ) spans .
there is a simpler explanation for the rank of the @xmath2 matrix .
recall from section [ sec2 ] that the scattered wave @xmath64 depends only on @xmath46 , amplitudes of the incoming wave at scatterers .
for all the infinitely degenerate incoming waves of energy @xmath21 , we can have a basis in which all but @xmath0 waves have @xmath3 amplitude at all @xmath40 s .
consequently , the system only scatters the @xmath0 waves which have non-@xmath3 amplitudes at @xmath40 s , and does not scatter the rest at all . in eigenmodes @xmath87 ( plus all other linearly independent functions of @xmath13 orthogonal to all @xmath87 s ) basis , the @xmath2 matrix has a diagonal form : @xmath100 the eigenmodes are actually eigenvectors corresponding to non-@xmath3 eigenvalues of the @xmath2 matrix .
in fact , matrix @xmath80 is representation of @xmath2 matrix in the ( nonorthogonal ) @xmath101 basis ( or @xmath102 is representation of @xmath2 matrix in the @xmath103 basis ) .
this can be verified if you view eq .
( [ basis ] ) as transformation of basis .
we can use eigenmode expansion to calculate the total cross section for an incoming plane wave .
optical theorem can also be verified .
the total scattering cross section not only depends on the energy ( or @xmath42 ) , but also the direction @xmath104 of the incoming wave .
this is what we would expect for a system without spherical symmetry .
the average of total cross section over all possible incoming directions for a given energy is independent of @xmath104 .
from now on , we will use `` total cross section '' in this sense .
the total cross section @xmath105 is a sum of contributions from each eigenmode .
see ref .
@xcite . ] .
numerical evidence shows that for scatterers with the properties of eq .
( [ bw ] ) , each phase angle @xmath91 monotonically increases from @xmath3 to @xmath106 as the energy goes from @xmath107 to @xmath108 .
therefore each eigenmode will contribute a resonance peak to total cross section ( if the peaks do not overlap ) .
the position of the peak is at where @xmath109 , and the width is determined by how fast @xmath91 passes through @xmath110 and nearby region as a function of energy . for a system of @xmath0 scatterers , there will be @xmath0 resonance peaks in total cross section , whose positions and widths are consistent with the @xmath0 poles of @xmath63 on complex @xmath24 plane .
first we consider the simplest case of multiple scattering : two identical scatterers separated by distance @xmath111 .
it is easy to verify that @xmath54 , @xmath112 , and @xmath76 all have the form of @xmath113 , so does @xmath114 .
thus without knowing @xmath115 explicitly , we can immediately say the two ( unnormalized ) eigenmodes are : @xmath116 corresponding to the two scatterers oscillating perfectly in and out of phase with the same strength .
( [ 2modes ] ) is valid at any energy .
[ fig1 ] shows an example of the total cross section and phase angles as a function of energy . for @xmath117 ,
i.e. , the scatterers are placed within each other s effective radius , the `` @xmath1-like '' symmetric @xmath118 mode becomes a broad band , and the `` @xmath119-like '' antisymmetric @xmath120 mode becomes a sharp proximity resonance peak , both shifted from the peak of single scatterer , as shown in fig .
[ fig1]a @xcite . in the limit of @xmath121 ,
@xmath118 mode rigorously becomes an @xmath1-wave and @xmath120 mode a @xmath119-wave .
solid lines : ( a ) total cross section @xmath5 and ( b ) phase angles @xmath91 of each eigenmode as a function of energy of a system of two identical scatterers separated by distance @xmath122 .
each scatterer is specified by eq .
( [ bw ] ) with @xmath123 and @xmath124 . for comparison ,
the total cross section and phase angle of a single scatterer are also shown ( dashed lines ) . ]
the three scatterer case turns out to have a richer phenomenology .
we consider three identical scatterers positioned on a straight line separated by equal distance : @xmath125 .
compared to the two scatterer case , the difference is that the relative amplitudes in each @xmath82 are now energy dependent . nonetheless , upon reflection about @xmath126 plane , there are always two symmetric modes @xmath118 , @xmath127 , and one antisymmetric mode @xmath120 . fig . [ fig2]b shows the phase angles as a function of energy . they determine the positions and widths of the peaks in total cross section ( fig .
[ fig2]a ) .
the eigenmodes at @xmath118 mode s peak ( @xmath128 ) are : @xmath129 we see that @xmath118 is @xmath1-like , @xmath120 @xmath119-like , and @xmath127 @xmath111-like . indeed numerical evidence shows that at low energies ( @xmath130 ) @xmath118 , @xmath120 , and @xmath127 converge to @xmath1- , @xmath119- , and @xmath111-waves when @xmath131 .
the @xmath118 mode gives most of the broad band and the other two modes give two sharp proximity resonance peaks .
( a ) total cross section @xmath5 and ( b ) phase angles @xmath91 of each eigenmode as a function of energy of a system of three identical scatterers placed on a straight line separated by equal distance @xmath122 , at @xmath132 , @xmath133 , and @xmath134 .
each scatterer is the same as used in fig .
[ fig1 ] . in ( b ) , solid lines : @xmath135 , dotted lines : @xmath136 , dashed lines : @xmath137 . ]
if we disregard the continuity of @xmath138 , it is determined up to an integral multiple of @xmath106 . for reasons that will become clear in a moment , we have drawn fig .
[ fig2]b with restriction @xmath139 . at @xmath140 , @xmath135 and @xmath137
undergo an avoided crossing : instead of one going sharply from some @xmath141 to @xmath142 and the other remaining at @xmath3 , @xmath135 goes sharply from @xmath141 to @xmath143 and @xmath137 goes sharply from @xmath144 to @xmath142 .
this is really an avoided crossing of eigenvalues @xmath145 and @xmath146 at @xmath147 .
tracking @xmath118 and @xmath127 around the avoided crossing region , we find @xmath118 is @xmath1-like and @xmath127 is @xmath111-like before the crossing ( @xmath148 ) , yet @xmath118 becomes @xmath111-like and @xmath127 becomes @xmath1-like after the crossing ( @xmath149 ) .
in contrast , @xmath136 rigorously crosses @xmath135 and @xmath137 , since it has a different symmetry . indeed
if we remove the symmetry of the system , there is always avoided crossing whenever two phase angles tend to cross at @xmath3 ( fig .
[ fig3]b ) .
the @xmath127 mode peak exhibits a fano lineshape ( fig .
[ fig2]a ) @xcite .
the dip around @xmath150 before the peak comes from the avoided crossing at @xmath3 .
sin@xmath151 will exhibit a fano lineshape if @xmath152 goes from some @xmath141 to @xmath142 , with minimum at @xmath153 and the peak still at @xmath154 .
therefore , there is always a dip when two phase angles tend to cross at @xmath3 .
the deepness of the dip depends on how narrowly they avoid crossing .
( a ) total cross section @xmath5 and ( b ) phase angles @xmath91 of each eigenmode as a function of energy of a system of three identical scatterers placed at @xmath132 , @xmath155 , and @xmath134 with @xmath122 .
each scatterer is the same as used in fig .
[ fig1 ] . in ( b ) , solid lines : @xmath135 , dotted lines : @xmath136 , dashed lines : @xmath137 .
the difference to fig .
[ fig2 ] is that the scatterer at the center is moved so there is no longer symmetry about the reflection of @xmath156 and @xmath126 plane . ]
numerical evidence shows that for a general system of @xmath0 identical scatterers confined well within each other s effective radius , there is always a broad band with resonance energy @xmath157 and some sharp proximity resonance peaks based on the high energy tail of the broad band ( fig .
[ fig4 ] ) @xcite . at peak
, the broad band corresponds to a mode ( say , @xmath118 ) in which all scatterers oscillate in phase , i.e. , all @xmath158 s have the same sign .
the peak is broad because the waves scattered from each scatterer add up constructively . in this sense
this is an @xmath1-like mode , corresponding to the superradiant state in scattering of one photon by a collection of atoms @xcite .
the remaining @xmath159 modes give @xmath159 sharp proximity resonance peaks ( if they are separate ) .
they correspond to subradiant states .
the scattered waves of these modes at their respective peaks have little @xmath1-wave component waves scattered from each scatterer tend to add up destructively . it has been reported that under certain conditions these narrow peak eigenmodes correspond to anderson localization @xcite .
@xmath118 mode is @xmath1-like at low energies . as a consequence of avoided crossing of phase angles ,
the @xmath1-like character will shift to different eigenmodes at different energies .
each sharp peak corresponds to a phase angle to sharply increasing from some @xmath141 ( @xmath160 ) on the broad band ( corresponding to @xmath1-like mode ) to @xmath142back onto the broad band with avoided crossing at @xmath3 , exhibiting a fano lineshape .
we may not see the dip if the phase angles do not come close enough in avoided crossing , but in principle a sharp proximity resonance peak is always unsymmetric ( fig .
[ fig2]a ( @xmath127 peak only ) , [ fig3]a and [ fig4]a ) .
( a ) total cross section as a function of energy for a system of seven identical scatterers randomly placed on a plane .
each scatterer is the same as used in fig .
the positions of the scatterers are shown in ( b).,title="fig : " ] ( a ) total cross section as a function of energy for a system of seven identical scatterers randomly placed on a plane .
each scatterer is the same as used in fig .
the positions of the scatterers are shown in ( b).,title="fig : " ] it is surprising to find numerically that , even for a typical nonsymmetric system of identical scatterers , at far off - resonant energies while still maintaining @xmath161 , @xmath87 s are very much like linear combinations of spherical harmonics of same @xmath62 , i.e. , they almost have well defined angular momenta .
@xmath87 s also have lowest @xmath62 s possible , so they are still @xmath1- , @xmath119- , @xmath111- , @xmath34 like . at resonant energies ,
the eigenmodes have mixed @xmath62 s in avoided crossing region .
we have shown that for a system of @xmath0 @xmath1-wave point scatterers , we can always find @xmath0 eigenmodes .
these eigenmodes or eigenchannels play the same role as spherical harmoics for a spherically symmetric target .
the @xmath2 matrix of the system is of rank @xmath0 and can be represented analytically .
the eigenmodes are eigenvectors corresponding to non-@xmath3 eigenvalues of the @xmath2 matrix .
the position and width of resonance peaks are determined by phase angles as a function of energy .
each phase angle monotonically increases from @xmath3 to @xmath106 as energy goes from @xmath107 to @xmath108 if each scatterer has one breit - wigner type resonance . for identical scatterers placed well within one wavelength of the incoming wave ,
the far off - resonant eigenmodes are always @xmath1- , @xmath119- , @xmath111- , @xmath34 like with lowest possible angular momenta . at resonant energies where
phase angles have avoided crossings , the eigenmodes have components of different @xmath62 s mixed .
there is always one broad band originating from an @xmath1-like mode whose peak is shifted lower in energy , and @xmath159 sharp proximity resonance peaks based on the high energy tail of the broad band . the broad band peak correspond to superradiant state in scattering of light , and the sharp peaks correspond to subradiant states .
the unsymmetric fano lineshape of a sharp peak comes from a phase angle sharply increasing from some @xmath141 on the broad band to @xmath142 ( back onto the broad band ) with avoided crossing at @xmath3 . in the future
, we will investigate the generalization of eigenmode expansion to systems of multi - channel point scatterers .
we will investigate further the dependence of the spectrum ( total cross section as a function of energy ) on the spatial distribution of scatterers , and statistical properties of the spectra of collections of a large number of randomly distributed scatterers .
we need to extend the theory to non - identical scatterers , and to multiphoton emission of collections of atoms ( i.e. true superradiance ) . | we show that for a general system of @xmath0 @xmath1-wave point scatterers , there are always @xmath0 eigenmodes . these eigenmodes or eigenchannels
play the same role as spherical harmonics for a spherically symmetric target
they give a phase shift only .
in other words , the @xmath2 matrix of the system is of rank @xmath0 and the eigenmodes are eigenvectors corresponding to non-@xmath3 eigenvalues of the @xmath2 matrix
. the eigenmode expansion approach can give insight to the total scattering cross section ; the position , width , and superradiant or subradiant nature of resonance peaks ; the unsymmetric fano lineshape of sharp proximity resonance peaks based on the high energy tail of a broad band ; and other properties .
off - resonant eigenmodes for identical proximate scatterers are approximately angular momentum eigenstates . |
in the past year there has been a remarkable proposal by arkani - hamed , dimopoulos and dvali ( add ) for solving the hierarchy problem @xcite . in their vision
the scale of strong quantum gravity is lowered to the weak scale .
a weak - scale valued planck scale @xmath10 and newton s constant are reconciled by allowing gravity to propagate in @xmath11 extra _ large _ compact dimensions .
an application of gauss law quantifies the relationship between @xmath12 , @xmath10 , and the volume of the compact space @xmath13 , to be @xmath14 in the case of 2 extra dimensions and the phenomenologically preferred value @xmath15tev , @xmath16 mm and future short - distance experiments may observe a transition in newton s law from a @xmath17 to @xmath18 force law . in the case of 6 extra dimensions and @xmath19tev , @xmath20mev , which is large compared to @xmath19tev .
the non - observation of kaluza - klein ( kk ) towers for the standard model ( sm ) particles , for example , requires that the sm particles not propagate in the large extra dimensions .
this requires that they be fixed to a 3@xmath3dimensional wall , with a thickness smaller than @xmath21 .
this may be accomplished using purely field - theoretic techniques @xcite , or by using the d - branes of string theory @xcite .
this remarkable proposal allows for many new interesting phenomena to be discovered at future experiments .
these include dramatic model - independent signals at high - energy colliders from bulk graviton production @xcite , signatures of the spin-2 structure of the kk gravitons @xcite , distortions in the drell - yan energy distribution @xcite , and enhancements in the @xmath22 , @xmath23 , @xmath24 , @xmath25 production cross - sections at linear colliders @xcite , as well as other interesting collider signatures @xcite and implications for gravitational processes at high energies @xcite .
future colliders may also discover kaluza - klein and string excitations of the standard model particles @xcite .
the existence of either _ large _ or tev@xmath1 sized dimensions also opens new territory for model - building . in an important early paper @xcite by antoniadis
, the author advocates the existence of tev@xmath1 sized dimensions to break supersymmetry at low - energies .
other prospects include new ideas for obtaining gauge coupling unification @xcite , suppressing proton decay , and flavor physics@xcite .
existing phenomena already strongly constrain @xmath10 .
these include astrophysical processes ( such as sn1987a in the case @xmath26 ) @xcite , bulk graviton production @xcite , and @xmath27 production at lep2 @xcite . in this paper
the bulk contributions at the one - loop level to the anomalous magnetic moment of the muon ( mmm ) are computed .
these include contributions from bulk gravitons , bulk @xmath7 gauge bosons , and , in a two higgs doublet model , those from bulk right - handed ( rh ) neutrinos .
the correction to the mmm is found to be @xmath28 , and is therefore in a first approximation _ independent of n _ , or , in the case of bulk fermions or vector bosons , independent of the _ dimension of the subspace _ inhabited by the bulk states .
now it may seem strange to compute a one - loop correction to the mmm when a correction from a tev - scale suppressed operator may be present at tree - level .
the important point is that the tree - level operator contains an unknown coefficient , whereas the one - loop corrections are present , calculable and essentially model - independent .
they thus represent a lower value to the full bulk correction to the mmm . in this paper
it is found that the `` expected '' one - loop correction to the mmm is typically @xmath2 for values of @xmath10 from @xmath29 400 gev to @xmath30 . for comparison , this is the same magnitude as the correction expected in low - energy supersymmetry at large @xmath31 @xcite .
the bulk correction to the mmm should be compared to its extremely well - measured value of @xcite @xmath32 the bnl experiment ( e821 ) hopes to lower the error to @xmath33 @xcite , a factor of @xmath34 below the existing sensitivity .
the prediction in the sm is @xcite @xmath35 and is in very good agreement with the current measurement .
the theoretical error is dominated by the hadronic contributions .
the error in the lowest order hadronic contribution is @xmath36 ppm of @xmath37 , but there is hope that this error can be reduced to @xmath38 ppm @xcite .
another large source of theoretical error is the hadronic `` light - on - light '' scattering correction to @xmath37 .
this error is estimated to be @xmath39 ppm @xcite .
these translate into an expected theoretical error of @xmath40 .
a main result of this paper is the computation of the one - loop correction from bulk gravitons and radions to the mmm .
the corrections from a single kk graviton and kk radion are found to each be miraculously finite .
the full one - loop bulk contribution is gotten by summing over all the kk states , and results in a correction to the mmm which is essentially model - independent and independent of @xmath11 , depending only on the scale of strong gravity .
requiring that the correction is smaller than the present experimental error implies , in the case @xmath41 , for instance , that the lower limit on @xmath10 is competitive with limits gotten from other physical processes .
it is also found that as a general rule the @xmath37 constraint provides a comparable limit to @xmath10 for _ larger _ numbers of extra dimensions .
this is in contrast to astrophysical or terrestrial constraints gotten from the direct production of gravitons , where the effect of a physical process with characteristic energy scale @xmath42 typically decouples as @xmath43 .
as such , the @xmath37 constraint provides complementary information .
bulk rh neutrinos were proposed in ref .
@xcite to generate small dirac neutrino masses , and can naturally account for the required mass splitting needed to explain the atmospheric neutrino anomaly .
it is found that the correction to the mmm from the tower of rh neutrinos in a two higgs doublet model is close to the future experimental sensitivity .
it is also found that while the constraint on the @xmath7 gauge coupling is quite impressive , a @xmath7 vector boson can still mediate isotope - dependent forces @xmath44 times stronger than gravity at sub - mm distance scales .
finally , kk excitations of the sm gauge bosons are also probed by the mmm and atomic parity violation measurements .
it is found that the stronger limits on the size of extra `` small '' dimensions are gotten from the atomic parity violation experiments .
the limit is @xmath45 tev @xmath46 tev @xmath47 for kk excitations into 1 ( 2 ) extra dimensions .
the outline of the paper is as follows .
section 1 discusses some machinery necessary to perform the one - loop graviton correction to the mmm .
section 2 discusses the results and limits gotten from the one - loop graviton computation .
section 3 discusses the bulk @xmath7 and kk photon contributions to the mmm , and concludes with a discussion of the kk @xmath48 correction to atomic parity violation measurements .
section 4 discusses the correction to the mmm in a recent proposal for generating small dirac masses within the context of a two higgs - doublet model .
the effective field theory approach described by sundrum in ref .
@xcite allows one to determine the couplings of the higher dimensional gravitons to some matter stuck on a wall .
the interactions are constrained by local 4-dimensional poincare invariance ( which interestingly enough , arise from the coordinate reparameterisation invariance of the wall embedding ) and the full @xmath49dimensional general coordinate invariance , as well as any local gauge symmetries on the wall .
at lowest order in @xmath50 , all the higher dimensional gravitons interact only through the induced metric on the wall .
since the interaction of the induced metric with on - the - wall matter is fixed by @xmath51d general coordinate invariance , the lowest order couplings of the kk gravitons to the on - the - wall matter are fixed by the ordinary graviton coupling to matter , and are obviously universal .
this is important , since the interactions of matter to the kk gravitons are therefore determined by replacing @xmath52 in the ordinary gravity - matter interactions . with @xmath53 , where @xmath54 , @xmath55 , the gravitons couple to the conserved symmetric stress - energy tensor with normalisation @xmath56 there is one subtlety though . the @xmath51d decomposition of the @xmath57 dimensional graviton results in a kk tower of spin-2 gravitons , @xmath58 vector bosons , and @xmath59 spin-0 bosons .
the vector bosons correspond to fluctuations in @xmath60 . in the linearised theory they couple only to @xmath61 , which vanishes if the wall has no momentum in the extra dimensions .
the zero mode of the vector bosons acquires a mass @xmath62 from the spontaneous breaking of translation invariance in the extra dimensions @xcite .
the zero mode of one of the spin-0 bosons corresponds to fluctuations in the size of the compact dimensions ( the `` radion '' ) and couples to @xmath63 .
it acquires a mass @xmath64ev to mev from the dynamics responsible for stabilising the size of the extra dimensions @xcite .
the remaining spin-0 fields do not couple to @xmath65 @xcite .
the end result is that only the tower of kk gravitons and radions couple to the matter on the wall . since @xmath63 is proportional to the fermion masses , it is important to include the radion and its tower of kk states in computing the correction to @xmath66 . a careful computation @xcite determines the normalisation of the coupling of the radion to @xmath63 .
the correct prescription is then to replace @xmath67 for a canonical normalisation of @xmath68 , @xmath69 .
the couplings of the kk tower of gravitons and radions to matter is then gotten from the ordinary graviton coupling to matter by using the substitution in eqn .
[ expansion ] .
next i present some of the machinery which is needed to perform the one - loop bulk graviton correction to the mmm .
the reader is referred to references @xcite for a far more careful treatment of this subject , and for early quantum gravity computations .
the recent references @xcite present a clear discussion of the proper procedure for determining the bulk graviton couplings to matter .
the lagrangian coupling gravity to on - the - wall photons and fermions is @xmath70 with @xmath71 , @xmath72 is the christoffel symbol , and @xmath73 is the @xmath74 gauge fixing parameter .
the @xmath75 fields are the vielbeins satisfying @xmath76 and @xmath77 .
the fermionic covariant derivative is @xmath78 , with @xmath79 $ ] . finally , the spin connection is @xmath80 inserting @xmath81 , @xmath82 , and expanding eqn .
[ l1 ] to @xmath83 gives @xmath84 the minimal coupling of gravity to the matter stress - energy tensor , eqn .
[ tuv ] , is gotten by integrating eqn .
[ lh1 ] by parts to remove the derivatives on @xmath85 .
the propagator for a massive spin-2 particle , with mass @xmath86 , is @xmath87 the tensor structure in the numerator is @xmath88 and is uniquely fixed by several requirements @xcite .
the first is that the graviton is the excitation of the metric , so @xmath89 is symmetric in @xmath90 and @xmath91 .
next , unitarity implies that on - shell @xmath89 is positive - definite and that the spin-1 and spin-0 components are projected out . finally , the unitarity requirement @xmath92 , where @xmath93 is one of 5 polarisation tensors , fixes the normalisation of @xmath89
. it will be useful to compare the ordinary massless graviton one - loop correction to @xmath37 @xcite to the massive spin-2 one - loop correction in the limit @xmath94 .
the massless spin-2 propagator is required for the former computation . in the harmonic gauge it is @xmath95 it will be important to note that the coefficient of @xmath96 in the numerator of the graviton propagator differs between the massless and massive cases .
each kaluza - klein ( kk ) mode @xmath11 of a bulk graviton contributes to @xmath37 .
it is conveinent to express the contribution of a feynman diagram @xmath97 , with internal kk mode @xmath11 , to @xmath37 as @xmath98 the fermion mass is @xmath99 .
the total correction to @xmath37 is then gotten by summing over all the kk modes and feynman diagrams .
the final result is @xmath100 this formula can be trivially modified to include bulk fermion or vector boson corrections to the mmm .
since the coupling of matter to the kk graviton , vector boson or fermion modes is universal , the only dependence of @xmath101 on the kk quantum number @xmath11 enters through the mass of the kk mode , @xmath102 . replacing the sum in eqn .
[ tr ] with an integral gives the final result @xmath103 } \int dm^2 ( m^2)^{(n/2 - 1 ) } { \cal a}_f(m^2_f , m^2 ) \hbox { } .
\label{fr}\ ] ] this is a good approximation since the mass splitting is tiny : @xmath104 for @xmath26 , @xmath105 for @xmath41 . in evaluating the one - loop contributions to the mmm , the dominant correction will occur from those states in the loop with masses close to the cutoff .
this is a direct consequence of the large multiplicity of states at large kk masses .
the total correction to the mmm is then well - approximated by substituting into eqn .
[ fr ] the large @xmath86 limit of @xmath106 , and cutting off the ( power - divergent ) integral over @xmath107 at kk masses @xmath108 . for the cases considered in this paper , in the large @xmath86 limit @xmath109 with @xmath110 or @xmath111 . then using the relation @xmath112 , a good approximation to eqn .
[ fr ] is , for @xmath110 say , @xmath113 } \frac{m^2_f}{m^2_*}. \label{cutoffsum}\ ] ] the unknown coefficient @xmath114 is a gross parameterisation of our ignorance about what is regulating the integral , and can only be computed in a finite theory of quantum gravity , e.g. string theory .
the expectation is that @xmath115 since , for example , in string theory the amplitude @xmath106 becomes exponentially suppressed at kk masses larger than the string scale .
so whereas @xmath114 is an unknown @xmath116 coefficient , the limits on @xmath10 gotten from the bulk graviton loop scale as @xmath117 and are therefore not so sensitive to its value . finally , it is consistent to keep the factors of @xmath118 , etc . in eqn .
[ cutoffsum ] , since this factor counts the degeneracy of states , which can only increase in the full theory .
_ a. _ using the lagrangian in eqn .
[ lh1 ] the feynman rules for the graviton and fermion , photon interactions can be gotten . using this , and the massive graviton propagator , eqn .
[ massiveprop1 ] , it is straightforward although tedious to compute the one - loop correction to @xmath37
. the feynman diagrams relevant to the calculation are shown in fig .
[ extradfig1 ] .
those terms in the numerator of the graviton propagator containing two or more @xmath119s do not contribute to @xmath37 at this order in @xmath120 because of the gravitational ward identity .
i have also verified this by an explicit computation . with this in mind , naive power counting implies that the contribution of each feynman diagram and for a given kk mode @xmath11 to @xmath37 is logarithmically divergent in the ultraviolet .
the explicit calculation presented below reveals though that these divergences cancel in the sum , so that the total contribution of each kk mode to @xmath37 is remarkably finite ! a similar miraculous cancellation was observed in the ordinary ( massless ) graviton one - loop correction to @xmath37 @xcite . in the notation of eqn .
[ amplitude ] @xmath121 , the partial amplitudes @xmath101 are @xmath122 @xmath123 where @xmath124 contains the only dependence of @xmath101 on the kk quantum number .
the other functions are @xmath125 , @xmath126 , @xmath127 , @xmath128 , @xmath129 , and @xmath130 . as promised above ,
the sum over all the 5 diagrams is finite : the coefficient of @xmath131 is @xmath132 ! the final result is @xmath133 where @xmath134 and @xmath135 .
an interesting limit is @xmath136 . in this case
the last term in eqn .
[ rf ] can be neglected , and @xmath137 .
note that the heavy kk states exhibit @xmath138 .
a short comment about this will be made later .
the one - loop correction from the ordinary massless graviton was computed years ago in ref .
@xcite . a comparison of the massive spin-2 correction , eqns .
[ r1 ] , [ r2 ] , [ r3 ] , in the limit @xmath139 to their results provides a non - trivial check on the calculation .
naively the contribution of each kk bulk graviton to @xmath37 in this limit should equal the ordinary massless graviton .
this is because the coupling of matter to all the graviton kk modes is universal , including the zero mode .
this expectation is incorrect though , since the tensor structure of the massive and massless graviton propagators are different .
essentially there are more helicity states flowing around the massive graviton loop than there are in the massless graviton loop .
specifically , the coefficient of the @xmath140 term in the numerator of the graviton propagator is @xmath141 and @xmath142 , respectively , and they are not the same . also note that the @xmath140 contribution is proportional to the radion correction , discussed below .
the massless graviton correction to @xmath37 is then gotten by adding and subtracting this `` @xmath143 '' contribution evaluated at @xmath144 , weighted by either @xmath141 or @xmath142 .
that is , the ordinary graviton correction to @xmath145 should be @xmath146 where @xmath147 is the contribution to @xmath145 , evaluated at @xmath148 , of the @xmath140 term in the massive graviton propagator .
that is , @xmath149 where @xmath150 is the contribution of a kk radion , discussed below . in this limit eqns .
[ r1 ] , [ r2 ] and [ r3 ] reduce to @xmath151 , @xmath152 , and @xmath153 .
now an explicit computation gives @xmath154 .
the correction from each diagram is evidently finite ! using eqn .
[ rel ] and the @xmath148 limit of eqns .
[ r1 ] , [ r2 ] and [ r3 ] presented above , one gets @xmath155 , @xmath156 and @xmath157 .
these results are in agreement with the computation given in ref .
@xcite . _
b. _ the contribution of the tower of kk radions to @xmath37
is gotten by computing the feynman diagrams in fig .
[ extradfig1 ] , using the lagrangian in eqn .
[ lh1 ] and the expansion of the induced metric , eqn .
[ expansion ] .
it is remarkable that , before summing over the kk states , the contribution of each feynman diagram to @xmath37 is finite ! in the notation of eqn .
[ amplitude ] @xmath158 , the result is @xmath159 where @xmath160 is the coupling of the radion to the trace of the stress - energy momentum tensor , @xmath69 .
the functions are : @xmath161 , @xmath162 , and @xmath163 . in the limit @xmath164 , @xmath165 . in the limit @xmath136 , @xmath166 , and @xmath167 .
so in this limit @xmath168 .
= 0.6 [ extradfig1 ] _ c. _ to constrain @xmath10 it is useful to consider the cases @xmath169 and @xmath170 . in the former case
@xmath171 reduces to @xmath172 , and @xmath173 reduces to @xmath174 . in the @xmath170 case @xmath175 , and @xmath176
. in view of the final sum over kk states , and
the fact that the sum in either limit is @xmath116 , the ( by far ) dominant contribution to @xmath37 occurs for those kk modes with large @xmath11 , since the degeneracy of states is much greater . in this case
the integral in eqn .
[ fr ] over the kk number phase space is trivial and power divergent . applying the formula of section [ section1 ] , eqn .
[ cutoffsum ] , and @xmath69 , the total correction to @xmath37 is @xmath177 } \frac{m^2_f}{m^2 _ * } \nonumber \\ & = & \frac{7\lambda}{12 \pi } \left ( \frac{m_{\mu}}{m_*}\right)^2 \hbox { , } n=2 \hbox { , } \nonumber \\ & = & \frac{25 \pi \lambda}{288 } \left ( \frac{m_{\mu}}{m_*}\right)^2 \hbox { , } n=6 . \label{mainresult}\end{aligned}\ ] ] the first contribution is from the tower of spin-2 kk states , and the second is the contribution from the tower of kk radion states .
note that this result scales as @xmath178 , and is _ independent of n_. the presence of the factor @xmath179 $ ] from the phase space integration substantially enhances the correction .
for instance , for @xmath180 this factor is @xmath181 and compensates the loop suppression .
requiring that @xmath182 is less than the current 2@xmath183 limit of @xmath184 gives the following 2@xmath183 limits : @xmath185 note that since the limit on @xmath186 , it is not so strongly sensitive to the value of @xmath114 .
the limit for @xmath26 is considerably weaker than existing accelerator limits .
for instance , the good agreement between existing lep2 data and the predicted sm cross - section for @xmath187 final states constrains the bulk graviton production rates @xcite , and implies @xmath188 for @xmath26 .
the @xmath37 constraint for @xmath41 is competitive with the limit from bulk graviton production which for @xmath41 is @xmath189 @xcite .
the future prospects are more promising though .
the bnl e821 experiment plans to decrease the experimental error in @xmath190 by a factor of @xmath34 , i.e. , down to @xmath191 .
this should be compared to the kk graviton and radion contribution to the mmm .
inserting the relevant numbers into eqn .
[ mainresult ] gives for @xmath26 ( upper ) and @xmath41 ( lower ) : @xmath192 so an @xmath193 correction to the mmm is not unreasonable .
finally , the final formulas for the correction to mmm , and the non - decoupling of the kk states , can be understood by computing the one - loop diagram in the full theory .
recall that the correction to the mmm requires a helicity flip .
this pulls out a factor of @xmath194 , lowering the degree of divergence by two
. then naive power counting gives @xmath195 for @xmath196 .
the other factors appearing in the integrand are also easily understood .
the first factor @xmath197 is the @xmath57 dimensional phase space factor .
the factors of @xmath198 and @xmath199 are from the the graviton and muon propagators .
finally , the third factor counts the momentum factors at the graviton vertices ; for example , in fig.1(a ) the graviton fermion
fermion interaction contains @xmath200 .
this naive explanation then reproduces eqn .
[ mainresult ] .
the kk states exhibit non - decoupling since their coupling to matter is stronger at higher energies .
_ a. _ in this section i compute one - loop correction of a bulk @xmath201 gauge boson to the mmm . a @xmath202 gauge field in the bulk that couples to leptons contributes to @xmath37 by the one - loop diagram in fig .
[ extradfig2]a .
it could be a gauged @xmath7 , or the higher kk modes of the photon , for example .
as with the bulk gravitons , the bulk vector boson interactions are constrained by the measured value of @xmath37 . in this case
the limits are more model - dependent .
each kk mode couples universally to leptons with an unknown 4-dimensional gauge coupling @xmath203 .
this bulk vector boson also does not have to live in the full @xmath11-dimensional bulk .
it could instead live in a @xmath204dimensional subspace with volume @xmath205 .
it will be shown though , that those regions of parameters resulting in @xmath2 corrections to the mmm will also be probed by future short - distance force experiments . while the limit from @xmath37 implies that the gauge coupling must be microscopic ( see eqn .
[ glimit ] below ) , it is also quite natural for the @xmath51dimensional coupling to be small @xcite . in the full @xmath206 dimensional gauge theory
the gauge coupling between a vector boson and fermions is @xmath207 . confining the fermions to a @xmath208dimensional wall , by either a topological defect or just using a @xmath204dimensional delta function , and performing the usual kk decomposition of the vector boson results in a universal coupling @xmath203 of each kk mode to fermions .
it also gives @xmath209 the relation between the fine structure constants in the 4-d effective theory and the fundamental @xmath204dimensional theory . as a result , it is very natural for the vector boson to be very weakly coupled .
for example , in the case @xmath210 , @xmath211 , independent of @xmath11 . in this case @xmath212 .
this also makes it very natural for the @xmath201 boson to be extremely light .
note that even if a @xmath7 gauge boson is lighter than the proton ( as will be the case here ) , the proton remains stable since the @xmath7 gauge boson does not carry @xmath7 charge .
these facts were pointed out in ref .
@xcite , where it was argued that a @xmath7 vector boson could mediate forces at the sub - mm distance scale @xmath213 times stronger than gravity . to avoid conflicts with these short - distance force experiments ,
the mass of the zero mode of a new @xmath202 coupled to baryons must be larger than @xmath214 .
this requires that the gauge symmetry is spontaneously broken in the bulk by the vev @xmath215 of some scalar field @xmath68 .
the mass term for the gauge field in the full theory , written using 4-d canonically normalised fields , is @xmath216 inserting the vev , the usual kk expansion for @xmath217 , integrating over the @xmath204volume , and using eqn .
[ finestructure ] gives @xmath218 for the mass of the zero mode .
if instead the scalar field @xmath68 is confined to a distant 3@xmath3d wall , then in eqn .
[ gaugebosonmass ] i should replace @xmath219 and insert the approximation @xcite @xmath220 .
this results in the same expression for the mass , eqn .
[ fmass ] .
consistency of this field theoretic approach requires that the vev must satisfy @xmath221 .
this implies an upper bound of @xmath222 to the mass of the zero mode . when this is combined with the constraint on @xmath203 from @xmath37 , an interesting _ upper bound _ to the mass of the zero mode is gotten .
the masses of the other kk states are @xmath223 an evaluation of fig .
[ extradfig2]a gives the contribution of one kk state to @xmath37 .
the result is finite and @xmath224 the normalisation of the @xmath202 charge is chosen such that the @xmath201 charge of a lepton is 1
. the function @xmath225 is defined in the previous section . in the limit @xmath94 eqn .
[ g2 gb ] reduces to the famous result from schwinger @xcite .
i now want to consider performing the sum over all the kk states .
the contribution from those states @xmath226 is obviously finite . in the opposite limit @xmath170 , @xmath227 . in this case
the correction decouples as @xmath228 . since the degeneracy of the kk states increases as @xmath229 , however , the resultant sum is power divergent as @xmath229 for @xmath230 , logarithmically divergent for @xmath231 , and finite for @xmath232 . as in the case of the bulk graviton loop ,
a cutoff @xmath233 is used to regulate these divergences , and an overall factor of @xmath115 is inserted to parameterise the result of an actual one - loop computation in a more complete theory of quantum gravity .
the contribution of those modes @xmath170 to @xmath37 is then , using a trivial modification of eqn .
[ cutoffsum ] , @xmath234}\right ) \left(\frac{\bar{m}_{pl } m_f}{m^2_*}\right)^2 \left(\frac{m^p _ * v_p}{m^n _ * v_n } \right ) \hbox { } , \hbox { } p>2 \\ \label{gauge5 } & = & 2 \lambda \frac{\alpha^{(p)}_x}{3 \pi } \left(\frac{2}{p-2 } \frac{\pi^{p/2 } } { \gamma[p/2]}\right ) \left(\frac { m_f}{m_*}\right)^2 \hbox { } , \hbox { } p>2 \\ \label{gauge4 } \delta a_{\mu } & = & \lambda \frac{\alpha_x}{3 } m^2_f r^2 \ln m^2 _ * /m^2_f \hbox { } , \hbox { } p=2.\end{aligned}\ ] ] to arrive at eqn .
[ gauge2 ] , @xmath232 , i have neglected the mass of the zero mode .
also note that in this case the result is finite .
for @xmath235 the symmetry breaking contribution is irrelevant since the resulting sum is divergent . to get from eqn .
[ gauge3 ] to eqn .
[ gauge5 ] , i have used the relation @xmath236 and the relation eqn .
[ finestructure ] between the @xmath51dimensional and @xmath237dimensional fine structure constants .
these expressions for the bulk vector boson contribution to @xmath37 can also be understood by considering computing the one loop diagram in the full theory .
recall that the correction to the mmm requires a helicity flip in the loop diagram .
this pulls out a factor of @xmath238 , lowering the degree of divergence by two .
then naive power counting gives for @xmath230 @xmath239 the first two factors of @xmath198 in the integrand are for the kk vector boson and fermion propagators . substituting @xmath240
further simplifies this result to @xmath241 , which is eqn .
[ gauge5 ] .
note that this result is independent of the dimension of the subspace .
this implies that the correction to mmm is independent of @xmath205 , and depends only on @xmath10 and @xmath242 , the fine structure constant in the full theory .
the dependence on @xmath205 only enters once the correction to the mmm is expressed in terms of the @xmath51dimensional fine structure constant .
= 0.5 [ extradfig2 ] _ b. _ requiring that @xmath243 gives the following @xmath244 limits for the interesting cases @xmath245 , and @xmath246 : @xmath247 the limit in the case @xmath26 is @xmath248 below the `` natural '' value of @xmath249 for @xmath2501 tev .
i emphasize that these tiny limits apply only to the situation where the vector boson occupies the whole bulk .
the limits on @xmath251 are considerably weaker if the vector boson lives in a subspace .
that is , for other choices of @xmath252 , the limit on @xmath251 is weakened by the amount @xmath253 . the limits in eqn .
[ glimit ] translate into the following @xmath254 upper bounds on the mass of the zero mode , using @xmath255 : @xmath256 in the case that @xmath257 , the gauge boson can mediate an isotope dependent force at sub - mm distance scales . even though the gauge coupling is microscopic
, the force can still be huge compared to the gravitational force at those distance scales .
the limits in eqn .
[ glimit ] quantify an upper limit to this signal . from the upper bound on the gauge boson mass , eqn .
[ masslimit ] , it is clear that there is still an allowed region where @xmath258 .
then the ratio of the forces is , for @xmath245 to @xmath246 , @xmath259 the conclusion is that even though the @xmath260 constraint is strong in the case @xmath210 , the @xmath7 gauge boson can still mediate forces @xmath8 times stronger than gravity at sub - mm distances . next i apply these results to the scenario of @xcite , where kk excitations of the sm gauge particles and higgs bosons ( and their superpartners ) result in a power law evolution of the gauge couplings .
this allows for a unification scale as low as few decades above the tev scale . in this picture the sm gauge bosons , for example , have kk excitations in one or two extra dimensions .
these will all contribute to the mmm . in what follows however , i only concentrate on the contribution from the kk photons for the reason that their contribution ( and hence also those from kk @xmath261 s and @xmath262 s ) is rather small ( see eqn . [ mmmr ] below ) .
it is then seen that the constraint from @xmath37 provides a rather weak upper limit to @xmath263 , the size of the subspace that the gauge bosons occupy .
using eqns .
[ gauge2 ] , [ gauge4 ] , @xmath264 , @xmath265 , and the @xmath244 experimental error on the muon @xmath145 gives @xmath266 the limit for @xmath232 is finite , whereas the case @xmath231 is logarithmically sensitive to the cutoff .
since these two constraints satisfy @xmath267 , the use of eqns .
[ gauge2 ] and [ gauge4 ] is consistent .
the rather weak bound for @xmath232 is due to the @xmath268 suppression , and the constant density of kk states .
finally , kk excitations of the @xmath48 gauge boson result in a tree level correction to the theoretical predictions for atomic parity violation parameters .
the measurement of atomic parity violation therefore constrain the size of some extra `` small '' dimensions .
these experiments measure a quantity @xmath269 which is the coherent sum of the @xmath261 boson vector couplings to the nucleus .
the measured value for cesium is @xcite @xmath270 which agrees very well with the theoretical prediction @xcite @xmath271 the exchange of a tower of kk @xmath48 bosons trivially modifies ( at tree - level ) the sm prediction to . ]
@xmath272 the @xmath273 contribution to the kk mass @xmath274 has been neglected .
note that the kk correction to @xmath269 pushes the theoretical prediction farther away from the measured value . to limit @xmath263
, i require that @xmath275 , or @xmath276 . using the measured value given above , and performing the sum over kk states results in the following @xmath244 limits @xmath277 so it is quite clear that the stronger limits on tev@xmath3sized extra dimensions are gotten from atmoic parity violation rather than the mmm .
compare the results in eqn .
[ atp ] to eqn .
the authors of ref .
@xcite proposed several mechanisms for generating @xmath278 sized neutrino masses . in one of their models which is the attention of this section , small dirac
neutrino masses are generated by introducing bulk rh neutrinos .
i now briefly explain their idea .
the interaction @xmath279 gives a dirac mass @xmath280 to the lh and rh zero mode neutrino .
the higher kk states also mix with the lh neutrino , but what will eventually concern us for the mmm are the extremely heavy kk states , which have very tiny mass mixing with the lh neutrinos . in their model this interaction arises from a higher dimension operator in the full theory . more specifically , the operator @xmath281 where i have allowed the fields to propagate in @xmath282 extra dimensions , and @xmath283 is a dimensionless constant . inserting the approximation @xmath284 for the sm zero modes @xcite , and the usual kk mode expansion for @xmath285 gives the relation between @xmath286 and @xmath283 : @xmath287 where to obtain the second relation i used @xmath112 and assumed @xmath288 for simplicity .
consequently , the coupling @xmath286 can be very tiny , and neutrino masses of the correct order of magnitude required to explain the atmospheric and solar neutrino anomalies can be obtained . to obtain a correction to the mmm at the one - loop level ,
i assume : \i ) the muon neutrino obtains a dirac mass by mixing with a sterile neutrino , as in eqn .
[ rhn1 ] ; \ii ) the higgs sector is extended to include an extra higgs doublet which also acquires a vev .
the second assumption is realised in low - energy supersymmetric extensions to the sm , for example .
the point of this assumption is to obtain a trilinear coupling between the physical charged higgs , the muon , and the bulk rh neutrinos .
consequently , these interactions contribute to the mmm at one - loop , as shown in fig .
[ extradfig2]b .
an evaluation of fig .
[ extradfig2]b gives the following correction to the mmm : @xmath289 where i have denoted by @xmath286 the trilinear coupling of the physical charged higgs , charged lh fermion , and bulk rh neutrinos .
it is related to a linear combination of the trilinear couplings in the gauge basis .
the function @xmath290 is @xmath291 , where @xmath292 is the mass of the charged higgs . in the limit @xmath293 , @xmath294 as in the case of the bulk gravitons and bulk vector bosons , the dominant correction to the mmm occurs from the superheavy kk
rh neutrinos .
in fact , the sum over kk states is identical ( up to numerical factors ) to the kk sum for the vector bosons . integrating eqn .
[ rhn3 ] for @xmath230 , and using trivial modifications to eqns [ fr ] and [ cutoffsum ] , gives @xmath295 } \frac{m^2_f}{m^2 _ * } \left(\frac{\bar{m}_{pl}}{m_*}\right)^{2p / n}.\ ] ] note that since @xmath296 , the strong dependence on @xmath297 cancels , leaving @xmath298 } \frac{m^2_f}{m^2_*}.\ ] ] choosing @xmath299 , for example , gives @xmath300 so since both @xmath283 and @xmath115 , it is possible to get @xmath301
. this may be within the reach of the bnl e821 experiment .
the author would like to thank n. arkani - hamed and m. suzuki for discussions and encouraging me to complete this project .
the author also thanks t. moroi for useful communications about the bnl e821 experiment .
this work was supported in part by the director , office of energy research , office of high energy and nuclear physics , division of high energy physics of the u. s. department of energy under contract de - ac03 - 76sf00098 .
the author also thanks the natural sciences and engineering research council of canada ( nserc ) for their support . n. arkani - hamed , s. dimopoulos and g. dvali , phys.lett . * b429 * ( 1998 ) 263 . i. antoniadis , n. arkani - hamed , s. dimopoulos and g. dvali , phys.lett.*b436 * ( 1998 ) 257 ; g. shiu and s. tye , phys.rev . * d58 * ( 1998 ) 106007 .
n. arkani - hamed , s. dimopoulos and g. dvali , phys . rev . *
d59 * ( 1999 ) 086004 .
g. giudice , r. rattazzi and j. wells , nucl . phys .
* b544 * ( 1999 ) 3 .
e. mirabelli , m. perelstein and m. peskin , phys .
rev . lett.*82 * ( 1999 ) 2236 .
j. l. hewett , phys .
* 82 * ( 1999 ) 4765 .
k. agashe and n. deshpande , phys . lett .
* b456 * ( 1999 ) 60 .
p. mathews , s. raychaudhuri , k. sridhar , phys .
lett . * b450 * ( 1999 ) 343 ; p. mathews , s. raychaudhuri , k. sridhar , phys . lett .
* b455 * ( 1999 ) 115 .
s. nussinov and r. shrock , phys . rev .
* d59 * ( 1999 ) 105002 .
i. antoniadis , phys .
b246 * ( 1990 ) 377 . i. antoniadis and k. benakli , phys . lett . *
b326 * ( 1994 ) 69 ; i. antoniadis , k. benakli and m. quirs , phys . lett . * b331 * ( 1994 ) 313 . k. dienes , e. dudas , t. gherghetta , phys . lett . *
b436 * ( 1998 ) 55 ; k. dienes , e. dudas , t. gherghetta , nucl . phys .
* b537 * ( 1999 ) 47 .
z. kakushadze , nucl.phys . *
b548 * ( 1999 ) 205 .
n. arkani - hamed and s. dimopoulos , hep - ph/9811353 ; z. berezhiani and g. dvali , phys .
* b40 * ( 1999 ) 24 . n. arkani - hamed , s. dimopoulos , g. dvali and j. march - russell , hep - ph/9811448 . t. moroi , phys.rev . * d53 * ( 1996 ) 6565 , erratum - ibid . *
d56 * ( 1997 ) 4424 .
et al . _ ,
particle data group : review of particle physics , euro .
j. * c3 * ( 1998 ) 1 .
by e821 bnl collaboration ( b. l. roberts _ et al .
_ ) , published in the proceedings of the xxviii international conference on high energy physics , warsaw , poland , 1996 , river edge , nj world scientific , 1997 , p. 1035
. t. kinoshita , phys .
* 75 * ( 1995 ) 4728 ; a. czarnecki , b. krause and w. j. marciano , phys .
* 76 * ( 1996 ) 3267 .
m. hayakawa , t. kinoshita and a. i. sanda , phys .
* 75 * ( 1995 ) 790 .
private communication with takeo moroi .
r. sundrum , phys . rev .
* d59 * ( 1999 ) 085009 ; r. sundrum , phys . rev .
* d59 * ( 1999 ) 085010 .
n. arkani - hamed , s. dimopoulos and j. march - russell , hep - th/9809124 . t. han , j. lykken and r - j zhang , phys . rev .
* d59 * ( 1999 ) 105006 .
m. j. g. veltman , `` quantum theory of gravitation '' in _ methods in field theory _ , north - holland , amsterdam ( 1976 ) ; m. t. grisaru , p. van nieuwenhiuzen and c. c. wu , phys .
bf d 12 ( 1975 ) 1813 ; a. van proeyen , nucl .
* b174 * ( 1980 ) 189 ; p. van nieuwenhiuzen , phy .
rep * 68 * , 121 ( 1981 ) .
f.a . berends and r. gastmans , phys .
* b55 * , 311 ( 1975 ) . j. schwinger , phys . rev . * 73 * ( 1948 ) 416 ; j. schwinger , phys .
rev * 75 * ( 1949 ) 898 . | it has been proposed recently that the scale of strong gravity can be very close to the weak scale .
dimensions of sizes anywhere from @xmath0 mm to @xmath0tev@xmath1 can be populated by bulk gravitons , vector bosons and fermions . in this paper
the one - loop correction of these bulk particles to the muon magnetic moment ( mmm ) are investigated . in all the scenarios considered here
it is found that the natural value for the mmm is @xmath2 .
one main result is that the contribution of each kaluza@xmath3klein graviton to the mmm is remarkably finite .
the bulk graviton loop implies a limit of @xmath4 gev on the scale of strong gravity .
this could be pushed up to @xmath5 tev , even in the case of six extra dimensions , if the bnl e821 experiment reaches an expected sensitivity of @xmath6 .
limits on a bulk @xmath7 gauge boson are interesting , but still allow for forces @xmath8 times stronger than gravity at mm@xmath1 distances .
the correction of a bulk right - handed neutrino to the mmm in one recent proposal for generating small dirac neutrino masses is considered in the context of a two higgs doublet model , and is found to be close to @xmath9 .
the contributions of all these bulk particles to the mmm are ( roughly ) independent of both the total number of extra dimensions and the dimension of the subspace occupied by the bulk states .
finally , limits on the size of `` small '' compact dimensions gotten from the mmm and atomic parity violation are determined and compared .
lbnl-42812 + ucb - pth-99/03 + * extra dimensions and the muon anomalous magnetic moment * .25 in m. l. graesser _ theoretical physics group , lawrence berkeley national laboratory + and + department of physics , university of california + berkeley , california 94720 _ |
the observation of ultra - high energy cosmic rays ( uhecrs ) has now spanned over forty years . over that period ,
many source models have been proposed to explain the origin of these remarkable events . in the past five years , theoretical models
have been suggested that would potentially produce dipole distributions oriented towards m87 @xcite or centaurus a @xcite .
in addition , the akeno giant air shower array ( agasa ) has reported findings suggesting a 4% dipole - like enhancement oriented towards the galactic center present in its events with energies around @xmath1 ev @xcite .
this result seemed to be corroborated by findings published by the fly s eye experiment in 1999 that suggested the possibility of an enhancement in the galactic plane also at energies around @xmath1 ev @xcite , and also by a re - analysis of data from the sugar array that was published in 2001 @xcite that showed an enhancement in the general vicinity of the galactic center .
however , both agasa and fly s eye are subject to a limiting factor ; they are both located too far north in latitude to directly observe the galactic center itself .
the re - analysis of sugar data actually demonstrated an excess that was offset from the galactic center by @xmath2 and was more consistent with a point source than a global dipole effect @xcite .
while the current high resolution fly s eye ( hires ) experiment is subject to a similar limitation in sky coverage as the agasa and fly s eye experiments , we will show that , by properly estimating the hires aperture and angular resolution , we can effectively exclude these dipole source models to a certain degree of sensitivity .
however , we are not able to completely exclude the findings of agasa or the theoretical predictions mentioned above . additionally , over the past decade , the search for sources of ultra - high energy cosmic rays ( uhecrs ) has also begun to focus upon small scale anisotropy in event arrival directions .
this refers to statistically significant excesses occurring at the scale of @xmath3 .
the interest in this sort of anisotropy has largely been fueled by the observations of the akeno giant air shower array ( agasa ) . in 1999
@xcite and again in 2001 @xcite , the agasa collaboration reported observing what eventually became seven clusters ( six `` doublets '' and one `` triplet '' ) with estimated energies above @xmath4 ev .
several attempts that have been made to ascertain the significance of these clusters returned chance probabilities of @xmath5 @xcite to 0.08 @xcite .
by contrast , the monocular ( and stereo ) analyses that have been presented by the high resolution fly s eye ( hires ) demonstrate that the level of autocorrelation observed in our sample is completely consistent with that expected from background coincidences @xcite .
any analysis of hires monocular data needs to take into account that the angular resolution in monocular mode is highly asymmetric .
it is very difficult to compare the results of the hires monocular and agasa analyses .
they are very different in the way that they measure autocorrelation .
differences in the published energy spectra of the two experiments suggest an energy scale difference of 30% @xcite . additionally , the two experiments observe uhecrs in very different ways .
the hires experiment has an energy - dependent aperture and an exposure with a seasonal variability @xcite .
these differences make it very difficult get an intuitive grasp of what hires should see if the agasa claim of autocorrelation is justified . in order to develop this sort of intuition ,
we apply the same analysis to both agasa and hires data .
our methods for detecting the presence of a dipole source model and small - scale clustering are based upon comparisons between the real data and a large quantity of events generated by our monte carlo simulation program .
the simulated data possess the same aperture and exposure as the actual hires - i monocular data set .
we show how the asymmetric angular resolution of a monocular air fluorescence detector can be accommodated in this method .
the data set that we consider consists of events that were included in the hires - i monocular spectrum measurement @xcite .
this set contains the events observed between may 1997 and february 2003 .
there were 1526 events with reconstructed energies above @xmath6 ev and 52 events with reconstructed energies above @xmath7 ev observed during this time period .
the data set represents a cumulative exposure of @xmath8 km@xmath9sr@xmath10yr at @xmath11 ev .
this data was subject to a number of quality cuts that are detailed in the above - mentioned papers @xcite .
we previously verified that this data set was consistent with monte carlo predictions in many ways including impact parameter ( @xmath12 ) distributions @xcite and zenith angle distributions @xcite .
for this study , we presumed an average atmospheric clarity @xcite . in order to perform anisotropy analysis on this subset of data , we must first parameterize the hires - i monocular angular resolution . for a monocular air fluorescence detector ,
angular resolution consists of two components , the plane of reconstruction , that is the plane in which the shower is observed , and the angle @xmath13 within the plane of reconstruction ( see figure [ figure : picture ] ) . [ cols="^ " , ] for this sample , we obtain @xmath14}=0.99352 $ ] .
to study the relative sensitivity of agasa and hires - i , we measure the value of @xmath14}$ ] for multiple simulated sets with a variable number of doublets inserted .
we then construct an interpolation of the mean value and standard deviation of @xmath14}$ ] from a given number of observed doublets for each experiment .
this will allow us to state the number of doublets required for each experiment in order for the 90% confidence limit of @xmath14}$ ] to be above the background value of 0.99250 .
in general , for a hires - i - like data set , the 90% confidence lower limit corresponds to the mean expected background signal with the inclusion of 6.25 doublets . for an agasa - like , the 90% confidence lower limit corresponds to the mean expected background signal with the inclusion of 5.5 doublets .
this demonstrates that while agasa has a slightly better ability to perceive autocorrelation , the sensitivity of the two experiments is comparable .
however , the observed hires - i signal corresponds to the 90% confidence upper limit with the inclusion of only 3.5 doublets beyond random background coincidence .
we conclude that the hires - i monocular detector sees no evidence of clustering in its highest energy events .
furthermore , the hires - i monocular data has an intrinsic sensitivity to global autocorrelation such that we can claim at the 90% confidence level that there can be no more than 3.5 doublets above that which would be expected by background coincidence in the hires - i monocular data set above @xmath7 ev . from this result , we can then derive , with a 90% confidence level , that no more than 13% of the observed hires - i events could be sharing common arrival directions .
this data set is comparable to the sensitivity of the reported agasa data set if one assumes that there is indeed a 30% energy scale difference between the two experiments .
it should be emphasized that this conclusion pertains only to point sources of the sort claimed by the agasa collaboration .
furthermore , because a measure of autocorrelation makes no assumption of the underlying astrophysical mechanism that results in clustering phenomena , we can not claim that the hires monocular analysis and the agasa analysis are inconsistent beyond a specified confidence level .
this work is supported by us nsf grants phy 9322298 , phy 9321949 , phy 9974537 , phy 0071069 , phy 0098826 , phy 0140688 , phy 0245428 , phy 0307098 by the doe grant fg03 - 92er40732 , and by the australian research council .
we gratefully acknowledge the contributions from the technical staffs of our home institutions .
we gratefully acknowledge the contributions from the university of utah center for high performance computing .
the cooperation of colonels e. fisher and g. harter , the us army and the dugway proving ground staff is appreciated .
p. l. biermann , e. j. ahn , g. medina tanco and t. stanev , nucl .
suppl . * 87 * , 417 ( 2000 ) [ arxiv : astro - ph/0008063 ] .
g. r. farrar and t. piran , arxiv : astro - ph/0010370 . l. a. anchordoqui , h. goldberg and t. j. weiler , phys .
lett . * 87 * , 081101 ( 2001 ) [ arxiv : astro - ph/0103043 ] .
n. hayashida _ et al .
_ [ agasa collaboration ] , arxiv : astro - ph/9906056 .
d. j. bird _
[ hires collaboration ] , arxiv : astro - ph/9806096 . j. a. bellido , r. w. clay , b. r. dawson and m. johnston - hollitt , astropart .
phys . * 15 * , 167 ( 2001 ) [ arxiv : astro - ph/0009039 ] .
m. takeda _ et al .
_ , apj * 522 * , 255 ( 1999 ) [ arxiv : astro - ph/9902239 ] .
m. takeda _ et al .
_ , proc . of 27th icrc ( hamburg ) ,
* 1 * , 337 ( 2001 )
. p. g. tinyakov and i. i. tkachev , jetp lett .
* 74 * , 1 ( 2001 ) [ pisma zh .
* 74 * , 3 ( 2001 ) ] [ arxiv : astro - ph/0102101 ] . c. b. finley and s. westerhoff , astropart
* 21 * , 359 ( 2004 ) .
[ arxiv : astro - ph/0309159 ] .
r. u. abbasi _ et al . _
[ high resolution fly s eye collaboration ] , phys .
lett . * 92 * , 151101 ( 2004 ) [ arxiv : astro - ph/0208243 ] .
m. takeda _ et al .
_ , phys .
lett . * 81 * , 1163 ( 1998 ) [ arxiv : astro - ph/9807193 ] .
t. abu - zayyad _ et al . _
[ high resolution fly s eye collaboration ] , submitted for publication in _ astroparticle physics _ [ arxiv : astro - ph/0208301 ] . | the high resolution fly s eye hires - i detector has now been in operation in monocular mode for over six years . during that time span ,
hires - i has accumulated a larger exposure to ultra - high energy cosmic rays ( uhecrs ) above @xmath0 ev than any other experiment built to date .
this presents an unprecedented opportunity to search for anisotropy in the arrival directions of uhecrs .
we present results of a search for dipole distributions oriented towards major astrophysical landmarks and a search for small - scale clustering .
we conclude that the hires - i data set is , in fact , consistent with an isotropic source model . |
nonequilibrium surface growth and interface dynamics represent an area of research that has received much attention in the last two decades @xcite .
a large number of discrete atomistic growth models @xcite and stochastic growth equations @xcite have been found @xcite to exhibit _ generic scale invariance _ characterized by power law behavior of several quantities of interest , such as the interface width as a function of time ( measured in units of deposited layers ) and space- and time
dependent correlation functions of the interface height .
much effort has been devoted to the classification of growth models and equations into different universality classes characterized by the values of the exponents that describe the dynamic scaling behavior implied by these power laws .
a variety of experimental studies @xcite have confirmed the occurrence of dynamic scaling in nonequilibrium epitaxial growth . among the various experimental methods of surface growth , molecular beam epitaxy ( mbe )
is especially important because it plays a crucial role in the fabrication of smooth semiconductor films required in technological applications . under usual mbe growth conditions ,
desorption from the film surface is negligible and the formation of bulk vacancies and overhangs is strongly suppressed .
it is generally believed that nonequilibrium surface growth under these conditions is well described by a _ conserved _ nonlinear langevin type equation @xcite and related atomistic models @xcite that form the so called `` mbe universality class '' .
surface growth is an example of a general class of problems involving the dynamics of non markovian , spatially extended , stochastic systems . in recent years ,
the concept of _ persistence _ @xcite has proven to be very useful in analyzing the dynamical behavior of such systems @xcite . loosely speaking , a stochastic variable is _ persistent _ if it has a tendency to maintain its initial characteristics over a long period of time .
the _ persistence probability _ @xmath4 is typically defined as the probability that a characteristic feature ( e.g. the sign ) of a stochastic variable does not change at all over a certain period of time @xmath5 .
although the mathematical concept of persistence was introduced a long time ago in the context of the `` zero crossing problem '' in gaussian stationary processes @xcite , it is only very recently that this concept has received attention in describing the statistics of first passage events in a variety of spatially extended nonequilibrium systems .
examples of such applications of the concept of persistence range from the fundamental classical diffusion equation @xcite to the zero temperature glauber dynamics of the ferromagnetic ising and @xmath6state potts models @xcite and phase ordering kinetics @xcite .
recently , a generalization of the persistence concept ( probability of persistent large deviations ) has been introduced @xcite . a closely related idea , that of sign time distribution , was developed in ref . @xcite .
an increasing number of experimental results are also available for persistence in systems such as coalescence of droplets @xcite , coarsening of two dimensional soap froth @xcite , twisted nematic liquid crystal @xcite , and nuclear spin distribution in laser polarized @xmath7 gas @xcite .
recent work of krug and collaborators @xcite has extended the persistence concept to the first
passage statistics of fluctuating interfaces .
persistence in the dynamics of fluctuating interfaces is of crucial importance in ultra small scale solid state devices .
as the technology advances into the nanometric regime , questions such as how long a particular perturbation that appears in an evolving interface persists in time and what is the average time required for a structure to first fluctuate into an unstable configuration become important
. the persistence probability can provide quantitative predictions on such questions .
recent experiments @xcite have demonstrated the usefulness of the concept of persistence in the characterization of the equilibrium fluctuations of steps on a vicinal surface .
analysis of experimental data on step fluctuations on al / si(111 ) @xcite and ag(111 ) @xcite surfaces has shown that the long time behavior of the persistence probability and the probability of persistent large deviations in these systems agrees quantitatively with the corresponding theoretical predictions .
these results show that the persistence probability and related quantities are particularly relevant for describing and understanding the long time dynamics of interface fluctuations . in the context of surface growth and fluctuations
, the persistence probability @xmath8 may be defined as the probability that starting from an initial time @xmath9 , the interfacial height @xmath10 at spatial position @xmath11 does not return to its original value at any point in the time interval between @xmath12 and @xmath13 .
this probability is clearly the sum of the probabilities of the height @xmath10 always remaining above ( the _ positive _ persistence probability @xmath14 ) and always remaining below ( the _ negative _ persistence probability @xmath15 ) its specific initial value @xmath16 for all @xmath17 .
this concept quantifies the tendency of a stochastic field ( in our case the interface height ) to persistently conserve a specific feature ( the sign of the interfacial height fluctuations ) .
the persistence probability @xmath8 would , in general , depend on both @xmath12 and @xmath5 .
in the early stage of the growth process starting from a flat interface ( transient regime ) , the interface gradually develops dynamical roughness @xcite due to the effect of fluctuations in the beam intensity . in this regime ,
the choice of the initial time @xmath12 is clearly important : it determines the degree of roughness of the configuration from which the interface evolves . at long times
, the growing interface enters into a new evolution stage , called the steady state regime , characterized by fully developed roughness that does not increase further in time .
in this regime , the choice of @xmath12 is expected to be unimportant . the work of krug _ et al . _
@xcite shows that for a class of _ linear _ langevin type equations for surface growth and atomistic models belonging in the same dynamical universality class as these equations , the persistence probability decays as a power law in time for long times in both transient and steady state regimes .
these power laws define the positive and negative persistence exponents , @xmath18 and @xmath19 , for positive and negative persistence in the transient and steady state regimes , respectively .
the @xmath20 symmetry of the linear growth equations implies that @xmath21 and @xmath22 in these systems . in ref .
, it was pointed out that the persistence exponent in the steady state of these linear models is related to the dynamic scaling exponent @xmath23 , that describes the growth of the interface width @xmath24 as a function of time @xmath5 in the transient regime ( @xmath25 ) , through the relation @xmath26 .
the validity of this relation was confirmed by numerical simulations .
since the exponent @xmath23 is the same for all models in the same dynamical universality class , this result implies that the persistence exponent in the steady state regime of these _ linear _ models is also universal .
numerical results for the persistence exponent in the transient regime , for which no analytic predictions are available , also indicate a similar universality .
kallabis and krug @xcite carried out a similar calculation for ( 1 + 1)dimensional kardar
zhang ( kpz ) @xcite interfaces .
they found that the nonlinearity in the kpz equation that breaks the @xmath20 symmetry is reflected in _
different _ values of the positive and negative persistence exponents , @xmath27 and @xmath28 , in the transient regime .
the values of the steady state persistence exponents @xmath29 and @xmath30 were found to be equal to each other , and equal to @xmath31 within the accuracy of the numerical results . this
is expected because the @xmath20 symmetry is dynamically restored in the steady state of the ( 1 + 1)dimensional kpz equation .
this is , however , a specific feature of the ( 1 + 1)dimensional kpz model , which for nongeneric reasons , turns out to be up down symmetric in the steady state .
nonlinear surface growth models ( e.g. the higher dimensional kpz model , the nonlinear mbe growth model ) are generically expected to have different values of @xmath32 in both transient and steady state regimes . in this paper , we present the results of a detailed numerical study of the persistence behavior of several atomistic , solid on solid ( sos ) models of surface growth in ( 1 + 1 ) and ( 2 + 1 ) dimensions .
while we concentrate on models in the mbe universality class , results for a few other models , some of which have been studied in refs . and are also presented for completeness .
the highly non trivial nature of the persistence probability , in spite of a deceptive simplicity of the defining concept , arises from the complex temporal non locality ( `` memory '' ) inherent in its definition .
in fact , there are very few stochastic problems where an analytical solution for the persistence probability has been achieved . these include the classical brownian motion @xcite , the random acceleration problem @xcite and the one dimensional ising and @xmath6-state potts models @xcite .
in general , the highly nonlocal nature of the temporal correlations in a non markovian stochastic process makes it extremely difficult to obtain exact results for the persistence probability even for seemingly simple stochastic processes . even for the simple diffusion equation , the persistence exponent is known only numerically , or within an independent interval approximation @xcite or series expansion approach @xcite .
however , it is fairly straightforward in most cases to directly simulate the persistence probability to obtain its stationary power law behavior at large times , and thus to numerically obtain the approximate value of the persistence exponent .
for this reason , we use stochastic ( monte carlo ) simulations of the atomistic growth models to study their temporal persistence behavior in the transient and steady state regimes .
these models are defined in terms of random deposition and specific cellular
automaton type local diffusion or relaxation rules .
some of these models are of the `` limited mobility '' type in the sense that the surface diffusion rules or local restrictions limit the characteristic length over which a deposited particle can diffuse to just one or a few lattice spacings .
the models in the mbe universality class considered in our study are : the das sarma tamborenea model @xcite , the wolf
villain model @xcite , the kim
das sarma model @xcite and its `` controlled '' version @xcite , and the restricted solid on solid ( rsos ) model of kim _ at al . _
we also present results for the family model @xcite that is known to belong to the edwards
wilkinson @xcite universality class and the restricted solid on solid ( rsos ) model of kim and kosterlitz @xcite that is in the kpz universality class .
the main objective of our study is to examine the effects of the nonlinearity in the mbe growth equation @xcite on the persistence behavior . unlike the ( 1 + 1)dimensional kpz equation
, the nonlinearity in the mbe growth equation persists in the steady state in the sense that the height profile exhibits a clear asymmetry between the positive and negative directions ( above and below the average height ) .
therefore , the positive and negative persistence exponents @xmath29 and @xmath30 are expected to have different values in these models . if this is the case , then the relation between the steady state persistence exponent and the dynamic scaling exponent @xmath23 found in linear models can not be valid for both @xmath29 and @xmath30 , indicating that at least one of these exponents is a new , nontrivial one not related to the usual dynamic scaling exponents .
the values of @xmath29 and @xmath30 and their relation to @xmath23 , as well as the values of the transient persistence exponents @xmath27 and @xmath28 are the primary questions addressed in our study .
we also investigate the universality of these exponents by measuring them for several models that are known to belong in the same universality class as far as their dynamic scaling behavior is concerned . to obtain accurate values of the exponents ,
the `` noise reduction '' technique @xcite is employed in some of the simulations of ( 1 + 1)dimensional models .
we also address some questions related to the methodology of calculating persistence exponents from simulations . since the value of the dynamical exponent @xmath33 is relatively large for models in the mbe universality class , the time required for reaching the steady state grows quickly as the sample size @xmath34
is increased ( @xmath35 ) . as a result
, it is difficult to reach the steady state in simulations for large @xmath34 .
it is , therefore , useful to find out whether the value of the steady state persistence exponents can be extracted from calculations of @xmath8 with @xmath36 .
another issue in this context involves the effects of the finiteness of the sample size @xmath34 and the sampling time @xmath37 ( the time interval between two successive measurements of the height profile ) on the calculated persistence probability .
an understanding of these effects is needed for extracting reliable values of the persistence exponents from simulations that _ always _ involve finite values of @xmath34 and @xmath38 .
understanding the effects of @xmath34 and @xmath37 on the persistence analysis is not only important for our simulations , but is also important in the experimental measurements of persistence which invariably involve finite system size and sampling time .
the main results of our study are as follows .
we find that the positive and negative steady state persistence exponents for growth models in the mbe universality class are indeed different from each other , reflecting the asymmetry of the interface arising from the presence of nonlinearities in the underlying growth equation .
our results for these exponent values are : @xmath0 and @xmath1 , respectively , in ( 1 + 1 ) dimensions ; @xmath2 and @xmath3 in ( 2 + 1 ) dimensions .
the values of the positive and negative persistence exponents for different models are clearly correlated with the asymmetry of the `` above '' and below ( defined relative to the mean interface height ) portions of the interface .
we show analytically that the _ smaller _ one of the two steady - state persistence exponents should be equal to @xmath39 . thus , the relation @xmath40 derived in ref .
@xcite for linear surface growth models is expected to be satisfied by @xmath41 for the nonlinear models considered here .
our numerical results are consistent with this expectation : we find that the positive persistence exponent is indeed close to @xmath39 , while the negative one is significantly higher .
similar asymmetry is found for the persistence exponents in the transient regime with @xmath42 in mbe growth . within the uncertainties in the numerically determined values of the exponents , they are universal in the sense that different models in the same dynamic universality class yield very similar values for these exponents . for the models in the edwards wilkinson and kpz universality classes ,
we find results in agreement with those of earlier studies @xcite .
our simulations also reveal that a measurement of the steady state persistence exponents is possible from simulations in which the initial time @xmath12 is much smaller than the time ( @xmath43 ) required for the interfacial roughness to saturate .
a similar result was reported in ref . where it was found that the steady state persistence exponent may be obtained from a calculation of @xmath8 with @xmath44 .
we find that the restriction @xmath45 is not necessary for seeing a power law behavior of @xmath46 a power law with the steady state exponents is found even if @xmath5 is close to or somewhat larger than @xmath12 .
we exploit this finding in some of our persistence simulations for ( 2 + 1)dimensional growth models which are more relevant to experiments .
these results , however , also imply that it would be extremely difficult to measure the transient persistence exponents from real surface growth experiments . finally , we show that the dependence of the steady state persistence probability on the sample size @xmath34 and the sampling time @xmath37 is described by a simple scaling function of the variables @xmath47 and @xmath48 .
this scaling description is similar to that found recently @xcite for a different `` persistence probability '' , the survival probability , that measures the probability of the height not returning to its _ average value _
( rather than the initial value ) over a certain period of time .
although the `` persistence '' and the `` survival '' @xcite probability seem to be qualitatively similar in their definitions , the two are mathematically quite unrelated , and in fact , no exponent can be defined for the survival probability . in this paper
we only discuss the persistence probability and the persistence exponent for surface growth processes .
the rest of the paper is organized as follows : in sec . [
a ] , we briefly discuss the main universality classes and their corresponding dynamic equations and scaling exponents relevant for surface growth phenomena .
section [ b ] contains a short overview of the persistence probability concept from the interface fluctuations perspective .
the discrete stochastic sos growth models considered in our study are described in sec .
[ models ] .
in addition , we briefly describe in this section the noise reduction technique which is employed in some of our simulations .
section [ sim ] contains a detailed description of our main results : in sec .
[ sim1 ] , our ( 1 + 1)dimensional simulation results for the transient and steady state persistence exponents are presented , focusing mostly on models described by the nonlinear mbe dynamical equation .
section [ analytic ] contains the analytic derivation of a relation between the smaller steady - state persistence exponent and the dynamic growth exponent . in sec .
[ sim - steady ] , we introduce an alternative approach for measuring the steady state persistence exponents , using a relatively short `` equilibration time '' that is much shorter than the time required for reaching the true steady state .
section [ sim2 ] contains the results of our ( 2 + 1)dimensional persistence calculation for a selection of linear and nonlinear models .
simulation results that establish a scaling form of the dependence of the persistence probability on the sample size and the sampling time are presented in sec .
[ fss ] . the final sec .
[ conclude ] contains a summary of our main results and a few concluding remarks .
the dynamic scaling behavior of stochastic growth equations may be classified into several universality classes .
each universality class is characterized by a set of scaling exponents @xcite which depend on the dimensionality of the problem .
these exponents are @xmath49 , where @xmath50 is the roughness exponent describing the dependence of the amplitude of height fluctuations in the steady state regime ( @xmath51 ) on the sample size @xmath34 , @xmath23 is the growth exponent that describes the initial power law growth of the interface width in the transient regime ( @xmath52 ) , and @xmath33 is the dynamical exponent related to the system size dependence of the time at which the interface width reaches saturation .
note that @xmath53 for all the models considered in this paper . to describe the interface evolution we use the single
valued function , @xmath54 , which represents the height of the growing sample at position @xmath55 and deposition time @xmath5 .
the interfacial height fluctuations are described by the root
mean squared height deviation ( or interface width ) which is a function of the substrate size @xmath34 and deposition time @xmath5 : @xmath56 where @xmath57 is the average sample thickness .
the width @xmath58 scales as @xmath59 for @xmath60 and @xmath61 for @xmath51 @xcite , @xmath62 being the equilibration time of the interface , when its stationary roughness is fully developed . since it is convenient to write the evolution equations in terms of the deviation of the height from its spatial average value , @xmath63 , from now on we will denote by @xmath54 the interface height fluctuation measured from the average height .
extensive studies of dynamic scaling in kinetic surface roughening ( for an extended review see ref .
@xcite ) have revealed the existence of ( at least ) four universality classes that are described , in the long wavelength limit , by the following continuum equations and sets of scaling exponents ( @xmath64 ) , shown for the 1 + 1(2 + 1)dimensional cases , respectively : \(1 ) the edwards wilkinson ( ew ) second order linear equation : @xmath65 @xmath66 @xmath67 ( 2 ) the kpz second order nonlinear equation : @xmath68 @xmath69 @xmath70 @xmath71 @xmath72 and \(4 ) the mbe fourth order nonlinear equation : @xmath73 @xmath74 @xmath75 where @xmath76 ( i=2 , 4 ) and @xmath77 ( j=2 , 22 ) are constant .
the quantity @xmath78 represents the noise term which accounts for the random fluctuations in the deposition rate .
we assume that the noise has gaussian distribution with zero mean and correlator @xmath79 d being a constant related to the strength of the bare noise .
note that we do not include the ( trivial ) constant external deposition flux term in the continuum growth equations since that is easily eliminated by assuming that the height fluctuation @xmath80 is always measured with respect to the average interface which is growing at a constant rate .
the concepts of universality classes and scaling exponents have been widely used in the literature to analyze the kinetics of surface growth and fluctuations . our study based on persistence probabilities
is motivated by the possibility that the concept of persistence may provide an additional ( and complementary ) tool to analyze the surface growth kinetics .
it addresses fundamental questions such as : is persistence an independent ( and new ) conceptual tool for studying surface fluctuations or essentially equivalent ( or perhaps complementary ) to dynamic scaling ? and does persistence lead to the definition of new universality classes on the basis of the values of the persistence exponent ? to answer these questions , we consider , for each of the four universality classes mentioned above ( i.e. eqs .
( [ ew_eq])([mbe_eq ] ) ) , at least one growth model and investigate how the associated persistence exponents are related to the dynamic scaling exponents mentioned above .
our goal is to calculate the positive and negative persistence probabilities ( @xmath81 ) for a growing ( fluctuating ) interface in the transient and steady state regimes . here
@xmath9 is the initial time , and we are interested in evaluating the probability of the height at a fixed position remaining persistently above ( @xmath14 ) or below ( @xmath15 ) its initial value ( i.e. its value at @xmath12 by definition ) during the time period between @xmath12 and @xmath13 . if one considers the special case @xmath82 , when the interface is completely flat , then the quantity of interest is the probability that the interfacial height ( measured from its spatial average ) does not return to its initial zero value up to time @xmath5 .
this case is known as the _ transient _ ( t ) regime .
for values of @xmath5 that are small compared to the time scale for saturation of the interface width ( @xmath83 ) , the persistence probabilities in this regime are expected to exhibit a power law decay in time : @xmath84 where @xmath85 are called the transient positive and negative persistence exponents . in the particular case of _ linear _ continuum growth equations , these exponents are equal because the symmetry under a change of sign of @xmath86 remains valid at all stages of the growth process .
however , in the case of dynamics governed by nonlinear continuum equations , the lack of this `` up down ''
interfacial symmetry implies that @xmath14 and @xmath15 ( and therefore , the exponents @xmath27 and @xmath28 ) would , in general , be different from each other .
no universal relationship between the transient positive and negative persistence exponents and the dynamic scaling exponents is known to exist for any one of the four universality classes mentioned above .
on the other hand , if one considers @xmath9 larger than @xmath87 , then the quantity of interest is the probability that the interfacial height at a fixed position does not return to its specific value at initial time @xmath12 during the subsequent time interval between @xmath12 and @xmath13 . instead of being flat ,
the interface morphology at time @xmath9 has completely developed roughness , which produces persistence exponents that are different from the transient exponents defined earlier .
this case is known as the _ steady state _ ( s ) regime . if @xmath88 , one expects to obtain in this regime the steady state persistence probability with a power law decay in time @xcite
@xmath89 where @xmath90 are the steady state positive and negative persistence exponents .
it has been pointed out by krug _
_ @xcite that for systems described by _ linear _ langevin equation , the steady state persistence exponents are related to the dynamic scaling exponent @xmath23 in the following way : @xmath91 the exponent @xmath23 is well known for linear langevin equations for surface growth dynamics , and is given in @xmath92dimensions by @xmath93 for nonconserved white noise ( eq . ( [ noise ] ) ) , where @xmath33 , the dynamical exponent , is here precisely equal to the power of the gradient operator entering the linear continuum dynamical growth equation [ i.e. @xmath94 in eq . ( [ ew_eq ] ) ; @xmath95 in eq . ( [ mh_eq ] ) ] . the relation defined by eq .
( [ theta_s ] ) holds true for the langevin equations of eqs .
( [ ew_eq ] ) and ( [ mh_eq ] ) , which are obviously linear , as well as for the special case of the ( 1 + 1)dimensional kpz equation of eq .
( [ kpz_eq])@xcite , which , despite its nonlinearity , behaves as the linear ew equation in the steady state . since the positive and negative exponents are expected to be different for general nonlinear langevin equations , the relation of eq .
( [ theta_s ] ) can not be valid for both @xmath29 and @xmath30 in systems described by such nonlinear equations .
therefore , at least one ( or perhaps both ) of these two persistence exponents must be non trivial in the sense that it is not related to the usual dynamic scaling exponents . for
this reason we pay particular attention to the mbe nonlinear equation and investigate whether its persistence exponents can be related to the dynamic scaling exponents .
in this paper , we use different atomistic limited mobility growth models for simulating surface growth processes . in these models ,
the substrate consists of a collection of lattice sites labeled by the index @xmath96 ( @xmath97 ) and the height variables @xmath98 take integral values .
the term `` limited mobility '' is meant to imply that in these models , each adatom is characterized by a finite diffusion length which is taken to be one lattice spacing in most of the models we consider here .
thus , a deposited atom can explore only a few neighboring lattice sites according to a set of specific mobility rules before being incorporated into the growing film .
the solid on solid constraint is imposed in all these models , so that defects such as overhangs and bulk vacancies are not allowed . in
most of the models considered in this work , the possibility of desorption is neglected , thereby making the models `` conserved '' in the sense that all deposited atoms are incorporated in the film ; the noise ( given by eq .
( [ noise ] ) ) is of course nonconserved since the system is open to the deposition flux .
the deposition process is described by a few simple rules in these models .
an atomic beam drops atoms on the substrate in a random manner .
once a lattice site on the substrate is randomly chosen , the diffusion rules of the model are applied to the atom dropped at the chosen site to determine where it should be incorporated .
the allocated site is then instantaneously filled by the adatom .
we consider both ( 1 + 1) and ( 2 + 1) dimensional models ( one or two spatial dimensions and one temporal dimension ) defined on substrates of length @xmath34 in units of the lattice spacing .
the deposition rate is taken to be constant and equal to @xmath99 particles per unit time in our simulations of the family ( f ) , larger curvature ( lc ) , das sarma tamborenea ( dt ) , wolf villain ( wv ) and controlled kim
das sarma ( ckd ) models ( see below ) . in these simulations ,
one complete layer is grown in each unit of time . in the rsos kim
kosterlitz ( kk ) and kim park kim ( kpk ) models described below , the diffusion rules are replaced by a set of local restrictions on nearest neighbors height differences , which have to be satisfied after the deposition .
the randomly chosen deposition site is rejected ( the atom is not deposited ) if these restrictions are not satisfied . as a consequence ,
the number of deposition attempts does not coincide with the number of successful depositions in the kk model , although they are linearly related .
all conserved growth models satisfy the conservation law @xmath100 where @xmath101 is the surface current and @xmath102 is the noise term . using different expressions , dictated primarily by symmetry considerations , for the current @xmath101 , one can obtain all the conserved langevin equations discussed in sec . [
the atomistic growth models considered in our work provide discrete realizations of these continuum growth equations .
it is known that some of the discrete growth models we study here have complicated transient behavior @xcite .
for this reason , obtaining the dynamic scaling exponents that show the true universality classes of these models is often quite difficult . to make this task easier , the noise reduction technique @xcite was introduced in simulations of such models .
it has been shown @xcite that this technique helps in suppressing high steps in the models and reduces the corrections in the scaling behavior , so that the true asymptotic universality classes of the growth models can be seen in simulations that cover a relatively short time .
this makes it interesting to examine whether the persistence probabilities in these discrete models also exhibit similar transient behavior , and whether the noise reduction technique can help in bringing out the true persistence exponents of these models . to investigate this ,
we have applied the noise reduction technique to some of the discrete models studied in this paper .
the noise reduction technique can be easily incorporated in the simulation of any discrete growth model by a small modification in the diffusion process @xcite . when an atom is dropped randomly ,
the regular diffusion rules for the growth model are applied and the final allocated site is chosen . instead of adding the atom at that final site ,
a counter at that site is increased by one but the height of that site remains unchanged .
when the counter of a lattice site increases to the value of a pre determined noise reduction factor , denoted by @xmath103 , the height at that lattice site is increased by one and the counter of that site is reset back to zero . the value of the noise reduction factor @xmath103 should be chosen carefully . if @xmath103 is too small , the suppression of the noise effect is not enough and the true universality class is not seen .
however , is @xmath103 is too large , the kinetically rough growth becomes layer by layer growth @xcite and the universality class of the model can not be determined .
the atomistic models considered in our work are defined below .
( i ) * family model * : the family ( f ) model @xcite is an extensively studied sos discrete stochastic model , rigorously known to belong to the same dynamical universality class as the ew equation .
it allows the adatom to explore within a fixed diffusion length to find the lattice site with the smallest height where it gets incorporated .
if the diffusion length is one lattice constant ( this is the value used in our simulations ) , the application of this deposition rule to a randomly selected site @xmath96 involves finding the local minimum height value among the set : @xmath104 and @xmath105 ( in ( 1 + 1)dimensions ) .
the height of the site with the minimum height is then increased by one .
\(ii ) * larger curvature model * : the kim
das sarma model @xcite is a more complex one which allows the atomic surface current @xmath101 to be written as a gradient of a scalar field @xmath106 , @xmath107 , which can depend on @xmath80 , @xmath108 , @xmath109 and so on . in the particular case
when @xmath110 , one obtains the so called larger curvature ( lc ) model .
as the name suggests , the diffusion rules applied to a randomly selected site @xmath96 allow the adatom to get incorporated at the site in the neighborhood of site @xmath96 where the local curvature ( given by @xmath111 in ( 1 + 1)dimensions ) has the largest value .
the lc model asymptotically rigorously belongs to mh universality class described by eq .
( [ mh_eq ] ) .
\(iii ) * wolf villain model * : the diffusion rules of the wolf
villain ( wv ) model @xcite allow the adatom to diffuse to its neighboring sites in order to maximize its local coordination number which , for the ( 1 + 1)dimensional case , varies between 1 and 3 when the bond with the atom lying below the site under consideration is taken into account .
in contrast to the f model , in this case the surface develops deep valleys with high steps almost perpendicular to the substrate . for the range of times and sample sizes used in the present study ,
the wv model may be considered to belong to the mbe universality class @xcite described by eq.([mbe_eq ] ) .
however , recent studies @xcite have shown that the asymptotic universality class of this model in ( 1 + 1)dimensions is the same as that of the ew equation .
in contrast , in ( 2 + 1)dimensions , studies based on the noise reduction technique @xcite have revealed that the wv model exhibits at very long times unstable ( mounded ) dynamic universality which can not really be described by any of the continuum equations ( eqs .
( [ ew_eq])-([mh_eq ] ) ) given above .
\(iv ) * das sarma tamborenea model * : the das sarma tamborenea ( dt ) model @xcite is characterized by diffusion rules that are slightly different from those in the wv model . in this case , the diffusing atom tries to _ increase _ its coordination number , not necessarily to _ maximize _ it .
for example , if a randomly selected deposition site has its local coordination number equal to 1 ( i.e. no lateral neighbor in ( 1 + 1)dimensions ) , and the two neighbors of this site have coordination numbers equal to 2 and 3 , the deposited atom does not necessarily move to the neighboring site with the larger local coordination number : it moves to one of the two neighboring sites with equal probability ( the atom would necessarily move to the site with coordination number 3 in the wv model ) .
this minor change in the local diffusion rules actually changes the asymptotic universality class : the ( 1 + 1)dimensional dt model belongs to the mbe universality class @xcite corresponding to the nonlinear continuum dynamical equation of eq .
( [ mbe_eq ] ) . however , the ( 2 + 1)dimensional dt model asymptotically belongs to the ew universality @xcite at very long times .
\(v ) * controlled kim
das sarma model * : the kim
das sarma model mentioned above provides a discrete realization of the continuum equation of eq .
( [ mbe_eq ] ) if the scalar field @xmath106 is chosen to be @xmath112 .
however , the discrete treatment of the spatial gradients produces strong instabilities in the growth process due to uncontrolled growth of isolated structures , such as pillars or grooves .
these instabilities can be easily controlled by introducing higher order nonlinear terms @xcite .
we call this new model the _ controlled _ kim das sarma ( ckd ) model . in this model ,
the scalar field @xmath106 is chosen to be @xmath113 , where the nonlinear function @xmath114 is given by @xmath115 with @xmath116 being the control parameter .
the ckd diffusion rules for a randomly chosen deposition site , @xmath96 , imply the minimization of the scalar field @xmath106 , using the standard discretization scheme for the lattice derivatives @xmath117 and @xmath118 : @xmath119 @xmath120 and @xmath121 @xcite , one can remove the nonlinear growth instabilities completely and ensure an overall behavior of the ckd model similar to that of the dt model .
\(vi ) * kim kosterlitz and kim park kim models * : for completeness , we also present in this paper the results for the rsos kim
kosterlitz ( kk ) @xcite and kim park kim ( kpk ) @xcite models which are known to belong asymptotically to the kpz and mbe universality classes , respectively .
the common feature of these two models is the replacement of the usual diffusion rules of the sos models described above by local restrictive conditions controlling nearest
neighbor height differences . in the kk model , deposition sites are randomly chosen , but the incorporation of the adatoms into the substrate is subject to a specific restriction : the deposition event occurs if and only if the absolute value of the height difference between the randomly selected deposition site , @xmath96 , and _ each _ of its nearest neighboring sites remains smaller than or equal to a positive integer @xmath122 after deposition ( our simulations were done for @xmath123 ) . if this strict constraint is not satisfied , the attempted deposition of an adatom is rejected , and the random selection of the deposition site is repeated until the deposition is successfully done . since every attempt to deposit an adatom is not successful , the definition of `` time '' in this model is not quite the same as that in the other models where every deposition attempt leads to the incorporation of a new adatom in the growing film . in the kk model , the `` time '' is equivalent to the average height , which is not the same as the number of attempted depositions per site ( these two quantities are the same in the other models considered here ) .
the kk model is known to belong to the kpz universality class , and in fact provides the most numerically efficient and accurate method for calculating the kpz growth exponents .
kim _ et al .
_ @xcite discovered that a slight change in the algorithm for choosing the incorporation site transforms the kk model into a new one , the kpk model , that belongs to the mbe universality class .
the change consists of extending the search for appropriate incorporation sites ( i.e , sites where the constraint on the absolute values of the nearest neighbor height differences would be satisfied after the incorporation of an adatom ) to the neighbors of the originally selected deposition site @xmath96 . if the original site does not satisfy the constraint , then the neighboring sites ( @xmath124 in ( 1 + 1 ) dimensions ) are checked , and an adatom is incorporated at one of these sites if the incorporation does not violate the constraint
otherwise , the search is extended to the next nearest neighbors of @xmath96 , and so on until a suitable incorporation site is found .
we mention that in our implementation of this process , if , for example , both the sites @xmath125 and @xmath126 are found to be suitable for incorporation , then one of them is chosen randomly without any bias .
application of this algorithm in ( 2 + 1)dimensions involves extending the search for suitable incorporation sites to those lying inside circles of increasing radii around the randomly selected deposition site @xmath96 . the diffusion and incorporation rules of the kpk model @xcite lead essentially to a conserved version of the kim kosterlitz rsos model @xcite , and as such the continuum growth equation corresponding to the kpk model is the conserved kpz equation ( with nonconserved noise ) , which is precisely the mbe equation ; eq .
( [ mbe_eq ] ) is the conserved version of eq .
( [ mh_eq ] ) with nonconserved noise in both .
simulations for ( 1 + 1)dimensional discrete growth models were carried out for @xmath127 , @xmath128 and @xmath129 .
the value @xmath127 corresponds to the f model that has a relatively small equilibration time ( of the order of @xmath130 ) .
the remaining conservative models , characterized by @xmath131 ( lc ) and @xmath132 ( wv , dt , ckd and kpk ) , have a much slower dynamics ( with @xmath33 values 4 or 3 ) . so their corresponding equilibration time intervals , required for the interface roughness to reach saturation , are of the order of @xmath133 and @xmath134 , respectively . for this reason , the largest values of @xmath34 for which the steady state could be reached in reasonable simulation time are considerably shorter in these models than in the f model .
the fastest equilibration occurs in the kk model ( @xmath135 ) where @xmath136 . in calculations of the transient persistence probabilities , the initial configuration of the height variables
is taken to be perfectly flat , i.e. @xmath137 ( @xmath138 ) .
the lattice size was in the range @xmath139 , and the duration of the deposition process , measured in units of number of grown monolayers ( ml ) , was @xmath140 .
the results were averaged over @xmath140 independent runs . for measurements in the steady state situation ,
a saturation of the interface roughness was first obtained by depositing a large number ( of the order of @xmath62 ) of monolayers and subsequent time evolution from one of the steady state configurations obtained this way was used for measuring the persistence probabilities .
a much smaller lattice length ( @xmath141 for the f model , @xmath142 for the kk model , @xmath143 for the kpk model , and @xmath144 for the lc , wv , dt and ckd models ) was used in these calculations in order to reach the steady state saturation within reasonable simulation times .
.positive and negative persistence exponents , @xmath145 and @xmath146 , for the transient ( @xmath147 ) regime , measured for seven different discrete growth models ( identified in the first column ) using kinetic monte carlo simulations with relatively large system sizes ( @xmath34 ) .
the measured growth exponent , @xmath23 , and the universality class of the model are indicated in the last two columns , respectively . [ cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] we now focus on the steady state persistence exponents which , as discussed above , are found using relatively small values of @xmath9 and @xmath148 . in fig .
[ fig14](a ) , we show that for @xmath149 ( e.g. for @xmath150 ml ) , the persistence probability of the f model does not exhibit a clear power law decay .
however panels b ) and c ) of fig .
[ fig14 ] reveal that once @xmath9 becomes of the order of the measurement time @xmath5 , the expected power law is recovered and in addition the steady state exponent for the linear f model , @xmath151 , which should be equal to @xmath152 ) with @xmath153 , is recovered .
the results for the dt model are presented in fig .
[ fig15 ] .
the steady state persistence exponents have been measured from the power law decays shown in fig .
[ fig15](b ) . in this temporal regime , as shown in panel d ) , the growth exponent is equal to the asymptotic value of @xmath154 .
the persistence behavior of the dt model in this regime is characterized by @xmath155 and @xmath156 , indicating that the relation of eq .
( [ sc1 ] ) holds reasonably well for the ( 2 + 1) dimensional nonlinear mbe dynamics , as in the @xmath157dimensional case .
it is important to mention that the same values of the persistence exponents have been obtained using a dt system with @xmath158 , equilibrated for @xmath159 ml , as required in the traditional method ( used in most of the ( 1 + 1)dimensional simulations ) of measuring the steady state persistence probabilities .
thus the `` quick and easy '' method of obtaining the steady state persistence exponent again agrees well with the exponent extracted from the actually saturated interface , as discussed above .
persistence probabilities for the ( 2 + 1)dimensional f model of size @xmath160 , averaged over 200 runs , obtained from simulations with different values of the initial time @xmath12 .
a ) @xmath150 ml .
b ) @xmath161 ml .
c ) @xmath162 ml .
the persistence probability curves in case c ) show the expected power law decay characterized by the exponent values @xmath163 and @xmath164 .
d ) log log plot of the interface width @xmath24 vs deposition time @xmath5 in units of ml .
the slope in the intermediate growth decade is @xmath165 and thereafter it decreases to zero , as expected.,width=604,height=453 ] persistence probabilities for the ( 2 + 1)dimensional dt model of size @xmath160 , averaged over 200 runs , obtained from simulations with different values of the initial time @xmath12 .
a ) @xmath161 ml : the persistence probabilities do no exhibit clear power law decay .
b ) @xmath166 ml : the persistence probability curves show the expected power law decay characterized by the exponent values @xmath167 and @xmath168 .
c ) results for the ( 2 + 1)dimensional ckd model , obtained using @xmath169 , @xmath170 ml , @xmath171 and @xmath172 .
d ) log log plot of the interface width @xmath24 vs deposition time @xmath5 in ml for the dt model .
the slope manifests a crossover from an initial value of @xmath173 to the asymptotic value of 0.20.,width=604,height=453 ] we have also performed some preliminary persistence calculations for the ( 2 + 1)dimensional ckd model in order to check the validity of our reported ( 2 + 1)dimensional mbe persistence exponents . using @xmath169 and @xmath174 ml , we find that the values of the positive and negative persistence exponents depend to some extent on the chosen values for the coefficient @xmath121 of the nonlinear term and the control parameter @xmath175 .
for example , we obtain @xmath176 and @xmath177 using @xmath178 and @xmath179 , and @xmath180 and @xmath181 using @xmath178 and @xmath172 .
both cases are characterized by a growth exponent of @xmath182 , in agreement with ref .
@xcite , which is consistent with the expected value of @xmath154 .
the results obtained in the latter case are displayed in panel c ) of fig .
[ fig15 ] for the purpose of illustrating the similarity between the dt and ckd models . from these observations
, we conclude that the ( 2 + 1)dimensional dt and ckd persistence results are consistent with each other and they clearly reflect the nonlinearity of the mbe dynamical equation in the difference between the values of the positive and negative persistence exponents as expected for the up down asymmetric generic nonlinear situation .
since all the results described above have been obtained from simulations of finite systems , it is important to address the question of how the persistence probabilities are affected by the finite system size .
we have already encountered such effects in our study of persistence probabilities for ( 1 + 1)dimensional models ( see table [ tab_dt ] ) , where it was found that the measured values of the persistence exponents in the steady state increase slightly as the system size @xmath34 is decreased , while the value of the growth exponent @xmath23 decreases with decreasing @xmath34 .
we did not investigate finite size effects in our study of the transient persistence probabilities because these studies were carried out for large values of @xmath34 and relatively small values of the time @xmath5 .
the qualitative dependence of the measured values of the steady state persistence exponents @xmath183 and the growth exponent @xmath23 on the sample size @xmath34 is not difficult to understand .
the steady state persistence probabilities @xmath184 exhibit a power law decay with exponent @xmath183 as long as the time @xmath5 is small compared to the characteristic time scale @xmath185 of the system which is proportional to @xmath186 .
the decay of @xmath187 becomes faster as @xmath5 approaches and exceeds this characteristic time scale .
since this departure from power law behavior occurs at earlier times for smaller systems , the value of the persistence exponent extracted from a power law fit to the decay of the persistence probability over a fixed time window is expected to increase as the system size is reduced . in a similar way
, the measured value of @xmath23 is expected to be smaller for relatively small values of @xmath34 because the precursor to the saturation of the width at long times occurs at shorter values of @xmath5 in smaller systems . thus ,
the general trends in the system size dependence of the persistence and growth exponents are reasonable .
however , it would be useful to obtain a more quantitative description of these trends . since the characteristic time scale @xmath185 ( `` equilibration '' or `` saturation '' time ) of a system of linear size @xmath34 is proportional to @xmath186 , one expects , in analogy with the theory of finite size scaling in equilibrium critical phenomena , that the steady state persistence probability @xmath188 ( in this discussion , we omit the initial time @xmath12 in the argument of @xmath189 because the steady state persistence probability is independent of the choice of @xmath12 ) would be a function of the scaling variable @xmath47 .
another time scale has to be taken into consideration in a discussion of the scaling behavior of the persistence probability .
this is the sampling time @xmath37 which is the time interval between two successive measurements of the height at a fixed spatial point . in our simulations of the atomistic growth models ,
the smallest value of @xmath37 is 1 ml because the heights are measured after each deposition of one complete ( ideal ) monolayer .
however , larger integral values of @xmath37 can also be used in the calculation of the persistence probabilities . since experimental measurements
are also carried out at discrete time intervals , the presence of a finite value of @xmath37 has to be accounted for in the analysis of experimental data also .
note that the persistence probability itself is mathematically defined , @xmath8 , for continuous values of time @xmath5 whereas measurements and simulations are necessarily done on discrete time .
it has been pointed out in ref .
@xcite that discrete time sampling of a continuous time stochastic process does affect the measured persistence probability .
such effects have been investigated in detail @xcite in the context of a different stochastic probability ( called the survival probability in ref .
@xcite ) that measures the probability of the interface height at a fixed position not returning to its time averaged value within time @xmath5 . in that work , it was found that the survival probability measured for a system of size @xmath34 with sampling interval @xmath37 is a function of the scaling variables @xmath47 and @xmath48 .
we expect a similar behavior for the steady state persistence probabilities measured in our simulations .
thus , the expected scaling behavior of @xmath190 is @xmath191 where the function @xmath192 should decay as @xmath193 for small @xmath194 and @xmath195 .
, for the f model shown for different system sizes with different sampling times .
panel a ) : double log plot showing three different persistence curves vs. time corresponding to : @xmath196 and @xmath197 , @xmath198 and @xmath199 , @xmath200 and @xmath201 , respectively .
panel b ) : finite size scaling of @xmath202 .
results for persistence probabilities for three different sizes ( as in panel a ) ) with the same value of @xmath48 ( i.e. @xmath203 ) are plotted versus @xmath204 ( @xmath94 ) .
the dotted ( dashed ) line is a fit of the data to a power law with an exponent of @xmath205 ( @xmath206).,title="fig:",width=642,height=245 ] [ fig16 ] to test the validity of this scaling ansatz , we have carried out calculations of the steady state persistence probability in the linear f model ( the positive and negative persistence probabilities are the same in this model ) using different values of @xmath34 and @xmath37 . due to the linearity of the f model , we have computed a persistence probability @xmath207 given by the average value of the positive and negative persistence probabilities . if the scaling description of eq .
( [ scaling ] ) is valid , then plots of @xmath208 versus @xmath47 for different values of @xmath34 and @xmath37 should coincide if the value of @xmath37 for the different sample sizes are chosen such that the ratio @xmath48 remains constant . as shown in fig .
16 where we present the data obtained from simulations of the ( 1 + 1)dimensional f model for three different values ( @xmath209 , @xmath210 and @xmath211 ) of @xmath34 and three correspondingly different values of @xmath37 ( 4 , 16 and 64 , so that @xmath48 with @xmath94 is held fixed at @xmath212 ) , plots of @xmath207 versus @xmath204 ( which is proportional to @xmath47 because @xmath37 is chosen to be proportional to @xmath186 ) for the three different sample sizes exhibit an excellent scaling collapse .
these results confirm the validity of the scaling form of eq .
( [ scaling ] ) . as shown in fig .
16 , the scaling function @xmath114 exhibits the expected power law behavior for relatively small values of @xmath204 .
our results also show signatures of a crossover to a power law decay with exponent 1 as @xmath5 approaches and exceeds the characteristic time scale @xmath185 ( this crossover occurs near @xmath213 in the f model ) .
we discuss below a possible explanation for this behavior .
height fluctuations at times @xmath12 and @xmath13 are expected to be completely uncorrelated if @xmath5 is large compared to @xmath185 .
therefore , the persistence probability @xmath207 for values of @xmath5 much larger than @xmath185 may be obtained by considering a collection of fluctuating variables which have the same probability distribution ( since the system is in the steady state ) , and which are completely uncorrelated with one another .
let @xmath214 represent such a collection of variables ( these variables may be thought of as the height at a particular site measured at regularly spaced times with spacing larger than @xmath185 ) . for simplicity , we assume that each @xmath215 is uniformly distributed between @xmath216 and @xmath217 . then , given a particular value of @xmath218 , the probability @xmath219 that all the variables @xmath220 are larger than @xmath218 may be easily obtained as @xmath221^n = [ ( a - x_0)/(2a)]^n . \label{calc1}\ ] ] the positive persistence probability @xmath222 is obtained by averaging this probability over the probability distribution of @xmath218 .
thus , we have @xmath223 which decays as a power law with exponent 1 for large @xmath122 .
this power law behavior does not depend on the form of the assumed probability distribution for the fluctuating variables @xmath224 . assuming a general probability distribution @xmath225 with @xmath226 , eq .
( [ calc2 ] ) can be written as @xmath227^n dx_0 = \int_{-\infty}^\infty p(x_0 ) \left [ 1-\int_{-\infty}^{x_0 } p(x ) dx \right ] ^n dx_0 . \label{calc3}\ ] ] for large @xmath122 , the quantity that multiplies @xmath228 in eq .
( [ calc3 ] ) is of order unity only for values of @xmath218 for which @xmath229 is of order @xmath230 .
physically , this means that the positive persistence probability is nonzero for large @xmath122 only if the initial value @xmath218 is very close to the lower limit of the allowed range of values .
this effectively restricts the upper limit of the integral over @xmath218 to @xmath231 where @xmath231 satisfies the requirement that @xmath232 where @xmath233 is a constant of order unity . since the quantity that multiplies @xmath228 in eq .
( [ calc3 ] ) is of order unity for such values of @xmath218 , it follows that @xmath234 this simple analysis shows that the simulation results for the behavior of the scaling function of fig .
16 for large values of @xmath204 are quite reasonable . while we have not carried out similar scaling analyses [ eq .
( [ scaling ] ) ] for other models , we expect the scaling form of eq .
( [ scaling ] ) to be valid in general .
we expect that such scaling analysis of the persistence probability as a function of the system size @xmath34 and the sampling time @xmath37 would be useful in the analysis of numerical and experimental data in the future .
in fact , we believe that a direct experimental verification of the scaling ansatz defined by eq .
( [ scaling ] ) will be valuable .
in this paper we have investigated the temporal first passage statistics , expressed in terms of temporal persistence probabilities , for a variety of atomistic models that provide discrete realizations of several linear and nonlinear langevin equations for the stochastic dynamics of growing and fluctuating interfaces . using extensive kinetic monte carlo simulations ,
we have obtained transient and steady state persistence exponents for these ( 1 + 1 ) and ( 2 + 1)dimensional sos and rsos growth models .
we have followed the methodology of krug _ et al .
_ @xcite and extended their numerical work to the nonlinear mbe dynamical equation by studying the persistence behavior of the atomistic dt , wv , ckd and kpk models . from these studies ,
we have identified two persistence exponents for each of the two temporal regimes ( transient and steady state ) of the persistence probability .
the difference between the values of the two exponents reflects the nonlinearity ( and the resulting lack of up down symmetry ) of the mbe dynamical equation . among the models studied here
, we find that in ( 1 + 1 ) dimensions and in the range of system sizes used in our simulations , wv and dt models are hardly distinguishable from the point of view of transient and steady state persistence behavior : the persistence exponents obtained for these two models are very close to each other .
we , therefore , conclude that in the range of simulation parameters used in this study , the ( 1 + 1)dimensional dt and wv models belong to the same universality class ( namely the mbe universality class ) as far as their persistence behavior is concerned .
a separate investigation is required in order to understand the universality class of the wv model in ( 2 + 1 ) dimensions .
the kpk model appears not to reflect well the nonlinear feature of the underlying dynamical equation in the values of the positive and negative persistence exponents .
this is probably due to strong finite size effects arising from the small lattice sizes used in our traditional steady state simulations ( i.e. using @xmath235 ) .
these finite size effects appear to lead to overestimated persistence exponents [ and underestimated growth exponent , consistent with eq .
( [ theta_s ] ) ] .
we have also investigated the ckd model , which is another discrete model belonging to the mbe universality class , our main goal being a closer examination of how the nonlinearity of the underlying continuum equation is reflected in values of the transient and steady state persistence exponents . in this case
we have obtained clearly different values for the positive and negative persistence exponents .
the predictions of the ckd model concerning the persistence exponents have been checked by applying the noise reduction technique to the dt model .
we found that for the mbe universality class , the steady state persistence exponents in ( 1 + 1 ) dimensions are : @xmath236 and @xmath237 .
these two values represent the average of the results obtained for the ckd and the noise reduced dt models .
these results suggest that measurements of persistence exponents can be profitably used to detect the presence of nonlinearity in the continuum equations underlying surface growth and fluctuation phenomena .
the observed difference between the positive and negative steady state persistence exponents for the models in the mbe universality class implies that the relation of eq .
( [ theta_s ] ) , known to be valid for linear langevin equations ( our results for the linear f and lc models are in agreement with this relation ) , can not be satisfied by both these exponents .
thus , it is clear that at least one of these steady state persistence exponents is not related to the usual dynamic scaling exponents in a simple way .
we have found that the relation of eq .
( [ theta_s ] ) is approximately satisfied ( within the error bars of our numerical results ) by the smaller one of the two steady state persistence exponents in all the ( 1 + 1) and ( 2 + 1)dimensional discrete stochastic growth models studied in this paper .
we have also shown analytically that this relation between the smaller persistence exponent and the growth exponent is , in fact , exact . the smaller exponent appears to correspond to the positive ( negative ) persistence probability if the top ( bottom ) part of the steady state interface profile is smoother .
this observation suggests a deep ( and potentially important ) connection between the surface morphology and the associated persistence exponent , which has no simple analog in the dynamic scaling approach where the critical exponents ( @xmath50 , @xmath23 , @xmath238 ) by themselves do not provide any information about the up down symmetry breaking in the surface morphology .
further investigation of this aspect would be very interesting and highly desirable , particularly if experimental information on persistence properties of nonequilibrium surface growth kinetics becomes available .
our investigation of the effects of the initial configuration on the persistence probabilities indicates that the transient persistence exponents can be obtained only if the interface is completely flat at the initial time .
this restriction puts severe limits on the possibility of measuring the transient persistence exponents in real experiments where it would be very difficult , if not impossible , to meet the requirement of zero initial roughness .
we have also found the surprising and useful result that the steady state persistence exponents can be accurately measured even if the initial configuration corresponds to a value of @xmath12 that is much smaller than the time required for the interface to reach saturation . in other words ,
the persistence probabilities exhibit their steady state behavior for measurement times comparable to the initial time @xmath12 even if the value of @xmath12 is much smaller than @xmath186 .
this behavior was found in both ( 1 + 1 ) and ( 2 + 1 ) dimensions , in all the linear and nonlinear models we studied .
this finding is very useful because it opens up the possibility of numerically calculating the steady state persistence exponents for large systems and for higher dimensions as well .
in fact , this observation enabled us to calculate the steady state persistence exponents for ( 2 + 1)dimensional models belonging to the ew and mbe universality classes .
for the mbe universality class , we have considered the dt model and found the positive and negative persistence exponents in the steady state to be @xmath239 and @xmath240 , respectively , in ( 2 + 1 ) dimensions .
we have also examined in detail the dependence of the measured steady state persistence probability in the ( 1 + 1)dimensional f model on the sample size @xmath34 and the sampling interval @xmath37 which is always finite in simulations and experimental measurements .
we found that this dependence is described by a simple scaling form .
the scaling function was found to exhibit power law decay with with exponent 1 for times larger than @xmath186 .
we have proposed a simple explanation for this behavior .
we believe that such scaling analysis would prove to be useful in future numerical and experimental studies of persistence properties .
we conclude from the results of this study that persistence probabilities provide a valuable set of tool for investigating the dynamics of nonequilibrium systems in general , and surface growth and fluctuations in particular .
recent experimental studies have shown that the concept of persistence can be applied to analyze the dynamics of fluctuating steps on al / si(111 ) , ag(111 ) and pb(111 ) surfaces @xcite recorded using stm methods .
we believe that in view of the importance of thermal and shot noise fluctuations in the dynamics of growing and fluctuating interfaces , theoretical and experimental studies of persistence would play an important role in the analysis of the dynamics of nonequilibrium surface growth .
barabasi and h. e. stanley , _ fractal concepts in surface growth _
( cambridge university , cambridge , 1995 ) ; a. mckane , m. droz , j. vannimenus and d. wolf , _ scale invariance , interfaces , and non equilibrium dynamics _
( new york and london , 1995 ) .
f. family , j. phys .
a : gen . * 19 * , l441 ( 1986 ) .
s. das sarma and p. tamborenea , phys .
lett . * 66 * , 325 ( 1991 ) ; p. tamborenea and s. das sarma , phys .
e * 48 * , 2575 ( 1993 ) .
d. wolf and j. villain , europhys . lett .
* 13 * , 389 ( 1990 ) .
j. m. kim and s. das sarma , phys .
lett . * 72 * , 2903 ( 1994 ) ; j. krug , phys .
lett . * 72 * , 2907 ( 1994 ) .
s. das sarma , s. v. ghaisas and j. m. kim , phys .
e * 49 * , 122 ( 1994 ) .
j. m. kim and j. m. kosterlitz , phys .
rev . lett .
* 62 * , 2289 ( 1989 ) ; j. m. kim , j. m. kosterlitz and t. ala - nissala , j. phys .
a * 24 * , 5569 ( 1991 ) .
y. kim , d. k. park and j. m. kim , j. phys .
a * 27 * , l533 ( 1994 ) .
w. w. mullins , j. appl .
* 28 * , 333 ( 1957 ) ; c. herring , j. appl . phys . * 21 * , 301 ( 1950 )
. s. f. edwards and d. r. wilkinson , proc .
london , ser.a * 381 * , 17 ( 1982 ) .
m. kardar , g. parisi and y. -c .
zhang , phys .
lett . * 56 * , 889 ( 1986 ) .
lai and s. das sarma , phys .
lett . * 66 * , 2348 ( 1991 ) .
j. villain , j. phys .
i ( france ) * 1 * , 19 ( 1991 ) . c. dasgupta ,
s. das sarma and j. m. kim , phys .
e * 54 * , r4552 ( 1996 ) ; c. dasgupta , j. m. kim , m. dutta and s. das sarma , phys .
e * 55 * , 2235 ( 1997 ) .
j. krug , adv .
phys . * 46 * , 139 ( 1997 ) .
j. krim and g. palasantzas , int .
* b9 * , 599 ( 1997 ) . s. n. majumdar , curr .
* 77 * , 370 ( 1999 ) .
s. n. majumdar , c. sire , a. j. bray and s. j. cornell , phys .
lett . * 77 * , 2867 ( 1996 ) ; b. derrida , v. hakim and r. zeitak , phys .
lett . * 77 * , 2871 ( 1996 ) ; b. derrida , a. j. bray and c. godreche , j. phys .
a * 27 * , l357 ( 1994 ) .
b. derrida , v. hakim and v. pasquier , phys .
* 75 * , 751 ( 1995 ) .
s. n. majumdar and c. sire , phys .
77 * , 1420 ( 1996 ) .
s. cueille and c. sire , eur .
j. b * 7 * , 111 ( 1999 ) .
j. krug , s. n. majumdar , s. j. cornell , a. j. bray and c. sire , phys .
e * 56 * , 2702 ( 1997 ) .
h. kallabis and j. krug , europhys .
lett * 45 * , 20 ( 1999 ) .
d. slepian , bell syst .
j. * 41 * , 463 ( 1962 ) .
i. dornic and c. godreche , j. phys .
a * 31 * , 5413 ( 1998 ) .
t. j. newman and z. toroczkai , phys .
e * 58 * , r2685 ( 1998 ) ; z. toroczkai , t. j. newman and s. das sarma , phys .
e * 60 * , r1115 ( 1998 ) .
m. marcos - martin , d. beysens , j. p. bouchaud , c. godreche and i. yekutieli , physica a * 214 * , 396 ( 1995 ) . w. y. tam , r. zeitak , k. y. szeto and j. stavans , phys .
lett . * 78 * , 1588 ( 1997 ) ; w. y. tam and k. y. szeto , phys . rev .
e. * 65 * , 042601 ( 2002 ) .
b. yurke , a. n. pargellis , s. n. majumdar and c. sire , phys .
e * 56 * , r40 ( 1997 ) .
g. p. wong , r. w. mair and r. l. walsworth , phys .
lett * 86 * , 4156 ( 2001 ) .
d. b. dougherty , i. l. lyubinetsky , e. d. williams , m. constantin , c. dasgupta and s. das sarma , phys .
lett * 89 * , 136102 ( 2002 ) .
d. b. dougherty , o. bondarchuk , m. degawa and e. d. williams , surf .
* 527 * l213 ( 2003 ) .
m. constantin , s. das sarma , c. dasgupta , o. bondarchuk , d. b. dougherty and e. d. williams , phys .
* 91 * , 086103 ( 2003 ) .
w. feller , _ introduction to probability theory and its applications _
vol . 1 ( wiley , new york , 1968 )
. y. g. sinai , theor .
90 * , 219 ( 1992 ) s. n. majumdar and a. j. bray , phys .
lett . * 81 * , 2626 ( 1998 ) .
the asymptotic long time behavior of the ( 1 + 1)dimensional wolf
villain model is known to be in the edwards wilkinson universality class .
however , its behavior in the relatively short time regime considered here is expected to be similar to that of other models in the mbe universality class .
p. punyindu and s. das sarma , phys .
e * 57 * , r4863 ( 1998 ) . c. dasgupta , m. constantin , s. das sarma and s. n. majumdar , phys .
e * 69 * , 022101 ( 2004 ) ; m. constantin and s. das sarma , to appear in phys .
e ; cond - mat/0312612 . f. family and t. vicsek , j. phys .
a * 18 * , l75 ( 1985 ) ; f. family , physica * 168a * , 561 ( 1990 ) .
s. das sarma , c. j. lanczycki , r. kotlyar , and s. v. ghaisas , phys .
e * 53 * , 359 ( 1996 ) .
s. das sarma and p. punyindu , phys .
e * 55 * , 5361 ( 1997 ) .
d. e. wolf and j. kertesz , europhys .
lett . * 4 * , 651 ( 1987 ) ; j. kertesz and d. e. wolf , j. phys .
a * 21 * , 747 ( 1988 ) .
p. devillard and h. e. stanley , phys
a * 38 * , 6451 ( 1988 ) .
p. punyindu chatraphorn and s. das sarma , phys .
e * 66 * , 041601 ( 2002 ) .
p. smilauer and m. kotrla , phys .
b * 49 * , 5769 ( 1994 ) .
s. das sarma , p. punyindu chatraphorn and z. toroczkai , phys .
e * 65 * , 036144 ( 2002 ) . | persistence probabilities of the interface height in ( 1 + 1) and ( 2 + 1)dimensional atomistic , solid on solid , stochastic models of surface growth are studied using kinetic monte carlo simulations , with emphasis on models that belong to the molecular beam epitaxy ( mbe ) universality class .
both the initial transient and the long time steady state regimes are investigated .
we show that for growth models in the mbe universality class , the nonlinearity of the underlying dynamical equation is clearly reflected in the difference between the measured values of the positive and negative persistence exponents in both transient and steady state regimes . for the mbe universality class , the positive and negative persistence exponents in the steady state are found to be @xmath0 and @xmath1 , respectively , in ( 1 + 1 ) dimensions , and @xmath2 and @xmath3 , respectively , in ( 2 + 1 ) dimensions .
the noise reduction technique is applied on some of the ( 1 + 1)dimensional models in order to obtain accurate values of the persistence exponents .
we show analytically that a relation between the steady state persistence exponent and the dynamic growth exponent , found earlier to be valid for linear models , should be satisfied by the smaller of the two steady - state persistence exponents in the nonlinear models .
our numerical results for the persistence exponents are consistent with this prediction .
we also find that the steady state persistence exponents can be obtained from simulations over times that are much shorter than that required for the interface to reach the steady state .
the dependence of the persistence probability on the system size and the sampling time is shown to be described by a simple scaling form . |
a supermassive black hole is believed to reside in the dynamical center of the galaxy ( eckart & genzel 1996 ; genzel et al .
1997 ; ghez et al . 1998 ; schdel et al . 2002 ; genzel et al .
2000 ; ghez et al .
2000 ; schdel et al . 2003 ; ghez et al .
it is associated with the compact radio source and dominates the gravitational potential at distances @xmath5 pc from the center .
the central mass has been estimated with a few different methods . in particular , the statistical analysis of sky - projected positions and velocities of stars in the central 0.5 pc gave estimates that range above @xmath6 ( genzel et al .
2000 ; schdel et al .
the systematic error of these estimates , however , is difficult to quantify as they depend on the assumed three - dimensional statistical model of the cluster .
independent estimates of the black hole mass were derived from observations of a few stars ( s - stars ) that move on very tight orbits around , at radii @xmath7 pc , and have orbital periods as short as a few decades . fractions of s - star orbits have been mapped out over the past decade ( schdel et al .
2002 ; schdel et al . 2003 ; ghez et al .
2003 ; ghez et al .
2005 ) . by checking which value of the central mass
provides an acceptable fit to the orbital data , eisenhauer et al .
( 2005 ) found @xmath8 , and ghez et al .
( 2005 ) found @xmath9 for the fiducial distance to the galactic center @xmath2 kpc .
same star orbits were used to derive @xmath10 kpc ( eisenhauer et al . 2005 ) , in agreement with previous methods ( e.g. reid et al .
1993 ) . in this paper , we report a new independent estimate of the central mass that uses the young massive stars at @xmath0 pc as test particles . a remarkable property of these stars has been discovered recently : they form a disk - like population ( levin & beloborodov 2003 ; genzel et al .
two stellar disks apparently coexist at these radii .
one of them rotates clockwise on the sky and the other one counter - clockwise .
new observations of paumard et al .
( 2006 ) have increased the number of known disk members by a factor of three and show clearly the two - disk structure of the young population .
the clockwise disk is especially well pronounced .
it has about 30 members and a small thickness @xmath11 . in this paper
, we investigate the orbital motions of stars in the clockwise disk and infer the central gravitating mass . as a tool of our mass estimate we use orbital roulette , a recently developed statistical method ( beloborodov & levin 2004 , hereafter bl04 ) .
this method is generally more precise than the estimators based on the virial theorem and provides a statistically well - defined way to quantify the uncertainty of the mass estimate .
the disk - like structure of the young stellar population is not evident when it is observed in projection : the line - of - sight coordinaties @xmath12 of the stars are unknown .
this structure , however , is evident in the distribution of the measured 3d velocities . in particular , the velocity vectors of stars that move clockwise on the sky cluster near a particular plane . this may be only if the orbits of stars are near this plane and form a disk - like population .
the clockwise disk has a small thickness at angular distances @xmath13 from .
we use the stars in this region to define the midplane of the disk .
it is defined as the midplane of the observed velocity vectors , and its normal vector is are standard coordinates : they are chosen so that is at the origin , the @xmath14 and @xmath15 axes are in the plane of the sky and directed along increasing right ascension and declination , respectively , and the @xmath12 axis is directed along the line of sight away from us . ]
the deviations of individual stellar velocities from the midplane are shown in figure 1 . hereafter in this paper we use the sample of stars between 0.7 `` and 5 '' with firmly detected clockwise rotation @xmath17 , where @xmath18 is the measurement error in @xmath19 .
the sample has 28 stars , which is sufficiently large to study the distribution of velocity deviations from the disk midplane ( fig . 2 ) .
the deviation angles @xmath20 are defined by @xmath21 .
the observed distribution of @xmath22 may be fitted by a disk population model .
we will use a simple gaussian model of orbital inclinations @xmath23 around the midplane .
the probability distribution of @xmath23 is then given by @xmath24 $ ] .
the gaussian model of the disk has one parameter the dispersion @xmath25 and predicts a certain probability distribution for @xmath22 .
we calculate this distribution using monte - carlo technique , including a simulation of observational errors .
we randomly draw @xmath26 orbits from the model population and choose a randomly directed vector of length @xmath27 in each orbital plane @xmath28 .
this gives 28 simulated velocities @xmath29 .
then observational errors @xmath30 are added to each @xmath29 ( the real @xmath27 and @xmath31 from paumard et al .
2006 are used ) . finally , deviations @xmath20 of velocities @xmath29 from the disk midplane are calculated .
this simulation is repeated @xmath32 times , which gives the accurate distribution of @xmath22 predicted by the model as well as the average values @xmath33 and @xmath34 as functions of @xmath25 .
the observed @xmath35 and @xmath36 are reproduced by the monte - carlo model with @xmath37 .
the model is consistent with the observed distribution of @xmath22 for @xmath38 .
hereafter we will use the best - fit model with @xmath39 .
the geometrical thickness of the disk may be described by the elevation of stars above / below the midplane , = || , where @xmath40 is the orbital phase of the star , and @xmath23 is the inclination of its orbit . assuming circular orbits for simplicity , one finds after averaging over the orbital phase , = ( ) ^1/2 .
this gives for the gaussian model ( approximating @xmath41 for small @xmath42 ) , [ eq : h ] = ( ) ^3/2 = 0.090.03 .
we now check whether the data is consistent with circular motion of the disk stars .
the assumption of circular motion implies that radius - vector @xmath43 of the @xmath44-th star satisfies _
i^cir_i=0 , i=1, ... ,n . this equation , together with the known position ( @xmath45 ) on the sky , determines @xmath46 the line - of - sight coordinate of the star .
the easiest way to test the circular - motion hypothesis is to look at the orientations of orbital planes implied by this hypothesis .
the inferred @xmath47 allow us to calculate the angular momentum vector for each star _ i^cir=_i^cir_i , i=1, ...
,n , and then find its deviation @xmath48 from the disk axis @xmath49 .
the dispersion of @xmath48 would be expected to be consistent with disperion @xmath37 of orbital planes inferred directly from @xmath29 .
the actual distribution of @xmath48 is shown in figure 3 .
one can see that the circular hypothesis implies orbital inclinations that are inconsistent with the found thickness of the disk .
a significant fraction of the disk stars must have non - negligible eccentricities .
a circular orbit would provide a straightforward estimate of the central mass @xmath3 .
indeed , circular motion implies a simple relation between @xmath3 , @xmath50 , and @xmath51 : @xmath52 .
an estimate of @xmath3 could be done by applying this relation to stars with small @xmath53 , which are possibly on circular orbits .
however , it is difficult to quantify the accuracy of such analysis .
a more accurate estimate of @xmath3 is made in 4 , without making any prior assumptions on eccentricities .
measurements of stellar positions relative to are limited to 2 dimensions the plane of the sky .
however , the stars that belong to the clockwise disk have an extra constraint on their 3d positions @xmath43 : they must be close to the midplane of the stellar disk .
this constraint may be written as ( n_0)_i=0 , i=1, ..
,n , where @xmath54 is the disk axis and @xmath55 represents the dispersion of orbital planes around the disk midplane . the line - of - sight coordinate of the @xmath44-th star is then given by @xmath56 the orbital plane @xmath57 may differ from the best - fit disk plane by @xmath58 , which induces an error in the inferred @xmath12-coordinate . to estimate the error we rock @xmath57 with a gaussian distribution around @xmath49 with a mean deviation of @xmath59 . the resulting @xmath46 and their uncertainties
are found using the following monte - carlo simulation .
we draw randomly @xmath57 and @xmath60 , @xmath61 from their measured gassian distributions
. then we determine @xmath62 from the condition @xmath63 .
the obtained realization of @xmath57 and @xmath64 is given a weight that corresponds to the level of consistency of @xmath62 with the measured gaussian @xmath65 .
( in the simulation , we get this weight by rejecting the realization if its @xmath62 is farther in the gaussian tail than a randomly chosen @xmath66 from the gaussian distribution . ) in this way , a realization @xmath57 , @xmath64 is given the weight that takes into account both the thickness of the disk and the error of the 3d velocity .
we find the mean @xmath46 and its dispersion @xmath67 using about @xmath68 realizations .
the deduced 3d positions together with the measured 3d velocities give 6d information for each star .
then the velocities @xmath69 and @xmath70 may be deduced , where @xmath51 and @xmath40 are polar coordinates in the orbital plane of the star .
the measurement errors of @xmath29 and the uncertainty in the individual orbital planes imply errors @xmath71 and @xmath72 .
we estimate the uncertainties assuming the gaussian dispersion in @xmath57 with @xmath39 and gaussian velocity errors @xmath73 .
the results are summarized in table 1 and illustrated in figure 4 .
the orbital periods of stars at @xmath74 pc are thousands of years , and it is presently impossible to trace any significant fraction of the orbits .
therefore , a statistical method must be applied using the observed _ instantaneous _ positions @xmath43 and velocities @xmath29 of the stars .
such a method must take into account the possible errors in @xmath43 and @xmath29 .
a traditional mass estimate would be based on the virial theorem , [ eq : vir ] m=. the virial relation gives @xmath75 .
one faces difficulties , however , when trying to define the error of this estimate ( see bl04 for discussion ) .
a major assumption in the derivation of the virial relation ( [ eq : vir ] ) is that the stars have random orbital phases . bl04
have shown that the random - phase ( `` roulette '' ) principle can be applied directly to data , without invoking the virial theorem , and this gives a more accurate and statistically complete estimator .
it judges any trial gravitational potential @xmath76 by reconstructing the orbits in this potential and checking the inferred orbital time phases .
the true potential must give a `` fair roulette '' a random ( flat ) distribution of the phases .
the 3d positions of the disk stars have uncertainties because their precise orbital planes are unknown . besides
, the 3d velocities are measured with errors .
we take into account the uncertainties using monte - carlo technique : many random realizations of the data set @xmath77 are generated and studied .
the possible @xmath77 are generated assuming the @xmath59 gaussian distribution of the orbital planes and gaussian errors in @xmath29 with dispersions @xmath31 given in paumard et al .
( 2006 ) .
each realization of the data set @xmath77 is analyzed by the roulette estimator as described in bl04 .
the analysis aims to accept or rule out a trial potential @xmath78 at a chosen confidence level . given @xmath79 and a data set @xmath77 one can reconstruct all @xmath26 orbits and find the time orbital phases @xmath80 of the stars .
the time phase @xmath81 is defined so that @xmath82 at the pericenter of the orbit and @xmath83 at the apocenter . to judge
@xmath79 we check the consistency of @xmath80 with the expected flat distribution between 0 and 1 ( and do that for many realizations of the data sets to take into account the uncertainties in the data ) .
the roulette algorithm is described in detail in bl04 .
here we use its simplest version based on the analysis of the mean phase @xmath84 .
the true @xmath3 must satisfy the condition [ eq : roul ] |g(m)=0.5(12n)^-1/2 .
if @xmath79 is too small then @xmath85 , i.e. all stars will be found near the pericenter , while too large @xmath79 will give @xmath86 , i.e. all stars will sit near apocenter .
one can accept only such @xmath79 that give @xmath87 consistent with the fair roulette ( eq . [ eq : roul ] ) . in our monte - carlo simulation , @xmath88 that is found for a realization of the data set @xmath77 is compared with @xmath89 found for a realization of a fair roulette @xmath90 .
the statistics of this comparison is accumulated for @xmath68 random realizations .
thus , the probability @xmath91 of @xmath92 is accurately found .
the probability @xmath93 is monotonically increasing with @xmath79 since larger masses imply larger orbital phases ( closer to the apocenter ) .
trial masses that give small @xmath91 are ruled out with significance @xmath91 ( bl04 ) . if @xmath94 , we look at @xmath95 .
trial masses that give small @xmath96 are ruled out with significance @xmath97 .
the found functions @xmath98 are shown in figure 5 .
the confidence intervals for @xmath3 are between the two curves .
for example , the 80% confidence interval ( @xmath99 ) is @xmath100 .
the confidence intervals approximately correspond to a gaussian distribution with standard deviation @xmath101 , and we can summarize the result as @xmath102 .
we have also checked the distribution of @xmath80 for the found @xmath3 .
it is consistent with the expected flat distribution ; this is verified using the anderson - darling measure ( see bl04 ) .
we conclude that the observed stellar motions are consistent with kepler potential @xmath103 with @xmath102 .
there are two sources of systematic error of our result .
first , we had to chose a disk - population model in the calculations .
we specified the midplane ( normal @xmath49 ) and the gaussian distribution of orbital inclinations around the midplane with dispersion @xmath25 .
the parameters @xmath49 and @xmath25 have been derived by fitting the 3d velocity data and have finite errors .
the obtained @xmath3 is most sensitive to @xmath104 which determines the disk inclination to the line of sight .
our best - fit disk has @xmath105 ; larger @xmath106 would imply smaller sizes of the orbits and hence smaller @xmath3 .
@xmath104 as large as @xmath107 may still be consistent with the velocity data ( paumard et al .
2006 ) , and it would change our estimate to @xmath108 .
the parameter @xmath25 has the error of @xmath109 .
the choice of @xmath110 instead of @xmath59 would broaden the confidence intervals by a factor @xmath111 .
the second possible source of error is the distance to the galactic center @xmath4 which we assumed to be 8 kpc .
the mass estimate depends on @xmath4 because the proper velocities @xmath112 and positions @xmath113 scale linearly with @xmath4 ( @xmath114 is not affected as its measurement is based on the doppler effect ) . the midplane of the disk varies little and may be assumed constant to a first approximation
. then the deduced radial positions @xmath51 of the stars scale linearly with @xmath4 .
the dependence of @xmath3 on @xmath4 may be derived using a toy problem with @xmath115 stars moving in the disk midplane on a circular orbit of radius @xmath116 .
the true and deduced masses are then given by @xmath117 @xmath118.\ ] ] after averaging over @xmath119 stars and taking into account that @xmath120 and @xmath121 , we find [ eq : scale ] m(0true)=m(8 kpc)^-1 .
we have empirically checked this dependence by repeating the roulette simulation for various @xmath4 ( as small as 7 kpc ) .
the obtained @xmath122 are well described by a simple shift of @xmath3 according to equation ( [ eq : scale ] ) . in particular
, @xmath123 kpc ( eisenhauer et al . 2005 ) would give @xmath124 .
table 1 shows the eccentricities of reconstructed star orbits for @xmath2 kpc and the best - fit central mass @xmath125 .
the errors @xmath126 ( @xmath127 ) are caused by uncertainties @xmath128 and @xmath129 .
the errors are found using the monte - carlo simulation that is similar to the estimation of @xmath67 ( see 2.3 ) .
this simulation gives the probability distribution @xmath130 for each @xmath131 .
a non - zero eccentricity is detected if @xmath132 and accurately measured if @xmath133 , where @xmath134 is the mean of @xmath130 and @xmath126 is its dispersion .
if distribution @xmath130 is broad ( large errors ) , its mean represents the errors rather than the true @xmath131 .
therefore , @xmath134 is an accurate estimate of eccentricity only if the distribution @xmath130 is sufficiently narrow , @xmath135 .
table 1 shows @xmath136 for stars with @xmath137 . for stars with @xmath138 we give @xmath139 as an upper limit .
the obtained values are consistent with the results of paumard et al .
( 2006 ) where eccentricities are estimated with another method .
the reconstructed orbits of stars in the clockwise disk depend on the assumed central mass @xmath3 . in particular , the star orbital phases depend on @xmath3 .
we have used this dependence to estimate @xmath3 : the true @xmath3 must give time phases consistent with a uniform distribution .
this is the first practical application of orbital roulette , a new method of estimating the gravitational potential from instantaneous motions of test bodies ( bl04 ) .
the obtained estimate @xmath140 is consistent with the recent estimates based on the maps of orbits of s stars in the very vicinity of the center ( @xmath141 pc ) .
the independent estimate presented here uses different stars at @xmath142 pc and is complementary to the small - scale estimates .
it measures the total mass within @xmath143 pc which may include stars and dark matter .
the approximate equality of the masses within the central 0.1 pc and 0.01 pc implies an upper bound on the mass between 0.01 and 0.1 pc : @xmath144 .
this constrains a possible cusp of the stellar population around the cenral black hole as well as the dark - matter content of the central region .
further observations may provide a better constraint or resolve @xmath145 .
the data shows an inner cutoff of the disk population at a radius @xmath146 pc ( see fig . 1 and paumard et al .
the observed peak of the population at @xmath143 pc supports the hypothesis that the young stars were born in a gravitationally unstable accretion disk ( lb03 ) .
the found geometrical thickness of the clockwise stellar population @xmath147 ( eq . [ eq : h ] ) does not imply that it was born with this @xmath148 .
the parent accretion disk was likely thinner , and its stellar remnant have thickened with time .
the age of the stellar population , @xmath149 million years , should be compared with the timescale of diffusion of orbital parameters .
the latter may be written as t_diff~0.1 , where @xmath150 is a characteristic mass of the objects that dominate fluctuations of the gravitational potential , @xmath151 is their mass density averaged over the considered region @xmath152 pc , and @xmath153 is the total number of these objects . in a uniform stellar core of total mass @xmath154 , the fluctuations would be dominated by low - mass stars , whose total number is @xmath155 . substituting the observed characteristic velocity , @xmath156 km / s , one would find @xmath157 yr@xmath158 .
however , the galactic center is not uniform .
the two stellar disks and stellar clusters are observed , which significantly perturb the gravitational potential .
these perturbations should strongly reduce @xmath159 .
for example ,
if 1/30 of the core mass resides in @xmath160 objects with @xmath161 then @xmath162 yr . the corresponding acquired thickness of the stellar disk is @xmath163 in agreement with the observed value . a similar estimate for @xmath159 may be derived from the interaction between the two young disks ( nayakshin et al .
2005 ) .
the eccentricities of disk stars must also diffuse with time .
if the stars were born on circular orbits then @xmath164 are expected to develop as the disk thickens to @xmath165 .
table 1 shows that a significant fraction of the young stars in the clockwise disk have relatively large eccentricities .
the data are at best marginally consistent with the hypothesis that the stars used to have circular orbits 6 million years ago , and we leave a detailed analysis to a future work .
note that the stars may have been born in a gaseous disk that was not completely circularized : the circularization time was longer than the timescale for star formation .
then the initial eccentricities @xmath166 of the stars are non - zero . besides
, the interaction of stars with the disk might have influenced @xmath166 before the disk was gone .
an alternative scenario assumes that the stars were originally bound to an inspiralling intermediate - mass black hole ( imbh ; hansen & milosavljevic 2003 ) , so that the observed disk plane is associated with the orbital plane of the imbh . in this scenario ,
the inspiralling young cluster was stripped by the tidal field of and left a disk of stars with a small dispersion of orbital planes .
calculations of levin , wu , & thommes ( 2005 ) show that the stripped stars can generally have eccentric orbits
. however , the inspiralling cluster scenario appears to be in conflict with the observed stellar distribution in the clockwise disk ( see also paumard et al .
2006 for discussion ) .
positions are offsets from .
+ @xmath14 , @xmath15 , @xmath12 , @xmath51 are in units of @xmath167 cm , @xmath69 and @xmath70 in km / s , @xmath22 and @xmath40 in degrees . + in order to express distances in arcseconds one should divide @xmath14 , @xmath15 , @xmath12 , @xmath51 in the table by 1.20 . + | two disks of young stars have recently been discovered in the galactic center .
the disks are rotating in the gravitational field of the central black hole at radii @xmath0 pc and thus open a new opportunity to measure the central mass .
we find that the observed motion of stars in the clockwise disk implies @xmath1 for the fiducial distance to the galactic center @xmath2 kpc and derive the scaling of @xmath3 with @xmath4 . as a tool for our estimate we use orbital roulette , a recently developed method .
the method reconstructs the three - dimensional orbits of the disk stars and checks the randomness of their orbital phases .
we also estimate the three - dimensional positions and orbital eccentricities of the clockwise - disk stars . |
in a preceding paper @xcite we analyzed the implications of physical positivity ( ` unitarity ' ) and positivity of the topological susceptibility for the two - point correlation function of the topological density .
in particular we found that the short distance singularity has to be softer than tree level perturbation theory would predict . in this article
we verify for the two - dimensional @xmath0 nonlinear sigma model that one - loop perturbation theory is indeed in agreement with this requirement .
such a softening of the short distance singularity is already predicted by including the running of the coupling constant in tree level perturbation theory @xcite .
but since we are dealing here with a composite operator , this check of internal consistency of perturbation theory is quite nontrivial .
the action of the ( ferromagnetic ) o@xmath1 spin model on a 2-dimensional square lattice @xmath2 is @xmath3 where the sum is over nearest neighbor pairs @xmath4 of lattice sites .
the partition function reads @xmath5 \ , \mathrm{e}^ { - \mathcal{s } } , \ ] ] with measure @xmath6 = \prod_x \mathrm{d}^3 { \vec{s}}_x \ , \delta ( { \vec{s}}_x^{\,2 } -1 ) , \ ] ] imposing the unit norm constraint for all spins , i.e. @xmath7 .
we will consider the thermodynamic limit as the @xmath8 limit of the model on an @xmath9 square lattice with periodic boundary conditions , so that translation invariance holds .
the origin 0 is then located at the center of @xmath10 .
unit lattice vectors are called @xmath11 and @xmath12 .
we want to study the two - point function of the topological charge density @xmath13 , to be denoted by @xmath14 ( with @xmath15 ) , @xmath16 \ , q_x q_0 \ , \mathrm{e}^ { - \mathcal{s } } .\ ] ] the topological susceptibility is defined by @xmath17 in this paper we will consider two lattice definitions of @xmath13 : a ` field theoretical ' one and the geometric one due to berg and lscher . both make use of a triangulation of the square lattice @xmath18 as in figure [ figure : triangles ] . in both cases , the topological charge density is written as @xmath19 where @xmath20 with @xmath21 a certain function .
in @xcite we used a symmetrization of this definition by introducing in addition a ` mirror ' triangulation @xmath22 and defining @xmath23 ; this was necessary there in order to maintain reflection symmetry . here , however , this symmetrization is unnecessary , because it has no effect in perturbation theory .
the 2-point function @xmath24 thus has the form @xmath25 with @xmath26 to make the notation more transparent , call @xmath27 and denote @xmath28 and analogously for @xmath29 , @xmath30 , @xmath31 ( see figure [ figure : triangles ] ) .
the indices just introduced are denoted by @xmath32 , @xmath33 , @xmath34 , etc . in the following sections ,
only formulae for @xmath35 and @xmath36 will be presented .
the full topological charge density can be recovered by means of ( [ action : densitydecomp ] ) and ( [ action : expressiontriangles ] ) , and the full two - point function by adding the other three contributions in ( [ expansion : twopointfndecomp ] ) , obtained by substituting @xmath37 , @xmath38 and @xmath39 for @xmath40 , respectively . moreover
, a numerical factor will be introduced in the definition of the correlator so as to work with simpler expressions : @xmath41 we first consider a symmetrized version of the field theoretical ( ft ) definition @xcite of the topological charge density @xmath42 which gives rise to the two - point function @xmath43 \left [ \vec{w}_1 \cdot ( \vec{w}_2 \wedge \vec{w}_3 ) \right ] \right\rangle = \left\langle \det_{i,\,j } ( \vec{z}_i \cdot \vec{w}_j ) \right\rangle .\ ] ] the perturbative treatment of the problem begins with the @xmath44 decomposition of the spins , @xmath45 substituting in the determinant , we get @xmath46 the perturbative expansion of ( [ expansion : twopointfnyetagain ] ) consists of a taylor expansion in @xmath47
. only integral powers of @xmath48 contribute . for our purposes , it will be enough to keep terms up to and including order @xmath49 .
then @xmath50 where , in general , perturbative contributions to magnitude @xmath51 are denoted by @xmath52 berg and lscher s ( bl ) definition of the topological charge density was introduced in @xcite . with each elementary triangle @xmath53 of the chosen triangulation one
associates a multiple of the signed area of the minimal spherical triangle determined by @xmath54 on the unit sphere .
explicitly , @xmath55 the corresponding perturbative two - point function is , up to and including order @xmath49 , @xmath56 substituting ( [ expansion : othreetootwo ] ) in ( [ bl : twopoint ] ) and taylor expanding the result , we arrive at @xmath57 where @xmath58 will be referred to as ` bl difference . ' in this section we use the bl topological charge density ( [ bl : defqbl ] ) and will comment on the difference with the ft definition when appropriate
. therefore we drop the bl labels and write @xmath60 for bookkeeping purposes we do the perturbative computation of the density correlator out keeping factors @xmath61 explicit , as for an o@xmath62 problem , although the topological density has a meaning only for @xmath63 . the only vertices relevant for feynman diagrams up to and including order @xmath49 are order @xmath48 : * two-@xmath64 vertex : @xmath65 * four-@xmath64 vertex : @xmath66 , } \end{array } } \ ] ] where subindices are lattice points , superindices are o@xmath62 indices , @xmath67 is the number of nearest neighbors of point @xmath68 , two - index deltas are kronecker , @xmath69 and @xmath70 the explicitly volume - dependent factor in ( [ computation : twopivertex ] ) comes from fixing one spin using the faddeev - popov method as in @xcite to eliminate the divergent zero mode that would otherwise appear in the gaussian measure defining tree level perturbation theory with periodic boundary conditions . in order to compute the green s functions relevant to the correlators being studied , we start from the corresponding quantity on the @xmath71 square lattice and put @xmath72 , @xmath73 .
@xmath74 where the momenta @xmath75 are summed over the first brillouin zone @xmath76 , the prime meaning exclusion of the zero mode @xmath77 .
the thermodynamic ( @xmath8 ) limit of ( [ green : lattice ] ) does not exist , since the sum diverges logarithmically in @xmath78 .
the divergence is independent of @xmath79 , however , and the thermodynamic limit of the difference @xmath80 does exist . according to a general argument due to david @xcite ,
the perturbation expansion of invariant observables should be free of infrared divergences ; the first sign of this is the fact that everything is expressed in terms of @xmath81 , which has a well defined thermodynamic limit given by the integral @xmath82 ( another possible source of infrared divergence is the sum over intermediate lattice sites , which will be addressed in section [ section : numeval ] ) .
obviously @xmath83 , @xmath84 .
this subtracted propagator was studied by other methods in @xcite , which we will use as a check for our results . due to parity , the numerator in the integrand of ( [ green : integralsubtracted ] ) can be replaced by @xmath85 .
integration in @xmath86 can be performed by the calculus of residues , with the result @xmath87 here t@xmath88 is chebyshev s polynomial of the @xmath89-th kind .
it is straightforward to check that ( [ green : integralchebyshev ] ) behaves asymptotically as @xmath90 with @xmath91 , @xmath92 constants .
according to @xcite , @xmath93 where @xmath94 is euler s constant .
similarly it can be seen that @xmath95 and @xmath96 for large @xmath97 and @xmath98 . to see this
, one subtracts from the fourier representation of the lhs of ( [ 2ndasymp ] ) and [ mixasymp ] the corresponding continuum expression with a suitable smooth high momentum cutoff ; this difference , being the fourier transform of a smooth ( @xmath99 ) function will decay faster than any power of @xmath97 .
the cutoff on the continuum expression , on the other hand , does not affect the leading long distance behavior , which is obtained straightforwardly .
expressions ( [ 2ndasymp ] ) and ( [ mixasymp ] ) , in which the constants @xmath91 and @xmath92 in ( [ green : asymptotics ] ) do not appear , will be useful in the computation of the charge correlator . in the following , we will denote green s functions by @xmath100 even when subtracted .
the tree level contribution to @xmath101 is given by @xmath102 the correlator in the rhs being evaluated in the gaussian measure with one spin fixed .
the corresponding diagrams have the structure of figure ( [ graph : treelevel ] ) .
contracting spin indices , @xmath103 the @xmath104 matrices @xmath105 and @xmath106 are symmetric , and therefore the second term in the rhs vanishes .
playing with indices , we can recast equation ( [ treelevel : afterdiagrams ] ) in the form @xmath107 \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } g _ { z_j w_m }
g _ { z_k w_n } , \ ] ] this can be written in terms of differences @xmath108 with @xmath109 ( which restricts ir divergences to summation effects , the ir divergences of green s functions being cancelled in the differences ) : @xmath110 } \\ & & \\ & & { \displaystyle \qquad \times \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } ( g _ { z_j w_m } - g _ { z_k w_m } ) ( g _ { z_k w_m } - g _ { z_k w_n } ) } \end{array } } \ ] ] we can now perform the sums over all indices : in general , for a @xmath111 matrix @xmath112 , @xmath113 \delta_{\ell i } + 2 ( a^2 ) _
{ \ell i } - 2 ( \tr a ) a_{\ell i } \right\ } } \\ & & \\ & = & { \displaystyle 3 \left [ ( \tr a)^2 - \tr(a^2 ) \right ] + 2 \ , \mathrm{s } ( a^2 ) - 2 ( \tr a ) \ , \mathrm{s}(a ) , } \end{array } } \ ] ] where @xmath114 denotes the sum of all entries in @xmath115 .
this identity can be obtained as follows : one starts from the well - known identity @xmath116 and expands it to order @xmath117 to obtain @xmath118 by ` polarization ' , i.e. the replacement of @xmath119 by @xmath120 on both sides and comparing the coefficient of @xmath121 eq .
( [ treelevel : sumsoverindices ] ) follows if we put @xmath122 and take for @xmath119 the matrix with all entries equal to 1 .
in our case , @xmath123 } \\ & & \\ & & { \displaystyle \quad \times \left\ { 3 \left [ ( \tr g _ { z_\cdot w_\cdot } ) ^2 - \tr ( g _ { z_\cdot w_\cdot } ^2 ) \right ] + 2 \ , \mathrm{s } ( g _ { z_\cdot w_\cdot } ^2 ) - 2 ( \tr g _ { z_\cdot w_\cdot } ) \ , \mathrm{s } ( g _ { z_\cdot w_\cdot } ) \right\ } , } \end{array } } \ ] ] the one - loop ( order @xmath49 ) correction to the correlator can be decomposed as @xmath124 according to the structures depicted in figure ( [ graph : loop ] ) .
@xmath125 ( i ) contains the contributions of the 4-legged , disconnected graphs with one 2-vertex : @xmath126 } \\ & & \\ & & { \displaystyle \qquad \times \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } g_{z_j w_m } \sum_p g_{z_k p } g_{w_n p } .
} \end{array } } \ ] ] @xmath125 ( ii ) contains the x - shaped contributions : @xmath127 \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } } \\ & & \\ & & \qquad { \displaystyle \times \nnsum{p}{q } \left ( g_{z_j p } g_{w_m p } - g_{z_j q } g_{w_m q } \right ) \left ( g_{z_k p } g_{w_n p } - g_{z_k q } g_{w_n q } \right ) . }
\end{array } } \ ] ] @xmath125 ( iii ) gathers the contributions of 4-legged graphs with tadpoles : @xmath128 \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } g_{z_j w_m } } \\ & & \\ & & { \displaystyle \qquad\qquad \times \nnsum{p}{q } ( g_{pp } - g_{qq } ) \left ( g_{z_k p } g_{w_n p } - g_{z_k q } g_{w_n q } \right ) } \\ & & \\ & & { \displaystyle - \ , \left [ ( n-1)^2 - ( n-1 ) \right ] \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } g_{z_j w_m } } \\ & & \\ & & { \displaystyle \qquad\qquad \times \nnsum{p}{q } \big [ g_{pp } g_{z_k p } g_{w_n p } + g_{qq } g_{z_k q } g_{w_n q } } \\ & & \\ & & { \displaystyle \qquad\qquad\qquad\qquad - g_{pq } \left ( g_{z_k p } g_{w_n q } + g_{z_k q } g_{w_n p } \right ) \big ] .
} \end{array } } \ ] ] the first term of ( iii ) vanishes by translation invariance .
@xmath125 ( iv ) consists of the six - legged graphs contributing to the correlator according to the ft definition of the charge density : @xmath129 \det g_{z_\cdot w_\cdot } } \\ & & \\ & & { \displaystyle - \ , \frac{n-1}{2 } \ , \left [ ( n-1)^2 - ( n-1 ) \right ] \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } g_{00 } g_{z_j w_m } g_{z_k w_n } } \\ & & \\ & & { \displaystyle - \ , \left [ ( n-1)^2 - ( n-1 ) \right ] } \\ & & \\ & & { \displaystyle \qquad \times \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } g_{z_j w_m } \left ( g_{z_i z_k } g_{z_i w_n } + g_{z_k w_\ell } g_{w_\ell w_n } \right ) . } \end{array } } \ ] ] the determinant term in ( iv)@xmath130 has a vanishing prefactor .
@xmath125 ( v ) is the bl difference , also consisting of six - legged graphs : @xmath131 \left [ \sum_{u , v } \left ( g_{z_u z_v } - g_{00 } \right ) \right ] } \\ & & \\ & & { \displaystyle \qquad \times \sum_{ijk\ell
mn } \varepsilon_{ijk } \varepsilon_{\ell mn } g_{z_j w_m } g_{z_k w_n } } \\ & & \\ & & { \displaystyle - \ , \frac{1}{4 } \
, \left [ ( n-1)^2 - ( n-1 ) \right ] \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } g_{z_j w_m } } \\ & & \\ & & { \displaystyle \qquad \times \sum_{u , v } \left [ g_{z_u z_k } \left ( g_{z_v w_n } - g_{z_u w_n } \right ) + g_{z_k w_u } \left ( g_{w_n w_v } - g_{w_n w_u } \right ) \right ] . } \end{array } } \ ] ] as happened with the tree level expression , the one - loop correction can be rewritten in terms of differences @xmath108 with @xmath109 .
this works independently for ( i ) , ( ii ) and ( v ) : @xmath132 \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } } \\ & & \\ & & { \displaystyle \qquad \times ( g_{z_j w_m } - g_{z_k w_m } ) \sum_p ( g_{z_k p } - g_{z_i p } ) ( g_{w_n p } - g_{w_m p } ) , } \end{array } } \ ] ] @xmath133 \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } } \\ & \\ & { \displaystyle \times \nnsum{p}{q } \big [ ( g_{z_j p } - g_{z_j q } ) - ( g_{w_m p } - g_{w_m q } ) \big ] \big [ ( g_{z_k p } - g_{z_k q } ) - ( g_{w_n p } - g_{w_n q } ) \big ] } \\ & \\ & \quad { \displaystyle \times \big [ ( g_{z_j p } - g_{z_k p } ) + ( g_{z_j q } - g_{z_k q } ) - ( g_{w_m p } - g_{w_n p } ) - ( g_{w_m q } - g_{w_n q } ) \big]^2 , } \end{array } } \ ] ] @xmath134 \left [ \sum_{u , v } \left ( g_{z_u z_v } - g_{00 } \right ) \right ] } \\ & & \\ & & \qquad { \displaystyle \times \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } ( g_{z_j w_m } - g_{z_k w_m } ) ( g_{z_k w_m } - g_{z_k w_n } ) } \\ & & \\ & & { \displaystyle - \ , \frac{1}{8 } \ ,
\left [ ( n-1)^2 - ( n-1 ) \right ] \sum_{ijk\ell mn } \varepsilon_{ijk } \varepsilon_{\ell mn } ( g_{z_j w_m } - g_{z_k w_m } ) } \\ & & \\ & & \qquad { \displaystyle \times \sum_{u , v } \big [ ( g_{z_u z_k } - g_{z_v z_k } ) ( g_{z_v w_n } - g_{z_u w_n } ) } \\ & & \\ & & \qquad\qquad\quad { \displaystyle + ( g_{z_k w_u } - g_{z_k w_v } ) ( g_{w_n w_v } - g_{w_n w_u } ) \big ] , } \end{array } } \ ] ] while ( iii)@xmath130 and ( iv)@xmath130 can not be brought to that form .
however , the sum ( iii)@xmath130 + ( iv)@xmath130 can , because , using translation invariance and the fact that @xmath135 , @xmath136 and @xmath137 the first terms in the rhs of ( [ thegraphs : graphiiidiff ] ) and ( [ thegraphs : graphivdiff ] ) cancel .
green s functions appearing ( up to 1-loop order ) in the perturbative expansion of @xmath139 are of the form @xmath140 , where * @xmath141 is one of @xmath142 , @xmath143 ( where @xmath144 is summed over all lattice points ) , @xmath145 and @xmath146 , * @xmath147 is a linear combination @xmath148 of unit lattice vectors with @xmath149 integers ( @xmath150 up to 1-loop order . )
the following notation proves useful when applying computer techniques to this perturbative problem : @xmath151 here @xmath152 is one of the _ letters _ @xmath153 , @xmath154 , @xmath155 or @xmath156 according to the vector @xmath141 . the subscript is the minimal _ word _ of the form @xmath157 , @xmath158 , matching vector @xmath147 under the rule @xmath159 ( obviously , only words of the forms @xmath160 , @xmath161 , @xmath162 , @xmath163 can be minimal . ) for instance , @xmath164 this notation is way more readable than the standard notation , and allows us to cope with the full form of the density correlator , including all four terms in ( [ expansion : twopointfndecomp ] ) .
a c program was used to generate the full analytic forms in @xmath138 notation for each contribution ( tree level and one - loop contributions i through v ) , its output being simplified by means of a computer algebra program .
the simplification thus achieved is dramatic , witness the results in next section .
the _ full _ tree level 2-point topological charge correlator between 0 and @xmath165 , both in standard and @xmath138 notation , reads : @xmath166 } \\ & & \\ & = & { \displaystyle 2 \bigg [ ( \zeta_{ab } + \zeta_{cd } - 2 \zeta ) ( \zeta_{ad } + \zeta_{cb } - 2 \zeta ) - ( \zeta_a - \zeta_b + \zeta_c - \zeta_d ) ^2 \bigg ] . } \end{array } } \ ] ] from this expression and the asymptotics ( [ 2ndasymp ] ) and ( [ mixasymp ] ) of the differences of green s functions , the leading term of the asymptotic behavior for @xmath167 can be easily computed : @xmath168 the full 1-loop contributions will only be presented in @xmath138 notation ( cancelling terms in ( iii ) and ( iv ) are omitted ) : @xmath125 the graphs with 2-vertices : @xmath169 } \\ & \\ & { \displaystyle \quad - \ , ( \zeta_{ab } + \zeta_{cd } - 2 \zeta ) \sum_p ( \xi_a - \xi_b ) ( \eta_a - \eta_b ) - \ , ( \zeta_{ad } + \zeta_{cb } - 2 \zeta ) \sum_p ( \xi_{ab } - \xi ) ( \eta_{ab } - \eta ) \big ] , } \end{array } } \ ] ] @xmath125 the x - shaped graphs : @xmath170 } \\ & & \\ & & { \displaystyle \qquad\quad \times [ ( \eta_{ab } - \eta ) ( \eta_{ab } - \eta_{aa } ) - ( \eta_a - \eta_b ) ( \eta_a - \eta_{aab } ) ] } \\ & & \\ & & { \displaystyle + \ , 2 \ , \sum_p [ ( \xi_{ab } - \xi ) ( \xi_{ab } - \xi_{bb } ) + ( \xi_a - \xi_b ) ( \xi_b - \xi_{abb } ) ] } \\ & & \\ & & { \displaystyle \qquad\quad \times [ ( \eta_{ab } - \eta ) ( \eta_{ab } - \eta_{bb } ) + ( \eta_a - \eta_b ) ( \eta_b - \eta_{abb } ) ] , } \end{array } } \ ] ] @xmath125 the tadpole graphs : @xmath171 } \\ & \\ & { \displaystyle \quad + ( \zeta_{ad } + \zeta_{cb } - 2 \zeta ) \sum_p \big [ - ( \xi_{ab } - \xi ) ( \eta_{aab } + \eta_{abb } - \eta_c - \eta_d ) - ( \xi_{aab } + \xi_{abb } - \xi_c - \xi_d ) ( \eta_{ab } - \eta ) \big ] } \\ & \\ & { \displaystyle \quad + ( \zeta_a - \zeta_b + \zeta_c - \zeta_d ) \sum_p \big [ ( \xi_{ab } - \xi ) ( \eta_{aa } + \eta_{ad } - \eta_{bb } - \eta_{cb } ) + ( \xi_{aa } + \xi_{ad } - \xi_{bb } - \xi_{cb } ) ( \eta_{ab } - \eta ) } \\ & \\ & { \displaystyle \qquad\qquad\qquad\quad\qquad\quad + ( \xi_a - \xi_b ) ( \eta_{aab } + \eta_{abb } - \eta_c - \eta_d ) + ( \xi_{aab } + \xi_{abb } - \xi_c - \xi_d ) ( \eta_a - \eta_b ) \big ] \big\ } , } \end{array } } \ ] ] @xmath125 graphs with six external legs in the ft definition : @xmath172 , \ ] ] @xmath125 the bl difference : @xmath173 .\ ] ] note that ( iv@xmath174 ) and ( v@xmath174 ) do not involve sums over lattice points , and that they are proportional to the tree level expression ( [ formulae : treelevel ] ) .
their sum reads @xmath175 .\ ] ] the simplicity of these expressions might be a hint of the existence of a more direct derivation .
the procedure we employ to evaluate the correlator numerically follows the conventional philosophy adopted in formal perturbation theory : the contributions to each order are first evaluated in a finite volume and then the thermodynamic limit is taken order by order .
it is hoped ( though far from proven ) that by this procedure one obtains an asymptotic expansion of the infinite volume correlator . for a more detailed discussion of the difficult issues involved
see for instance @xcite .
the various perturbative contributions to the charge density correlator @xmath176 were computed , by means of c programs , for @xmath165 in a finite @xmath177 sublattice of the whole @xmath178 .
determines the range of @xmath97 for which topological charge correlators @xmath14 are computed .
@xmath179 is a cutoff on the sum over green s functions @xmath180 .
@xmath181 determines the range for which exact values of @xmath180 , as opposed to asymptotic approximations , are used . ]
lattice symmetry allows to reconstruct all correlators in the @xmath182 square lattice ( in our case , @xmath183 ) from those in the shaded region .
as explained in the next sections , the computation of @xmath14 involves sums over green s functions @xmath180 with @xmath97 arbitrarily large , and a cutoff @xmath179 on the summation range is imposed .
also , since we have an asymptotic expression for @xmath180 , we only need to compute green s functions exactly for @xmath97 in a smaller region determined by another cutoff @xmath184 ; for larger @xmath97 the asymptotic expression is used .
figure [ figure : green ] shows the ranges @xmath78 , @xmath181 , @xmath179 involved in the computation .
it is impractical to compute and store an array of all possible values of ( subtracted ) green s functions @xmath180 according to the exact expression ( [ green : integralchebyshev ] ) .
it is far more convenient to store only such values as differ noticeably from the approximate expression ( [ green : asymptotics ] ) .
we chose to compute exact values for @xmath180 with @xmath97 less than a certain value @xmath185 ( see figure [ figure : green ] ) , and approximate values for the remaining ones as they were needed .
the constant appearing in ( [ green : asymptotics ] ) must be computed for this procedure to be useful . working with @xmath186 ,
exact green s functions @xmath187 were computed for the triangular region @xmath188 , @xmath189 .
all those in the region @xmath190 can be obtained from them by symmetry .
exact values of @xmath181 at the edge @xmath191 of the triangle were used to fit the green s function space dependence to ( [ green : asymptotics ] ) , since this edge is the furthest removed from the centre and therefore provides the best agreement with the asymptotic form . a value @xmath192 is obtained for the constant ( see figure [ fig : greens ] ) , which agrees with the analytical expression ( [ green : valuesofcd ] ) taken from @xcite . for @xmath193 , @xmath194 .
the dashed line corresponding to the fitted dependence is indistinguishable from the computed points in this range . ] for @xmath186 , the numerical results for the correlators are stable with respect to 10% changes in @xmath181 .
the tree level correlator ( [ formulae : treelevel ] ) , as well as one - loop contributions ( [ formulae : oneloopfour ] ) and ( [ formulae : oneloopfive ] ) , do not involve sums over lattice points , and are readily computed for each @xmath165 .
one - loop contributions ( [ formulae : oneloopone ] ) , ( [ formulae : onelooptwo ] ) and ( [ formulae : oneloopthree ] ) involve sums over points arbitrarily far away from this sublattice , and a cut - off must be introduced .
we choose to restrict the sum to points within the circle of radius @xmath179 about the origin . at @xmath195 , results are stable with respect to 10% changes in @xmath179 or in the shape of the summation domain ( circle vs. square ) .
this indicates the absence ( or cancellation ) of any infrared divergences connected with the summation over intermediate lattice points .
results are presented for the computations outlined in the previous sections .
range parameters are @xmath183 , @xmath186 , @xmath195 . , at tree level and one loop , as a function of @xmath196.,title="fig : " ] + , at tree level and one loop , as a function of @xmath196.,title="fig : " ] figure [ fig : nearorigin ] shows the ( nonuniversal ) near - origin tree level and one loop contributions to the topological charge density correlator @xmath197 as a function of @xmath196 . note that both contributions are positive at the origin , negative everywhere else , and are suppressed for large @xmath196
the peak at the origin results , in the continuum limit , in the singular contact term discussed e.g. in @xcite .
the values of the peak at tree level and one loop level are @xmath198 the tree level peak can also be calculated analytically from ( [ green : integralchebyshev ] ) and ( [ formulae : treelevel ] ) , yielding @xmath199 .
since this is a perturbative analysis , the topological susceptibility ( [ action : susceptibility ] ) should vanish order by order in @xmath48 .
summing the computed correlators over the whole @xmath200 region , the following results are obtained : @xmath201 which are compatible with zero and serve as a check of our computations .
note , in particular , that the problems deemed to arise with @xmath202 in the continuum limit of the @xmath0 model do not show up in this perturbative treatment .
figure [ fig : tree ] shows the tree level contribution multiplied by @xmath203 , as a function of @xmath196 .
this product approaches for large @xmath196 the constant -0.2026 , compatible with ( [ formulae : asympttreezeta ] ) , so the tree level result ( including all numerical factors ) is @xmath204 for @xmath205 . multiplied by @xmath203 as a function of @xmath196 . ]
the one - loop contributions ( [ formulae : oneloopone ] ) through ( [ formulae : oneloopfive ] ) to the correlator were computed and their behavior for large @xmath196 analyzed : @xmath206 with @xmath207 , @xmath208 , @xmath209 .
these anstze were obtained by inspection of the plots and trial and error , except for the last one , which stems directly from the proportionality of ( [ formulae : oneloopfourplusfive ] ) to ( [ formulae : treelevel ] ) and the fact that @xmath210 , @xmath211 , which results in the asymptotics in ( [ results : itovasympt ] ) with @xmath212 . the leading terms in ( i@xmath174 ) and ( iii@xmath174 ) cancel , but subleading terms @xmath213 and @xmath214 remain .
the sum of all one - loop contributions ( see figure [ fig : loop ] ) behaves as @xmath215 with constants @xmath216 note that these are not identical to @xmath217 and @xmath92 in ( [ results : itovasympt ] ) , because the subleading terms in the asymptotic behavior of ( i@xmath174 ) and ( iii@xmath174 ) contribute .
multiplied by @xmath203 as a function of @xmath218 , together with a linear fit for large @xmath196 . ]
the same analysis was performed along the diagonal of points @xmath219 with identical results for the asymptotic forms of the correlator : full rotational invariance is restored in this regime .
we now discuss the continuum limit , again in the spirit of formal perturbation theory , i.e. termwise in the perturbative expansion ; no claim can be made that the resulting expansion is asymptotic to a nonperturbatively defined continuum limit .
up to one loop we have found for the correlator on the unit lattice and large lattice distances @xmath220 latexmath:[\[\mathcal{g}_{\mathbf{x}}\sim - \frac{1}{\beta^2}\ \frac{1}{32\pi^4{|\mathbf{x}}|^4}- \frac{1}{\beta^3}\ \frac{1}{64\pi^2{|\mathbf{x}}|^4}(c_0+c_1 \ln to obtain a continuum limit we have to introduce the lattice spacing @xmath115 , rescale @xmath222 and make @xmath223 dependent on the cutoff @xmath115 .
we also have to rescale the correlator @xmath224 by a factor @xmath225 according to its engineering dimension , to obtain the correlator @xmath226 in continuum normalization @xmath227 valid asymptotically for @xmath228 .
a continuum limit should exist if we let @xmath223 depend logaritmically on @xmath115 according to the one - loop callan - symanzik @xmath223 function for the @xmath0 model @xcite : @xmath229 inserting this in ( [ contnormcorr ] ) and reexpanding to order @xmath230 we see that the terms proportional to @xmath231 cancel if @xmath232 , which is consistent with the value 0.0645 produced by our numerical computation ; the continuum limit of the correlator is then @xmath233 } \right ) + o \left ( \beta_0^{-4 } \right)\ .\ ] ] by choosing a renormalization scale @xmath234 we obtain @xmath235 if we had chosen the ft instead of the bl definition , the only difference wuld have been a change in @xmath236 , corresponding to different renormalization scale , so to the order considered the two definitions are related by a finite renormalization . equation ( [ contcorr ] ) is now valid for all @xmath237 , since in the limit @xmath238 only the asymptotic behavior of @xmath14 survives .
we can compare this result with the renormalization group improved tree level result @xcite @xmath239 if we reexpand ( [ rgtree ] ) to order @xmath230 , using the one - loop rg flow @xmath240 the expressions ( [ contcorr ] ) and ( [ rgtree ] ) agree to order @xmath241 if the renormalization scales are chosen appropriately .
so we have established that our one - loop calculation supports the softening of the short distance behavior predicted by the rg improved tree level result .
we computed the two - point functions of the topological charge density perturbatively to one loop and found consistency with the rg improved tree level perturbative result , indicating a softening of the short distance singularity compared to the naive tree level result .
this means that at the level of formal one - loop perturbation theory the requirements of the two positivities analyzed in @xcite are indeed satisfied .
but we should point out again that a mathematical justification of the procedures of formal perturbation theory ( interchange of the weak coupling limit with the thermodynamic and continuum limits ) does not exist , therefore it would be very interesting to see if this softening is really present at the nonperturbative level .
numerical checks using monte carlo simulations are presumably very difficult , however , since the check requires identifying logarithmic corrections to a rather large power .
we are grateful to peter weisz for a critical reading of the manuscript and for drawing our attention to reference @xcite .
99 m. aguado and e. seiler , _ the clash of positivities in topological density correlators _ , mpp-2005 - 18 , hep - lat/0503015 . j. balog and m. niedermaier , _ nucl .
_ * b500 * ( 1997 ) 421 .
e. vicari , _ nucl .
_ * b554 * ( 1999 ) 301 .
a. di giacomo , f. farchioni , a. papa and e. vicari , _ phys .
_ * b276 * ( 1992 ) 148 ; _ phys .
_ * d46 * ( 1992 ) 4630 .
b. berg and m. lscher , _ nucl .
_ * b190[fs3 ] * ( 1981 ) 412 .
p. hasenfratz , _ phys .
_ * b141 * ( 1984 ) 385 .
f. david , _ phys .
_ * b96 * ( 1980 ) 371 .
shin , _ nucl .
_ * b525 * ( 1998 ) 457 .
a. patrascioiu and e. seiler , _ phys .
_ * d57 * ( 1998 ) 1394 .
e. seiler and i. o. stamatescu , _ some remarks on the witten - veneziano formula for the eta - prime mass _ , mpi - pae / pth 10/87 .
e. brzin and j. zinn - justin , _ phys .
_ * b14 * ( 1976 ) 3110 . | to check the consistency of positivity requirements for the two - point correlation function of the topological charge density , which were identified in a previous paper , we are computing perturbatively this two - point correlation function in the two - dimensional @xmath0 model .
we find that at the one - loop level these requirements are fulfilled .
mpp-2005 - 42 + perturbation theory for @xmath0 topological charge correlators + |
the iron peak elements are synthesized in supernova environments . the abundance data for these elements in galactic halo and disk stars can provide important constraints on the conditions ( _ i.e. _ , the elemental content of the ejecta , the supernova mass cut , explosive energies , etc .
) that occur in explosive nucleosynthesis in , for example , type ii supernovae ( sne ii ) .
abundance trends with metallicity for these elements , _
i.e. _ galactic chemical evolution studies , can also provide insight into the individual contributions from both type i ( sne ia ) and sne ii nucleosynthesis and into the progenitor masses for these objects .
an examination of these iron - peak elements and the associated abundance trends , comparing stars with varying metallicities , also provides direct insights into the chemical evolution of the galaxy - quantifying the nature and frequency of the sources of the production of these elements .
these studies of iron - peak elements , unfortunately , have suffered due to a lack of precise abundance values .
for example , it has been possible to obtain stellar photospheric abundances of cr from the moderately - strong ( neutral ) cr i lines , which have accurately determined atomic properties , _
i.e. _ , log gf values . on the other hand ,
the weaker ionic cr ii lines commonly employed in abundance analyses tend to not have as well determined atomic properties ( measurement associated errors ; sobeck et al .
there are significant discrepancies in the abundances determined using cr i lines as opposed to cr ii lines ( especially in metal deficient stars ) , which may be due to errors in the atomic properties of cr ii or in the atmospheric models and line transfer codes employed to determine these abundances ( sobeck et al .
2007 ) .
the abundance trends discussed in this paper were originally noticed by @xcite , who measured abundances of numerous elements in stars down to a metallicity of [ fe / h]=-4 . in particular , they found that below [ fe / h]=-2.4 cr / fe decreased and co / fe increased with decreasing fe / h .
thus , the co / cr ratio increases overall as fe / h drops .
they speculated that this particular behavior could be explained by the effects of metallicity or progenitor mass on stellar yields .
in addition , alpha freeze - out , in which a high photon to baryon ratio in metal - poor stars results in an elevated volume density of alpha particles , favors the synthesis of nuclei heavier than fe , e.g. co , at the expense of lighter nuclei , e.g. cr . at the same time , @xcite point out that the mn / fe ratio also increases with fe / h and suggest that this may be linked to the contribution of type ia supernovae , sne ia . the extensive study by @xcite employed detailed chemical evolution models and the massive star yields of @xcite in an attempt to model the evolution of numerous elements , including those of the fe - peak , using extant data available at the time as constraints .
since the data did not extend below -3 in [ fe / h ] , the trends noted by @xcite were not noticeable in the plots presented in @xcite .
however , their models were successful in modeling the available data
. the trends in cr / fe , co / fe , and mn / fe were considered in light of supernova parameters by @xcite . noting that stable co and fe
are produced by isotopes originating in the complete silicon burning region during explosive nucleosynthesis , while cr and mn precursors come from the incomplete silicon burning zone farther out , these authors examined the effects of mass cut position , neutron excess , explosion energy , and progenitor mass on these element ratios .
they found that the trends in the element ratios could be duplicated if the position of the mass cut migrated outward as metallicity increased .
varying the other three parameters independently appeared to have a less dramatic effect .
a major effort was made in this regard by @xcite , who compiled data for 12 elements , including six fe - peak elements ti , cr , mn , fe , co , and ni , and analyzed the trends present in [ x / fe ] vs. [ fe / h ] plots .
their analysis centered on the computation of a detailed two - infall ( timescale ) chemical evolution model for the milky way .
this particular model was originally developed by @xcite , who used independent formation time scales for the halo - thick disk and the thin disk as well as a surface density threshold for star formation to reproduce numerous features of the milky way galaxy .
@xcite employed the massive star yields of @xcite for solar composition along with their own yield scaling factors to force matches to the data trends .
their final product is a table of recommended massive star yields derived empirically in this way , where the yield in solar masses is given as a function of stellar mass .
a summary of their recommended changes to the original @xcite yields was also provided .
more recently , @xcite used new yield calculations to compute a chemical evolution model of the galaxy .
their predictions for fe - peak elements are generally in line with observations , although the specific behavior of cr , co , and mn at low metallicity are not closely reproduced . in summary ,
the trends found in the fe - peak elements have inspired several promising theories about how they arise . at the same time , it is clear that the uncertainties involved in the computation of massive star yields are significant ; the mass cut position and explosion energy for supernova models in particular would seem to be relatively unconstrained at this point .
this situation makes it difficult to derive a reliable set of stellar yields .
the goal of this paper is to approach the problem from the reverse direction . by initially adopting the idea of @xcite
, we employ detailed chemical evolution models to derive empirically a set of robust integrated yields , i.e. values arrived at by integrating stellar yields over an initial mass function . in this way we produce quantities which serve as real targets for future yield computations .
that is , regardless of the assumptions going into the stellar yield computations regarding mass cut position or explosion energy , the theoretical predictions , when integrated over an imf , must match the inferred integrated yields .
the latter then serve as major constraints for yield studies for the fe - peak elements studied here . as is explained below , this is largely possible because at [ fe / h ] values below about -2 , where the interesting trends appear , the instantaneous recycling approximation is a good one , and thus details of star formation history do not play a significant role .
we also note that we have included many more data in our analysis than were available , particularly at higher metallicities , for earlier studies , for example those by mc william ( 1997 ) .
the sources for the observational data used in this study , along with the abundance trends , are discussed in section [ data ] .
sections [ analysis ] and [ discussion ] contain the results of our model studies of the observed trends . in section [ summary ]
we summarize our work and list our conclusions .
for our pool of element abundance data , we draw from the following references : @xcite , @xcite , @xcite , @xcite , @xcite and @xcite . in general
, our sample consists of high - resolution spectral data and ( combined ) covers a metallicity range of -4.0@xmath0[fe / h]@xmath0 + 0.5 . with regard to exact element abundances , there are data for these species : ti
i / ii , v i / ii , cr
i / ii , mn i , fe i / ii , co i , and ni i. the specific observational characteristics associated with each of the data sources are listed in table [ sobeck ] .
all of the chosen references employ the fundamental assumptions of plane - parallel geometry , one - dimensionality , and local thermodynamic equilibrium ( lte ) .
we also remark that each of the studies utilize different telescope - instrument set - ups , data reduction techniques , stellar model atmospheres , and line transfer codes ( there is no single , consistent approach for our sample ) .
we point out that abundance analyses , which employ a three - dimensional or non - lte methodology , provide a rigorous treatment of radiative transfer in stars . however , such approaches up to this point have been limited in scope and have not yet been employed to determine abundances over the entire
_ observable range _ of metallicity or in the many different stellar populations ( i.e. there is a limited amount of available data ) .
furthermore , the magnitude of the change in the abundance data that occurs with the implementation of these methodologies depends upon a variety of factors including effective temperature , surface gravity , and metallicity ( i.e. not all stellar abundances are affected _ similarly _ ) . hence in order to assemble a large and cohesive data sample , we employ data exclusively from lte analyses .
figure [ 6plot_notracks ] displays the results of our data compilation for ti i , v i , cr i , mn i , co i , and ni i , where in each panel we plot [ x / fe ] vs. [ fei / h ] .
symbol color indicates the source of abundances , as defined in the legend .
below we discuss each of the general element abundance trends as a function of metallicity and remark upon any anomalous data points / behavior .
_ titanium _ : in the upper left panel , the [ ti / fe ] abundance ratio appears to increase from its solar value to [ ti / fe ] @xmath1 + 0.4 in the range -1.0@xmath2[fe / h]@xmath2 + 0.40 .
it then hovers around [ ti / fe]=+0.4 as metallicity is decreased with some scatter at the extremely metal - poor end ( especially from the @xcite data ) .
this trend in the ti data is also seen in the abundances compilations from @xcite and @xcite over a similar metallicity range .
_ vanadium _ : as shown in the upper middle panel , [ v / fe ] remains roughly at its solar value across the full range of metallicity .
there are four aberrantly high data points present from the @xcite sample at about [ fe / h]=-3.0 . we note that there are no vanadium abundances from @xcite .
a flat trend is likewise displayed in the data plotted by @xcite . _ chromium _ : the upper left panel features cr data . for -2.1@xmath2[fe / h]+0.4 , cr stays roughly constant with [ cr / fe]@xmath10 . below [ fe / h]=-2.1 ,
the [ cr / fe ] ratio declines steadily with decreasing metallicity ( again note these data are from cr i transitions only ) . a similar cr behavior is exhibited in the @xcite and @xcite compilations .
_ manganese _ : the mn trend is displayed in the bottom left panel . for the relatively metal - rich regime of -1.0@xmath3[fe / h]@xmath3 + 0.4 , mn decreases from a super - solar value to approximately [ mn / fe]@xmath1 - 0.4 .
it seems that the mn abundance is flat in the region -2.5@xmath3[fe
/ h]@xmath3 - 1.0 and then , there is a slight indication of a further downturn at [ fe / h]@xmath1 - 2.5 ( there is large scatter in the extremely metal - poor stars especially from the @xcite sample ) .
results in both @xcite and @xcite show a flat , _ subsolar _ behavior for [ mn / fe ] in the range -4.0@xmath2[fe / h]@xmath2 - 1.0 ; above a metallicity of [ fe / h]=-1 , an upward [ mn / fe ] trend is evident .
_ cobalt _ : shown in the bottom center panel , co behavior remains relatively flat at [ co / fe]@xmath10 in the range -2.2@xmath2[fe / h]@xmath2 + 0.4 .
then , the [ co / fe ] ratio proceeds to rise sharply as metallicity continues to decrease .
a duplicate co trend is seen in @xcite while @xcite lack co data below [ fe / h]@xmath1 - 2.5 .
_ nickel _ : as displayed in the bottom right panel , the [ ni / fe ] ratio hovers around its solar value for -2.0@xmath2[fe / h]@xmath2 + 0.4 . for metallicities lower than [
fe / h]=-2.0 , the ni behavior shows considerable scatter with little discernible trend .
this is in general agreement with what is exhibited in both @xcite and @xcite .
two trends which are of particular interest to us are those of cr and co in the range -4.0@xmath2[fe / h]@xmath2 - 2.2 . the abundances of these two elements seem to mirror one another in this metallicity regime see further discussion below in [ discussion ] .
we now use the trends shown in fig . [ 6plot_notracks ] along with chemical evolution models in order to derive the integrated yield @xmath4 as a function of metallicity for each of the six elements .
we define an integrated yield of element _ x _ as @xmath5 where @xmath6 is the stellar yield , @xmath7 is the initial mass function , and @xmath8 and @xmath9 , are , respectively , the upper and lower limits to the mass range of all stars formed .
@xmath4 is then the mass fraction of all stars formed within the mass range that is eventually expelled as new element @xmath10 .
the benefit of integrated yields is that they provide a convenient means of directly comparing different sets of published stellar yield predictions , where normally those yields are presented as a function of stellar mass .
the stellar yield as a function of mass may vary widely , depending upon the assumptions made regarding mass cut , explosion energy , or other free stellar parameters .
but in the end , a set of yields when integrated over a mass function , i.e. , the integrated yield , must provide an adequate amount of each element to explain observed abundance patterns , regardless of the assumptions which went into the original yield calculations . to derive integrated yields from the observed trends discussed above , we employed a chemical evolution code to compute a track through each [ x / fe ] vs. [ fe / h ] plot , adjusting the yields by scaling factors until we achieved suitable agreement between observation and theory . at that point the yields for each element were integrated over an imf to produce a value for the integrated yield .
our chemical evolution code is the one used most recently by @xcite and is described in detail in the appendix of that paper .
briefly , the code carries out a time integration for a single zone characterized by a star formation history specified by an infall rate , an imf which was a slightly flattened ( @xmath11=-1.20 ) salpeter relation @xcite=-1.10 ) to reproduce some of the n/@xmath11 versus @xmath11/h data reported in @xcite .
] , a star formation efficiency , and a star formation law . at each time
step the new production of each element , obtained by integrating the stellar yield as a function of mass over the effective mass range , is added to the present level in order to update the current abundance level of that element in the interstellar medium .
thus , the program keeps track of the abundance of each element as a function of time and metallicity . generally speaking
, the relative contribution of stars of a particular mass is directly linked to the rate of star formation at the time in history when these stars were formed .
the elements included in the calculation were h , he , c , n , o , ne , si , s , ti , v , cr , mn , fe , co , and ni . the values for the basic set of input parameters used in all models , unless otherwise noted , are provided in table [ parameters ] .
many of these values were adopted directly from @xcite and our models were then checked for consistency against their results .
we were able to reproduce the metallicity distribution function in their fig .
38 , the age - metallicity relationship in their fig . 7 , as well as their tracks in the [ x / fe ] versus [ fe / h ] plots relevant to the elements which we are investigating in this paper ( adopting their yields ) .
we began the modeling by using the combination of massive star yields by @xcite for h through s and @xcite for ti through ni ( see the list above ) . for low and intermediate mass stars we adopted the yields of he ,
c , and n by @xcite . to determine the contributions of type ia supernovae we employed the prescriptions of @xcite along with the yields published by @xcite . for the massive stars , kobayashi et al .
publish yields for both type ii supernovae and hypernovae .
we assumed , as they did , that these two event types occur with equal frequency , and thus the stellar yield at a particular mass was set equal to the average of the supernova and hypernova yield . note that the choice of the kobayashi yields was arbitrary , as they only provided a starting point in our search for an empirical set of yields .
our outcome is not based upon our choice of the initial set of trial yields .
figure [ 6plot_tracks ] shows the observed abundances in fig . [ 6plot_notracks ] but now with model tracks added .
the solid green lines show the model results using the kobayashi yields as published and unscaled .
one can see that in the cases of ti / fe , v / fe , co / fe , and perhaps ni / fe the model underpredicts the observed abundances , particularly at low metallicity . on the other hand , the same model predicts cr / fe levels above those observed but does well with mn / fe . in the case of v
/ fe , the data are ambiguous , as it is unclear whether the ratio increases below [ fe]=-2 , as indicated by a few of the @xcite points , or remains constant , as suggested by some data points from mcwilliam et al . and @xcite .
in an alternative version of this model we employed the scheme for calculating the contributions of snia explosions discussed by @xcite .
this method , based upon empirical evidence , assumes the existence of a bimodal distribution of delay times , wherein 35 - 50% of the type ia progenitors have lifetimes around 10@xmath12 years , while the remaining systems involve small mass progenitors which require more time to evolve and have a broad distribution of delay times .
this bimodal feature results in a significant number of type ia events beginning to occur at metallicities as low as [ fe / h]=-2.0 , i.e. , significantly earlier than the prescription of @xcite would predict .
the result of using the @xcite scheme is shown with the dashed green line .
clearly there is no general improvement in the match between data and theory .
we note the significant scatter at low metallicity for most of the elements in fig . [
6plot_tracks ] , which makes interpretation of the trends more difficult .
the possible explanations for this scatter include the lack of sufficient data , the presence of the ejecta from individual nucleosynthetic events such as the first stars , and insufficient mixing of that ejecta with the interstellar medium .
we next calculated a model using the massive star yields for ti - ni by @xcite , who listed isotope yields prior to decay .
[ recall that @xcite used these yields in their analysis . ]
the isotopes whose individual yields contribute to the final yield for a particular fe - peak element , either directly or through decay , are listed in table [ isotopes ] .
the solid red line shows the results in fig . [
6plot_tracks ] , and we point out that our tracks agree closely with the model results of @xcite , as expected .
the quality of the match between observation and theory varies from element to element . for ti / fe
, our computed track closely follows the one employing the @xcite yields but neither reproduces the upward trend with declining metallicity .
roughly the same can be said for the case of v / fe .
for cr / fe our model reproduces the flat behavior of the data at metallicities above -2 , but below that point the model has less of a downturn than the data .
the model produces a good match with the data in the case of mn / fe , but falls well below the co / fe trends .
finally , in the case of ni / fe the ww95 model is consistent with the data at both the high and low metallicity ends , while sagging below the observations in the mid - range .
in order to improve the model fit to the observations and thus allow us to compute useful integrated yields , we applied scaling factors to the kobayashi yields , thereby forcing a fit to the data .
these factors were derived through trial and error and are listed in table [ scalingfactors ] . in particular , for the fe peak elements other than fe itself , we adjusted the massive star yields in a succession of trial models until we obtained a reasonable eyeball fit to the data over the metallicity range from -4 @xmath3 [ z ] @xmath3 + 0.5 .
the model results in this case are shown in fig . [ 6plot_tracks ] with a solid violet line .
we see that in the cases of ti , v , mn , and co the trends are matched fairly well .
we note that for v , a few of the points from the data of @xcite suggest an upward trend below a metallicity of -2 .
thus , we used a second scaling factor to produce a model in which the behavior of those points in particular was reproduced . in the case of ni ,
the scatter in the data below -2 is too large to speculate about the success of the model .
finally , the situation with cr will be discussed further below as it relates to co , but we simply point out here that while the apparent plateau below -3 is unexplained by our force fit model , the calculations employing either the @xcite or @xcite yields do produce a flattening in that low metallicity region .
we next derived the values for massive star integrated yields of the fe - peak elements based upon our force fit model just described .
we did this by rerunning the model incorporating the scaled kobayashi yields and at each time ( metallicity ) point integrated the scaled yields over the same imf as that used in the model calculations , i.e. a slightly flattened ( @xmath11=-1.20 ) salpeter imf , according to equation [ integratedyield ] .
the integrated yields derived in this manner are provided in table [ integratedyields ] .
note that the integrated yields for v correspond to the upper model track .
the first two columns of the table give the log of the metallicity and its value normalized to the sun @xcite , respectively , while subsequent columns list the log of the individual integrated yields .
we emphasize that these integrated yields provide useful constraints for assessing published stellar yields , since any set of yields when integrated over an imf should closely match them . note that the imf may have any form as long as the integral of the yields over that imf gives the proper value for the integrated yield that is consistent with the observations .
the integrated scaled kobayashi yields are plotted in fig .
[ p7 ] as a function of the log of metallicity ( violet tracks ) along with analogous values for the unscaled yields from the same authors ( green tracks ) .
( since we did not use scaled fe yields , only the unscaled integrated yield track is plotted for that element . ) clearly the largest offsets occur at metallicities below -2 . in the cases of ti , v , and
co , we find a clear decrease in the integrated yields as metallicity rises .
this contrasts with the predictions of @xcite , who find relatively little metallicity sensitivity for the yields of these three elements . on the other hand , for cr and mn we infer a direct relation with metallicity for the yields of these elements , while again @xcite find little metallicity effect . for ni ,
our results are qualitatively similar to those of @xcite below a metallicity of -1 , after which we find that ni yields decline while they infer a rise with increased metallicity .
quantitatively , the theoretical yields fall below our values at metallicities below solar .
while our general approach to yield evaluations resembles that of @xcite , the details of the studies are different enough that a direct comparison of results is difficult if not impossible to make .
for example , they scaled the yields of @xcite for solar metallicity only , while we employed the @xcite yields and accounted for metallicity effects .
( note that the latter set of yields was unavailable at the time of the franois et al . study . ) obviously , a valid comparison would be possible only if the two research teams had based their studies on the same set of theoretical yields .
we do point out , however , that both groups were successful at producing a model with scaled yields that resulted in good fits to the observations .
finally , we compare our integrated yields from table [ integratedyields ] , of all seven of the fe - peak elements we are considering , in fig .
[ px ] as a function of metallicity log z. here we see some interesting trends which we will attempt to interpret in the next section .
first , cr production increases while co production decreases as a function of metallicity , resulting in the two tracks crossing at around log z / z@xmath13 of -3 .
this result clearly is linked directly to the observed behavior of these two elements with metallicity ( see data for these two elements in fig . [ 6plot_notracks ] ) .
other interesting trends include the indirect behavior of v and ti with metallicity . under conditions such as we have at low metallicity where the instantaneous recycling approximation applies , one can show using standard chemical evolution equations @xcite that the only factors besides the integrated yields which might substantially influence the evolution of an element ratio are gas infall or outflow , through their diluting / concentrating effects . therefore , we ran two variations of our force fit model in which we employed star formation time scales of 2 gyr and 7 gyr , where we previously used 4 gyr in the basic model ( see table [ parameters ] ) .
we found no perceptible difference between the three models , suggesting that our integrated yield values are robust within a wide range of galaxy formation conditions and regions .
as discussed in the introduction , the particular behavior of cr / co with increasing metallicity was originally pointed out by @xcite and deserves special attention .
this behavior is illustrated in fig .
[ cr2covfel ] , where to the data displayed in figs .
[ 6plot_notracks ] and [ 6plot_tracks ] we have added abundances from @xcite indicated with green diamonds .
also shown in the figure are five well - studied r - process - rich halo stars , indicated by red circles .
the abundance ratios of cr / co in all of these high - resolution , high s / n studies ( , [ fe / h ] = 3.1 , sneden ( 2003 ) ; , [ fe / h ] = 3.0 , westin ( 2000 ) ; hd 122563 , [ fe / h ] = 2.7 , westin ( 2000 ) ; , [ fe / h ] = 2.1 , cowan ( 2002 ) ; , [ fe / h ] = 2.2 , ivans ( 2006 ) ) are consistent with the behavior of the other sample stars illustrated here . the data together show a steady increase by a factor of more than 10 in cr / co between -4 and -1.5 of [ fe / h ] , followed by a leveling off at roughly -1.5 as metallicity continues to increase . for comparison ,
data presented in @xcite , which extend down only to -3 in metallicity , show a slight decline in cr / fe , while co / fe remains level at their lowest metallicities , although the scatter is enough that it is difficult to discern a downward trend in cr / co [ @xcite do not explicitly plot cr / co ] . in fig .
[ cr2covfel ] we also show the predictions of several chemical evolution models , where the line types and colors are consistent with those used in fig . [ 6plot_tracks ] . here
we see that the model which employed the scaled kobayashi yields and successfully matched the cr and co trends with fe in fig .
[ 6plot_tracks ] satisfactorily reproduces the trend in cr / co in fig .
[ cr2covfel ] ( solid violet track ) , as expected .
this is true at least out to -3.5 in metallicity , below which the data suggest the presence of a plateau , which is not predicted by our model .
the model using the kobayashi yields as published ( solid green track ) , along with two variations of it in which all core collapse events are either snii ( solid maroon track ) or hypernovae ( solid orange track ) , are generally consistent with the data above a metallicity of -2 but fail to reproduce the downward trend at lower metallicities .
also , models in which the snia scheme of @xcite was used are indicated with dashed lines of the same color as the associated model that makes use of the same stellar yields and input parameters .
clearly , this change does not produce an improvement to the original models .
finally , the solid red line represents the model in which we employed the yields and input parameters of @xcite and @xcite , respectively .
this track does not match the data well except near solar metallicities , due in large part to the predicted behavior of co / fe ( see fig . [ 6plot_tracks ] ) .
in general , the yields of @xcite coupled with either an equal mix of snii and hypernovae or 100% hypernovae give reasonable matches to the data at metallicities above -1.5 . at metallicities below this , however , the nearly level predicted values of cr / co for these two models fail to match the downward trend . also , the all - snii model predicts cr / co values which are generally too high at all metallicities .
an additional comment concerns the apparent plateau below about -3.5 in metallicity .
clearly , our forced fit using scaled kobayashi yields does not reproduce this behavior .
at the same time , however , we note that the model using the ww95 yields does predict a roughly constant value for [ cr / co ] over this metallicity range .
combining the ww95 results below -3 with the scaled kobayashi model results above that level would clearly produce a good fit to the data .
finally , we see that the general upward trend of [ cr / co ] with metallicity for [ fe / h ] below -1.5 in fig .
[ cr2covfel ] is explained by increasing cr integrated yields and decreasing co integrated yields below [ z]=-1.2 in fig .
above that same point , while cr yields appear to level off , the co yields actually reverse direction and go up . according to the discussion in @xcite ,
the behavior of cr and co yields below -1.2 may be linked to a systematic outward migration of the mass cut with metallicity during the sn explosion .
the same models featured in the above analysis of [ cr / co ] are shown in fig .
[ ti2fevfe ] , which displays the data and models relevant to [ ti / fe ] .
again we see that the scaled kobayashi model successfully reproduces the trend in the data , while the others are less successful . in particular
, the model employing the @xcite yields ( solid red track ) fails by two orders of magnitude to reproduce the [ ti / fe ] behavior . referring again to fig .
[ px ] we see that the observed downward trend in [ ti / fe ] is consistent with a similar trend in the integrated yield of ti .
there have been two recent laboratory determinations of atomic data ( e.g. oscillator strengths ) for both the neutral ( sobeck et al . 2007 ) and first - ionized species ( nilsson et al .
2006 ) of chromium .
the study by lai et al .
( 2008 ) was the only one of our selected sample to employ these new data .
they found that [ cri / fe ] ratio decreased in the region -4.3 @xmath2 [ fe / h ] @xmath2 -2.0 while the [ crii / fe ] ratio remained solar over the same metallicity range .
as a follow - up , they generated plots of cr i and cr ii as a function of effective temperature ( t@xmath14 ; a stellar atmospheric parameter ) and discovered a trend in the cr i data .
their finding may indicate that the neutral cr abundances are affected by departures from lte .
similarly , there have been up - to - date laboratory determinations of oscillator strengths for both the neutral ( nitz et al . 1999 ) and first - ionized species ( crespo lopez urratia et al . 1994 ) of cobalt .
the abundance investigation of 17 stars by bergemann ( 2008 ) , which employed a non - lte methodology and these new atomic data , found that the [ coi / fe ] ratio increases as metallicity decreases in the range -2.5 @xmath2 [ fe / h ] @xmath2 0.0 .
this result contradicts the data from our selected studies and signifies that co i abundances are susceptible to non - lte effects .
it is evident that further examination of both the cr and co abundance trends with [ fe / h ] over the full extent of metallicity and in a statistically - significant data sample is warranted to validate these two recent findings . as discussed above and indicated by red , filled circles in fig .
[ cr2covfel ] , we included five well - studied , metal - poor r - process - rich stars in our abundance comparisons of cr / co . since the suggested site for the r - process is core - collapse ( massive star ) supernovae ( see _ e.g. _ , sneden , cowan , & gallino 2008 ) , the products of ( early galactic ) nucleosynthesis in these halo stars , including the iron - peak elements , have not had major contributions from type ia sne - those events presumably arising from objects that require much longer stellar evolutionary timescales than type ii sne . there have been several attempts to explain the behavior of cr / co versus metallicity based upon type ii supernova nucleosynthesis .
thus , for example it has been shown that co and fe result from complete si burning , while cr is synthesized in these models from incomplete si burning ( nakamura _ et al . _ 1999 ) .
we see in fig .
[ 6plot_notracks ] that [ co / fe ] and [ cr / fe ] have opposite slopes with respect to metallicity .
we note that the results of nakamura et al .
supernova models depend critically upon the mass cuts between the nascent neutron star ( or black hole ) and the ejected envelope .
the yields of the synthesized material depends upon the ejected mass , and thus directly upon these mass cuts .
how these mass cuts depend upon progenitor masses and upon the explosion energies is also critical to determining the elemental yields .
in addition there may be other contributing factors in these models such as rotation energy that could affect the nucleosynthesis products .
the selected abundance investigations do have a few weak points .
for instance , they do not employ the most recent determinations of atomic data ( with exception of v i , v ii , and fe i ) .
additionally , there is a general presumption of iron as a reliable and robust abundance indicator .
however recently , lai et al . (
2008 ) found a correlation between excitation potential and both metallicity and effective temperature for fe i transitions ( in a sample of extremely metal deficient stars ) .
as an alternative , @xcite suggest the exclusive employment of fe ii transitions to derive metallicity ( and thereby , avoid the issues associated with the use of fe i lines ) .
none of the chosen investigations did this in their respective determinations of metallicity . also for the majority of stars examined ( with low temperature and metallicity ) , the dominant form of all of the elements is the first - ionized species ( especially the elements with an ionization potential less than 7 ev : ti , v , cr ) .
yet , the selected studies largely rely upon transitions only from the neutral species of the element ( and in fact , do not present any data for mn ii or co ii ) .
then , the derivation of abundances from fe - peak elements is sensitive to departures from lte ( effects such as overionization and resonance scattering ; e.g. shchukina & trujillo bueno 2001 ) . for example , the abundance investigation of 14 stars by bergemann & gehren ( 2008 ) , which employed a non - lte methodology , found that the [ mni / fe ] ratio remained solar in a metallicity range -2.5 @xmath2 [ fe / h ] @xmath2 0.0 .
this finding contradicts the results of the selected abundance studies .
note though , that the magnitude of these non - lte effects ( for each of the elements ) has yet to be quantified in rigorous and consistent analysis of a large stellar sample .
the principal goal of this paper has been to produce a set of integrated stellar yields which can be used by theorists in the future to test their yield predictions for fe - peak elements .
utilizing an extensive data base of published stellar iron peak elemental abundances for the solar vicinity , we have produced plots of the form [ x / fe ] versus [ fe / h ] for the elements ti , v , cr , mn , co , and ni . we next employed detailed one - zone chemical evolution models to evaluate the massive star yields of @xcite and @xcite in their ability to reproduce these trends . finally , we scaled the latter yield set and used it as input to our models in order to force a fit to the data for each plot . from this scaled set we then derived our empirical integrated yields , using a slightly flattened time - independent salpeter initial mass function .
our analysis led to the following conclusions : * as recognized previously by other authors , there are clear upward trends of [ ti / fe ] and [ co / fe ] and downward trends of [ co / fe ] and [ mn / fe ] as metallicity decreases .
this is especially the case below [ fe / h ] of roughly -2 . *
the above trends appear to be the result of changes in massive star yields with metallicity and are unrelated to star formation history .
* models utilizing yields of @xcite and @xcite generally reproduce the data near solar metallicity . however , the agreement between observation and theory at lower metallicities is generally less satisfactory . *
a set of integrated yields as a function of metallicity was derived from global data trends for six elements . in applying scaling factors to the yields of @xcite to force a fit to the observations , we make the following comparisons between the empirical and theoretical yields .
our results suggest that actual yields of ti , v , and co are generally larger as well as anticorrelated with metallicity compared to the @xcite predictions .
we also find the yields of cr and mn to be generally smaller and directly correlated with metallicity compared to the theoretical results .
our results for ni are consistent with theory , although our model suggests that all ni yields should be scaled up slightly . one clear problem with our analysis involves the scatter in the data and in some cases the difficulty in establishing the nature or direction of the trend .
a good example can be seen in the plot of [ v / fe ] versus [ fe / h ] in fig . [
6plot_notracks ] , where we see both a possible flat as well as an upward trend below [ fe / h]=-2 .
a second example appears in the graph of [ cr / fe ] versus [ fe / h ] at metallicities below -3 .
here we see an apparent bifurcation in the data as one branch remains horizontal while the other continues trending downward . in the case of v
, we have attempted to match both branches , while in the cr case we ignore the horizontal branch for now . at these low metallicities
it is likely that the abundance pattern observed in a star is the result of expelled material from only one earlier - generation star instead of a well - mixed contribution from many such stars .
thus , star to star variations echo analogous differences in the yields of the earlier stars .
this is not an effect which our models are designed to take into account .
rather homogeneous mixing is assumed .
future work should attempt to explore the causes of the observed scatter and its possible link to apparent bifurcations in the data .
for now we have attempted only to analyze what we see as global trends in each plot in fig . [
6plot_notracks ] .
we thank andy mcwilliam , chris sneden and friedel thielemann for helpful discussions . this work has been supported by the national science foundation through grants ast 08 - 06577 to rbch and ast 07 - 07447 to jjc .
gratton @xmath15 sneden 1991 & cat / ces & 19 & 4122 - 6434 & 150 & 50000 & -2.7@xmath3[fe / h]@xmath3 - 0.2 & ti i / ii;v i / ii;cr
i / ii;mn i;fe i / ii;co i;ni i + mcwilliam et al .
1995 & lco / echelle & 33 & 3600 - 7600 & 12 - 47 & 22000 & -4.1@xmath3[fe / h]@xmath3 - 1.9 & ti i / ii;v i / ii;cr i;mn i;fe i / ii;co i;ni i + feltzing @xmath15 gustafsson 1998 & mcd/2-d coud & 47 & 3900 - 7000 & 200 & 100000 & -0.1@xmath3[fe / h]@xmath3 + 0.5 & ti i;v i / ii;cr
i / ii;mn i;fe i / ii;co i;ni i + cayrel et al .
2004 & vlt / uves & 70 & 3300 - 10000 & 95 - 430 & 45000 & -4.1@xmath3[fe / h]@xmath3 - 2.7 & ti i / ii;cr i;mn i;fe i / ii;co i;ni i + sobeck et al .
2006 & mcd/2-d coud ; keck / hires & 216 & 6010 - 6025 & 40 - 240 & 40000 - 60000 & -2.7@xmath3[fe / h]@xmath3 + 0.1 & mn i + lai et al .
2008 & keck / hires & 28 & 3020 - 7665 & 8 - 425 & 40000 & -4.2@xmath3[fe / h]@xmath3 - 2.6 & ti i / ii;v i / ii;cr
i / ii;mn i / ii;fe i / ii;co i;ni i + lr star formation efficiency & 2.8 gyr@xmath16 + star formation power law exponent & 1.5 + initial mass function exponent , @xmath11 & -1.20 + stellar mass range lower limit & 0.08 m@xmath13 + stellar mass range upper limit & 40 m@xmath13 + infall time scale & 4 gyr + current total mass density & 50 m@xmath13 pc@xmath17 + current age & 13 gyr + type ia factor , c & 0.007 + hypernova fraction , @xmath18 & 0.5 lr ti & @xmath19ti , @xmath20ti , * @xmath21ti * , @xmath22ti , @xmath23ti , @xmath20v , @xmath21v , @xmath22v , @xmath21cr , @xmath22cr + v & @xmath23v , * @xmath24v * , @xmath24cr , @xmath24mn + cr & @xmath23cr , * @xmath25cr * , @xmath26cr , @xmath27cr , @xmath25mn , @xmath26mn , @xmath27mn , @xmath25fe + mn & * @xmath28mn * , @xmath28fe , @xmath28co + fe & @xmath27fe , * @xmath29fe * , @xmath30fe , @xmath31fe , @xmath29co , @xmath30co , @xmath31co , @xmath29ni + co & @xmath32fe , * @xmath32co * , @xmath32ni , @xmath32cu + ni & @xmath33co , @xmath34co , * @xmath31ni * , @xmath33ni , @xmath34ni , @xmath35ni , @xmath36ni , @xmath33cu lrr ti & full range & 0.15 - 0.063x + v & x@xmath37 - 2 & -0.34 - 0.32x + v & x@xmath38 - 2 & + 0.30 + cr & x@xmath37 - 2 & + 0.50 + 0.30x + cr & x@xmath38 - 2 & -0.10 + mn & x@xmath37 - 1.5 & + 0.36 + 0.18x + mn & x@xmath38 - 1.5 & + 0.10 + co & x@xmath37 - 1.75 & -0.39 - 0.89x + co & x@xmath38 - 1.75 & 0.0 + ni & x@xmath37 - 1.0 & -0.75-x + ni & x@xmath38 - 1.0 & + 0.25 lcccccccc -@xmath39&-@xmath39&-5.70&-5.83&-6.32&-6.52&-3.32&-5.34&-4.69 + -6.0&-4.3&-5.76&-6.14&-6.03&-6.34&-3.30&-5.58&-4.71 + -4.0&-2.3&-5.86&-6.67&-5.51&-6.04&-3.23&-5.94&-4.69 + -3.0&-1.3&-5.89&-6.83&-5.32&-5.87&-3.15&-6.05&-4.61 + -2.0&-0.3&-5.91&-5.83&-5.33&-5.70&-3.12&-5.95&-4.76 + -1.7&0.0&-5.88&-6.83&-5.32&-5.62&-3.13&-5.78&-4.96 + -1.3&+0.4&-5.88&-6.82&-5.31&-5.61&-3.11&-5.76&-5.44 versus [ fe / h ] , showing observed values published in those sources indicated by symbol color defined in the legend . the specific ratio representing the dependent variable is indicated in the upper left corner of each panel.,width=576 ] but with model results included .
the green tracks are from models which employ the original fe - peak yields of @xcite , while the violet tracks are from models in which the scaled kobayashi yields have been used .
the red tracks refer to a model that employs the massive star fe - peak yields of @xcite .
the green dashed tracks show the results when details of the prescription of @xcite for snia events is employed .
the legend shows the symbol and color for each data set of observed abundances .
, width=576 ] versus log z / z@xmath13 for the seven elements of the fe peak under consideration .
green tracks refer to unadjusted yields from @xcite , while violet tracks show values for the adjusted yields .
the value for z@xmath13 was assumed to be 0.0122 [ @xcite.],width=576 ] versus [ fe / h ] .
data plotted are neutral species of the elements .
data sources are distinguished by symbol shape and color and are identified in the legend .
curves show the results of several chemical evolution models , also identified in the legend and explained in the text .
the dashed green and violet lines show the effect of using the snia prescription of @xcite instead of that of @xcite.,width=576 ] but for [ ti / fe ] versus [ fe / h ] .
data sources are distinguished by symbol shape and color and are identified in the legend .
curves show the results of several chemical evolution models , also identified in the legend and explained in the text .
the dashed green and violet lines show the effect of using the snia prescription of @xcite instead of that of @xcite.,width=576 ] | we present here the initial results of a new study of massive star yields of fe - peak elements .
we have compiled from the literature a database of carefully determined solar neighborhood stellar abundances of seven iron - peak elements , ti , v , cr , mn , fe , co , and ni , and then plotted [ x / fe ] versus [ fe / h ] to study the trends as functions of metallicity .
chemical evolution models were then employed to force a fit to the observed trends by adjusting the input massive star metallicity - sensitive yields of @xcite .
our results suggest that yields of ti , v , and co are generally larger as well as anticorrelated with metallicity , in contrast to the @xcite predictions .
we also find the yields of cr and mn to be generally smaller and directly correlated with metallicity compared to the theoretical results .
our results for ni are consistent with theory , although our model suggests that all ni yields should be scaled up slightly .
the outcome of this exercise is the computation of a set of integrated yields , _
i.e. _ , stellar yields weighted by a slightly flattened time - independent salpeter initial mass function and integrated over stellar mass , for each of the above elements at several metallicity points spanned by the broad range of observations .
these results are designed to be used as empirical constraints on future iron - peak yield predictions by stellar evolution modelers .
special attention is paid to the interesting behavior of [ cr / co ] with metallicity these two elements have opposite slopes as well as the indirect correlation of [ ti / fe ] with [ fe / h ] .
these particular trends , as well as those exhibited by the inferred integrated yields of all iron - peak elements with metallicity , are discussed in terms of both supernova nucleosynthesis and atomic physics . |
let @xmath0 be an abelian category . in @xcite we learned a one - to - one correspondence between complete cotorsion pairs in @xmath0 and abelian model structures on @xmath0 .
we assume the reader is familiar with this correspondence .
let s recall that an abelian model structure on @xmath0 is equivalent to a triple of classes @xmath9 for which @xmath10 is thick and which admits two complete cotorsion pairs @xmath11 and @xmath12 .
this data is equivalent to an abelian model structure on @xmath0 where @xmath10 are the trivial objects , @xmath5 are the cofibrant objects and @xmath7 are the fibrant objects . in practice
virtually all the abelian model structures we encounter also have the extra condition that they are _
hereditary_. this just means that @xmath5 is closed under taking kernels of epimorphisms between objects in @xmath5 and @xmath7 is closed between taking cokernels of monomorphisms between objects in @xmath7 .
we will assume this throughout this note , that the cotorsion pairs are hereditary .
now let s denote these cotorsion pairs by @xmath13 it is easy to note the following properties of these two complete hereditary cotorsion pairs : 1 .
@xmath3 and @xmath14 .
2 . @xmath15 .
remarkably , there is a converse to this as we will prove in this note .
so our goal is to prove the following theorem .
[ them - main ] let @xmath1 and @xmath2 be two complete hereditary cotorsion pairs satisfying conditions ( 1 ) and ( 2 ) above .
then there is a unique thick class @xmath10 for which @xmath9 is a hovey triple .
moreover , this thick class @xmath10 can be described in the two following ways : @xmath16 the proof we give is both elementary and very general .
it holds in the general setting of when @xmath0 is a weakly idempotent complete exact category .
the author showed in @xcite that hovey s correspondence carries over to this setting .
so while we state the theorems in the setting of abelian categories we have particularly constructed the proofs so that they hold in a general weakly idempotent complete exact category .
the author again wishes to thank hanno becker and the referee of his paper @xcite .
the construction of hovey triples that we give here is a direct generalization of a construction of becker from @xcite . in particular , when the two given cotorsion pairs are _ injective _ , then our construction is exactly becker s right bousfield localization construction from @xcite . on the other hand , when the two given cotorsion pairs are _ projective _ , then our construction coincides with becker s left bousfield localization construction from @xcite . the main difference in our proof , when comparing it to becker s proof ,
is in how we show that the class of trivial objects @xmath10 is thick . our proof , while longer , is more direct and uses only elementary properties of short exact sequences .
finally , the author wishes to thank mark hovey , who the author has been fortunate to work with over the years .
it was during a recent exchange that the author found theorem [ them - main ] , and it solves the problem of finding the gorenstein ac - flat model structure on the category of ( left ) modules over an arbitrary ring .
this is the flat analog of the gorenstein ac - injective and gorenstein ac - projective model structures of @xcite .
these applications , and others , will appear elsewhere .
but the author finds theorem [ them - main ] to be of interest in its own right .
we wish to prove theorem [ them - main ] .
so assume @xmath1 and @xmath2 are two complete hereditary cotorsion pairs satisfying conditions we start by showing that the two classes below that define @xmath10 do coincide .
@xmath16 so say @xmath17 is in the top class , that is , that there is a short exact sequence @xmath18 where @xmath19 and @xmath20 . since @xmath21 has enough projectives we can find a short exact sequence @xmath22 where @xmath23 and @xmath24 .
we take a pullback and from ( * ? ? ?
* ; * ? ? ?
* proposition 2.12 ) we get the commutative diagram below with exact rows and columns and whose lower left corner is a bicartesian ( pushpull ) square . @xmath25{r } \arrow[tail]{d } & r ' \arrow[tail]{d } & \\
p \arrow[tail]{r } \arrow[two heads]{d } & q ' \arrow[two heads]{r } \arrow[two heads]{d } & q \arrow[equal]{d } \\ x \arrow[tail]{r } & r \arrow[two heads]{r } & q \end{tikzcd}\ ] ] since @xmath8 is closed under extensions , by the assumption ( 2 ) @xmath15 , we deduce that @xmath26 .
now since @xmath27 and the cotorsion pairs are hereditary we conclude @xmath28 .
now the left vertical column shows that @xmath17 is in the bottom class describing @xmath10 . a similar argument
will show that any @xmath17 in the bottom class must be in the top class .
so the two descriptions of @xmath10 coincide .
we start by showing @xmath10 is closed under retracts .
so let @xmath30 and @xmath31 be such that @xmath32 .
we wish to show @xmath33 .
start by writing @xmath34 and also apply the fact that @xmath2 has enough injectives to get a short exact sequence @xmath35 . now construct a commutative diagram as shown below : @xmath36 the map @xmath37 exists because @xmath38 and similarly the map @xmath39 exists because @xmath40 .
next , the two right vertical maps exist simply by the universal property of @xmath41 and @xmath42 being cokernels .
next denote the map @xmath43 by @xmath44 , and its cokernel @xmath45 by @xmath46 .
then we see @xmath47 .
so again the universal property of @xmath41 being the cokernel of @xmath44 gives us a map @xmath48 such that @xmath49 .
this proves that @xmath50 factors through @xmath41 , but now we argue that this implies @xmath50 actually factors through an object of @xmath15 . indeed using that @xmath2 has enough injectives we find @xmath51 , but this time it follows that @xmath52 . now using that @xmath53 , we see that @xmath48 extends over @xmath54 .
so we see that we can find maps @xmath55 and @xmath56 such that @xmath57 and @xmath58 .
thus the composition below is the identity @xmath59 . @xmath60 but this just means @xmath61 is a retract of @xmath62 . since @xmath8 is closed under direct sums and retracts get that @xmath19 .
this proves @xmath33 and we are done .
we now turn to the two out of three property and our next immediate goal is to show closure of @xmath10 under extensions .
note now that @xmath10 clearly contains both @xmath6 and @xmath8 , for we will use this ahead .
we start by making the following claim . .
suppose @xmath63 is exact with @xmath19 and @xmath30 .
then there exists a commutative diagram as below where @xmath64 and @xmath65 .
@xmath66{r } \arrow[tail]{d } & \widetilde{r } '' \arrow[two heads]{r } \arrow[tail]{d } & \widetilde{r } ' \arrow[tail]{d } \\
\widetilde{q } \arrow[tail]{r } \arrow[two heads]{d } & \widetilde{q } '' \arrow[two heads]{r } \arrow[two heads]{d } & \widetilde{q } ' \arrow[two heads]{d } \\
r \arrow[tail]{r } & y \arrow[two heads]{r } & w \end{tikzcd}\ ] ] indeed since @xmath67 there are short exact sequence @xmath68 and @xmath69 with @xmath70 and @xmath71 .
but in this case we also have @xmath72 and so @xmath73 .
this means there exists a lift as shown : @xmath74{d } \arrow[dashed]{dl } \\
r \arrow[tail]{r } & y \arrow[two heads]{r } & w \end{tikzcd}\ ] ] this lift allows for the construction , analogous to the usual horseshoe lemma of homological algebra , of a commutative diagram as below .
@xmath66{r } \arrow[tail]{d } & \widetilde{r } '' \arrow[two heads]{r } \arrow[tail]{d } & \widetilde{r } ' \arrow[tail]{d } \\
\widetilde{q } \arrow[tail]{r } \arrow[two heads]{d } & \widetilde{q } \oplus \widetilde{q } ' \arrow[two heads]{r } \arrow[two heads]{d } & \widetilde{q } ' \arrow[two heads]{d } \\
r \arrow[tail]{r } & y \arrow[two heads]{r } & w \end{tikzcd}\ ] ] since any class that is part of a cotorsion pair is closed under direct sums and extensions we now have @xmath75 and @xmath76 , and so we have proved our first claim . .
@xmath10 is closed under extensions .
now suppose @xmath77 is exact with @xmath78 .
we need to prove that @xmath79 too .
since @xmath30 we may now find an exact sequence @xmath80 where @xmath19 and @xmath20 . now form the pushout diagram shown .
@xmath81{r } \arrow[tail]{d } & y \arrow[two heads]{r } \arrow[tail]{d } & w ' \arrow[equal]{d } \\
r \arrow[tail]{r } \arrow[two heads]{d } & p \arrow[two heads]{r } \arrow[two heads]{d } & w ' \\ q \arrow[equal]{r } & q \end{tikzcd}\ ] ] note the second row is the type of row from claim 1 .
so the horseshoe argument provides a diagram as shown below where @xmath64 and @xmath65 .
@xmath82 & & & & & \widetilde{r } ' \arrow[tail]{dd } \\ & & & \widetilde{r } '' \arrow[tail]{dd } \arrow[two heads]{urr } & & \\ & \widetilde{r } \arrow[tail]{dd } \arrow[tail]{urr } & & & & \widetilde{q } ' \arrow[two heads]{dd } \\ & & & \widetilde{q } '' \arrow[two heads]{dd } \arrow[two heads]{urr } & w ' \arrow[equal]{dr } & \\ & \widetilde{q } \arrow[two heads]{dd } \arrow[tail]{urr } & y \arrow[tail]{dr } \arrow[two heads]{urr } & & & w ' \\ w \arrow[tail]{dr } \arrow[tail]{urr } & & & p \arrow[two heads]{urr } & & \\ & r
\arrow[tail]{urr } & & & & \\ \end{tikzcd}\ ] ] now pullback the entire diagram over the original exact sequence @xmath77 and we get what we want . in particular , the pullback in the middle of the diagram leads to the following bicartesian ( pushpull ) square and the hereditary property of @xmath6 gives us @xmath83 .
@xmath84{r } \arrow[tail]{d } & \widetilde{r } '' \arrow[tail]{d } & \\
l \arrow[tail]{r } \arrow[two heads]{d } & \widetilde{q } '' \arrow[two heads]{r } \arrow[two heads]{d } & q \arrow[equal]{d } \\
y \arrow[tail]{r } & p \arrow[two heads]{r } & q \end{tikzcd}\ ] ] .
next suppose @xmath85 is exact with @xmath78 .
we need to prove that @xmath86 too .
start by writing @xmath87 and again forming a pushout diagram as shown .
@xmath81{r } \arrow[tail]{d } & w ' \arrow[two heads]{r } \arrow[tail]{d } & z \arrow[equal]{d } \\ { \widetilde{{\mathcal{r } } } } \arrow[tail]{r } \arrow[two heads]{d } & p \arrow[two heads]{r } \arrow[two heads]{d } & z \\ { \widetilde{{\mathcal{q } } } } \arrow[equal]{r } & { \widetilde{{\mathcal{q } } } } \end{tikzcd}\ ] ] since we have shown @xmath10 is closed under extensions we get @xmath88 .
so now we can write @xmath89 .
we now take yet another pushout , the pushout of @xmath90 to get yet another diagram as shown .
@xmath91{r } \arrow[tail]{d } & { \widetilde{{\mathcal{r } } } } \arrow[tail]{d } & \\ p \arrow[tail]{r } \arrow[two heads]{d } & { \widetilde{{\mathcal{r } } } } ' \arrow[two heads]{r } \arrow[two heads]{d } & { \widetilde{{\mathcal{q } } } } ' \arrow[equal]{d } \\ z \arrow[tail]{r } & l \arrow[two heads]{r } & { \widetilde{{\mathcal{q } } } } ' \end{tikzcd}\ ] ] ( by ( * ? ? ?
* ; * ? ? ?
* proposition 2.12 ) we get that the lower left square is bicartesian . ) now since @xmath1 is an hereditary cotorsion pair we get @xmath92 .
so the short exact sequence @xmath93 finishes the proof of the claim .
( @xmath9 is a hovey triple . )
we just need to show @xmath95 and @xmath96 .
the proof of each is similar , so we will just show @xmath96 . here , we clearly have @xmath97 , so we just need to show @xmath98 .
so let @xmath99 . using that @xmath17 is in @xmath10 we may write a short exact sequence @xmath100 .
but since @xmath101 and @xmath102 is a cotorsion pair , this sequence must split .
so @xmath17 is a retract of @xmath103 and so must be in @xmath8 .
this completes the proof that @xmath9 is a hovey triple .
the uniqueness of @xmath10 follows from a general fact : the class of trivial objects in a hovey triple is always unique by ( * ? ? ?
* proposition 3.2 ) . | let @xmath0 be an abelian category , or more generally a weakly idempotent complete exact category , and suppose we have two complete hereditary cotorsion pairs @xmath1 and @xmath2 in @xmath0 satisfying @xmath3 and @xmath4 .
we show how to construct a ( necessarily unique ) abelian model structure on @xmath0 with @xmath5 ( respectively @xmath6 ) as the class of cofibrant ( resp .
trivially cofibrant ) objects and @xmath7 ( respectively @xmath8 ) as the class of fibrant ( resp .
trivially fibrant ) objects . |
ultraluminous infrared galaxies ( ulirgs ) with infrared luminosity at 81000 @xmath3 m greater than 10@xmath4 @xmath5 ( soifer et al .
1987 ) are extremely energetic objects in the universe .
although they contribute little to the infrared luminosity density in the local universe due to small numbers , they become cosmologically important at z @xmath6 1 ( e.g. , le floch et al . 2005 ; magnelli et al .
their enormous infrared luminosity comes from dust heated by hot young stars ( starburst ) , a supermassive black hole rapidly accreting matter ( active galactic nucleus , agn ) , or a mixture of these two ( see sanders & mirabel 1996 and lonsdale et al .
2006 for a review ) .
these starburst and/or agn activities can be triggered by tidal interactions between galaxies and associated shocks ( e.g. , bushouse 1987 ; liu & kennicutt 1995 ; barnes 2004 ) .
in fact , numerous observational and theoretical studies suggested that ulirgs are mergers of gas - rich disk galaxies ( e.g. , clements et al .
1996 ; mihos & hernquist 1996 ; veilleux et al .
2002 ; younger et al .
2009 ; hwang et al .
2010a ) , and evolve into quasars ( e.g. , sanders et al . 1988
; dasyra et al . 2006 ; hopkins et al .
2006 ; yuan et al .
2010 ) or intermediate - mass elliptical galaxies ( e.g. , genzel et al .
2001 ; tacconi et al .
2002 ) .
to better understand the origin and evolution of ulirgs , it is essential to find out what their primary energy source is .
a standard spectroscopic method to distinguish between starburst and agn is to use so - called bpt diagrams ( baldwin et al .
1981 ) , which are based on optical emission line ratios sensitive to the photoionization source .
this optical diagnostic was revised by veilleux & osterbrock ( 1987 ) and an alternative scheme was proposed by kewley et al .
( 2006 ) . in heavily obscured galaxies like ulirgs , additional observations at other wavelengths
are helpful to characterize their dominant energy source ( e.g. , x - ray : franceschini et al .
2003 , teng et al . 2009
; infrared : risaliti et al .
2006 , farrah et al .
2007 , imanishi et al . 2010
; radio : nagar et al . 2003 , sajina et al . 2008 ) . however , it is still difficult to determine the relative contribution of starburst and agn within individual galaxies .
ulirgs were discovered in large numbers by the _ infrared astronomical satellite _ ( _ iras _ ; neugebauer et al . 1984 )
and the number of ulirgs increased with the advent of wide - field galaxy redshift surveys ( goto 2005 ; pasquali et al .
2005 ; cao et al .
2006 ; hwang et al .
2007 , 2010a ; hou et al .
however , previous studies based on optical spectra of the large sample are limited to the sloan digital sky survey ( sdss ; york et al .
2000 ) , which mainly covers the northern hemisphere .
the optical spectral properties of ulirgs in the southern hemisphere remain to be studied . in this study , we present the analysis of optical spectra for about a hundred southern ulirgs in the catalogue given by hwang et al .
the structure of this paper is as follows . the survey data and our observations are described in section 2 .
section 3 explains the procedures to analyse these data and to classify the ulirgs . in section 4 ,
the results of our study are discussed , and are compared with those of previous studies .
we summarize and conclude in section 5 . throughout
, we adopt @xmath7 = 70 km s@xmath8 mpc@xmath8 and a flat @xmath9cdm cosmology with density parameters @xmath10 = 0.3 and @xmath11 = 0.7 .
hwang et al . ( 2007 ) identified 324 ulirgs by cross - correlating the _ iras _ faint source catalogue version 2 ( moshir et al . 1992 ) with the spectroscopic catalogues of galaxies in the fourth data release of sdss ( adelman - mccarthy et al . 2006 . ) , the final data release of the 2df galaxy redshift survey ( 2dfgrs ; colless et al .
2001 ) , and the second data release of the 6df galaxy survey ( 6dfgs ; jones et al .
2004 , 2005 ) .
the spectral types of ulirgs in the sdss among them were determined in hou et al .
( 2009 ) . in this study , we focus on 198 ulirgs that were not covered by the sdss but covered by the 2dfgrs and 6dfgs .
we used the optical spectra of these ulirgs available online for 6dfgs , http://www-wfau.roe.ac.uk/6dfgs/ ] .
the optical spectra of 2dfgrs galaxies were taken with the two - degree field ( 2df ) multi - object spectrograph on the anglo - australian telescope .
this spectrograph has 140 @xmath3 m diameter fibres corresponding to 2.16 at the plate centre and 1.99 at the edge and covers 36008000 with a spectral resolution of @xmath12 9 .
the 6dfgs galaxies were observed with the six - degree field ( 6df ) multi - object spectrograph having 6.7 diameter fibres on the united kingdom schmidt telescope .
the original grating in the 6dfgs spectrograph spans 40008400 and gives a resolution of 512 .
the improved grating , used after 2002 october , spans 39007500 with a resolution of 4.96.6 .
it is noted that the 2dfgrs and 6dfgs spectra are not properly flux - calibrated , but the flux ratio of adjacent lines is still useful ( to be discussed in section 4 ) .
in addition , we conducted optical spectroscopy of 15 ulirgs in the survey sample plus one ulirg , _ iras _ 09022@xmath23615 , in sanders et al .
( 2003 ) at the cerro tololo inter - american observatory ( ctio ) .
the total number of our ulirg sample with spectra was increased to 199 .
the additional ulirgs were observed on 2008 february 2021 with the ritchey - chretien spectrograph and the loral 3k ccd at the ctio 4-m telescope ( 0.5 pixel@xmath8 ) . a slit with a width of 1.5 ( a slightly larger than the seeing size ) was adopted , and was positioned in the east - west direction ( p.a .
we used 316 lines mm@xmath8 grating to cover the spectral range 450010500 with a resolution of 5.6 ( @xmath12 2.8 pixels ) .
three exposures were taken for each object and the integration times ranged from 360 to 3000 seconds depending on its brightness .
data reduction was performed using the iraf package .
this involved bias subtraction , flat fielding , sky subtraction , wavelength and flux calibration .
galaxy spectra were extracted using an aperture width corresponding to a constant linear scale of 5 kpc at the redshift of each galaxy . a he - ne - ar lamp and standard star hiltner 600 ( massey et al .
1988 ) taken nightly were used for wavelength and flux calibration , respectively . in fig . 1 , we display 13 spectra with median signal - to - noise ratio ( s / n ) per pixel for the continuum greater than 3 . typical emission lines
( , @xmath13 , + @xmath14 , and @xmath15 ) are clearly seen in most spectra . among the sample of 199 ulirgs considered in this study , we analyse the spectra of 115 ulirgs with continuum s / n @xmath6 3 . in the case
that there are more than one spectrum for the same object , the spectrum with a higher s / n of flux is chosen for the final classification .
if is not available from any spectra , the spectrum with a higher s / n of flux has priority .
after transforming the spectrum to the rest - frame and subtracting the local continuum defined by a linear fit around emission - lines , we measure the line widths and fluxes via gaussian profile fit using the mpfit / idl package based on the levenberg - marquardt method ( markwardt 2009 ) .
a single gaussian is used to obtain the line width of @xmath0 . for the line
fluxes , we fit the @xmath16 line , @xmath15 doublet , + @xmath14 and + @xmath13 line complexes with one gaussian , two gaussians , four gaussians , and six gaussians , respectively .
the widths of the doublet are kept the same and the height ratio of @xmath17 to @xmath18 is fixed to 1/3 , as required by the energy level structure of the ion ( osterbrock & ferland 2006 ) .
the doublet is fitted in the same way but each of the lines is modelled with two gaussians because line often has a blue , asymmetric wing that is perhaps from outflows of gas with opaque clouds ( e.g. , heckman et al 1981 ; greene & ho 2005 ) .
some agn host galaxies show both narrow and broad balmer lines ( e.g. , osterbrock & mathews 1986 ; hao et al . 2005 ) and the narrow balmer lines can have extended bases even in star - forming galaxies , which are probably due to wolf - rayet stars ( e.g. , osterbrook & cohen 1982 ; brinchmann et al . 2008 ) . to take this into account
, we fit each balmer line with two gaussians .
if the + line complex can not be decomposed into individual lines , the line fluxes are not kept .
meanwhile , the decomposition of doublet is not important because we are interested only in the total flux of the lines .
uncertainties in the line measurements , comes from mpfit routine , are typically 1030 per cent .
we correct the line fluxes for galactic extinction using the foreground reddening maps provided by schlegel et al .
( 1998 ) and the extinction law of cardelli et al .
( 1989 ) .
balmer emission lines can be strongly affected by stellar absorption features .
we remove the stellar absorption effects following the method discussed in hopkins et al .
( 2003 ) : @xmath19 , where @xmath20 is the stellar absorption - corrected line flux , @xmath21 is the measured line flux after foreground reddening correction , @xmath22 is the equivalent width of the line , and @xmath23 is a correction factor .
we adopted @xmath23 values ( 2.6 and 3.2 for and , respectively ) of sc type galaxies ( miller & owen 2002 ) .
if both and are measured , we can correct the line fluxes for internal extinction using the balmer decrement and the extinction curve with an assumption of an intrinsic / line ratio of 2.85 for star - forming galaxies and 3.1 for agn galaxies ( osterbrock & ferland 2006 ) .
we do not apply this correction when the observed ratio is smaller than the theoretical value .
an observed line width is the convolution of intrinsic line width and instrumental response .
since the instrumental resolution of each survey is significantly different , the observed line width should be corrected for a fair comparison . to correct the full width at half maximum ( fwhm ) of @xmath0 line using the quadrature method
, we adopt 250 km s@xmath8 as a finite resolution of ctio spectra .
then we determine that the resolutions are roughly 450 , 600 , and 400 km s@xmath8 for the 2dfgrs , original 6dfgs , and improved 6dfgs grating , respectively , by considering that the lines from different spectra have the same intrinsic width .
note that corrected line widths below 100 km s@xmath8 should be treated with caution due to their large uncertainties ( @xmath6 200 km s@xmath8 ) .
diagram for the sample b ulirgs , which are denoted by open symbols .
the ulirgs already classified in the diagnostic diagrams are overplotted by filled symbols ( black : seyfert 2 plus liner ; dark - grey : composite ; light - grey : star - forming ) .
circles , squares , and triangles represent ctio , 2dfgrs , and 6dfgs ulirgs , respectively .
stars and diamonds represent ulirgs from the 1 jy sample ( veilleux et al .
1999 ; yuan et al . 2010 ) and the sdss sample ( hou et al . 2009 ) , respectively .
the dashed and solid lines show the criteria of zakamska et al .
( 2003 ) and our new boundary separating narrow - line agns ( nlagns ) and non - agns . in this diagram ,
/ line ratios are corrected only for galactic extinction and stellar absorption.,width=302 ] * table 1 .
* basic information and spectral types of 115 southern ulirgs . + [ cols=">,^,^,^,>,>,>,^,<,^,^,<",options="header " , ] the small aperture spectroscopy could not always contain enough light of an extended source to determine its spectral type .
for reliable classification , it is suggested to use an aperture covering more than @xmath12 20% of the galaxy light .
the minimum aperture covering fraction of 20% corresponds to 2.1 kpc ( see fig . 6 in kewley et al .
all spectra of ulirgs in this study satisfy this condition .
therefore , aperture - related effects on our classification are expected to be negligible . in the 2dfgrs and 6dfgs spectra
, the flux for individual line can be unreliable because these spectra are not properly flux - calibrated . nevertheless , the flux ratio between lines with similar wavelengths is still useful ( e.g. , mouhcine et al . 2005 ; owers et al .
2007 ) . to ensure this , in fig .
4 , we compare / line ratios derived from both calibrated ( i.e. , ctio or sdss ) and uncalibrated ( i.e. , 2dfgrs or 6dfgs ) spectra .
it shows that two measurements agree well within the errors .
the measurements between the 2dfgrs and 6dfgs spectra also agree well .
among the 115 ulirgs , there are 22 objects whose spectral types have been previously determined in the literature .
the spectral types for 19 out of 22 ulirgs determined in this study are consistent with those in the literature .
if we compare ulirgs whose subclasses ( see columns 11 and 12 in table 1 ) were determined , 9 out of 13 ulirgs show a good agreement .
although there are three ulirgs with different types ( or four ulirgs with different subclasses ) , all of these ulirgs are composite galaxies either in this study or in the previous studies .
3 are presented . the one - to - one relation ( solid line )
is overplotted.,width=340 ] in our sample , 50% ( 57/115 ) of the ulirgs are found to host agn .
the agn fraction depends on the infrared luminosity and the _ iras _ flux density ratios ( hereafter _ iras _ colours ) as follows : 48% for ulirgs with 12.0 @xmath24 @xmath25 @xmath26 12.15 , 50% for [ 12.15 , 12.4 ) , and 57% for [ 12.4 , 13.0 ) .
40% for ulirgs with @xmath21.4 @xmath24 @xmath27 @xmath26 @xmath20.9 , 47% for [ @xmath20.9 , @xmath20.5 ) , and 92% for [ @xmath20.5 , @xmath20.1 ) .
36% for ulirgs with @xmath20.7 @xmath24 @xmath28 @xmath26 @xmath20.3 , 33% for [ @xmath20.3 , @xmath20.1 ) , and 52% for [ @xmath20.1 , 0.3 ) .
the ulirgs with a flux upper limit at 25 ( 100 ) @xmath3 m are not included in the calculations for _ iras _ 25@xmath260 ( 60@xmath2100 ) @xmath3 m colour dependence .
these results are summarized in table 4 together with those from the 1 jy ( kim & sanders 1998 ) and sdss ( hou et al . 2009 ) samples .
note that no ulirgs in the 1 jy sample have colours of @xmath28 @xmath26 @xmath20.3 due to the selection criteria in kim & sanders ( 1998 ) .
the agn fractions for each sample are comparable in the sense that the differences are less than 1.6 @xmath29 by assuming poisson errors . in all samples , there is a tendency for the ulirgs with higher infrared luminosity , warmer _ iras _ 25@xmath260 @xmath3 m colour to be more agn - like . these findings are consistent with those in previous works ( e.g. , veilleux et al . 1999 ; kewley et al . 2001 ; goto 2005 ; cao et al .
2006 ; hou et al . 2009 ; yuan et al .
2010 ) . in fig .
5 , we present infrared luminosity , _ iras _ 25@xmath260 @xmath3 m colour , and _ iras _
60@xmath2100 @xmath3 m colour distributions of the combined sample ( 432 ulirgs ) including this study , kim & sanders ( 1998 ) , and hou et al .
based on the kolmogorov - smirnov ( k - s ) test , we find that the infrared luminosity distribution of broad - line agn ulirgs is significantly different from those of non - agn and narrow - line agn ulirgs , while those of non - agn and narrow - line agn ulirgs are not so different .
we also find that the three types of ulirgs are significantly different from each other in _ iras _
25@xmath260 @xmath3 m colour distribution but they are indistinguishable in _ iras _ 60@xmath2100 @xmath3 m colour distribution . in _ iras _
25@xmath260 @xmath3 m colour distribution , the significant difference between agn and non - agn ulirgs strengthens that mid - infrared colour is a good indicator of agn activity in infrared galaxies ( e.g. , de grijp et al .
1985 ; neff & hutchings 1992 ) .
the little differences in _ iras _
60@xmath2100 @xmath3 m colour distribution reinforce that star formation dominates the emission of agn in the far - infrared regime ( e.g. , elbaz et al . 2010 ; hatziminaoglou et al .
2010 ; hwang et al . 2010b
; shao et al .
2010 ) . on the other hand , narrow - line agn ulirgs
have in general lower infrared luminosity and cooler _ iras _
25@xmath260 @xmath3 m colour than broad - line agn ulirgs do .
if we assume that the infrared emission in galaxies is emitted isotropically , and does not depend on viewing angle ( e.g. , mulchaey et al .
1994 ; schartmann et al .
2008 ; gandhi et al .
2009 ) , the difference between narrow - line agn and broad - line agn ulirgs seems not to be compatible with the predictions of the orientation - dependent unification model of agns in which narrow - line and broad - line agns are intrinsically same objects observed from different angles ( antonucci 1993 ) .
this conflict can be explained if narrow - line agn ulirgs host a central engine deeply buried by extended , dusty regions of star formation as proposed by several authors ( e.g. , genzel et al 1998 ; gerssen et al .
2004 ) . in fig .
6 , we show the relations between optical ( / line ratio or line width ) and infrared ( infrared luminosity , _ iras
_ 25@xmath260 @xmath3 m colour or _ iras
_ 60@xmath2100 @xmath3 m colour ) properties of the combined ulirg sample .
broad - line agn ulirgs are not plotted because their line widths and fluxes are not properly measured in many cases .
since the spectral types are determined by the optical emission lines , narrow - line agns and non - agns are well separated along the y - axis .
as expected , clear correlations are found only in the middle panels ( see the values of spearman s rank correlation coefficients ) .
we checked these relations using various infrared colours from the _ akari _ all - sky survey point source catalogues but we could not draw any meaningful results due to the small number of data points .
we studied optical spectral properties of 115 southern ulirgs using the spectra obtained from our ctio observations , 2dfgrs , and 6dfgs .
for ulirgs with measurement , we classified them in the standard diagnostic diagrams .
we classified the other ulirgs using the @xmath0/h@xmath1 line ratio against @xmath0 line width diagram with an empirically determined criterion .
main results are summarized below . 1 .
our new criterion , @xmath30 ( / ) @xmath6 @xmath20.13 ( fwhm@xmath31 /100 km s@xmath8 ) @xmath32 0.85 , is successful to separate agn ulirgs from non - agn ulirgs with completeness and reliability of about 90% .
2 . in our sample of the 115 ulirgs , there are 8 broad - line agns , 49 narrow - line agns , and 58 non - agns .
the agn fraction is 50% and changes as a function of infrared luminosity and _ iras
_ 25@xmath260 @xmath3 m colour .
these results are consistent with those in previous studies .
3 . using the combined ulirg sample
, we show that the colour distributions of agn and non - agn ulirgs are significantly different in _ iras _
25@xmath260 @xmath3 m colour and are indistinguishable in _ iras _ 60@xmath2100 @xmath3 m colour .
these results support that mid - infrared colour is sensitive to agn activity , whereas far - infrared colour is dominated by star formation .
we also show that broad - line agn ulirgs differ from narrow - line agn ulirgs in infrared luminosity and _ iras _ 25@xmath260 @xmath3 m colour .
this presents a challenge to the simple unification model of agns .
we thank the ctio staff for their help on the observations .
we also thank l. g. hou for providing the line width data of the sdss ulirgs .
we are grateful to anonymous referee whose comments helped to improve the original manuscript .
this work was supported by mid - career research program through nrf grant funded by the mest ( no.2010 - 0013875 ) .
h.s.h acknowledges the support of the centre national detudes spatiales ( cnes ) .
s.c.k is a member of the dedicated researchers for extragalactic astronomy ( dream ) team in korea astronomy and space science institute ( kasi ) . | we present a study of the optical spectral properties of 115 ultraluminous infrared galaxies ( ulirgs ) in the southern sky . using the optical spectra obtained at ctio 4 m and provided by the 2df galaxy redshift survey and the 6df galaxy survey
, we measure emission line widths and fluxes for spectral classification . we determine the spectral types of ulirgs with measurement using the standard diagnostic diagrams .
for ulirgs without measurement , we determine their spectral types using the plane of flux ratio between @xmath0 and h@xmath1 versus line width based on our new empirical criterion .
this criterion is efficient to distinguish active galactic nuclei ( agns ) from non - agn galaxies with completeness and reliability of about 90 per cent .
the sample of 115 ulirgs is found to consist of 8 broad - line agns , 49 narrow - line agns , and 58 non - agns .
the agn fraction is on average 50 per cent and increases with infrared luminosity and _ iras _ 25@xmath260 @xmath3 m colour , consistent with previous studies .
the _ iras _ 25@xmath260 @xmath3 m colour distributions are significantly different between agn and non - agn ulirgs , while their _ iras _
60@xmath2100 @xmath3 m colour distributions are similar .
[ firstpage ] galaxies : active galaxies : general galaxies : starburst infrared : galaxies |
safety critical systems , such as aircraft , satellites , and electricity grids , often rely on sensors to measure their state and their environment .
the true state of the system may not be measurable , or may be corrupted by noise , as quantified by a so - called observation process .
the controller must choose actions based only on the information contained in the observation process .
the nature of safety critical systems dictates a need for formal methods to accurately assess the system s ability to meet rigorous safety requirements , and also to synthesize controllers that guarantee performance to a desired level ( correct by design ) .
it is therefore paramount that the controller exploit information from the observation process , to obtain theoretical safety guarantees that are as accurate as possible .
reachability analysis , which determines whether a system s state remains within a given safe region and/or reaches a target set within some time horizon , has been used extensively as a tool for verification and controller synthesis for hybrid systems @xcite , @xcite , @xcite and extended to stochastic hybrid systems ( shs ) @xcite , @xcite .
there has been little focus , however , on reachability analysis for partially observable shs .
while there has been some work on deterministic hybrid systems with hidden modes @xcite or uncertain systems with imperfect information on a partial order @xcite , reachability analysis of a partially observable shs has been approached only theoretically @xcite , @xcite .
this , along with our previous work @xcite provides the first computational results for both controller synthesis and verification of safety specifications for partially observable shs .
existing computational results for reachability analysis of fully observable shs are also limited . the safety problem for a discrete time shs ( dtshs ) , which considers only whether the state of the system can be controlled to remain within a safe region of the state space , can be formulated as a multiplicative cost stochastic optimal control problem @xcite , and solved in the same manner as a markov decision process ( mdp ) .
unfortunately , solutions via dynamic programming @xcite require evaluation of a value function over all possible states , which is infinite when those states are continuous .
discretization procedures can be employed to impose a finite number of states , as in @xcite and @xcite , which present rigorous uniform and adaptive gridding methods for verification of dtshs .
similarly , approximate abstractions of the original stochastic model to an equivalent system that has the same properties are presented in @xcite , @xcite , and @xcite .
even so , current applications are limited to those with only a few discrete and continuous states .
the safety problem for a partially observable dtshs ( podtshs ) can similarly be formulated as a partially observable mdp ( pomdp ) .
however , pomdps are plagued by dimensionality on an even greater scale than mdps .
the common approach to solving pomdps is to replace the growing history of observations and actions by a sufficient statistic , often called the belief state , which , for a pomdp with an additive cost function , is the distribution of the current state conditioned on all past observations and actions @xcite .
this belief state is treated as the perfectly observed true state , and mdp solution methods can then be applied . however , with a continuous state space , the belief state is a function defined over an infinite domain , and it is impossible to enumerate over all such functions .
therefore the study of efficient , approximate solutions to pomdps is essential .
although finding the solution to a general pomdp is hard @xcite , many algorithms for approximating solutions to finite state pomdps have been developed .
these mainly rely on point - based value iteration ( pbvi ) schemes that only consider a subset of the belief space to update the value function used in the dynamic program ( for a survey of pbvi algorithms , see @xcite ) . because the value function is piecewise - linear and convex @xcite ( and so equivalently represented by a finite set of vectors ) , sampling from the belief state provides a systematic way of storing a finite subset of those vectors .
such methods must be tailored to continuous state pomdps because of the dimensionality of the belief state .
other than discretizing the state space and solving an equivalent finite state pomdp , many existing methods for continuous state pomdps assume the belief state is gaussian ( e.g. @xcite , @xcite ) , and represent the belief state in a parameterized form which is then discretized and solved as a discrete state mdp .
when the belief state can not adequately be represented using a single gaussian , a gaussian mixture may be used instead . an equivalent point - based algorithm for continuous - state pomdps using gaussian mixtures
is presented in @xcite , and demonstrated on a stochastic hybrid system with hidden modes in @xcite . the safety problem for a podtshs is further complicated because the belief state is not the conditional distribution of the current state of the system @xcite , @xcite , but must also include the distribution of a binary variable that indicates whether the state of the system has remained within a safe region up to the previous time step .
this , coupled with the stochastic hybrid system dynamics , makes accurately representing the belief state as a single gaussian impossible .
we formulate the safety problem for a podtshs as a pomdp , and investigate representations of the belief state in either vector or gaussian mixture form through finite- and continuous - state approximations to the podtshs .
these representations allow us to exploit point - based methods developed for pomdps .
this paper extends our previous work @xcite in several ways .
first , we validate the use of pomdp solution techniques for reachability analysis of a podtshs , by demonstrating that the value function is convex and admits a function representation related to the piecewise - linear vector representation of a finite state pomdp .
second , we present a finite state approximation to the dtshs ( presented in @xcite without proofs ) that allows the belief state to take vector form under certain conditions , and show convergence for the approximation .
third , we preserve the continuity in the hybrid state space through a gaussian mixture representation for the belief state , and approximate the indicator function that represents the safe region using gaussian radial basis functions . in this case , we provide an error bound as a function of the integrated error ( 1-norm in the function space @xmath0 ) of the indicator function approximation .
our solution method converges to the true solution from below , using either the finite or continuity - preserving belief state .
we demonstrate both approaches on a temperature regulation problem .
the rest of the paper is organized as follows .
section [ sec : background ] relates the safety problem for a podtshs to optimal control of a pomdp .
section [ sec : abstraction ] justifies the use of pomdp solution techniques , and presents the finite and gaussian mixture approximations to the safety problem for a podtshs ( as well as error bounds ) .
section [ sec : pbvi ] describes the use of point - based approximation techniques , through sampling of belief states and discretization of the observations .
we present a numerical example in section [ sec : example ] , and concluding remarks and directions for future work in section [ sec : conc ] .
all proofs can be found in the appendix .
a probability space @xmath1 consists of a sample space @xmath2 , a @xmath3-algebra @xmath4 defined over @xmath2 , and a probability measure @xmath5 that assigns probabilities to events in @xmath4 . for @xmath6 ,
we presume @xmath7 , the borel @xmath3-algebra on @xmath8 .
the probability measure @xmath5 maps elements @xmath9 to the interval @xmath10 $ ] .
the density @xmath11 associated to @xmath5 is defined according to the lebesgue measure as @xmath12 .
we denote expected value by @xmath13 . a probability measure or expected value induced by a control policy @xmath14 ( to be defined later ) , is @xmath15 or @xmath16 , respectively . for a space @xmath17
, @xmath18 is the @xmath19-times product space of @xmath17 .
the state of a system at time @xmath19 , @xmath20 , takes values in state space @xmath17 .
we use @xmath21 to denote absolute value for @xmath22 and cardinality for a set @xmath2 .
a vector or function norm is denoted with @xmath23 . unless otherwise specified , @xmath24 .
a random variable @xmath25 follows a gaussian distribution with mean @xmath26 and covariance @xmath27 , and @xmath28 represents a gaussian density with mean @xmath26 and covariance @xmath27 evaluated at @xmath29 .
finally , @xmath30 denotes an inner product , with @xmath31 for vectors @xmath32 , and @xmath33 or @xmath34 for appropriately defined functions @xmath35 , @xmath36 .
the main results of this paper rely on framing a podtshs as a pomdp with a hybrid state space , and drawing upon existing results for finite state pomdps with additive cost objectives .
we therefore present an overview of pomdps and efficient approximation techniques for their optimal control , and then express a podtshs with a safety objective as a pomdp with a multiplicative cost function .
[ pomdpdef](pomdp @xmath37 ) a pomdp is a tuple @xmath38 where 1 .
@xmath39 is a finite set of states 2 .
@xmath40 is a finite set of possible control inputs 3 .
@xmath41 is a finite set of observations 4 .
@xmath42 $ ] is a state transition function that assigns a probability measure to state @xmath43 given state @xmath44 and control @xmath45 for all @xmath19 : @xmath46 5 .
@xmath47 $ ] is an observation function that assigns a probability measure to observation @xmath48 given state @xmath44 and control @xmath49 for all @xmath19 : @xmath50 6 .
@xmath51 $ ] is an initial probability measure over the state space @xmath39 : @xmath52 the state evolves stochastically and is markovian ( e.g. , the state at the next time step depends only on the current state and action ) .
the information available to the controller at time @xmath19 is @xmath53 ; that is , the controller can not directly observe the state .
the control input at each time step is selected according to a control policy @xmath14 , which maps the available information at each time @xmath19 onto @xmath54 .
[ def : policy ] for a pomdp @xmath37 , a policy @xmath14 for some time horizon @xmath55 is a sequence of functions , @xmath56 , such that @xmath57 .
we consider non - randomized policies , i.e. ones that assign a single control input to each possible @xmath58 , which are sufficient for the problem we consider @xcite .
a control policy @xmath14 induces a probability space @xmath59 over the pomdp with state space @xmath60 , @xmath3-algebra @xmath61 on @xmath2 , and probability measure @xmath15 based on @xmath62 , @xmath63 , @xmath64 , and @xmath14 .
the execution of a pomdp is as follows . at time @xmath65 , state @xmath66 is produced from initial distribution @xmath67 . at each subsequent time @xmath68 , an observation @xmath48 is produced according to @xmath69 , and added to the list of past observations and control inputs to produce @xmath70 @xmath71 .
the control input is chosen according to @xmath72 , and cost @xmath73 is accrued .
the next state @xmath43 is then generated according to @xmath74 .
the goal is to minimize the expected sum of costs accrued according to function @xmath75 over a time horizon @xmath55 by optimally choosing control actions according to the policy @xmath14 .
@xmath76\ ] ] equation can be solved using dynamic programming , much like for a markov decision process @xcite .
the value function , @xmath77 , represents the expected sum of costs accrued from time @xmath19 to @xmath55 given that @xmath58 has been recorded thus far , and is computed recursively backwards in time . however , since the size of vector @xmath58 increases with @xmath19 , and is difficult to store , the optimal control input and value function can instead be expressed as a function of a belief state .
the belief state is a _
sufficient statistic _ for the information vector @xmath58 because it condenses all information necessary for making optimal decisions without sacrificing optimality . for an additive cost function @xmath75
, the belief state is a function that describes the probability of being in state @xmath78 given all past observations and actions , @xmath79 $ ] @xcite . by treating the belief state as the true state of the system ,
can be equivalently solved as an mdp over the belief state .
the optimal policy @xmath80 is hence defined in terms of the belief state , with @xmath81 the space of all beliefs .
the optimal policy can be found using a value function over the belief space , @xmath82 the transition operator @xmath83 provides the next belief state @xmath84 given the current observation , action , and belief state , @xmath85 and the likelihood of the observation is @xmath86 .
sondik @xcite first showed that for a finite horizon @xmath87 , the value function at each time @xmath19 is piecewise linear and convex in the belief state , and thus can be expressed as @xmath88 the functions @xmath89 , or `` @xmath90-vectors '' , characterize the value of being in state @xmath91 at time @xmath19 when a specific action @xmath92 is taken , plus the expected sum of future rewards assuming all subsequent actions are chosen optimally .
because each @xmath90-vector is associated with a specific action , selection of the optimal @xmath90-vector in defines the optimal policy for belief @xmath93 at time @xmath19 .
the set @xmath94 of @xmath90-vectors needed to exactly represent the value function @xmath95 at time @xmath19 is finite , but grows exponentially , since computing @xmath94 requires calculation of @xmath96 @xmath90-vectors .
an exact @xmath90-vector representation is therefore often infeasible , and approximate solutions are required .
we draw upon methods in point - based value iteration ( pbvi ) @xcite , @xcite because they provide a lower bound to the value function , which is key for our safety verification problem . in the most general pbvi method ,
a finite subset @xmath97 is selected , then one @xmath90-vector is generated for each belief point @xmath98 , @xmath99 , so that @xmath100 .
the value at some @xmath93 not necessarily in @xmath101 can be approximated by @xmath102 ( as compared to ) with @xmath103 , since we presume that for any @xmath93 in a neighborhood of @xmath104 the same action will likely be optimal @xcite .
hence @xmath105 can be generated recursively from @xmath106 without enumeration over all possible combinations of observations and subsequent @xmath90-vectors in @xmath106 , by using the backup operation @xmath107 for each @xmath108 . essentially , the pbvi algorithm consists of selecting a set of belief points @xmath109 , and repeatedly applying to each element of @xmath101 . for @xmath55 finite ,
the backup operator is applied to @xmath101 @xmath55 times . for pbvi methods that have been extended to pomdps with a continuous state space and discrete observation and action spaces @xcite ,
the @xmath90-vectors are replaced by @xmath90-functions defined over the continuous space @xmath110 .
since observations and actions are finite , the number of @xmath90-functions remains finite , and the value function is piecewise - linear and convex under the @xmath90-function representation .
a podtshs has discrete and continuous states with interacting dynamics that are characterized by stochastic kernels , and observations that may also be hybrid .
it can be expressed within the pomdp framework by extending definition [ pomdpdef ] .
[ podtshs](podtshs @xmath111 ) .
a podtshs is a class of pomdp , i.e. a tuple @xmath112 , in which 1 .
@xmath113 is a hybrid state space with @xmath114 a finite set of states with cardinality @xmath115 and continuous state dimension @xmath116 2 .
@xmath40 is a finite set of control inputs affecting both discrete and continuous state transitions 3 .
@xmath117 is a hybrid observation space with @xmath118 and continuous observation dimension @xmath119 4 .
@xmath120 $ ] is a stochastic transition kernel that assigns a probability measure over @xmath121 at time @xmath122 given @xmath44 and @xmath45 for all @xmath19 : @xmath123 with @xmath124 5 .
@xmath125 $ ] is an observation function that assigns a probability measure over @xmath126 at time @xmath19 given @xmath44 for all @xmath19 : @xmath127 6 .
@xmath128 $ ] is an initial probability measure over @xmath129 : @xmath130 while we presume @xmath131 is finite , a continuous or hybrid input set can be approximated as a finite set when computing safety probabilities and the optimal policy .
the state transition kernel comprises a discrete component @xmath132 that governs mode updates , and a kernel @xmath133 for continuous state transitions . for modeling purposes , we choose to order such that the discrete mode @xmath134 updates first at each time step , and the subsequent mode @xmath135 influences the evolution of @xmath20 to @xmath136 .
@xmath137 the functions @xmath63 and @xmath64 are also separated into discrete and continuous components , @xmath138 functions @xmath133 , @xmath139 , and @xmath140 are borel - measurable stochastic transition kernels over @xmath141 , and @xmath132 , @xmath142 , and @xmath143 are standard probability distributions over finite state spaces .
we consider specifically a switched affine system , such that the continuous state @xmath29 evolves according to @xmath144 the @xmath145 are independent and identically distributed gaussian random variables for all @xmath19 , @xmath146 .
the matrix @xmath147 and function @xmath36 change according to the mode @xmath135 .
we assume the discrete observations @xmath148 depend only on @xmath149 , and the observations @xmath150 depend linearly on @xmath29 , corrupted by additive gaussian noise @xmath151 , with @xmath151 independent and identically distributed for all @xmath19 , @xmath152 .
@xmath153 kernels @xmath133 , and @xmath139 admit gaussian densities @xmath154 , and @xmath155 .
we also assume @xmath140 admits a gaussian density @xmath156 .
for ease of notation , we let @xmath157 , @xmath158 , and @xmath159 @xmath160 for @xmath161 , @xmath162 , and @xmath163 . we require that the following lipschitz properties hold , which are guaranteed for @xmath164 , @xmath165 , and @xmath166 , given that they are gaussian densities and @xmath131 is finite .
@xmath167 we define the maximum values @xmath168 and @xmath169 .
while we impose assumptions of linearity and additive gaussian noise in and to facilitate subsequent derivations , these assumptions can be relaxed in certain cases , which will be highlighted where appropriate .
we use stochastic optimal control to find both a control policy that maximizes the probability of the state remaining within a safe region of the state space , as well as an estimate of that probability @xcite . for a compact borel set @xmath170 , terminal time @xmath55 , and predefined policy @xmath14 , the objective to optimize is @xmath171 $ ] .
this objective can be expressed more commonly as an expected value of a multiplicative cost function , since for a random variable @xmath172 and event @xmath147 , @xmath173 = \mathbb{e}[{\bf 1}_a(x)]$ ] , with the indicator function @xmath174 if @xmath175 and @xmath176 otherwise , as shown in @xcite : @xmath177.\ ] ] the maximal safety probability and optimal policy @xmath178 are given by @xmath179 in the fully observable case , @xcite gives a dynamic programming formulation for optimizing , which returns both the maximal safety probability and optimal policy .
we would like to take a similar approach to find both and .
formally , we would like to solve the following problem .
[ probstate ] for a podtshs @xmath180 with a safe set @xmath181 and time horizon @xmath182 , we wish to 1 . compute the maximal probability of remaining within @xmath183 for @xmath55 time steps .
2 . compute the optimal policy @xmath178 given by .
if the maximal probability and optimal policy can not be computed exactly ( which is quite likely @xcite ) , an approximation that produces a suboptimal policy and lower bound on the maximal safety probability is desired .
we exploit the pbvi method to solve problem [ probstate ] , by transforming problem [ probstate ] into an optimal control problem for a pomdp .
hence we first show the safety problem for @xmath111 can be reduced to a dynamic program , despite a non - standard belief state .
we then show that the @xmath90-functions and belief states can be approximately represented in closed form and that finite collections of each may be generated and used to approximate , similar to a point - based pomdp solver .
we present two approximations of problem [ probstate ] for the podtshs @xmath111 : the first discretizes @xmath39 to produce a finite state pomdp , and the second preserves continuity in @xmath39 by using a gaussian mixture approach , thus characterizing the podtshs by a collection of weights , means , and covariances .
the multiplicative nature of the cost function for the safety problem renders the belief state for an additive cost pomdp inapplicable , and we derived a different sufficient statistic @xmath184 for problem [ probstate ] in @xcite .
this sufficient statistic produces a modified conditional distribution of the current state that includes the probability that all past states are in the safe set @xmath183 .
@xmath185\ ] ] we define the _ information state _ as the function @xmath186 ( where @xmath0 is the space of integrable functions ) associated with the probability distribution produced by @xmath187 , so that @xmath188 for all @xmath189 , @xmath190 .
note that the information state is distinct from the belief state ( e.g. the conditional distribution of the current state ) .
the information state updates recursively with a bounded linear operator @xmath191 ( for proof see @xcite ) @xmath192 where @xmath193 is given by @xmath194 in comparing to , the latter integrates over the compact hybrid set @xmath183 , as opposed to a summation over finite set @xmath39 .
we define a dynamic programming recursion over @xmath3 as @xmath195 \end{aligned}\ ] ] with solution @xmath196 .
the optimal policy is @xmath197 , with @xmath198 for all @xmath199 $ ] .
[ lem : sigalpha ] for any @xmath19 , the value function can be written as @xmath200 lemma [ lem : sigalpha ] leads to the following representation of the value function evaluated at some @xmath201 .
@xmath202 with @xmath203 used to denote the observation - dependent optimal @xmath90-function in @xmath204 that maximizes @xmath205 ( the value function @xmath206 ) for a specific observation @xmath207 .
the set of @xmath90-functions at time @xmath19 is @xmath208 with @xmath209 the control inputs chosen according to .
[ lem : sigconvex ] the value function is convex in @xmath3 for all @xmath210 , @xmath211 , @xmath212 @xmath213 and @xmath214 : @xmath215 lemmas [ lem : sigalpha ] and [ lem : sigconvex ] show that the value function is convex and admits an @xmath90-function representation , hence @xmath111 is amenable to pomdp solution techniques . however
, we can not use lemma [ lem : sigalpha ] to solve problem [ probstate ] directly , since @xmath94 is not finite and the @xmath90-functions and information states have no common structure .
we first consider a finite state pomdp @xcite , whose solution converges to the true safety probability and optimal policy .
the state space @xmath39 is discretized to obtain a vector representation of @xmath90 and @xmath3 .
the observation space is unchanged ( i.e. hybrid ) , because the set of observations only affects the finiteness of sets @xmath94 and @xmath216 .
we defer discussion of producing finite collections of @xmath94 and @xmath216 to section [ sec : pbvi ] . given safe set @xmath217 , let @xmath218 .
denote @xmath219 , the finite lebesgue measure of @xmath220 .
each @xmath221 is partitioned into a finite number of subsets , so that @xmath222 , with @xmath223 pairwise disjoint ( i.e. @xmath224 for all @xmath225 ) .
finally , let @xmath226 be the diameter of partition @xmath223 so that @xmath227 , with @xmath228 .
the partition of @xmath183 is denoted by @xmath229 with @xmath230 .
each element @xmath231 has a representative point @xmath232 and the set @xmath233 is the discrete representation of @xmath183 .
we do not consider here how the points @xmath234 are selected , but an example is provided in section [ sec : example ] .
the function @xmath235 maps a state @xmath236 to its representative point @xmath234 and the function @xmath237 is the set - valued map from discrete point @xmath234 to its corresponding set @xmath231 .
we will abuse notation slightly and interchangeably write @xmath238 and @xmath239 , since @xmath149 is mapped to itself , and hence the property of interest is the mapping from @xmath29 to @xmath240 .
the discrete state space is defined as @xmath241 , with @xmath242 a single variable that represents all states @xmath243 .
[ pomdpapprox](pomdp approximation to podtshs , @xmath244 ) .
the pomdp approximation is a tuple @xmath245 where 1 .
@xmath246 is a finite set of discrete states 2 .
@xmath41 is as defined in definition 3 3 .
@xmath40 is as defined in definition 3 4 .
@xmath247 $ ] is a discrete state transition function that assigns probabilities to elements of @xmath246 5 . @xmath248 $ ] is as defined in definition 3 , but is conditioned only on values @xmath249 rather than @xmath250 6 .
@xmath251 $ ] is a function that assigns probabilities to elements of @xmath246 at time zero we define the transition function as @xmath252 with @xmath253 , and the initial distribution @xmath254 on @xmath246 as @xmath255 the discrete probability space is @xmath256 with @xmath257 , @xmath258 the @xmath3-algebra on @xmath259 , and @xmath260 the probability measure uniquely defined by @xmath254 , @xmath63 , @xmath261 , and a control policy @xmath262 , @xmath263 , with @xmath264 the set of all information states @xmath265 defined on @xmath246 at time @xmath19 .
we further define the operator @xmath266 and the intermediate vector @xmath267 for any @xmath268 , @xmath269 , @xmath270 as @xmath271 the safety problem for @xmath244 is to find latexmath:[$p_{\mathrm{safe},\delta}^n(\rho_{\delta } ; k_{\delta } ) = \sup_{\pi_{\delta}\in\pi_{\delta } } \mathbb{p}^{\pi_{\delta}}\left [ z_n \in k_{\delta } , \ , \forall\ , n=0,\dotsc , n state @xmath265 and the value function @xmath273 $ ] for @xmath210 .
the discrete information state represents a probability mass function over @xmath246 , and can be expressed as an integral over an equivalent density ( just as @xmath274 ) .
@xmath275 with @xmath276 given by @xmath277 this can be verified by substituting the expression for @xmath261 in terms of @xmath278 into and using an induction argument .
the value function is @xmath279 the maximum probability of remaining within @xmath280 over @xmath55 time steps is @xmath281 the safety probability for the finite state approximation @xmath244 converges to the true solution as grid size parameter @xmath282 tends to zero . to show this , we first describe the error between the continuous information state @xmath283 and the vector approximation @xmath265 .
[ thm : sigdiscrete ] the density @xmath284 defined in satisfies @xmath285 for all @xmath250 , @xmath286 , and @xmath287 given by @xmath288 with @xmath289 $ ] , @xmath290 . to prove convergence of the value function @xmath291 to @xmath292
, we must first show that integration over the infinite spaces @xmath293 and @xmath294 results in a bounded solution .
consider the following two lemmas regarding integration of @xmath164 and @xmath165 over unbounded sets @xmath295 and @xmath296 , respectively .
[ lem : finiteinty ] for any @xmath297 , for all @xmath298 , @xmath189 , the following holds : @xmath299 \delta_{i , q}^x\ ] ] with @xmath300 and @xmath301 .
in other words , @xmath302 is the lebesgue measure of region @xmath223 mapped to the observation space @xmath303 via the matrix @xmath304 , and increased by @xmath305 in all directions , with @xmath306 the largest eigenvalue of @xmath307 .
a similar result holds for the integral of @xmath165 over @xmath308 .
[ lem : finiteintx ] for any @xmath297 , for all @xmath298 , @xmath189 , and any @xmath269 , @xmath309 , the following holds : @xmath310 \delta_{i , q}^x,\ ] ] @xmath311 and @xmath312 , with @xmath313 the largest eigenvalue of @xmath314 . in order to show convergence of to , we require some additional definitions .
first , similarly to @xmath315 , we define piecewise constant function @xmath316 as @xmath317 , so that @xmath318 we also define @xmath319 in the same manner as @xmath320 , except that it is directly related to @xmath321 , i.e. uses the same optimal control input @xmath92 , and the same combination of @xmath322-functions ( determined by @xmath203 ) . in other words , @xmath323 is identical to @xmath321 in terms of the optimal policy tree from time @xmath19 to @xmath55 , but the values are calculated using @xmath324 and @xmath325 , @xmath326 for a specific @xmath90-function @xmath327 associated with @xmath328 .
the superscript @xmath327 for @xmath329 and @xmath330 indicates that the same choice of @xmath92 and combination of @xmath331 are used for both @xmath332 and @xmath333 .
a bound on the difference between @xmath332 and @xmath333 is given in the following lemma .
[ lem : alphatilde ] for any @xmath334 $ ] , and any function @xmath335 , the associated function @xmath336 defined in satisfies @xmath337\delta^x\ ] ] for all @xmath91 .
the constants @xmath338 and @xmath339 are equal to @xmath340 and @xmath341 from lemmas [ lem : finiteinty ] and [ lem : finiteintx ] , respectively .
we now can show convergence of the approximate safety probability over the discretized state space to the true safety probability .
[ thm : valuefuncerrordisc ] for any time @xmath334 $ ] , and any @xmath342 , @xmath343 , the error between the value function and the value function based on the finite state approximation is bounded above by @xmath344 with @xmath345 .
specifically , the safety probability for podtshs @xmath111 over time horizon @xmath55 satisfies @xmath346\delta^x.\ ] ] theorem [ thm : valuefuncerrordisc ] shows that the finite state approximation @xmath244 provides a means to approximately compute through the safety probability for @xmath244 , . as @xmath347 , the finite state safety probability converges to the true value , and the policy @xmath348 converges to @xmath178 . the linearity and additive gaussian noise assumptions for the continuous state dynamics and observations and are required only in the proofs of lemmas [ lem : finiteinty ] and [ lem : finiteintx ] , and also in the sense that they guarantee the lipschitz properties used in all proofs .
therefore , nonlinear dynamics and non - gaussian noise are permissible , so long as is satisfied , and an equivalent bound on the integrals in lemmas [ lem : finiteinty ] and [ lem : finiteintx ] can be derived .
we now consider a different approximation by representing the information state and @xmath90-functions as gaussian mixtures , such that each are characterized by a finite set of weights , means , and covariances , dependent on the mode @xmath149 .
difficulty arises from the incorporation of the indicator function @xmath349 in and .
integration over the compact set @xmath183 rather than all of @xmath39 violates the preservation of the gaussian form of @xmath3 under operator @xmath350 , and similarly for the @xmath90-functions . to preserve the gaussian mixture structure , we therefore propose a radial basis function ( rbf ) approximation @xcite to the indicator function , using gaussians as the basis function . for each @xmath221
, we set @xmath351 typically , both the centers @xmath352 and the number of components @xmath353 are fixed beforehand . the weights and covariances can then be chosen to optimize the approximation .
for simplicity we will denote @xmath354 by @xmath355 .
this approximation is valid since the rbfs are dense in @xmath356 for @xmath357 @xcite , i.e. given any function @xmath35 in @xmath356 , a weighted combination of rbfs can approximate @xmath35 to arbitrary accuracy given enough components , and @xmath349 is in @xmath0 .
however , the discontinuity in @xmath358 produces the gibbs phenomenon at the boundary of @xmath221 in the rbf approximation .
although these oscillations will always exist for a finite number of components , they could possibly be mitigated @xcite .
the oscillations can be constrained to a smaller region of @xmath183 ( shorter wavelength ) with the addition of more components , indicating that the integrated error may be reduced even if the pointwise error is not .
because we are interested only in integrating over @xmath183 , this works to our advantage .
we define a new operator @xmath359 and a new @xmath90-function @xmath360 by inserting the rbf approximation into and , respectively .
we do not include the scaling factor @xmath361 in the operator @xmath359 , which is difficult to incorporate into a gaussian mixture .
if the information state only represented the probability distribution over the current state of the system , we could simply ensure the weights all sum to one , but it includes the probability that past states were in @xmath183 , and hence is not guaranteed to integrate to one .
once we implement pbvi in section iv , we are only concerned with finding the @xmath90-function that maximizes the inner product @xmath362 , which is unaffected by a constant scaling factor , and hence we lose nothing by disregarding it .
we also presume continuous observations , as in section [ sec : abs_finite ] . @xmath363\tau(s'|s , u)\sigma_g(s)\,dx \label{eq : rbfphi}\\ \alpha^g_{y , u,\sigma}(s ) & = \int_{\mathcal{s } } \alpha_{n+1,g}^{*(y)}(s')\gamma(y|s')\tau(s'|s , u)\,ds'\left[\sum_{i=1}^{i_{q } } w_i(q ) \phi_i(x ) \right]\label{eq : rbfalpha}\end{aligned}\ ] ] we provide two lemmas stating that the operator @xmath364 and the @xmath90-function update @xmath360 preserve the gaussian mixture representation of @xmath365 and @xmath366 for all @xmath19 .
these lemmas rely on the fact that @xmath165 and @xmath164 are gaussian , and that a product of gaussian distributions is again a gaussian .
we need some additional assumptions , however , to simplify the derivations and make the following lemmas valid .
[ ass : gauss1 ] the transition kernel @xmath132 does not depend on @xmath29 , i.e. the mode update is dependent only on the previous mode : @xmath367 .
[ ass : gauss3 ] the matrices @xmath368 and @xmath304 in and are invertible for all @xmath369 . without the first assumption
, we would need to represent @xmath370 by a gaussian mixture , which can not be exact since we are approximating a function over discrete states by a continuous function @xcite .
the second assumption allows us to express @xmath371 and @xmath372 as gaussian densities over @xmath29 . while these assumptions are quite restrictive , they can be relaxed at a cost of introducing additional error , as discussed later in remark [ rem : gauss ] .
[ lem : phigauss ] the operator @xmath364 is closed under gaussian mixtures , i.e. for @xmath373 a gaussian mixture with @xmath374 components , @xmath375 is also a gaussian mixture with @xmath376 components for any @xmath269 , @xmath268 .
[ lem : alphagauss ] the expression for @xmath360 is closed under gaussian mixtures , i.e. if @xmath377 is a gaussian mixture with @xmath378 components , @xmath379 is also a gaussian mixture with @xmath380 components , for any @xmath269 , @xmath268 , @xmath201 .
lemma [ lem : phigauss ] implies that we can approximate @xmath3 through a gaussian mixture and use the equivalent update operator @xmath364 , hence the gaussian mixture approximation of @xmath3 is @xmath381 .
similarly , the gaussian mixture of any @xmath90-function is written as @xmath382 for all @xmath19 .
@xmath383 similarly , the gaussian mixture approximation of any @xmath90-function is written : @xmath384 the weights , means , and covariances are defined recursively .
their exact representations can be found in the appendix .
note that although the gaussian mixture representation of @xmath360 has a finite number of components given that the representation of @xmath385 is finite , the actual @xmath90-function , @xmath366 , is expressed as the integral of @xmath360 over @xmath41 .
therefore , without the assumption that @xmath41 is finite , @xmath366 must have an infinite number of components ( by breaking the integral over @xmath41 into a summation over regions of size @xmath386 and taking the limit as @xmath387 ) .
we take some liberty in overlooking this discrepancy , because it does not affect the proofs in this section .
we impose a finite set @xmath41 in section [ sec : pbvi ] , which makes the gaussian mixture representation of the @xmath90-functions indeed valid , and discuss additional error implications .
the subscript @xmath36 denotes that we are computing an estimate based on the gaussian mixture approximation .
the value function @xmath388 is described through the recursion @xmath389 the safety problem for the gaussian mixture approximation is defined as @xmath390 since @xmath165 , @xmath164 , and @xmath166 are gaussian , and based on assumptions [ ass : gauss1 ] - [ ass : gauss3 ] made above , the gaussian mixture representations of @xmath90 and @xmath3 are exact , aside from the approximation of @xmath349 using rbfs . to quantify the error incurred from calculating @xmath391 as opposed to @xmath392 ( from integration of over @xmath39 rather than over @xmath183 ) , we define the error @xmath393 we first analyze the error between @xmath283 and @xmath365 , which is stated in terms of the 1-norm rather than the @xmath394-norm , to be consistent with considering the integrated error in .
[ thm : sigconterror ] the gaussian mixture approximation @xmath365 of @xmath283 satisfies @xmath395 for any @xmath334 $ ] , @xmath268 , and @xmath269 , with @xmath396 and @xmath397 . to show convergence of to , we define the function @xmath398 which utilizes the same policy tree as @xmath399 for a specific @xmath400 .
this is equivalent to , except that @xmath398 is defined by replacing @xmath401 with .
@xmath402\ ] ] with @xmath329 the optimal control input associated with @xmath332 and @xmath330 indicating that @xmath403 is chosen according to the indices selected by @xmath203 for @xmath404 .
note that the @xmath90-functions are no longer guaranteed to be equal to zero outside of @xmath183 , and also are not guaranteed to be bounded above by one .
so long as is of bounded height , however , the @xmath90-functions also remain bounded , and we write @xmath405 .
we could also adjust the weights in so that the @xmath90-functions can not exceed one , although this may increase the error @xmath406 .
the following lemma describes the relation between @xmath321 and @xmath319 .
[ lem : alphgtil ] for any @xmath334 $ ] , and any @xmath335 , the associated function @xmath407 defined in satisfies @xmath408 where @xmath409 .
[ thm : valuefuncerrorgauss ] for any time @xmath334 $ ] , and any @xmath342 , @xmath410 , the error between the value function given @xmath283 and the value function given @xmath365 using the gaussian mixture approximation is bounded above by @xmath411 with @xmath412 .
specifically , the safety probability for podtshs @xmath111 over time horizon @xmath55 satisfies @xmath413 theorem [ thm : valuefuncerrorgauss ] shows that the convergence of the gaussian mixture approximation of both @xmath3 and the value function depends on the integrated error between the indicator function over @xmath183 and the rbf approximation , rather than the pointwise error . although the pointwise error may not converge to zero for a finite number of components in the rbf , the integral of the error can be small , as we will show in section [ sec : example ] .
[ rem : gauss ] the linearity and additive gaussian noise assumptions on the dynamics and , as well as assumptions [ ass : gauss1 ] - [ ass : gauss3 ] , are used in the gaussian mixture approximation to ensure that the only error in the value function and information state approximations comes from the approximation of the indicator function @xmath401 by a gaussian mixture . dropping these assumptions
requires that we approximate @xmath165 , @xmath132 , and @xmath164 by gaussian mixtures , which is possible but introduces additional error that we have chosen not to consider .
a numerical solution of problem [ probstate ] via either a discrete or gaussian mixture approximation additionally requires sets @xmath94 and @xmath216 to be finite , whereas we have sets of infinite size because of the uncountable nature of @xmath41 .
however , a lower bound on the safety probabilities and can still be obtained , by characterizing the error that results from using @xmath414 and @xmath415 , finite collections of @xmath90-functions and information states , respectively .
we again exploit point - based approximation methods described in section [ sec : back_pomdp ] .
we examine the generation of subsets of the information states and @xmath90-functions , and prove that each guarantees a lower bound to the safety probability of whichever approximation of section [ sec : abstraction ] we choose .
in contrast to most point - based solvers , we do not assume a finite set of observations , and hence discretize the observations merely for the computation of the @xmath90-functions .
combining belief space sampling with discretized observations assures a lower bound to the safety probability .
we characterize the error from using a sampled subset of @xmath216 for performing backup operations ( as in ) .
presume that a finite set of information states @xmath416 has been generated according to one of the many methods available @xcite .
we generate a finite set @xmath105 of @xmath90-functions , one for each @xmath417 .
the convexity of the value functions guarantees that the subset @xmath105 provides a lower bound on @xmath292 .
further , we can show that the error between the approximate value function @xmath418 ( represented by @xmath105 ) and the true value function @xmath292 ( represented by the complete set of @xmath90-functions @xmath94 ) depends on how densely we sample @xmath216 .
we define an intermediate value function @xmath419 that generates @xmath105 recursively from @xmath204 , i.e. that introduces one point - based backup from the full set @xmath204 .
then @xmath316 is written as a function of @xmath420 rather than @xmath421 .
the value function @xmath418 is formally defined as @xmath422 @xmath423 so that @xmath418 is characterized by the finite set @xmath105 at each time step .
we also define an intermediate value function @xmath424 that generates @xmath105 recursively from @xmath204 , i.e. that introduces one point - based backup from the full set @xmath204 .
then @xmath316 is written as a function of @xmath420 rather than @xmath421 .
@xmath425 we let @xmath426 denote the maximum hausdorff distance over @xmath19 between points in @xmath416 and points in @xmath216 with respect to the metric @xmath427 .
@xmath428 in the following , we do not distinguish between the vector and gaussian mixture representations of @xmath3 and @xmath90 , because the results apply to both cases .
[ lemma3 ] for any @xmath334 $ ] and @xmath286 , the error introduced in one iteration of point - based value iteration is at most @xmath426 .
@xmath429 we now use lemma [ lemma3 ] to derive a bound between the true value function and the point - based approximation at any time @xmath19 . [ thm1 ] for a set of information states @xmath216 , sampled set @xmath416 , and any time @xmath334 $ ] and @xmath286 , the error from using point - based value iteration versus full value iteration is bounded above by @xmath430 thus the error between the point - based approximation and the actual value function is directly proportional to how densely @xmath416 is sampled , and converges to zero as @xmath416 approaches @xmath216 .
the proofs of lemma [ lemma3 ] and theorem [ thm1 ] are a straightforward extension of those appearing in @xcite , and are omitted . over the uncountably infinite space @xmath41 , we can not calculate @xmath431 for all @xmath268 , despite a finite set of @xmath92 and @xmath3 .
we therefore compute a subset of @xmath431 for the finite set @xmath432 , to approximate @xmath433 as @xmath434 .
we discretize @xmath41 in a similar fashion to the discretization of @xmath39 in section [ sec : abs_finite ] . however , since @xmath303 is not compact , we consider an expanded set @xmath435 defined so that the probability of observing a value @xmath207 for @xmath436 that is outside of @xmath437 is approximately zero , i.e. @xmath438 , @xmath439 .
the sets @xmath440 are divided into disjoint subsets @xmath441 , @xmath442 .
we also define @xmath443 , such that @xmath444 . the partition of @xmath437 is denoted @xmath445 with @xmath446 .
the diameter of partition @xmath441 is @xmath447 , with maximum diameter @xmath448 .
each partition @xmath449 has a representative element @xmath450 and a set @xmath451 .
the function @xmath452 maps observation @xmath268 to its representative value @xmath450 ; the set - valued function @xmath453 maps the point @xmath454 to the corresponding set @xmath455 .
the finite observation space is @xmath456 . for the finite state approximation , the transition function @xmath457 $ ]
is defined as @xmath458 for the gaussian mixture approximation , we define the transition function @xmath459 in the same fashion as , but with @xmath460 so that the @xmath90-functions will also be gaussian mixtures at each time step .
note that @xmath461 , @xmath462 is a set of mesh points inside @xmath455 associated with @xmath463 , and @xmath464 is a weight proportional to the mesh spacing ( determined , e.g. , by the trapezoidal rule for numerical integration ) .
we use @xmath465 and @xmath466 to denote the approximation using a finite subset of @xmath467 , with the important distinction that the subset is now generated by a finite collection of observations ( as opposed to @xmath416 , i.e. we assume here that the set @xmath216 is finite ) .
the value function is then @xmath468 , where @xmath469 defined as in section [ sec : abs_finite ] , only with @xmath470 replaced by @xmath471 .
@xmath472 with @xmath473 we again define the intermediate value function @xmath474 , with @xmath475 calculated using @xmath476 ( as opposed to @xmath477 ) to capture the error introduce in one backup iteration using discretized observations .
we can then bound the error introduced in one iteration of approximating the @xmath90-vectors through discretized observations .
[ lem : discobsalpha ] for any time @xmath334 $ ] and @xmath343 , the error between @xmath478 and @xmath474 satisfies @xmath479 given that the discretized observations @xmath463 are chosen so that @xmath480 and with @xmath481 the largest lebesgue measure of sets @xmath440 .
lemma [ lem : discobsalpha ] requires defining the representative points @xmath461 so that the integral of @xmath482 over @xmath483 is greater than a piecewise constant approximation integrated over @xmath483 , which can be satisfied by picking the points @xmath484 where the gaussian density @xmath164 is minimized within cell @xmath441 . without this requirement , finding @xmath485 at a finite set of points
still guarantees a lower bound to the value function for any time @xmath19 , and is intuitively more accurate as @xmath486 .
lemma [ lem : discobsalpha ] leads to the following theorem regarding the error between @xmath478 ( based on continuous observations ) and @xmath487 ( based on discretized observations ) .
we again use the notation @xmath488 to indicate that @xmath488 is represented by the set @xmath489 of @xmath90-functions calculated using the discretized observations .
[ thm : discobsvf ] given discretized observation process @xmath490 with transition function , for any time @xmath334 $ ] , the error between @xmath478 calculated according to @xmath41 and @xmath487 calculated according to @xmath490 satisfies @xmath491 for any @xmath343 , with @xmath481 the largest lebesgue measure of sets @xmath440 .
specifically , the safety probability for @xmath244 satisfies @xmath492 the results of discretizing the observations for the gaussian mixture abstraction are nearly identical to those for the finite state abstraction .
the main difference arises in approximating the integral @xmath493 with a gaussian sum . to ensure the approximate value function
provides a lower bound to @xmath494 , we must under - approximate the integral @xmath495 for each @xmath463 .
we define @xmath496 similarly to @xmath466 , using discretized observations and the rbf approximation to the indicator function , @xmath497\sigma_{n , g}(x , q)\,dx\right\ } \end{aligned}\ ] ] and @xmath498 is the intermediate value function that finds the optimal @xmath499 , rather than in @xmath500 .
we can bound the error between @xmath494 and @xmath498 , and between @xmath494 and @xmath496 , equivalently to lemma [ lem : discobsalpha ] and theorem [ thm : discobsvf ] , respectively .
[ lem : gaussobsalpha ] for any time @xmath334 $ ] and @xmath410 , the error between @xmath388 and @xmath501 satisfies @xmath502 given that the observations @xmath463 are chosen so that @xmath503 and with @xmath481 the largest lebesgue measure of sets @xmath440 .
[ thm : gaussobsvf ] given discretized observation process @xmath490 with transition function , for any time @xmath334 $ ] , the error between @xmath388 calculated according to @xmath41 and @xmath504 calculated according to @xmath490 satisfies @xmath505 for any @xmath410 , with @xmath481 the largest lebesgue measure of sets @xmath440 .
specifically , the safety probability for the gaussian mixture approximation satisfies @xmath506 the proofs of lemma [ lem : gaussobsalpha ] and theorem [ thm : gaussobsvf ] follow directly from the proofs of lemma [ lem : discobsalpha ] and theorem [ thm : discobsvf ] . to summarize , given either the finite state or gaussian mixture approximation , we can subsequently 1 ) sample @xmath207 from @xmath41 and @xmath92 from @xmath40 to generate the progressive subsets @xmath507 or @xmath508 , and 2 ) discretize @xmath41 and use the set @xmath490 to calculate @xmath509 or @xmath510 , which are then used to generate @xmath511 and @xmath512 .
using sets @xmath507 and @xmath465 in place of @xmath264 and @xmath467 provides a lower bound to the safety probability @xmath513 that converges as @xmath426 and @xmath514 approach zero ( and similarly for @xmath508 and @xmath515 ) .
the temperature regulation problem is a benchmark example for hybrid systems , and a stochastic version with perfect state information is presented in @xcite .
we consider the case of one heater , which can either be turned on to heat one room , or turned off .
the temperature of the room at time @xmath19 is given by the continuous variable @xmath20 , and the discrete state @xmath516 indicates the heater is on at time @xmath19 , and @xmath517 denotes the heater is off .
the stochastic difference equation governing the temperature is given by @xmath518 with constants @xmath519 , @xmath520 , and @xmath521 , and @xmath145 i.i.d .
gaussian random variables with mean zero and variance @xmath522 .
the control input is given by @xmath523 with @xmath524 , but the chosen control is not always implemented with probability @xmath525 .
instead , @xmath134 is updated probabilistically , dependent on @xmath49 and @xmath526 , with transition function @xmath527 . so while function @xmath528 deterministically returns a single control input , control input @xmath529 may not always be implemented . to model this as a partially observable problem ,
assume the actual temperature is unknown , and only a noisy measurement is available to the controller .
the controller does , however , know whether the heater is on or off at time @xmath19 ( i.e. @xmath134 is perfectly observed ) .
the observation @xmath530 is given by @xmath531 , with @xmath151 i.i.d .
gaussian random variables with mean zero and variance @xmath532 .
it is desirable to keep the temperature of the room between @xmath533 and @xmath534 degrees celsius at all times , hence the safe region @xmath535 $ ] does not depend on the discrete state @xmath134 ( so @xmath536 ) .
we consider the probability of remaining within @xmath183 for @xmath537 time steps given initial temperature distribution @xmath166 normally distributed with varying mean @xmath538 and variance @xmath539 .
the initial mode is given as @xmath540 .
the finite state and gaussian mixture approximations are used in a pbvi algorithm in the style of perseus @xcite .
we consider a uniform grid ( @xmath541 constant for all @xmath327 , @xmath149 ) over the region @xmath542 for the finite state approximation , with representative points at the end - point of each grid cell .
for example , setting @xmath543 gives @xmath544 for @xmath545 and @xmath546 , and a total of @xmath547 cells @xmath223 .
the function @xmath548 maps @xmath149 to itself , and maps @xmath29 to the nearest @xmath240 in absolute value .
the gaussian mixture approximation utilizes an rbf approximation of the indicator function calculated using matlab s _ gmdistribution _ function .
we used a reduction process to limit the number of components of each @xmath90 and @xmath3 for the gaussian mixture approximation .
similar gaussians are combined into a single component based on the 2-norm distance between functions @xcite .
each mixture was limited to 30 components to reduce overall computation time without overly sacrificing accuracy .
this number can easily be changed , however , depending on the importance of speed versus accuracy .
both approximations employ a sequence of sampled sets @xmath416 and a finite set of observations to calculate the @xmath90-functions for the pbvi algorithm . to generate the sets @xmath416 , we initialized a set of 40 states @xmath549 normally distributed with variance @xmath550 and mean @xmath551 randomly chosen uniformly on @xmath183 .
each @xmath549 was updated according to @xmath364 or @xmath552 with @xmath92 chosen randomly and @xmath207 sampled from the corresponding @xmath549 ( i.e. @xmath553 ) .
this process was repeated @xmath55 times , so that for each time step we had a set of 40 sampled @xmath3s .
the finite set of observations were produced by a uniform grid over @xmath554 $ ] , again using end - points as the representative observations . to compare performance of the finite state and gaussian mixture approximations , we present computation times and safety probability estimates for each , with varying @xmath282 , @xmath514 , and number of components in the indicator approximation .
safety probabilities for varying initial distributions @xmath555 are presented in figs .
[ fig : discretexcompare ] and [ fig : gaussindcompare ] for the finite state approximation and gaussian mixture approximation , respectively .
the optimal policy at time zero is shown for varying @xmath555 in figs .
[ fig : discretexcompareu ] and [ fig : gaussindcompareu ] for the finite and gaussian approximations , respectively .
computation times for the gaussian mixture approximation are given in table [ table : gausstime ] , and for the finite state approximation in table [ table : discretetime ] .
we also show sample rbf approximations to the indicator function @xmath349 in fig .
[ fig : indcompare ] with varying numbers of components @xmath353 .
the error between the rbf approximation and @xmath349 for varying @xmath353 is shown in fig .
[ fig : l1err ] .
as the number of components increases , the approximation becomes more accurate , although as seen in fig .
[ fig : indcompare ] , oscillations remain at the endpoints of @xmath183 .
the increasing accuracy is most apparent in fig .
[ fig : l1err ] , and demonstrates the convergence towards zero of the error @xmath406 with increasing @xmath353 .
we show safety probabilities for @xmath556 .
decreasing @xmath514 causes a slight increase in the safety probabilities , as expected , but there is not a significant improvement in the probability estimates , although as seen in tables [ table : gausstime ] and [ table : discretetime ] , the increase in computation time is significant .
this is likely problem - specific , and the value of @xmath514 may have a greater impact for some applications .
the safety probability estimates for the finite state approximation are in general greater than for the gaussian mixture approximation .
the mixture reduction method employed , as well as the indicator function approximation , make the gaussian method seemingly less accurate than the finite state approximation .
however , over a finer mesh @xmath282 , the finite state method results in greater computation time .
note also that for @xmath557 , the safety probabilities decrease relative to the probabilities for @xmath543 , which highlights that although sampling from the information space and discretizing the observations guarantees a lower bound , no such guarantee exists when discretizing the state space , and there is a chance the safety estimates do not bound the true safety probabilities from below .
although the coarse grid produces similar results to the fine grid ( @xmath558 versus @xmath559 ) , in higher dimensional problems the number of grid cells becomes prohibitive even when @xmath282 is large , and the gaussian mixture approximation may be more computationally tractable .
all scenarios produce a nearly identical optimal thresh - hold policy based on the initial mean @xmath551 , indicating that an optimal policy may be computed fairly quickly using any of the above methods .
the computation time is unfortunately still quite high in both cases , and at this time each method is likely applicable to systems with only a few modes and continuous state dimensions of no more than two or three .
it is likely , however , that computation time can be improved by using more sophisticated point - based solvers , such as @xcite .
further gains may be possible through adaptive gridding techniques , similar to those in @xcite , to decrease the number of finite states required without sacrificing accuracy . .
[ cols="^,>,>,^ , > , > " , ] interestingly , increasing the number of components in the rbf approximation to the indicator function only slightly improves the safety estimates of the gaussian mixture approximation , although the error from increasing the number of components to 30 drops significantly .
this may be caused by the mixture reduction technique , leading to a loss in the added benefit of an increased number of components when that number is again reduced . however , although the error with @xmath560 is large , we obtain safety estimates that are quite similar to the finite state approximation .
this requires further investigation , but may help in decreasing computation time without losing significant accuracy by choosing @xmath353 to be small .
we have presented the first numerical results for verification and controller synthesis for safety objectives , given a partially observable dtshs . we have considered two approximations that enable the use of a well - known pomdp optimization technique .
the first approximation discretizes the state space over a compact set @xmath183 and enables a vector representation of the information states and @xmath90-functions .
the second approximates the indicator function over compact set @xmath183 using a finite set of gaussian radial basis functions and enables a gaussian mixture representation of the information states and @xmath90-functions
. we can apply point - based value iteration to either approximation , and guarantee a lower bound to the safety probability , which is proven to converge to the true safety probability of the original podtshs .
a simple numerical example shows that both methods provide similar safety estimates .
the finite state approximation is faster when a coarse discretization is used , but quickly becomes slower than the gaussian mixture approximation with a finer discretization .
therefore , although the gaussian mixture produces lower safety estimates , it may be better suited to higher dimensional problems .
although we present a switched affine system with additive gaussian noise , both approximations may be extended to non - gaussian systems .
convergence results for the finite state approximation apply to arbitrary transition kernels @xmath133 and @xmath139 , given they still satisfy certain lipschitz conditions .
the gaussian mixture approximation further requires approximating @xmath133 , @xmath132 , and @xmath139 with gaussian mixtures , and introduces additional error .
we also focus on the safety problem , although the computational techniques presented will apply to other verification properties such as reachability , reach - avoid objectives , and others , by modifying the information state and @xmath90-functions slightly .
we are currently working to formally extend these results to other verification objectives and more complex applications .
because both methods are relatively slow , we plan to continue to refine them to decrease computation time , which is possible through the use of more sophisticated existing point - based solvers .
we are also exploring more efficient computation by exploiting problem structure , through the use of adaptive gridding schemes and other representations of @xmath90 and @xmath3 beyond vectors and gaussian mixtures . by induction . at time @xmath55 , @xmath561 by defining @xmath562 , we obtain the desired result . note that this definition of @xmath563 is in line with the definition given in section [ sec : back_pomdp ] , because although it does not represent a full policy tree ( being at the terminal time , there are no more branches on the tree ) , it does represent the immediate value of being in state @xmath564 , given by @xmath565 .
next , assuming @xmath566 , @xmath292 can be written as @xmath567\notag\\ & = \max_{u\in\mathcal{u}}\int_{\mathcal{y}}\sup_{\alpha_{n+1}\in\gamma_{n+1}}\langle \alpha_{n+1 } , \phi_{y , u}\sigma_n \rangle \mathbb{p}(dy|\sigma_n , u ) \\ & = \max_{u\in\mathcal{u}}\int_{\mathcal{y}}\sup_{\alpha_{n+1}\in\gamma_{n+1}}\left[\int_k\int_{{\mathcal{s } } } \alpha_{n+1}(s')\gamma(y|s ' ) \tau(s'|s , u)\sigma_n(s)\,ds'\,ds\right]dy.\label{eq : vn } \end{aligned}\ ] ] then for a specific observation @xmath207 , action @xmath92 , and function @xmath568 ( index @xmath327 used to identify a particular @xmath569 ) , the function @xmath570 can be defined as @xmath571 because @xmath570 does not depend on @xmath283 , we can redefine the supremum over all @xmath204 to be over all @xmath570 .
@xmath572 for a specific @xmath3 , @xmath92 , and @xmath207 , we define @xmath573 with @xmath203 denoting the index @xmath327 of the @xmath90-function @xmath570 that maximizes the inner product .
we further simplify @xmath292 as @xmath574 therefore , the set of all @xmath575 is described by @xmath576 with @xmath209 the control inputs chosen according to , and @xmath292 may be written as @xmath577 by induction . at time @xmath65 , @xmath578 and
the inequality is trivially satisfied . for all @xmath579 , assume @xmath580 . at time @xmath581 , for any @xmath268 and any @xmath269 , @xmath582 we add and subtract @xmath583 and @xmath584 , apply the triangle inequality , and use the lipschitz inequalities .
@xmath585\notag\\ & \hspace{10mm}\leq \min\left\ { \frac{1}{p(y|\sigma_n , u ) } , \frac{1}{p(y|\hat{\sigma}_n , u)}\right\ } \left[\vphantom{\sum_{q\in\mathcal{q}}\int_{k_q } } h_y^{(2)}\|x ' - \xi(x')\|\phi_v^ * + \phi_w^*h_x^{(2)}\|x ' - \xi(x')\|+ \phi_w^*h_q\|x - \xi(x)\|\phi_v^*\right . \notag\\ & \hspace{35 mm } \left.+ \phi_w^*|\sigma_n - \hat{\sigma}_n|\sum_{q\in\mathcal{q}}t_q(q'|s , u)\int_{k_q}\tau_x(x'|q',\xi(s),u)\,dx\right ] \label{eq : thmsig2 } \end{aligned}\ ] ] since @xmath165 is bounded by @xmath586 , and the lebesgue measure of @xmath221 is at most @xmath587 , we obtain @xmath588 . \\
\end{aligned}\ ] ] combining terms and applying the induction hypothesis gives the desired result .
we exploit properties of the derivative of a gaussian distribution , which bounds the lipschitz constants for @xmath164 from above . for clarity of notation , we will write @xmath75 rather than @xmath304 , with dependence on @xmath149 implicit .
the constant @xmath589 is the maximum value of the derivative of @xmath590 with respect to @xmath29 : @xmath591 recall that @xmath592 for @xmath147 a symmetric matrix .
therefore @xmath593 . taking the derivative again with respect to @xmath594 and setting equal to zero
, we see that the maximum of occurs at @xmath595 .
although @xmath596 , we create a tighter bound for the case in which @xmath597 is greater than @xmath305 ( for @xmath598 such that there exists @xmath599 for which @xmath600 , the upper bound @xmath589 is attained ) using the following function .
@xmath601 then , @xmath602 we use the change of variable @xmath603 , with @xmath604 , to rewrite the second term of , and apply an identity for integrals of polynomials . @xmath605 inserting into proves the lemma . by induction . at time @xmath55 , @xmath606 since for any @xmath436 , by definition @xmath607 .
assume for all @xmath608 , @xmath609\delta^x$ ] .
for @xmath610 , separate terms and apply the triangle inequality as in the proof of theorem [ thm : sigdiscrete ] .
@xmath611\delta^x+ n_q\left[h_q + \beta_1^xh_x^{(2 ) } + \beta_2^x\right]\delta^x\label{eq : alphatil2 } \end{aligned}\ ] ] the second term of comes from lemma [ lem : finiteinty ] and noting that @xmath612 represents a probability that is bounded above by one .
the third term comes from lemma [ lem : finiteintx ] and the lipschitz inequality for @xmath132 .
the term @xmath401 does not affect the bound , and only indicates that both @xmath321 and @xmath319 are equal to zero for @xmath613 . applying the induction hypothesis to gives the desired result . by construction .
at any time @xmath334 $ ] , given @xmath614 and @xmath615 , we can rewrite the value function evaluated at @xmath3 in terms of @xmath90-functions .
@xmath616 assume without loss of generality that @xmath617 .
then , because @xmath618 by definition of the optimality of @xmath619 , we can write @xmath620 applying lemma [ lem : alphatilde ] to the first term of , and noting that the integral in the second term is in fact taken over @xmath183 rather than @xmath294 since @xmath621 is zero for all @xmath622 , we obtain @xmath623\delta^x + n_q\lambda\eta_n^{\sigma}\delta^x\ ] ] which completes the proof .
the proof of both lemma [ lem : phigauss ] and lemma [ lem : alphagauss ] require the following gaussian density identities .
first , the product of two gaussian densities is again a gaussian density , up to a constant factor .
@xmath624 with @xmath625 second , for invertible matrix @xmath147 , constant @xmath93 , and variables @xmath29 and @xmath207 , @xmath626 both identities are easily shown directly .
the proof then follows by construction .
given @xmath627 , observation @xmath628 , and control input @xmath629 , the operator @xmath630 produces @xmath631\tau(s'|s , u)\sigma_{n , g}(s)\,dx.\end{aligned}\ ] ] we use the notation @xmath632 , @xmath633 , and @xmath634 for the weights , means , and covariances of the indicator function approximation to distinguish them from the other gaussian mixtures representing @xmath3 and @xmath90 .
replacing @xmath365 by its gaussian mixture representation in , and expanding @xmath482 and @xmath278 , gives @xmath635 t_q(q'|q , u)\phi(x';ax+g(q',u),{\mathcal{v } } ) \notag\\ & \hspace{30 mm}\times\left[\sum_{l=1}^l w^{\sigma}_{l , n}(q)\phi(x ; \mu^{\sigma}_{l , n}(q ) , { \mathcal{p}}^{\sigma}_{l , n}(q))\right]\\ & = \sum_{q=1}^{n_q}\sum_{l=1}^l\sum_{i=1}^{i_q}w_i^i(q)w_{m , n}^{\sigma}(q)y_{q}(y^q|q')t_q(q'|q , u)|c^{-1}|\phi(x';c^{-1}y^x ; c^{-1}{\mathcal{w}}c^{-t})\notag\\ & \hspace{20 mm}\times \int_{\mathbb{r}^m}\phi(x ; \mu_i^i(q ) , { \mathcal{p}}_i^i(q))\phi(x ; \mu_{l , n}^{\sigma}(q ) , { \mathcal{p}}_{l , n}^{\sigma}(q))\notag\\ & \hspace{30 mm}\times|a^{-1}|\phi(x ; a^{-1}(x'-g(q',u ) ) , a^{-1}{\mathcal{v}}a^{-t})\,dx\label{eq : phi1}\\ & = \sum_{q=1}^{n_q}\sum_{l=1}^l\sum_{i=1}^{i_q}|a^{-1}||c^{-1}|w_i^i(q)w_{m , n}^{\sigma}(q)y_{q}(y^q|q')t_q(q'|q , u)\phi(x';c^{-1}y^x ; c^{-1}{\mathcal{w}}c^{-t})\notag\\ & \hspace{20 mm}\times\phi(\mu_i^i(q ) ; \mu_{l , n}^{\sigma}(q ) , { \mathcal{p}}_i^i(q ) + { \mathcal{p}}_{l , n}^{\sigma}(q))\int_{\mathbb{r}^m}\phi(x ; \tilde{\mu},\tilde{{\mathcal{p}}})\notag\\ & \hspace{30 mm}\times \phi(x ; a^{-1}(x'-g(q',u ) ) , a^{-1}{\mathcal{v}}a^{-t})\,dx \label{eq : phi2}\\ & = \sum_{q=1}^{n_q}\sum_{l=1}^l\sum_{i=1}^{i_q}|a^{-1}||c^{-1}|w_i^i(q)w_{m , n}^{\sigma}(q)y_{q}(y^q|q')t_q(q'|q , u)\phi(x';c^{-1}y^x , c^{-1}{\mathcal{w}}c^{-t})\notag\\ & \hspace{20 mm}\times \phi(\mu_i^i(q ) ; \mu_{l , n}^{\sigma}(q ) , { \mathcal{p}}_i^i(q ) + { \mathcal{p}}_{l , n}^{\sigma}(q))\phi(a^{-1}(x'-g(q',u ) ) ; \tilde{\mu } , \tilde{{\mathcal{p } } } + a^{-1}{\mathcal{v}}a^{-t})\label{eq : phi3}\end{aligned}\ ] ] line follows from , line from combining @xmath636 and @xmath637 according to , and from a second application of and setting the integral of a gaussian density over @xmath638 equal to one . a final application of and gives @xmath639 which is again a gaussian mixture with @xmath640 components . specifically , @xmath641 with @xmath642,\\ { \mathcal{p}}_{q , l , i , n+1}^{\sigma}(q ' ) & = \left[c^t{\mathcal{w}}^{-1}c + \left(a\tilde{{\mathcal{p}}}a^t+{\mathcal{v}}\right)^{-1}\right]^{-1},\end{aligned}\ ] ] and @xmath643,\\ \tilde{{\mathcal{p } } } & = \left[\left({\mathcal{p}}_i^i(q)\right)^{-1 } + \left({\mathcal{p}}_{l , n}^{\sigma}(q)\right)^{-1}\right].\end{aligned}\ ] ] by construction .
given @xmath644 for some observation @xmath628 , then @xmath645 for control input @xmath629 is written @xmath646\label{eq : alph}\\ & = \sum_{q'\in\mathcal{q}}\int_{\mathbb{r}^m}\left[\sum_{m=1}^m w_{m , n+1}^{\alpha , y}(q')\phi(x';\mu_{m , n+1}^{\alpha , y}(q ' ) , { \mathcal{p}}_{m , n+1}^{\alpha , y}(q'))\right]\notag\\ & \hspace{20 mm}\times y_q(y^q|q')\phi(y^x ; cx',{\mathcal{w}})t_q(q'|q , u)\notag \\ & \hspace{30 mm}\times\phi(x';ax+g(q',u),{\mathcal{v}})\,dx ' \left[\sum_{i=1}^{i_{q } }
w_i^i(q ) \phi_i^i(x ) \right]\label{eq : alph1}\\ & = \sum_{q'=1}^{n_q}\sum_{m=1}^m\sum_{i=1}^{i_q}w_{m , n+1}^{\alpha , y}(q')w_i^i(q)y_q(y^q|q')t_q(q'|q , u)\notag\\ & \hspace{20 mm}\times \phi(x;\mu_i^i(q),{\mathcal{p}}_i^i(q))|c^{-1}|\phi(c^{-1}y^x ; \mu_{m , n+1}^{\alpha , y}(q ' ) , c^{-1}{\mathcal{w}}c^{-t } + { \mathcal{p}}_{m , n+1}^{\alpha , y}(q'))\notag\\ & \hspace{30 mm}\times\int_{\mathbb{r}^m}\phi(x';\hat{\mu } , \hat{{\mathcal{p}}})\phi(x';ax+g(q',u),{\mathcal{v}})\,dx'\label{eq : alph2}\\ & = \sum_{q'=1}^{n_q}\sum_{m=1}^m\sum_{i=1}^{i_q}w_{m , n+1}^{\alpha , y}(q')w_i^i(q)y_q(y^q|q')t_q(q'|q , u)\notag\\ & \hspace{20 mm}\times \phi(x;\mu_i^i(q),{\mathcal{p}}_i^i(q))\phi(y^x ; c\mu_{m , n+1}^{\alpha , y}(q ' ) , { \mathcal{w}}+ c{\mathcal{p}}_{m , n+1}^{\alpha , y}(q')c^t)\notag\\ & \hspace{30 mm}\times |a^{-1}|\phi(x ; a^{-1}(\hat{\mu}-g(q',u ) ) , a^{-1}(\hat{{\mathcal{p}}}+{\mathcal{v}})a^{-t})\label{eq : alph3 } \end{aligned}\ ] ] @xmath647 line follows from one application of , line from and another application of , and a final product of gaussian densities gives . hence @xmath360 is a gaussian mixture with @xmath648 components . recalling that @xmath649 , it follows that @xmath650 with @xmath651\\ { \mathcal{p}}_{y , q',m , i , n+1}^{\alpha}(q ) & = \left[\left({\mathcal{p}}_i^i(q)\right)^{-1 } + a^t\left(\hat{{\mathcal{p}}}+{\mathcal{v}}\right)^{-1}a\right]^{-1},\end{aligned}\ ] ] and @xmath652\\ \hat{{\mathcal{p } } } & = \left[c^t{\mathcal{w}}^{-1}c + \left({\mathcal{p}}_{m , n+1}^{\alpha , y}(q')\right)^{-1}\right]^{-1}.\end{aligned}\ ] ] by induction .
at time zero , @xmath653 , so that @xmath654 .
assume that @xmath655 for all @xmath656 .
then at time @xmath122 we have , for some @xmath268 and @xmath269 , @xmath657\notag\\ & \leq \phi_w^*\left [ \vphantom{\sum_{i=1}^{i_q}}\left\|\sigma_n - \sigma_{n , g}\right\|_1 + \sum_{q\in{\mathcal{q}}}\int_{\mathbb{r}^m}\left|{\bf 1}_{k_q}(x ) - \sum_{i=1}^{i_q}w_i(q)\phi_i(x)\right|\sigma_{n , g}(x , q)\,dx\right]\label{eq : thmconterror1 } \end{aligned}\ ] ] the first term of line follows because the integral over @xmath183 is less than the integral over all of @xmath39 , since @xmath183 is a compact subset of @xmath39 .
the induction hypothesis completes the proof . by construction . for any time
@xmath334 $ ] , given @xmath286 and @xmath410 , we can rewrite the value function evaluated at @xmath3 in terms of @xmath90-functions . @xmath663 as in the proof of theorem [ thm : valuefuncerrordisc ] , assume without loss of generality that @xmath664 .
@xmath665 define @xmath437 such that @xmath666 . then @xmath667 \notag\\ & \hspace{10 mm } + \int_{\mathcal{y}\backslash\overline{k}}\sum_{z , z'\in k_{\delta}}\alpha_{n+1,\delta}^{*(y)}(z')\gamma(y|z')\tau_{\delta}(z'|z , u^*)\sigma_{n,\delta}(z)\,dy \notag\\ & \hspace{10 mm}- \int_{\mathcal{y}\backslash\overline{k}}\sum_{z , z'\in k_{\delta}}\alpha_{n+1,\delta}^{*(\psi_y)}(z')\gamma(y|z ' ) \tau_{\delta}(z'|z , u^*)\sigma_{n,\delta}(z)\,dy\notag\\ & \leq \int_{\overline{k}}\sum_{z , z'\in k_{\delta}}\left[\alpha_{n+1,\delta}^{*(y)}(z')\gamma(y|z')\tau_{\delta}(z'|z , u^*)\sigma_{n,\delta}(z)\,dy\right.\notag\\ & \hspace{10 mm}\left
- \alpha_{n+1,\delta}^{*(\theta(y))}(z')\gamma(y|z')\tau_{\delta}(z'|z , u^*)\sigma_{n,\delta}(z)\,dy \right ] + \frac{\epsilon}{n } \label{eq : lemdiscobs2 } \end{aligned}\ ] ] note that is nonnegative , meaning that using @xmath668 produces a lower bound to the actual value function given by @xmath669 .
this follows because @xmath670 is chosen optimally for only a subset of @xmath41 ( at the points @xmath671 ) , and for all other @xmath268 , @xmath672 is suboptimal , producing a lower value .
next , we can bound @xmath673 from below based on how the points @xmath463 are defined .
@xmath674\,dy + \frac{\epsilon}{n}\\ & \leq n_q\overline{\lambda}h_y^{(1)}\delta^y + \frac{\epsilon}{n } \end{aligned}\ ] ] by induction . at time @xmath55 ,
@xmath675 since @xmath676 .
assume for all @xmath677 that @xmath678 .
then , at time @xmath19 , @xmath679 applying the induction inequality and combining terms completes the proof .
m. prandini and j. hu , _
stochastic reachability : theoretical foundations and numerical approximation _ ,
lecture notes in control and information sciences.1em plus 0.5em minus 0.4emspringer verlag , 2006 , pp .
107139 . i.
mitchell and j. templeton , `` a toolbox of hamilton - jacobi solvers for analysis of nondeterministic continuous and hybrid systems , '' in _ hybrid systems : computation and control _ , 2005 , vol . 3414 , pp .
480494 .
r. ghaemi and d. del vecchio , `` control for safety specifications of systems with imperfect information on a partial order , '' _ ieee transactions on automatic control _
59 , no . 4 , pp . 982
995 , 2014 .
s. soudjani and a. abate , `` adaptive and sequential gridding procedures for the abstraction and verification of stochastic processes , '' _ siam journal on applied dynamical systems _
12 , no . 2 ,
pp . 921956 , 2013 .
e. brunskill , l. kaelbling , t. lozano - perez , and n. roy , `` planning in partially - observable switching - mode continuous domains , '' _ annals of mathematics and artificial intelligence _ ,
185216 , 2010 . | assuring safety in discrete time stochastic hybrid systems is particularly difficult when only noisy or incomplete observations of the state are available .
we first review a formulation of the probabilistic safety problem under noisy hybrid observations as a dynamic program over an equivalent information state .
two methods for approximately solving the dynamic program are presented .
the first method approximates the hybrid system as an equivalent finite state markov decision process , so that the information state is a probability mass function .
the second approach approximates an indicator function over the safe region using radial basis functions , to represent the information state as a gaussian mixture . in both cases , we discretize the hybrid observation process and generate a sampled set of information states , then use point - based value iteration to under - approximate the safety probability and synthesize a suboptimal control policy .
we obtain error bounds and convergence results in both cases , assuming switched affine dynamics and additive gaussian noise on the continuous states and observations .
we compare the performance of the finite state and gaussian mixture approaches on a simple numerical example . |
magnetic field penetrates a sample of layered superconductor forming pancake vortices in every layer of the sample .
calculation of statistical properties of the pancake matter is a very complicated question for several reasons of which the following is the most important .
the pancakes interact through two mechanisms , magnetic and josephson .
while magnetic mechanism is reduced to pairwise interaction of the pancakes , potential energy due to the interlayer josephson coupling can not be brought down to such simple form .
available methods lead to a non - local description @xcite or rely on the duality transformation @xcite which maps the pancake system onto a system of quasi - two - dimensional coulomb - like plasma .
both approaches are technically complicated and difficult to generalize .
alternatively , many authors neglect josephson mechanism assuming that for extremely anisotropic superconductors its influence is small compared to the effects of the magnetic interaction @xcite . in this paper
we show that under certain circumstances it is possible to include josephson coupling into consideration with the help of general physical devices . at sufficiently high field
the josephson and magnetic interlayer coupling get smaller than the intralayer interaction . in such a situation
the system is viewed as a collection of weakly coupled 2d layers . if pancakes in every layer form a structure with short range crystalline order both magnetic and josephson mechanisms could be accounted for .
the resultant model is a version of the sine - gordon field theory .
it could be analyzed with the help of usual field theoretical tools .
the proposed approach allows one to find an upper bound for the pancake lattice melting temperature .
another interesting application is that it becomes possible to compare contributions of the magnetic and josephson interactions to the statistical properties of the pancake matter .
thus , we could determine under what circumstances it is possible to drop the josephson coupling from the description and when it can not be omitted .
the paper is organized as follows . in sect .
ii starting from the vortex hamiltonian for the anisotropic superconductor we reproduce the well - known result for the magnetic potential energy of the pancake lattice . in sect . iii the energy due to the josephson coupling is derived . in sect .
iv we compare josephson and magnetic interaction energies .
the purpose of the latter comparison is to determine when it is permissible to ignore the josephson mechanism and when it is not . in sect .
v we apply the derived hamiltonian to the problem of pancake lattice melting . sect .
vi contains conclusions .
we start with the vortex hamiltonian @xmath0 @xcite : @xmath1 the symbol @xmath2 stands for @xmath3 ; vector field @xmath4 denotes vorticity .
the component @xmath5 in the @xmath6th superconducting layer coincides with the pancake density in that layer : @xmath7 where @xmath8 is the position of the @xmath9th vortex in the @xmath6th layer .
the parallel components of the vorticity @xmath10 correspond to the josephson vortices between superconducting layers .
the josephson vorticity between @xmath6th and @xmath11th layer is equal to : @xmath12\sigma \right ) .
\nonumber\end{aligned}\ ] ] here @xmath13 is the spacing between neighboring layers , @xmath14 is 2d coordinate vector .
function @xmath15 is defined in such a way that it is localized on a straight line connecting points @xmath16 and @xmath17 .
physically , it is non - zero on the josephson vortex which connects a pancake in @xmath6th layer and another pancake in @xmath11th layer .
since a vortex can not terminate inside the superconductor we have @xmath18 or , equivalently , in fourier space @xmath19 let us rewrite the kernel @xmath20 in the following manner : @xmath21 note , that @xmath22 contains no @xmath23 . in other words
, it does not depend on the josephson coupling between layers .
it describes magnetic interaction between pancakes . on the contrary ,
kernel @xmath24 vanishes when the josephson energy vanishes .
now , using 3d - transversality of the vorticity ( eq.([div ] ) ) , we establish the following : @xmath25 this allows us to rewrite the hamiltonian ( [ h ] ) : @xmath26 in other words , only the 2d - transverse part @xmath27 of the josephson vorticity @xmath28 contributes to the interaction energy .
magnetic part of the energy is the easiest to handle .
imagine that in every layer the pancakes form 2d crystal with short range hexagonal lattice characterized by the lattice constant @xmath29 .
we assume that it is possible to choose the length scale @xmath30 in such a way that within @xmath31 the lattice translational invariance is undisturbed .
although , we do not need the exact value of @xmath31 for our calculations , yet , our ability to set such a scale is imperative .
if the above prerequisite is met a pancake lattice in @xmath6th layer could be described locally by a 2d displacement vector field @xmath32 .
this field does not change much on the scale @xmath31 .
the pancake density is expressed as : @xmath33 the sum is upon vectors of the reciprocal lattice .
the 2d divergence is defined as : @xmath34 the quantity @xmath35 denotes the average density of the pancakes : @xmath36 observe , that two first terms of ( [ vz ] ) vary slowly on the scale of the lattice constant @xmath29 while the third term quickly oscillates as function of @xmath14 : @xmath37 we have for the magnetic interaction energy : @xmath38 where the symbol @xmath39 stands for @xmath40 .
the term proportional to @xmath41 describes the statics of the longitudinal displacements
. we will not be interested in them for they are suppressed @xcite .
the remaining parts are : @xmath42 the notation @xmath43 was adopted in the formula above . to avoid clutter
we omit the subscript ` @xmath44 ' which denoted the transverse displacement mode : we write here @xmath32 rather than @xmath45
. we will continue to do so , yet , one has to remember that the displacement vector field we consider is pure shear .
the first term in the above expression gives usual @xmath46 : @xmath47 to proceed further with the second term we need to calculate @xmath22 : @xmath48 if @xmath49 which is true for @xmath50 the above expression for @xmath22 could be simplified : @xmath51 thus , we get for the second term in ( [ hmag0 ] ) : @xmath52 here we neglected the weak dependence of @xmath53 on @xmath54 since @xmath55 . finally :
@xmath56 here @xmath57 is the magnitude of the smallest reciprocal lattice vectors : @xmath58 properties of the hamiltonian similar to ( [ hmag1 ] ) was investigated in @xcite
. at finite temperature the above expression could be simplified .
it is enough to notice that at @xmath59 contribution of a term @xmath60 is proportional to @xmath61 .
thus , when fluctuations of the lattice are substantial it is permissible to retain in ( [ hmag1 ] ) lowest @xmath62 terms only .
consequently , it is not necessary to sum over all possible @xmath63 in the above formula .
it is sufficient to keep only six terms corresponding to six elementary reciprocal lattice vectors whose absolute values are equal to @xmath57 .
namely , the last term of ( [ hmag1 ] ) might be written as : @xmath64 our model becomes a vector version of a sine - gordon field theory .
now we add the josephson coupling to our hamiltonian .
this means that we have to include the josephson vorticity as well . at first
it seems like an impossible task since we have to account for the josephson vorticity fluctuations .
we will demonstrate that under rather broad conditions these fluctuations could be included by simple renormalization of the josephson coupling parameter .
[ h ] on fig.1 an example of josephson vorticity fluctuation is shown . in the absence of fluctuations
two pancake vortices in neighboring layers would be connected by a straight josephson vortex .
fluctuations create an outgrowth on the vortex line .
the size of the outgrowth is @xmath65 .
the deformed vortex is represented by a solid line .
the deformed vortex could be decomposed " into a vortex loop and a straight vortex connecting two pancakes .
thus , the problem of the josephson vorticity fluctuations could be reduced to the problem of the thermally induced vortex loop gas .
thermal fluctuations infuse superconductor with loops of josephson vortices of typical size @xmath65 .
this size @xmath65 diverges at the superconducting transition temperature @xmath66 and vanishes at @xmath67 .
it is finite at @xmath68 . loosely speaking
, @xmath65 characterizes a spacial scale above which
mean - field " josephson coupling could be defined and fluctuations are unimportant .
therefore , if we are in the regime @xmath69 the fluctuations might be omitted .
in such a case fluctuations result in small vibrations of the josephson vortex .
the only consequence of the fluctuations would be renormalization of effective value of the josephson coupling , which is equivalent to renormalization of the penetration depth from its bare value @xmath70 to experimentally measurable value of @xmath71 . of course
, it is desirable to generalize the argumentation beyond ( [ flucti ] ) .
how this could be done is discussed in appendix [ energy_of_loop ] .
once we settle the issue of the fluctuations we could proceed with the derivation of the josephson contribution to the energy of the pancake system . as in the previous section
we assume that in every layer the pancakes form a 2d structure with ( at least ) short range crystalline order .
remember also that the longitudinal displacements of the pancakes are suppressed . in absence of the longitudinal displacements
the josephson vorticity is a periodic function of vector @xmath72 defined as ( fig.[fig2 ] ) : @xmath73 [ h ] two - dimensional vector field @xmath74 could be expressed as a sum of smooth and oscillating parts : @xmath75 where the smooth part is equal to : @xmath76 vector function @xmath77 is defined like so : @xmath78 the lattice vector @xmath79 is chosen to deliver minimum to the expression @xmath80 ( see fig.([fig3 ] ) ) .
function @xmath81 is invariant under lattice translations . for any @xmath82 vector @xmath77
always lies within the primitive lattice cell .
physically , eq.([vort ] ) means that when two lattices are shifted only slightly with respect to each other @xmath83 .
however , when @xmath82 does not fit into the primitive cell the josephson vortices rearrange themselves to minimize their own length . in this case @xmath84 .
for example , if @xmath85 is the lattice vector then @xmath86 .
[ b ] next , we calculate the oscillating part .
full josephson vorticity field between @xmath6th and @xmath11th layers is given by eq.([j_vort ] ) with @xmath87 ( @xmath88 is the lattice vector which shows the undisturbed position of the @xmath9th pancake ) and @xmath89 .
the fourier transform of the vorticity vector field due to a single vortex @xmath9 is : @xmath90 the total vorticity is : @xmath91 it is non - zero only when @xmath63 is the reciprocal lattice vector . if @xmath63 belong to the reciprocal lattice , one prove @xmath92 thus , we have for the oscillating part : @xmath93 where the sum runs over the reciprocal lattice vectors .
the above expression is explicitly periodic with respect to both @xmath32 and @xmath94 .
this means that if one of the layers is shifted by a lattice vector the josephson vorticity remains unaffected by such a shift . to calculate the energy we need the transverse component of the josephson vorticity : @xmath95 } { iq_\|^2(\bf q_\| g(s_{\it n}))}\times\\ & & \left ( e^{i{\bf q}_\| { \bf u}_n({\bf r } ) } - e^{i { \bf q}_\| { \bf u}_{n-1}({\bf r } ) }
\right ) e^{- i{\bf q}_\| { \bf r}}.\nonumber\end{aligned}\ ] ] ( note : in 2d vector product is a scalar , not a vector . ) as for @xmath96 , it is transverse as long as @xmath32 s are transverse .
we are in position now to write down the josephson energy : @xmath97 } { q_\|(\bf q_\| g(s_{\it n}))}\right|^2 \times \label{hj2}\\ \left ( 2 - 2\cos { \bf q}_\|{\bf s}_n\right ) , \nonumber\\ h_{\rm j3 } = \frac{\phi_0 ^ 2 \rho_0 ^ 2}{8\pi } \sum_{n \ne m } \int_{\bf r } \sum_{\bf q_\|\ne 0 } { \cal k}_{\rm j } ( { \bf q}_\| , n - m ) \times \nonumber \\
\frac{[{\bf q}_\| \times \bf g(s_{\it n } ) ] [ { \bf q}_\| \times \bf g(s_{\it m } ) ] } { q^2_\|(\bf q_\| g(s_{\it n}))(\bf q_\| g(s_{\it m } ) ) } \times \nonumber \\ \left ( e^{i{\bf q}_\| { \bf u}_n } - e^{i { \bf q}_\| { \bf u}_{n-1 } } \right ) \left ( e^{-i{\bf q}_\| { \bf u}_{m } } - e^{-i { \bf q}_\| { \bf u}_{m-1 } } \right),\end{aligned}\ ] ] where we used the notation @xmath98 the term @xmath99 describes the kinetic energy of supercurrents induced by the josephson vorticity
. the kernel @xmath100 is non - zero for @xmath101 ( see appendix [ k ] ) since such currents spread over large regions .
therefore , two josephson vortices apart from each other interact via these currents .
the second term is purely local .
it describes the increase of the free energy of an interlayer josephson junction due to insertion of vortices .
the third term describes similar effect : the modification of the free energy of an interlayer josephson junction by a vorticity in another layer . as we will show below
, this effect is small . at low temperature , where it is possible to write @xmath102 , hamiltonian @xmath103 is responsible for the josephson contribution to the tilt modulus @xmath104 of the vortex lattice @xcite .
the first term corresponds to the @xmath105 piece of @xmath104 while the second and the third terms contribute to the so - called single - vortex part ( @xmath106 ) of the tilt modulus . at sufficiently high temperature
we can not approximate @xmath77 by @xmath85 .
instead we must follow different type of analysis .
the term @xmath107 may be disregarded right away if @xmath108 since at low josephson coupling we could neglect correlations of the displacement fields in different layers : @xmath109 .
thus , for @xmath108 @xmath110 since @xmath111 .
the latter statement is obviously true for @xmath77 is an odd function .
we have : @xmath112 where notation ( [ notation ] ) is used and @xmath113 are fourier transform coefficients of the odd periodic function @xmath81 .
as one notices from this formula , @xmath99 is a product of four exponents of the form @xmath114 . it will be shown that @xmath115 depends on two exponents only . at high enough temperatures
@xmath99 may be neglected in comparison with @xmath115 .
this is because @xmath116 is proportional to @xmath117 while @xmath118 is proportional to @xmath119 and the expectation value of the exponent vanishes when @xmath44 grows .
one could adopt a more formal approach . as shown in appendix [ scaling_dim ] at sufficiently large @xmath44 operator
@xmath99 becomes irrelevant in the renormalization group sense while @xmath115 remains relevant up until higher temperature .
in this temperature interval it is possible to neglect @xmath99 : its contribution to the thermodynamics is purely perturbative while the contribution of the relevant second term is singular .
the term @xmath120 could be also disregarded for the following reason .
the kernel @xmath121 may be simplified provided that the josephson length is bigger than the lattice constant : @xmath122 namely , we can write : @xmath123 the term @xmath124 is smaller than unity as long as ( [ small ] ) is valid . indeed , @xmath125 and @xmath126 .
this means that @xmath127 is smaller than @xmath128 .
therefore , @xmath120 ( corresponds to @xmath127 ) is much less than @xmath115 ( corresponds to @xmath128 ) .
thus , we retain @xmath115 only : @xmath129 } { q_\|(\bf q_\| g(s_{\it n}))}\right|^2\times \\ \left ( 2 - 2\cos { \bf q}_\|{\bf s}_n\right ) , \nonumber\\ { \cal k}_{\rm j } = \frac{1}{d_c \lambda_c^2 q^2_\|}.\end{aligned}\ ] ] we rewrite this expression in a more compact way : @xmath130 } { ( \bf q_\| g(s))}\right|^2 \left ( 1 - \cos { \bf q}_\|{\bf s}\right ) \label{h_def}.\end{aligned}\ ] ] function @xmath131 is dimensionless and periodic with the period of the hexagonal lattice . its contour graph is shown on fig.[fig4 ] . due to its periodicity @xmath131
can be expanded in a fourier series : @xmath132 [ b ] using this expansion it is possible to simplify the expression for @xmath103 at finite temperatures with the help of the trick already discussed at the end of sec.ii : @xmath133)\},\nonumber\\ & & \hat h_{q_0 } \approx -3.7,\\ & & \frac{3 |\hat h_{q_0}|}{256 \pi^5 } \approx 1.4\times 10^{-4},\end{aligned}\ ] ] here the sum over vectors @xmath63 runs only over the shortest of them .
the number @xmath134 is the fourier coefficient of @xmath135 corresponding to such reciprocal vectors .
this is the final equation for the josephson energy of the vortex system .
let us estimate coupling constants of our hamiltonian . that way we can understand what are the largest interaction in our system and when the perturbative treatment of interactions is permissible .
first , we calculate the elastic constant @xmath136 .
to find it we measure area in units of @xmath137 .
thus : @xmath138 for bscco @xmath139 assuming : @xmath140 the josephson constant corresponding to @xmath115 : @xmath141 the coupling constant corresponding to @xmath99 : @xmath142 using the estimate for the fourier coefficients @xmath143 we find : @xmath144 ratios of these two coupling constants to the elastic energy scale : @xmath145 and @xmath146 are both smaller than unity when ( [ small ] ) holds .
this explains the physical significance of ( [ small ] ) .
the latter is essentially a criterion for our vortex system to be viewed as a quasi-2d pancake gas .
if ( [ small ] ) is violated the system is better described in terms of the abrikosov vortices rather than pancakes .
the interlayer magnetic interaction constant could be estimated directly from ( [ hmag1 ] ) : @xmath147 the quantity @xmath148 is an estimate for number of layers coupled by the magnetic interaction with a given layer . the ratio of the magnetic and the josephson constants is : @xmath149 the last result is important for it allows one to judge when the josephson coupling could be neglected and when it must be retained .
as one can see from the above formula , the ratio is independent of the magnetic field and , in that sense , it is a material constant " .
if we take for bscco : @xmath150 then the ratio equals to 1.5 .
that is , even for such an extremely anisotropic material the josephson coupling is of the same order as magnetic .
in this section we find the upper bound @xmath151 on a melting temperature @xmath152 .
the gist of the following calculations is that at sufficiently high temperature the interlayer coupling becomes irrelevant and the system could be thought of as a collection of 2d layers decoupled from each other . in these layers
no long - range crystalline order is possible .
the temperature found in this fashion is not necessary a true melting point for melting could occur at even lower temperatures through a first order phase transition .
in certain situation our upper bound could serve as a reasonable estimate for the melting temperature .
this happens if the josephson coupling is sufficiently strong and the difference between the solid and the liquid phases of the pancake matter is small .
when this difference is exceeded by the bare josephson energy scale the phase transition is controlled by the josephson energy . since the latter vanishes at @xmath151 the melting line lies close to @xmath151 .
we assume that the temperature is large so that josephson term @xmath99 is irrelevant ( see appendix [ scaling_dim ] ) .
only @xmath115 part of the josephson coupling and the magnetic coupling , eq.([hmag2 ] ) , have to be accounted for .
in such a regime it is possible to get @xmath153 using simple perturbative argument which we are ready to present .
the crystalline phase is characterized by a non - zero order parameter : @xmath154 with the help of @xmath155 the interlayer interactions ( [ hmag2 ] ) and ( [ hj2 ] ) could be written in a mean - field manner as @xmath156 the above transformations reduces our multilayer problem to a problem for a single layer with the hamiltonian : @xmath157 it is presumed that @xmath158 is small : @xmath159 this guarantee the applicability of approximation ( [ mf ] ) .
it is equivalent to ( [ small ] ) .
let us transform the interlayer interaction term as follows : @xmath160 after such substitution the effective hamiltonian becomes quadratic .
thus , the problem could be solved exactly . however , when solving it we must keep in mind the self - consistency condition : @xmath161=\nonumber\\ \frac{q_0 ^ 2 t}{8\pi c_{66 } } \left [ \ln \frac{c_{66 } } { \alpha } + \langle ( { \bf q_\| u}_{n } ) ^2 \rangle + { \rm const.}\right].\nonumber\end{aligned}\ ] ] the last equation has positive solution for @xmath162 only when @xmath163 above @xmath151 the order parameter @xmath155 is zero .
since @xmath164 is a function of temperature itself , to find @xmath151 it is in fact necessary to solve the equation : @xmath165 assume that the temperature dependence of @xmath164 is given by a phenomenological formula : @xmath166 where the superconducting transition temperature @xmath66 equals to 100k for bscco .
eq.([tu_eq ] ) , thus , gives : @xmath167 let us reiterate , that this result is `` high - field '' .
namely , it is derived under the assumption that ( [ small ] ) is satisfied .
the latter inequality is equivalent to : @xmath168 thus , there is no contradiction if at low fields ( @xmath169 ) the melting temperature is higher than @xmath151 found above .
we can also find the contribution of the interlayer ( josephson and magnetic ) coupling to the total free energy density of the pancake lattice : @xmath170 where the following expression for @xmath171 was used : @xmath172 the latter expression is a trivial consequence of ( [ self - cons ] ) .
note that @xmath173 and @xmath174 .
thus , the formula for @xmath175 could be rewritten : @xmath176 interlayer contribution to the free energy , as expected , vanishes at @xmath177 .
this result is useful only if the lattice remains stable close to @xmath151 , that is , if @xmath152 is close to @xmath151 .
in this paper we proposed a description of the quasi-2d pancake lattice .
the description is a version of an elasticity theory : the hamiltonian is a functional of the displacement field only .
such a transparent structure allows easy generalization .
the most obvious extension is for the case of non - zero pinning .
this is an important direction for future studies .
another advantage of the method is that at hight magnetic field ( eq.([high_b ] ) ) all non - harmonic terms are small .
therefore , consistent perturbative analytical tools are applicable . at present we do not include dislocations and vacancies into our model . yet ,
to certain extent the effect of both dislocations and vacancies may be accounted for even at this stage . unlike the interlayer coupling , these two disorder the lattice .
the interlayer coupling tries to eliminate them by binding into a topologically neutral pairs . in turn
, dislocations and vacancies attempt to destroy the interlayer coupling by breaking translational order .
which side wins this competition could be determined through comparison of corresponding singular corrections to the free energy .
at least conceptually , this is a straightforward task .
the main goal of this paper is the derivation of the model .
however , we d like to discuss the most elementary of its consequences here .
first of all , our model allows to compare quantitatively the magnitude of the magnetic and josephson pancake interactions .
we have seen above that even for bscco ( @xmath178 ) both of these are of the same order .
therefore , models which neglect the josephson interaction are only qualitatively accurate , at best .
next , let us briefly discuss the application of formula ( [ tu_eq ] ) for pancake lattice melting in real layered superconductors . probably , it has no physical significance for bscco due to its extremely small josephson coupling .
the total interlayer coupling is also small since the magnetic interaction is of the same order as the josephson .
the transition into the liquid phase occurs because the free energy of the liquid becomes less than the free energy of the solid .
small contribution of the interlayer interaction could change the melting line slightly but otherwise is insignificant .
we would like to conjecture cautiously that ( [ tu_eq ] ) may be of relevance for pb - doped bscco .
this superconductor has much smaller anisotropy @xmath179 @xcite .
presumably , this indicates that its josephson coupling parameter is higher than that of pure bscco
. in @xcite pb - doped bscco was studied by means of magneto - optic technique .
a depinning transition was observed at @xmath180 .
the authors of @xcite interpreted the transition as a pancake lattice melting .
remarkably , they see that this transition is field independent .
this is a strong argument in favor of ( [ tu_eq ] ) being applicable since @xmath151 does not depend on the magnetic field .
another field - independent melting transition , 2d dislocation unbinding , is an unlikely candidate for it occurs at much smaller temperatures .
pb - doped bscco has @xmath181 k and @xmath182 nm . in @xcite @xmath183
was reported to be of the order of 180 nm .
thus : @xmath184 this is close to the transition at @xmath185 k reported in @xcite .
the fact , that @xmath151 and @xmath186 differs by about 25% is not by itself discouraging since penetration depth @xmath164 shows strong sample - to - sample variations @xcite .
it is very well possible that they are responsible for the deviation of the experimental transition temperature from its theoretical estimate .
yet , the final judgment about the applicability of the presented theory is to be postponed until further investigation . in conclusion
, we proposed a simple model for the pancake lattice .
it could be used to study pinning and thermodynamics of the pancake matter at sufficiently high fields and temperatures .
numerous fruitful discussions with a.l .
rakhmanov are appreciated .
the author is grateful for support provided by the dynasty foundation , by rfbr through grant 03 - 02 - 16626 and by russian federal program `` leading scientific schools '' nsh-1694.2003.2 .
in this appendix we will try to estimate the characteristic size @xmath187 of the thermally induced josephson vortex loops at given temperature @xmath44.we will also discuss what happens when @xmath65 grows bigger than @xmath29 . to achieve the first goal we calculate the energy of a single josephson vortex loop @xmath188 as a function of its radius @xmath189 and compare this energy to the temperature .
we will assume that @xmath190 is determined by the condition @xmath191 to proceed with the calculation we first find the fourier transformation of the vorticity field of a single loop whose radius is @xmath189 : @xmath192 = 2\pi r j_1 ( q_\| r ) \frac{[\hat { \bf z } \times { \bf q}_\| ] } { q_\|}.\end{aligned}\ ] ] the energy is given by : @xmath193 the integral in this formula could be transformed as follows : @xmath194 where @xmath195 is the superconducting coherence length . after a substitution @xmath196
the integral becomes : @xmath197 in deriving the above equation we used the approximation @xmath198 .
this is accurate for @xmath199 or , equivalently , @xmath200 .
finally , confining ourselves to the range @xmath201 we write : @xmath202 therefore : @xmath203 where @xmath151 is given by ( [ tu_eq ] ) . using ( [ xi_eq ] ) one
determines : @xmath204 at @xmath177 which is the highest temperature at which the model proposed in this paper is still applicable we have @xmath205 this give us the upper bound for the characteristic size of the thermally induced josephson vorticity loops . as it was discussed in the body of the paper we would like to have @xmath206 ( eq.([flucti ] ) ) .
consequently , @xmath207 it is not difficult to notice that the latter condition is not always compatible with the requirement @xmath208 which has to be imposed to observe the quasi-2d effects of the pancake physics ( eq.([small ] ) ) .
what does this mean for the analysis undertaken in this paper ?
the answer to this question is that it is possible to recover virtually all of the physics described in the paper even if ( [ flucti ] ) is not satisfied as long as @xmath209 .
( the quantity @xmath31 was defined in the paper as a size within which the 2d pancake lattice could be viewed as almost ideal . )
let us examine how this is done .
presumably , the violation of ( [ flucti ] ) might lead to problems with ( [ hj_comp ] ) .
it was calculated under assumption that it is possible to define a single - vortex line tension on a scale @xmath210 .
when vorticity fluctuation scale @xmath65 exceeds @xmath29 it becomes difficult to assign a well - defined line tension to a josephson vortex .
thus , the analysis at the scale of @xmath29 fails .
it is , however , possible to study the system at the scale of @xmath31 . at this scale
we need not to be concerned with the properties of a single josephson vortex . instead
, we will think in terms of average local vorticity which pierces interlayer josephson junction .
imagine first the situation when pancake lattices in the @xmath6th and @xmath11th layers coincide at @xmath211 .
this implies that for any @xmath14 , @xmath212 , the average josephson vorticity between these two layers is absent .
thus , the local density of the josephson energy is zero .
now we shift the lattices with respect to each other : @xmath213 .
the average vorticity @xmath214 is no longer zero as well .
we have to disentangle its contribution from contribution of the fluctuation - induced vorticity @xmath215 in the situation where fluctuations locally are strong : @xmath216 . to do so we coarse - grain the system description up to the scale @xmath190 .
after this coarse - graining @xmath217 and the bare value of the penetration depth @xmath70 has to be replaced by @xmath71 .
( the latter is the penetration depth measured experimentally . ) thus , at the scale @xmath218 the vorticity fluctuations could be neglected and it is possible to work with the effective josephson coupling instead of including fluctuations explicitly .
the `` mean - field '' vorticity @xmath219 is a periodic function of @xmath72 .
this implies that the pancake lattice energy density due to josephson coupling is a periodic function of @xmath72 as well : @xmath220 .
\label{hj_est}\end{aligned}\ ] ] this expression is completely general : it is the lowest @xmath63 terms of periodic function fourier series .
the final question now is the estimation of @xmath221 .
the latter , however , is fairly trivial .
as it was discussed in the body of the paper @xmath115 comes from the increase of the interlayer josephson junction energy due to non - zero @xmath219 .
thus , we have match ( [ hj_est ] ) and the interlayer josephson junction free energy : @xmath222\end{aligned}\ ] ] when @xmath223 both are zero .
when @xmath224 the josephson junction energy @xmath225 is no longer zero but rather equals to @xmath226 .
it is because the cos term in the above expression effectively vanishes due to fast oscillation of the phase difference .
this implies that @xmath227\right\ } \sim \frac{\phi_0 ^ 2}{16 \pi^3 \lambda_c^2 d_c}. \label{ej_est0}\end{aligned}\ ] ] since maximum of the sum in this formula is equal to 9 the coupling constant @xmath221 is : @xmath228 equations ( [ hj_est ] ) and ( [ ej_esti ] ) reproduce ( [ hj2 ] ) up to a numerical constant of order unity .
the derivation of ( [ hj2 ] ) developed in this appendix is more general than that found in the body of the paper for it does not rely on @xmath229 condition .
the advantage of the latter , however , it its ability to give the numerical constant accurately .
the fact that numerical coefficients of ( [ hj2 ] ) and ( [ ej_esti ] ) are very close is pure luck .
obviously , the estimates like ( [ ej_est0 ] ) have `` order of magnitude '' precision only .
numerical constant in ( [ ej_esti ] ) should not be taken too seriously .
we re - derived @xmath230 above .
operator @xmath231 may be re - derived in the similar fashion .
in this appendix we determine the scaling dimension ( sd ) of the operators ( [ n = m ] ) and ( [ hj ] ) .
we start by solving an auxiliary problem : we calculate sd of an exponent @xmath232 .
the easiest way to find sd is to evaluate the correlation function : @xmath233)^2 \rangle}.\end{aligned}\ ] ] the correlation function of the displacement field is found : @xmath234)^2\rangle = \int_{{\bf k } } \frac{t({\bf q}_\| { \bf k})^2}{c_{66}k^4 } \left ( 1 - e^{i{\bf k } ( { \bf r}^{\vphantom { ' } } - { \bf r } ' ) } \right).\end{aligned}\ ] ] the integral over @xmath235 was evaluated in @xcite ( eq .
( 9.3.33 ) and problem ( 9.9 ) of the latter reference ) : @xmath236 where ellipsis stand for terms which remain bound at @xmath237 .
consequently : @xmath234)^2\rangle = \frac{tq_0 ^ 2 } { 4\pi c_{66 } } \ln |{\bf r } - { \bf r}'|/a_0 + \ldots.\end{aligned}\ ] ] therefore : @xmath238 from this equation we can read off sd of the exponent : @xmath239 = t / t_u,\end{aligned}\ ] ] where @xmath151 is given by ( [ tu ] ) .
thus , sd of ( [ hj ] ) is equal to @xmath240 since it is a product of two exponents .
this sd becomes equal to 2 when @xmath177 . at @xmath241 operator ( [ hj ] )
is irrelevant , otherwise it is relevant . as usual , this mark the transition point from ordered to disordered phase driven by ( [ hj ] ) .
eq.([tu ] ) establishes the same fact derived by other means . as for ( [ n = m ] ) , its operator density is proportional to : @xmath242 here the factor @xmath243 comes from the kernel @xmath244 .
the kernel @xmath245 is calculated in appendix [ k ] .
sd of @xmath77 is @xmath240 as it is obvious from ( [ n = m ] ) , sd of the integration measure is -2 , sd of @xmath243 is 1 . totally , we have @xmath246 = 4t / t_u - 1.\end{aligned}\ ] ] equating this sd with 2 we discover that for @xmath247 operator ( [ n = m ] ) is irrelevant .
therefore , is is permissible to neglect it for these temperatures . at the same time
, operator ( [ hj ] ) remains relevant up until the temperature has grown to @xmath151 .
thus , the discussion of the sect.[upper_bound ] is valid for @xmath248
in this appendix we calculate the kernel @xmath249 required for eq.([n = m ] ) . we must evaluate the integral : @xmath250 in most layered superconductors the screening length @xmath23 is very big .
therefore , we are interested in distances @xmath251 . for such @xmath252
it is possible to neglect unity under the square root and write : @xmath253 next we perform the angular integration : @xmath254 the upper limit in this integral is assumed to be large : @xmath255 . if @xmath256 then arctg@xmath257 could be approximated by @xmath258 for almost whole integration interval .
the contribution of the interval where this approximation is invalid could be evaluated separately : @xmath259 the first integral is @xmath260 .
we neglect it .
the second integral must be integrated by parts : @xmath261 consequently : @xmath262 \text{\ for\ } r \ll \lambda_{\rm j}.\end{aligned}\ ] ] in the opposite limit @xmath263 we define a function : @xmath264 and write : @xmath265 the integral may be transformed in the following manner : @xmath266 in the last line the first integral is smaller than the second one .
indeed , the first one is @xmath267 while the second is @xmath268 .
therefore : @xmath269 note that : @xmath270 thus , for @xmath271 we have : @xmath272 | we study the pancake lattice system induced by application of transverse magnetic filed to the layered superconductor .
a simple statistical field theory for the pancake lattice is derived .
it incorporates effects of the magnetic interaction between pancakes as well as those of the interlayer josephson coupling .
the proposed description enables us to estimate the pancake crystal melting temperature from above .
also , it allows us to compare directly the effects of the magnetic and josephson energies on the statistical properties of the pancake lattice .
we demonstrate that even for such an anisotropic material as bscco the pancake interaction induced by the josephson coupling is of the same order as the magnetic interaction . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.